From 91c13e54b5d631b65e2f8344d5e0abd058f78ba1 Mon Sep 17 00:00:00 2001 From: rsc Date: Sun, 29 Feb 2004 22:11:15 +0000 Subject: Move utf, fmt. Small header file changes. --- include/complete.h | 5 +- include/lib9.h | 2 + include/plumb.h | 1 + include/thread.h | 14 +- src/lib9/fmt/LICENSE | 19 + src/lib9/fmt/NOTICE | 19 + src/lib9/fmt/README | 19 + src/lib9/fmt/charstod.c | 85 ++++ src/lib9/fmt/dofmt.c | 558 +++++++++++++++++++++ src/lib9/fmt/dorfmt.c | 61 +++ src/lib9/fmt/errfmt.c | 28 ++ src/lib9/fmt/fltfmt.c | 610 +++++++++++++++++++++++ src/lib9/fmt/fmt.c | 221 +++++++++ src/lib9/fmt/fmtdef.h | 121 +++++ src/lib9/fmt/fmtfd.c | 46 ++ src/lib9/fmt/fmtfdflush.c | 33 ++ src/lib9/fmt/fmtlock.c | 28 ++ src/lib9/fmt/fmtprint.c | 47 ++ src/lib9/fmt/fmtquote.c | 262 ++++++++++ src/lib9/fmt/fmtrune.c | 40 ++ src/lib9/fmt/fmtstr.c | 65 +++ src/lib9/fmt/fmtvprint.c | 46 ++ src/lib9/fmt/fprint.c | 28 ++ src/lib9/fmt/mkfile | 57 +++ src/lib9/fmt/nan.h | 4 + src/lib9/fmt/nan64.c | 76 +++ src/lib9/fmt/pow10.c | 57 +++ src/lib9/fmt/print.c | 28 ++ src/lib9/fmt/runefmtstr.c | 65 +++ src/lib9/fmt/runeseprint.c | 30 ++ src/lib9/fmt/runesmprint.c | 30 ++ src/lib9/fmt/runesnprint.c | 31 ++ src/lib9/fmt/runesprint.c | 30 ++ src/lib9/fmt/runevseprint.c | 39 ++ src/lib9/fmt/runevsmprint.c | 37 ++ src/lib9/fmt/runevsnprint.c | 38 ++ src/lib9/fmt/seprint.c | 27 + src/lib9/fmt/smprint.c | 27 + src/lib9/fmt/snprint.c | 28 ++ src/lib9/fmt/sprint.c | 27 + src/lib9/fmt/strtod.c | 539 ++++++++++++++++++++ src/lib9/fmt/strtod.h | 4 + src/lib9/fmt/test.c | 39 ++ src/lib9/fmt/vfprint.c | 31 ++ src/lib9/fmt/vseprint.c | 37 ++ src/lib9/fmt/vsmprint.c | 36 ++ src/lib9/fmt/vsnprint.c | 37 ++ src/lib9/utf/LICENSE | 13 + src/lib9/utf/NOTICE | 13 + src/lib9/utf/README | 13 + src/lib9/utf/lib9.h | 17 + src/lib9/utf/mkfile | 31 ++ src/lib9/utf/rune.c | 177 +++++++ src/lib9/utf/runestrcat.c | 25 + src/lib9/utf/runestrchr.c | 35 ++ src/lib9/utf/runestrcmp.c | 35 ++ src/lib9/utf/runestrcpy.c | 28 ++ src/lib9/utf/runestrdup.c | 30 ++ src/lib9/utf/runestrecpy.c | 32 ++ src/lib9/utf/runestrlen.c | 24 + src/lib9/utf/runestrncat.c | 32 ++ src/lib9/utf/runestrncmp.c | 37 ++ src/lib9/utf/runestrncpy.c | 33 ++ src/lib9/utf/runestrrchr.c | 30 ++ src/lib9/utf/runestrstr.c | 44 ++ src/lib9/utf/runetype.c | 1152 +++++++++++++++++++++++++++++++++++++++++++ src/lib9/utf/utfdef.h | 14 + src/lib9/utf/utfecpy.c | 36 ++ src/lib9/utf/utflen.c | 38 ++ src/lib9/utf/utfnlen.c | 41 ++ src/lib9/utf/utfrrune.c | 46 ++ src/lib9/utf/utfrune.c | 45 ++ src/lib9/utf/utfutf.c | 41 ++ 73 files changed, 5771 insertions(+), 3 deletions(-) create mode 100644 src/lib9/fmt/LICENSE create mode 100644 src/lib9/fmt/NOTICE create mode 100644 src/lib9/fmt/README create mode 100644 src/lib9/fmt/charstod.c create mode 100644 src/lib9/fmt/dofmt.c create mode 100644 src/lib9/fmt/dorfmt.c create mode 100644 src/lib9/fmt/errfmt.c create mode 100644 src/lib9/fmt/fltfmt.c create mode 100644 src/lib9/fmt/fmt.c create mode 100644 src/lib9/fmt/fmtdef.h create mode 100644 src/lib9/fmt/fmtfd.c create mode 100644 src/lib9/fmt/fmtfdflush.c create mode 100644 src/lib9/fmt/fmtlock.c create mode 100644 src/lib9/fmt/fmtprint.c create mode 100644 src/lib9/fmt/fmtquote.c create mode 100644 src/lib9/fmt/fmtrune.c create mode 100644 src/lib9/fmt/fmtstr.c create mode 100644 src/lib9/fmt/fmtvprint.c create mode 100644 src/lib9/fmt/fprint.c create mode 100644 src/lib9/fmt/mkfile create mode 100644 src/lib9/fmt/nan.h create mode 100644 src/lib9/fmt/nan64.c create mode 100644 src/lib9/fmt/pow10.c create mode 100644 src/lib9/fmt/print.c create mode 100644 src/lib9/fmt/runefmtstr.c create mode 100644 src/lib9/fmt/runeseprint.c create mode 100644 src/lib9/fmt/runesmprint.c create mode 100644 src/lib9/fmt/runesnprint.c create mode 100644 src/lib9/fmt/runesprint.c create mode 100644 src/lib9/fmt/runevseprint.c create mode 100644 src/lib9/fmt/runevsmprint.c create mode 100644 src/lib9/fmt/runevsnprint.c create mode 100644 src/lib9/fmt/seprint.c create mode 100644 src/lib9/fmt/smprint.c create mode 100644 src/lib9/fmt/snprint.c create mode 100644 src/lib9/fmt/sprint.c create mode 100644 src/lib9/fmt/strtod.c create mode 100644 src/lib9/fmt/strtod.h create mode 100644 src/lib9/fmt/test.c create mode 100644 src/lib9/fmt/vfprint.c create mode 100644 src/lib9/fmt/vseprint.c create mode 100644 src/lib9/fmt/vsmprint.c create mode 100644 src/lib9/fmt/vsnprint.c create mode 100644 src/lib9/utf/LICENSE create mode 100644 src/lib9/utf/NOTICE create mode 100644 src/lib9/utf/README create mode 100644 src/lib9/utf/lib9.h create mode 100644 src/lib9/utf/mkfile create mode 100644 src/lib9/utf/rune.c create mode 100644 src/lib9/utf/runestrcat.c create mode 100644 src/lib9/utf/runestrchr.c create mode 100644 src/lib9/utf/runestrcmp.c create mode 100644 src/lib9/utf/runestrcpy.c create mode 100644 src/lib9/utf/runestrdup.c create mode 100644 src/lib9/utf/runestrecpy.c create mode 100644 src/lib9/utf/runestrlen.c create mode 100644 src/lib9/utf/runestrncat.c create mode 100644 src/lib9/utf/runestrncmp.c create mode 100644 src/lib9/utf/runestrncpy.c create mode 100644 src/lib9/utf/runestrrchr.c create mode 100644 src/lib9/utf/runestrstr.c create mode 100644 src/lib9/utf/runetype.c create mode 100644 src/lib9/utf/utfdef.h create mode 100644 src/lib9/utf/utfecpy.c create mode 100644 src/lib9/utf/utflen.c create mode 100644 src/lib9/utf/utfnlen.c create mode 100644 src/lib9/utf/utfrrune.c create mode 100644 src/lib9/utf/utfrune.c create mode 100644 src/lib9/utf/utfutf.c diff --git a/include/complete.h b/include/complete.h index 7ee2b18a..ae8003d7 100644 --- a/include/complete.h +++ b/include/complete.h @@ -1,5 +1,7 @@ +/* #pragma lib "libcomplete.a" #pragma src "/sys/src/libcomplete" +*/ typedef struct Completion Completion; @@ -7,7 +9,8 @@ struct Completion{ uchar advance; /* whether forward progress has been made */ uchar complete; /* whether the completion now represents a file or directory */ char *string; /* the string to advance, suffixed " " or "/" for file or directory */ - int nfile; /* number of files that matched */ + int nmatch; /* number of files that matched */ + int nfile; /* number of files returned */ char **filename; /* their names */ }; diff --git a/include/lib9.h b/include/lib9.h index 8d5bdf96..94936e9e 100644 --- a/include/lib9.h +++ b/include/lib9.h @@ -742,6 +742,7 @@ extern void abort(void); extern int p9access(char*, int); extern long p9alarm(ulong); extern int await(char*, int); +extern int awaitnohang(char*, int); /* extern int bind(char*, char*, int); give up */ /* extern int brk(void*); */ extern int p9chdir(char*); @@ -790,6 +791,7 @@ extern int segfree(void*, ulong); extern int p9sleep(long); /* extern int stat(char*, uchar*, int); give up */ extern Waitmsg* p9wait(void); +extern Waitmsg* waitnohang(void); extern int p9waitpid(void); /* extern long write(int, void*, long); diff --git a/include/plumb.h b/include/plumb.h index 4a315467..a84344ac 100644 --- a/include/plumb.h +++ b/include/plumb.h @@ -42,6 +42,7 @@ struct Plumbattr int plumbsend(int, Plumbmsg*); Plumbmsg* plumbrecv(int); +Plumbmsg* threadplumbrecv(int); char* plumbpack(Plumbmsg*, int*); Plumbmsg* plumbunpack(char*, int); Plumbmsg* plumbunpackpartial(char*, int, int*); diff --git a/include/thread.h b/include/thread.h index b4a1c946..6c12648e 100644 --- a/include/thread.h +++ b/include/thread.h @@ -80,8 +80,8 @@ int nbsendul(Channel *c, unsigned long v); int proccreate(void (*f)(void *arg), void *arg, unsigned int stacksize); int procrfork(void (*f)(void *arg), void *arg, unsigned int stacksize, int flag); void** procdata(void); -void procexec(Channel *, int[3], char *, char *[]); -void procexecl(Channel *, int[3], char *, ...); +void threadexec(Channel *, int[3], char *, char *[]); +void threadexecl(Channel *, int[3], char *, ...); int recv(Channel *c, void *v); void* recvp(Channel *c); unsigned long recvul(Channel *c); @@ -93,6 +93,8 @@ int threadcreateidle(void (*f)(void*), void*, unsigned int); void** threaddata(void); void threadexits(char *); void threadexitsall(char *); +void threadfdwait(int, int); +void threadfdwaitsetup(void); int threadgetgrp(void); /* return thread group of current thread */ char* threadgetname(void); void threadint(int); /* interrupt thread */ @@ -100,12 +102,20 @@ void threadintgrp(int); /* interrupt threads in grp */ void threadkill(int); /* kill thread */ void threadkillgrp(int); /* kill threads in group */ void threadmain(int argc, char *argv[]); +void threadfdnoblock(int); void threadnonotes(void); int threadnotify(int (*f)(void*, char*), int in); int threadid(void); int threadpid(int); +long threadread(int, void*, long); +long threadreadn(int, void*, long); +int threadread9pmsg(int, void*, uint); +int threadrecvfd(int); +long threadwrite(int, const void*, long); +int threadsendfd(int, int); int threadsetgrp(int); /* set thread group, return old */ void threadsetname(char *name); +void threadsleep(int); Channel* threadwaitchan(void); int tprivalloc(void); void tprivfree(int); diff --git a/src/lib9/fmt/LICENSE b/src/lib9/fmt/LICENSE new file mode 100644 index 00000000..5dc21cb5 --- /dev/null +++ b/src/lib9/fmt/LICENSE @@ -0,0 +1,19 @@ +/* + * The authors of this software are Rob Pike and Ken Thompson. + * Copyright (c) 2002 by Lucent Technologies. + * Permission to use, copy, modify, and distribute this software for any + * purpose without fee is hereby granted, provided that this entire notice + * is included in all copies of any software which is or includes a copy + * or modification of this software and in all copies of the supporting + * documentation for such software. + * THIS SOFTWARE IS BEING PROVIDED "AS IS", WITHOUT ANY EXPRESS OR IMPLIED + * WARRANTY. IN PARTICULAR, NEITHER THE AUTHORS NOR LUCENT TECHNOLOGIES MAKE ANY + * REPRESENTATION OR WARRANTY OF ANY KIND CONCERNING THE MERCHANTABILITY + * OF THIS SOFTWARE OR ITS FITNESS FOR ANY PARTICULAR PURPOSE. +*/ + +This is a Unix port of the Plan 9 formatted I/O package. + +Please send comments about the packaging +to Russ Cox . + diff --git a/src/lib9/fmt/NOTICE b/src/lib9/fmt/NOTICE new file mode 100644 index 00000000..5dc21cb5 --- /dev/null +++ b/src/lib9/fmt/NOTICE @@ -0,0 +1,19 @@ +/* + * The authors of this software are Rob Pike and Ken Thompson. + * Copyright (c) 2002 by Lucent Technologies. + * Permission to use, copy, modify, and distribute this software for any + * purpose without fee is hereby granted, provided that this entire notice + * is included in all copies of any software which is or includes a copy + * or modification of this software and in all copies of the supporting + * documentation for such software. + * THIS SOFTWARE IS BEING PROVIDED "AS IS", WITHOUT ANY EXPRESS OR IMPLIED + * WARRANTY. IN PARTICULAR, NEITHER THE AUTHORS NOR LUCENT TECHNOLOGIES MAKE ANY + * REPRESENTATION OR WARRANTY OF ANY KIND CONCERNING THE MERCHANTABILITY + * OF THIS SOFTWARE OR ITS FITNESS FOR ANY PARTICULAR PURPOSE. +*/ + +This is a Unix port of the Plan 9 formatted I/O package. + +Please send comments about the packaging +to Russ Cox . + diff --git a/src/lib9/fmt/README b/src/lib9/fmt/README new file mode 100644 index 00000000..5dc21cb5 --- /dev/null +++ b/src/lib9/fmt/README @@ -0,0 +1,19 @@ +/* + * The authors of this software are Rob Pike and Ken Thompson. + * Copyright (c) 2002 by Lucent Technologies. + * Permission to use, copy, modify, and distribute this software for any + * purpose without fee is hereby granted, provided that this entire notice + * is included in all copies of any software which is or includes a copy + * or modification of this software and in all copies of the supporting + * documentation for such software. + * THIS SOFTWARE IS BEING PROVIDED "AS IS", WITHOUT ANY EXPRESS OR IMPLIED + * WARRANTY. IN PARTICULAR, NEITHER THE AUTHORS NOR LUCENT TECHNOLOGIES MAKE ANY + * REPRESENTATION OR WARRANTY OF ANY KIND CONCERNING THE MERCHANTABILITY + * OF THIS SOFTWARE OR ITS FITNESS FOR ANY PARTICULAR PURPOSE. +*/ + +This is a Unix port of the Plan 9 formatted I/O package. + +Please send comments about the packaging +to Russ Cox . + diff --git a/src/lib9/fmt/charstod.c b/src/lib9/fmt/charstod.c new file mode 100644 index 00000000..ec403b11 --- /dev/null +++ b/src/lib9/fmt/charstod.c @@ -0,0 +1,85 @@ +/* + * The authors of this software are Rob Pike and Ken Thompson. + * Copyright (c) 2002 by Lucent Technologies. + * Permission to use, copy, modify, and distribute this software for any + * purpose without fee is hereby granted, provided that this entire notice + * is included in all copies of any software which is or includes a copy + * or modification of this software and in all copies of the supporting + * documentation for such software. + * THIS SOFTWARE IS BEING PROVIDED "AS IS", WITHOUT ANY EXPRESS OR IMPLIED + * WARRANTY. IN PARTICULAR, NEITHER THE AUTHORS NOR LUCENT TECHNOLOGIES MAKE ANY + * REPRESENTATION OR WARRANTY OF ANY KIND CONCERNING THE MERCHANTABILITY + * OF THIS SOFTWARE OR ITS FITNESS FOR ANY PARTICULAR PURPOSE. + */ +#include +#include +#include "utf.h" +#include "fmt.h" +#include "fmtdef.h" + +/* + * Reads a floating-point number by interpreting successive characters + * returned by (*f)(vp). The last call it makes to f terminates the + * scan, so is not a character in the number. It may therefore be + * necessary to back up the input stream up one byte after calling charstod. + */ + +double +fmtcharstod(int(*f)(void*), void *vp) +{ + double num, dem; + int neg, eneg, dig, exp, c; + + num = 0; + neg = 0; + dig = 0; + exp = 0; + eneg = 0; + + c = (*f)(vp); + while(c == ' ' || c == '\t') + c = (*f)(vp); + if(c == '-' || c == '+'){ + if(c == '-') + neg = 1; + c = (*f)(vp); + } + while(c >= '0' && c <= '9'){ + num = num*10 + c-'0'; + c = (*f)(vp); + } + if(c == '.') + c = (*f)(vp); + while(c >= '0' && c <= '9'){ + num = num*10 + c-'0'; + dig++; + c = (*f)(vp); + } + if(c == 'e' || c == 'E'){ + c = (*f)(vp); + if(c == '-' || c == '+'){ + if(c == '-'){ + dig = -dig; + eneg = 1; + } + c = (*f)(vp); + } + while(c >= '0' && c <= '9'){ + exp = exp*10 + c-'0'; + c = (*f)(vp); + } + } + exp -= dig; + if(exp < 0){ + exp = -exp; + eneg = !eneg; + } + dem = __fmtpow10(exp); + if(eneg) + num /= dem; + else + num *= dem; + if(neg) + return -num; + return num; +} diff --git a/src/lib9/fmt/dofmt.c b/src/lib9/fmt/dofmt.c new file mode 100644 index 00000000..d26f7158 --- /dev/null +++ b/src/lib9/fmt/dofmt.c @@ -0,0 +1,558 @@ +/* + * The authors of this software are Rob Pike and Ken Thompson. + * Copyright (c) 2002 by Lucent Technologies. + * Permission to use, copy, modify, and distribute this software for any + * purpose without fee is hereby granted, provided that this entire notice + * is included in all copies of any software which is or includes a copy + * or modification of this software and in all copies of the supporting + * documentation for such software. + * THIS SOFTWARE IS BEING PROVIDED "AS IS", WITHOUT ANY EXPRESS OR IMPLIED + * WARRANTY. IN PARTICULAR, NEITHER THE AUTHORS NOR LUCENT TECHNOLOGIES MAKE ANY + * REPRESENTATION OR WARRANTY OF ANY KIND CONCERNING THE MERCHANTABILITY + * OF THIS SOFTWARE OR ITS FITNESS FOR ANY PARTICULAR PURPOSE. + */ +#include +#include +#include "utf.h" +#include "fmt.h" +#include "fmtdef.h" + +/* format the output into f->to and return the number of characters fmted */ +int +dofmt(Fmt *f, char *fmt) +{ + Rune rune, *rt, *rs; + int r; + char *t, *s; + int n, nfmt; + + nfmt = f->nfmt; + for(;;){ + if(f->runes){ + rt = (Rune*)f->to; + rs = (Rune*)f->stop; + while((r = *(uchar*)fmt) && r != '%'){ + if(r < Runeself) + fmt++; + else{ + fmt += chartorune(&rune, fmt); + r = rune; + } + FMTRCHAR(f, rt, rs, r); + } + fmt++; + f->nfmt += rt - (Rune *)f->to; + f->to = rt; + if(!r) + return f->nfmt - nfmt; + f->stop = rs; + }else{ + t = (char*)f->to; + s = (char*)f->stop; + while((r = *(uchar*)fmt) && r != '%'){ + if(r < Runeself){ + FMTCHAR(f, t, s, r); + fmt++; + }else{ + n = chartorune(&rune, fmt); + if(t + n > s){ + t = (char*)__fmtflush(f, t, n); + if(t != nil) + s = (char*)f->stop; + else + return -1; + } + while(n--) + *t++ = *fmt++; + } + } + fmt++; + f->nfmt += t - (char *)f->to; + f->to = t; + if(!r) + return f->nfmt - nfmt; + f->stop = s; + } + + fmt = (char*)__fmtdispatch(f, fmt, 0); + if(fmt == nil) + return -1; + } + return 0; /* not reached */ +} + +void * +__fmtflush(Fmt *f, void *t, int len) +{ + if(f->runes) + f->nfmt += (Rune*)t - (Rune*)f->to; + else + f->nfmt += (char*)t - (char *)f->to; + f->to = t; + if(f->flush == 0 || (*f->flush)(f) == 0 || (char*)f->to + len > (char*)f->stop){ + f->stop = f->to; + return nil; + } + return f->to; +} + +/* + * put a formatted block of memory sz bytes long of n runes into the output buffer, + * left/right justified in a field of at least f->width charactes + */ +int +__fmtpad(Fmt *f, int n) +{ + char *t, *s; + int i; + + t = (char*)f->to; + s = (char*)f->stop; + for(i = 0; i < n; i++) + FMTCHAR(f, t, s, ' '); + f->nfmt += t - (char *)f->to; + f->to = t; + return 0; +} + +int +__rfmtpad(Fmt *f, int n) +{ + Rune *t, *s; + int i; + + t = (Rune*)f->to; + s = (Rune*)f->stop; + for(i = 0; i < n; i++) + FMTRCHAR(f, t, s, ' '); + f->nfmt += t - (Rune *)f->to; + f->to = t; + return 0; +} + +int +__fmtcpy(Fmt *f, const void *vm, int n, int sz) +{ + Rune *rt, *rs, r; + char *t, *s, *m, *me; + ulong fl; + int nc, w; + + m = (char*)vm; + me = m + sz; + w = f->width; + fl = f->flags; + if((fl & FmtPrec) && n > f->prec) + n = f->prec; + if(f->runes){ + if(!(fl & FmtLeft) && __rfmtpad(f, w - n) < 0) + return -1; + rt = (Rune*)f->to; + rs = (Rune*)f->stop; + for(nc = n; nc > 0; nc--){ + r = *(uchar*)m; + if(r < Runeself) + m++; + else if((me - m) >= UTFmax || fullrune(m, me-m)) + m += chartorune(&r, m); + else + break; + FMTRCHAR(f, rt, rs, r); + } + f->nfmt += rt - (Rune *)f->to; + f->to = rt; + if(m < me) + return -1; + if(fl & FmtLeft && __rfmtpad(f, w - n) < 0) + return -1; + }else{ + if(!(fl & FmtLeft) && __fmtpad(f, w - n) < 0) + return -1; + t = (char*)f->to; + s = (char*)f->stop; + for(nc = n; nc > 0; nc--){ + r = *(uchar*)m; + if(r < Runeself) + m++; + else if((me - m) >= UTFmax || fullrune(m, me-m)) + m += chartorune(&r, m); + else + break; + FMTRUNE(f, t, s, r); + } + f->nfmt += t - (char *)f->to; + f->to = t; + if(fl & FmtLeft && __fmtpad(f, w - n) < 0) + return -1; + } + return 0; +} + +int +__fmtrcpy(Fmt *f, const void *vm, int n) +{ + Rune r, *m, *me, *rt, *rs; + char *t, *s; + ulong fl; + int w; + + m = (Rune*)vm; + w = f->width; + fl = f->flags; + if((fl & FmtPrec) && n > f->prec) + n = f->prec; + if(f->runes){ + if(!(fl & FmtLeft) && __rfmtpad(f, w - n) < 0) + return -1; + rt = (Rune*)f->to; + rs = (Rune*)f->stop; + for(me = m + n; m < me; m++) + FMTRCHAR(f, rt, rs, *m); + f->nfmt += rt - (Rune *)f->to; + f->to = rt; + if(fl & FmtLeft && __rfmtpad(f, w - n) < 0) + return -1; + }else{ + if(!(fl & FmtLeft) && __fmtpad(f, w - n) < 0) + return -1; + t = (char*)f->to; + s = (char*)f->stop; + for(me = m + n; m < me; m++){ + r = *m; + FMTRUNE(f, t, s, r); + } + f->nfmt += t - (char *)f->to; + f->to = t; + if(fl & FmtLeft && __fmtpad(f, w - n) < 0) + return -1; + } + return 0; +} + +/* fmt out one character */ +int +__charfmt(Fmt *f) +{ + char x[1]; + + x[0] = va_arg(f->args, int); + f->prec = 1; + return __fmtcpy(f, (const char*)x, 1, 1); +} + +/* fmt out one rune */ +int +__runefmt(Fmt *f) +{ + Rune x[1]; + + x[0] = va_arg(f->args, int); + return __fmtrcpy(f, (const void*)x, 1); +} + +/* public helper routine: fmt out a null terminated string already in hand */ +int +fmtstrcpy(Fmt *f, char *s) +{ + int p, i; + if(!s) + return __fmtcpy(f, "", 5, 5); + /* if precision is specified, make sure we don't wander off the end */ + if(f->flags & FmtPrec){ + p = f->prec; + for(i = 0; i < p; i++) + if(s[i] == 0) + break; + return __fmtcpy(f, s, utfnlen(s, i), i); /* BUG?: won't print a partial rune at end */ + } + + return __fmtcpy(f, s, utflen(s), strlen(s)); +} + +/* fmt out a null terminated utf string */ +int +__strfmt(Fmt *f) +{ + char *s; + + s = va_arg(f->args, char *); + return fmtstrcpy(f, s); +} + +/* public helper routine: fmt out a null terminated rune string already in hand */ +int +fmtrunestrcpy(Fmt *f, Rune *s) +{ + Rune *e; + int n, p; + + if(!s) + return __fmtcpy(f, "", 5, 5); + /* if precision is specified, make sure we don't wander off the end */ + if(f->flags & FmtPrec){ + p = f->prec; + for(n = 0; n < p; n++) + if(s[n] == 0) + break; + }else{ + for(e = s; *e; e++) + ; + n = e - s; + } + return __fmtrcpy(f, s, n); +} + +/* fmt out a null terminated rune string */ +int +__runesfmt(Fmt *f) +{ + Rune *s; + + s = va_arg(f->args, Rune *); + return fmtrunestrcpy(f, s); +} + +/* fmt a % */ +int +__percentfmt(Fmt *f) +{ + Rune x[1]; + + x[0] = f->r; + f->prec = 1; + return __fmtrcpy(f, (const void*)x, 1); +} + +/* fmt an integer */ +int +__ifmt(Fmt *f) +{ + char buf[70], *p, *conv; + uvlong vu; + ulong u; + int neg, base, i, n, fl, w, isv; + + neg = 0; + fl = f->flags; + isv = 0; + vu = 0; + u = 0; + /* + * Unsigned verbs + */ + switch(f->r){ + case 'o': + case 'u': + case 'x': + case 'X': + fl |= FmtUnsigned; + break; + } + if(f->r == 'p'){ + u = (ulong)va_arg(f->args, void*); + f->r = 'x'; + fl |= FmtUnsigned; + }else if(fl & FmtVLong){ + isv = 1; + if(fl & FmtUnsigned) + vu = va_arg(f->args, uvlong); + else + vu = va_arg(f->args, vlong); + }else if(fl & FmtLong){ + if(fl & FmtUnsigned) + u = va_arg(f->args, ulong); + else + u = va_arg(f->args, long); + }else if(fl & FmtByte){ + if(fl & FmtUnsigned) + u = (uchar)va_arg(f->args, int); + else + u = (char)va_arg(f->args, int); + }else if(fl & FmtShort){ + if(fl & FmtUnsigned) + u = (ushort)va_arg(f->args, int); + else + u = (short)va_arg(f->args, int); + }else{ + if(fl & FmtUnsigned) + u = va_arg(f->args, uint); + else + u = va_arg(f->args, int); + } + conv = "0123456789abcdef"; + switch(f->r){ + case 'd': + case 'i': + base = 10; + break; + case 'u': + base = 10; + break; + case 'x': + base = 16; + break; + case 'X': + base = 16; + conv = "0123456789ABCDEF"; + break; + case 'b': + base = 2; + break; + case 'o': + base = 8; + break; + default: + return -1; + } + if(!(fl & FmtUnsigned)){ + if(isv && (vlong)vu < 0){ + vu = -(vlong)vu; + neg = 1; + }else if(!isv && (long)u < 0){ + u = -(long)u; + neg = 1; + } + }else{ + fl &= ~(FmtSign|FmtSpace); /* no + for unsigned conversions */ + } + p = buf + sizeof buf - 1; + n = 0; + if(isv){ + while(vu){ + i = vu % base; + vu /= base; + if((fl & FmtComma) && n % 4 == 3){ + *p-- = ','; + n++; + } + *p-- = conv[i]; + n++; + } + }else{ + while(u){ + i = u % base; + u /= base; + if((fl & FmtComma) && n % 4 == 3){ + *p-- = ','; + n++; + } + *p-- = conv[i]; + n++; + } + } + if(n == 0){ + if(!(fl & FmtPrec) || f->prec != 0){ + *p-- = '0'; + n = 1; + } + fl &= ~FmtSharp; + } + for(w = f->prec; n < w && p > buf+3; n++) + *p-- = '0'; + if(neg || (fl & (FmtSign|FmtSpace))) + n++; + if(fl & FmtSharp){ + if(base == 16) + n += 2; + else if(base == 8){ + if(p[1] == '0') + fl &= ~FmtSharp; + else + n++; + } + } + if((fl & FmtZero) && !(fl & (FmtLeft|FmtPrec))){ + for(w = f->width; n < w && p > buf+3; n++) + *p-- = '0'; + f->width = 0; + } + if(fl & FmtSharp){ + if(base == 16) + *p-- = f->r; + if(base == 16 || base == 8) + *p-- = '0'; + } + if(neg) + *p-- = '-'; + else if(fl & FmtSign) + *p-- = '+'; + else if(fl & FmtSpace) + *p-- = ' '; + f->flags &= ~FmtPrec; + return __fmtcpy(f, p + 1, n, n); +} + +int +__countfmt(Fmt *f) +{ + void *p; + ulong fl; + + fl = f->flags; + p = va_arg(f->args, void*); + if(fl & FmtVLong){ + *(vlong*)p = f->nfmt; + }else if(fl & FmtLong){ + *(long*)p = f->nfmt; + }else if(fl & FmtByte){ + *(char*)p = f->nfmt; + }else if(fl & FmtShort){ + *(short*)p = f->nfmt; + }else{ + *(int*)p = f->nfmt; + } + return 0; +} + +int +__flagfmt(Fmt *f) +{ + switch(f->r){ + case ',': + f->flags |= FmtComma; + break; + case '-': + f->flags |= FmtLeft; + break; + case '+': + f->flags |= FmtSign; + break; + case '#': + f->flags |= FmtSharp; + break; + case ' ': + f->flags |= FmtSpace; + break; + case 'u': + f->flags |= FmtUnsigned; + break; + case 'h': + if(f->flags & FmtShort) + f->flags |= FmtByte; + f->flags |= FmtShort; + break; + case 'L': + f->flags |= FmtLDouble; + break; + case 'l': + if(f->flags & FmtLong) + f->flags |= FmtVLong; + f->flags |= FmtLong; + break; + } + return 1; +} + +/* default error format */ +int +__badfmt(Fmt *f) +{ + char x[3]; + + x[0] = '%'; + x[1] = f->r; + x[2] = '%'; + f->prec = 3; + __fmtcpy(f, (const void*)x, 3, 3); + return 0; +} diff --git a/src/lib9/fmt/dorfmt.c b/src/lib9/fmt/dorfmt.c new file mode 100644 index 00000000..cdaee8a5 --- /dev/null +++ b/src/lib9/fmt/dorfmt.c @@ -0,0 +1,61 @@ +/* + * The authors of this software are Rob Pike and Ken Thompson. + * Copyright (c) 2002 by Lucent Technologies. + * Permission to use, copy, modify, and distribute this software for any + * purpose without fee is hereby granted, provided that this entire notice + * is included in all copies of any software which is or includes a copy + * or modification of this software and in all copies of the supporting + * documentation for such software. + * THIS SOFTWARE IS BEING PROVIDED "AS IS", WITHOUT ANY EXPRESS OR IMPLIED + * WARRANTY. IN PARTICULAR, NEITHER THE AUTHORS NOR LUCENT TECHNOLOGIES MAKE ANY + * REPRESENTATION OR WARRANTY OF ANY KIND CONCERNING THE MERCHANTABILITY + * OF THIS SOFTWARE OR ITS FITNESS FOR ANY PARTICULAR PURPOSE. + */ +#include +#include +#include "utf.h" +#include "fmt.h" +#include "fmtdef.h" + +/* format the output into f->to and return the number of characters fmted */ + +int +dorfmt(Fmt *f, const Rune *fmt) +{ + Rune *rt, *rs; + int r; + char *t, *s; + int nfmt; + + nfmt = f->nfmt; + for(;;){ + if(f->runes){ + rt = f->to; + rs = f->stop; + while((r = *fmt++) && r != '%'){ + FMTRCHAR(f, rt, rs, r); + } + f->nfmt += rt - (Rune *)f->to; + f->to = rt; + if(!r) + return f->nfmt - nfmt; + f->stop = rs; + }else{ + t = f->to; + s = f->stop; + while((r = *fmt++) && r != '%'){ + FMTRUNE(f, t, f->stop, r); + } + f->nfmt += t - (char *)f->to; + f->to = t; + if(!r) + return f->nfmt - nfmt; + f->stop = s; + } + + fmt = __fmtdispatch(f, fmt, 1); + if(fmt == nil) + return -1; + } + return 0; /* not reached */ +} diff --git a/src/lib9/fmt/errfmt.c b/src/lib9/fmt/errfmt.c new file mode 100644 index 00000000..21847054 --- /dev/null +++ b/src/lib9/fmt/errfmt.c @@ -0,0 +1,28 @@ +/* + * The authors of this software are Rob Pike and Ken Thompson. + * Copyright (c) 2002 by Lucent Technologies. + * Permission to use, copy, modify, and distribute this software for any + * purpose without fee is hereby granted, provided that this entire notice + * is included in all copies of any software which is or includes a copy + * or modification of this software and in all copies of the supporting + * documentation for such software. + * THIS SOFTWARE IS BEING PROVIDED "AS IS", WITHOUT ANY EXPRESS OR IMPLIED + * WARRANTY. IN PARTICULAR, NEITHER THE AUTHORS NOR LUCENT TECHNOLOGIES MAKE ANY + * REPRESENTATION OR WARRANTY OF ANY KIND CONCERNING THE MERCHANTABILITY + * OF THIS SOFTWARE OR ITS FITNESS FOR ANY PARTICULAR PURPOSE. + */ +#include +#include +#include +#include "utf.h" +#include "fmt.h" +#include "fmtdef.h" + +int +__errfmt(Fmt *f) +{ + char *s; + + s = strerror(errno); + return fmtstrcpy(f, s); +} diff --git a/src/lib9/fmt/fltfmt.c b/src/lib9/fmt/fltfmt.c new file mode 100644 index 00000000..234f03d9 --- /dev/null +++ b/src/lib9/fmt/fltfmt.c @@ -0,0 +1,610 @@ +/* + * The authors of this software are Rob Pike and Ken Thompson. + * Copyright (c) 2002 by Lucent Technologies. + * Permission to use, copy, modify, and distribute this software for any + * purpose without fee is hereby granted, provided that this entire notice + * is included in all copies of any software which is or includes a copy + * or modification of this software and in all copies of the supporting + * documentation for such software. + * THIS SOFTWARE IS BEING PROVIDED "AS IS", WITHOUT ANY EXPRESS OR IMPLIED + * WARRANTY. IN PARTICULAR, NEITHER THE AUTHORS NOR LUCENT TECHNOLOGIES MAKE ANY + * REPRESENTATION OR WARRANTY OF ANY KIND CONCERNING THE MERCHANTABILITY + * OF THIS SOFTWARE OR ITS FITNESS FOR ANY PARTICULAR PURPOSE. + */ +#include +#include +#include +#include +#include +#include +#include +#include "fmt.h" +#include "fmtdef.h" +#include "nan.h" + +enum +{ + FDEFLT = 6, + NSIGNIF = 17 +}; + +/* + * first few powers of 10, enough for about 1/2 of the + * total space for doubles. + */ +static double pows10[] = +{ + 1e0, 1e1, 1e2, 1e3, 1e4, 1e5, 1e6, 1e7, 1e8, 1e9, + 1e10, 1e11, 1e12, 1e13, 1e14, 1e15, 1e16, 1e17, 1e18, 1e19, + 1e20, 1e21, 1e22, 1e23, 1e24, 1e25, 1e26, 1e27, 1e28, 1e29, + 1e30, 1e31, 1e32, 1e33, 1e34, 1e35, 1e36, 1e37, 1e38, 1e39, + 1e40, 1e41, 1e42, 1e43, 1e44, 1e45, 1e46, 1e47, 1e48, 1e49, + 1e50, 1e51, 1e52, 1e53, 1e54, 1e55, 1e56, 1e57, 1e58, 1e59, + 1e60, 1e61, 1e62, 1e63, 1e64, 1e65, 1e66, 1e67, 1e68, 1e69, + 1e70, 1e71, 1e72, 1e73, 1e74, 1e75, 1e76, 1e77, 1e78, 1e79, + 1e80, 1e81, 1e82, 1e83, 1e84, 1e85, 1e86, 1e87, 1e88, 1e89, + 1e90, 1e91, 1e92, 1e93, 1e94, 1e95, 1e96, 1e97, 1e98, 1e99, + 1e100, 1e101, 1e102, 1e103, 1e104, 1e105, 1e106, 1e107, 1e108, 1e109, + 1e110, 1e111, 1e112, 1e113, 1e114, 1e115, 1e116, 1e117, 1e118, 1e119, + 1e120, 1e121, 1e122, 1e123, 1e124, 1e125, 1e126, 1e127, 1e128, 1e129, + 1e130, 1e131, 1e132, 1e133, 1e134, 1e135, 1e136, 1e137, 1e138, 1e139, + 1e140, 1e141, 1e142, 1e143, 1e144, 1e145, 1e146, 1e147, 1e148, 1e149, + 1e150, 1e151, 1e152, 1e153, 1e154, 1e155, 1e156, 1e157, 1e158, 1e159, +}; + +static double +pow10(int n) +{ + double d; + int neg; + + neg = 0; + if(n < 0){ + if(n < DBL_MIN_10_EXP){ + return 0.; + } + neg = 1; + n = -n; + }else if(n > DBL_MAX_10_EXP){ + return HUGE_VAL; + } + if(n < (int)(sizeof(pows10)/sizeof(pows10[0]))) + d = pows10[n]; + else{ + d = pows10[sizeof(pows10)/sizeof(pows10[0]) - 1]; + for(;;){ + n -= sizeof(pows10)/sizeof(pows10[0]) - 1; + if(n < (int)(sizeof(pows10)/sizeof(pows10[0]))){ + d *= pows10[n]; + break; + } + d *= pows10[sizeof(pows10)/sizeof(pows10[0]) - 1]; + } + } + if(neg){ + return 1./d; + } + return d; +} + +static int +xadd(char *a, int n, int v) +{ + char *b; + int c; + + if(n < 0 || n >= NSIGNIF) + return 0; + for(b = a+n; b >= a; b--) { + c = *b + v; + if(c <= '9') { + *b = c; + return 0; + } + *b = '0'; + v = 1; + } + *a = '1'; /* overflow adding */ + return 1; +} + +static int +xsub(char *a, int n, int v) +{ + char *b; + int c; + + for(b = a+n; b >= a; b--) { + c = *b - v; + if(c >= '0') { + *b = c; + return 0; + } + *b = '9'; + v = 1; + } + *a = '9'; /* underflow subtracting */ + return 1; +} + +static void +xaddexp(char *p, int e) +{ + char se[9]; + int i; + + *p++ = 'e'; + if(e < 0) { + *p++ = '-'; + e = -e; + } + i = 0; + while(e) { + se[i++] = e % 10 + '0'; + e /= 10; + } + if(i == 0) { + *p++ = '0'; + } else { + while(i > 0) + *p++ = se[--i]; + } + *p++ = '\0'; +} + +static char* +xdodtoa(char *s1, double f, int chr, int prec, int *decpt, int *rsign) +{ + char s2[NSIGNIF+10]; + double g, h; + int e, d, i; + int c2, sign, oerr; + + if(chr == 'F') + chr = 'f'; + if(prec > NSIGNIF) + prec = NSIGNIF; + if(prec < 0) + prec = 0; + if(__isNaN(f)) { + *decpt = 9999; + *rsign = 0; + strcpy(s1, "nan"); + return &s1[3]; + } + sign = 0; + if(f < 0) { + f = -f; + sign++; + } + *rsign = sign; + if(__isInf(f, 1) || __isInf(f, -1)) { + *decpt = 9999; + strcpy(s1, "inf"); + return &s1[3]; + } + + e = 0; + g = f; + if(g != 0) { + frexp(f, &e); + e = (int)(e * .301029995664); + if(e >= -150 && e <= +150) { + d = 0; + h = f; + } else { + d = e/2; + h = f * pow10(-d); + } + g = h * pow10(d-e); + while(g < 1) { + e--; + g = h * pow10(d-e); + } + while(g >= 10) { + e++; + g = h * pow10(d-e); + } + } + + /* + * convert NSIGNIF digits and convert + * back to get accuracy. + */ + for(i=0; i= NSIGNIF-2) { + strcpy(s2, s1); + d = e; + s1[NSIGNIF-2] = '0'; + s1[NSIGNIF-1] = '0'; + xaddexp(s1+NSIGNIF, e-NSIGNIF+1); + g = fmtstrtod(s1, nil); + if(g == f) + goto found; + if(xadd(s1, NSIGNIF-3, 1)) { + e++; + xaddexp(s1+NSIGNIF, e-NSIGNIF+1); + } + g = fmtstrtod(s1, nil); + if(g == f) + goto found; + strcpy(s1, s2); + e = d; + } + + /* + * convert back so s1 gets exact answer + */ + for(d = 0; d < 10; d++) { + xaddexp(s1+NSIGNIF, e-NSIGNIF+1); + g = fmtstrtod(s1, nil); + if(f > g) { + if(xadd(s1, NSIGNIF-1, 1)) + e--; + continue; + } + if(f < g) { + if(xsub(s1, NSIGNIF-1, 1)) + e++; + continue; + } + break; + } + +found: + errno = oerr; + + /* + * sign + */ + d = 0; + i = 0; + + /* + * round & adjust 'f' digits + */ + c2 = prec + 1; + if(chr == 'f'){ + if(xadd(s1, c2+e, 5)) + e++; + c2 += e; + if(c2 < 0){ + c2 = 0; + e = -prec - 1; + } + }else{ + if(xadd(s1, c2, 5)) + e++; + } + if(c2 > NSIGNIF){ + c2 = NSIGNIF; + } + + *decpt = e + 1; + + /* + * terminate the converted digits + */ + s1[c2] = '\0'; + return &s1[c2]; +} + +/* + * this function works like the standard dtoa, if you want it. + */ +#if 0 +static char* +__dtoa(double f, int mode, int ndigits, int *decpt, int *rsign, char **rve) +{ + static char s2[NSIGNIF + 10]; + char *es; + int chr, prec; + + switch(mode) { + /* like 'e' */ + case 2: + case 4: + case 6: + case 8: + chr = 'e'; + break; + /* like 'g' */ + case 0: + case 1: + default: + chr = 'g'; + break; + /* like 'f' */ + case 3: + case 5: + case 7: + case 9: + chr = 'f'; + break; + } + + if(chr != 'f' && ndigits){ + ndigits--; + } + prec = ndigits; + if(prec > NSIGNIF) + prec = NSIGNIF; + if(ndigits == 0) + prec = NSIGNIF; + es = xdodtoa(s2, f, chr, prec, decpt, rsign); + + /* + * strip trailing 0 + */ + for(; es > s2 + 1; es--){ + if(es[-1] != '0'){ + break; + } + } + *es = '\0'; + if(rve != NULL) + *rve = es; + return s2; +} +#endif + +static int +fmtzdotpad(Fmt *f, int n, int pt) +{ + char *t, *s; + int i; + Rune *rt, *rs; + + if(f->runes){ + rt = (Rune*)f->to; + rs = (Rune*)f->stop; + for(i = 0; i < n; i++){ + if(i == pt){ + FMTRCHAR(f, rt, rs, '.'); + } + FMTRCHAR(f, rt, rs, '0'); + } + f->nfmt += rt - (Rune*)f->to; + f->to = rt; + }else{ + t = (char*)f->to; + s = (char*)f->stop; + for(i = 0; i < n; i++){ + if(i == pt){ + FMTCHAR(f, t, s, '.'); + } + FMTCHAR(f, t, s, '0'); + } + f->nfmt += t - (char *)f->to; + f->to = t; + } + return 0; +} + +int +__efgfmt(Fmt *fmt) +{ + double f; + char s1[NSIGNIF+10]; + int e, d, n; + int c1, c2, c3, c4, ucase, sign, chr, prec, fl; + + f = va_arg(fmt->args, double); + prec = FDEFLT; + fl = fmt->flags; + fmt->flags = 0; + if(fl & FmtPrec) + prec = fmt->prec; + chr = fmt->r; + ucase = 0; + if(chr == 'E'){ + chr = 'e'; + ucase = 1; + }else if(chr == 'F'){ + chr = 'f'; + ucase = 1; + }else if(chr == 'G'){ + chr = 'g'; + ucase = 1; + } + if(prec > 0 && chr == 'g') + prec--; + if(prec < 0) + prec = 0; + + xdodtoa(s1, f, chr, prec, &e, &sign); + e--; + if(*s1 == 'i' || *s1 == 'n'){ + if(ucase){ + if(*s1 == 'i'){ + strcpy(s1, "INF"); + }else{ + strcpy(s1, "NAN"); + } + } + fmt->flags = fl & (FmtWidth|FmtLeft); + return __fmtcpy(fmt, (const void*)s1, 3, 3); + } + + /* + * copy into final place + * c1 digits of leading '0' + * c2 digits from conversion + * c3 digits of trailing '0' + * c4 digits after '.' + */ + c1 = 0; + c2 = prec + 1; + c3 = 0; + c4 = prec; + switch(chr) { + default: + chr = 'e'; + break; + case 'g': + /* + * decide on 'e' of 'f' style convers + */ + if(e >= -4 && e <= prec) { + c1 = -e; + c4 = prec - e; + chr = 'h'; /* flag for 'f' style */ + } + break; + case 'f': + c1 = -e; + if(c1 > prec) + c1 = prec + 1; + c2 += e; + break; + } + + /* + * clean up c1 c2 and c3 + */ + if(c1 < 0) + c1 = 0; + if(c2 < 0) + c2 = 0; + if(c2 > NSIGNIF) { + c3 = c2-NSIGNIF; + c2 = NSIGNIF; + } + + /* + * trim trailing zeros for %g + */ + if(!(fl & FmtSharp) + && (chr == 'g' || chr == 'h')){ + if(c4 >= c3){ + c4 -= c3; + c3 = 0; + }else{ + c3 -= c4; + c4 = 0; + } + while(c4 && c2 > 1 && s1[c2 - 1] == '0'){ + c4--; + c2--; + } + } + + /* + * calculate the total length + */ + n = c1 + c2 + c3; + if(sign || (fl & (FmtSign|FmtSpace))) + n++; + if(c4 || (fl & FmtSharp)){ + n++; + } + if(chr == 'e' || chr == 'g'){ + n += 4; + if(e >= 100) + n++; + } + + /* + * pad to width if right justified + */ + if((fl & (FmtWidth|FmtLeft)) == FmtWidth && n < fmt->width){ + if(fl & FmtZero){ + c1 += fmt->width - n; + }else{ + if(__fmtpad(fmt, fmt->width - n) < 0){ + return -1; + } + } + } + + /* + * sign + */ + d = 0; + if(sign) + d = '-'; + else if(fl & FmtSign) + d = '+'; + else if(fl & FmtSpace) + d = ' '; + if(d && fmtrune(fmt, d) < 0){ + return -1; + } + + /* + * copy digits + */ + c4 = c1 + c2 + c3 - c4; + if(c1 > 0){ + if(fmtzdotpad(fmt, c1, c4) < 0){ + return -1; + } + c4 -= c1; + } + d = 0; + if(c4 >= 0 && c4 < c2){ + if(__fmtcpy(fmt, s1, c4, c4) < 0 || fmtrune(fmt, '.') < 0) + return -1; + d = c4; + c2 -= c4; + c4 = -1; + } + if(__fmtcpy(fmt, (const void*)(s1 + d), c2, c2) < 0){ + return -1; + } + c4 -= c2; + if(c3 > 0){ + if(fmtzdotpad(fmt, c3, c4) < 0){ + return -1; + } + c4 -= c3; + } + + /* + * strip trailing '0' on g conv + */ + if((fl & FmtSharp) && c4 == 0 && fmtrune(fmt, '.') < 0){ + return -1; + } + if(chr == 'e' || chr == 'g') { + d = 0; + if(ucase) + s1[d++] = 'E'; + else + s1[d++] = 'e'; + c1 = e; + if(c1 < 0) { + s1[d++] = '-'; + c1 = -c1; + } else + s1[d++] = '+'; + if(c1 >= 100) { + s1[d++] = c1/100 + '0'; + c1 = c1%100; + } + s1[d++] = c1/10 + '0'; + s1[d++] = c1%10 + '0'; + if(__fmtcpy(fmt, s1, d, d) < 0){ + return -1; + } + } + if((fl & (FmtWidth|FmtLeft)) == (FmtWidth|FmtLeft) && n < fmt->width){ + if(__fmtpad(fmt, fmt->width - n) < 0){ + return -1; + } + } + return 0; +} diff --git a/src/lib9/fmt/fmt.c b/src/lib9/fmt/fmt.c new file mode 100644 index 00000000..976bf7ca --- /dev/null +++ b/src/lib9/fmt/fmt.c @@ -0,0 +1,221 @@ +/* + * The authors of this software are Rob Pike and Ken Thompson. + * Copyright (c) 2002 by Lucent Technologies. + * Permission to use, copy, modify, and distribute this software for any + * purpose without fee is hereby granted, provided that this entire notice + * is included in all copies of any software which is or includes a copy + * or modification of this software and in all copies of the supporting + * documentation for such software. + * THIS SOFTWARE IS BEING PROVIDED "AS IS", WITHOUT ANY EXPRESS OR IMPLIED + * WARRANTY. IN PARTICULAR, NEITHER THE AUTHORS NOR LUCENT TECHNOLOGIES MAKE ANY + * REPRESENTATION OR WARRANTY OF ANY KIND CONCERNING THE MERCHANTABILITY + * OF THIS SOFTWARE OR ITS FITNESS FOR ANY PARTICULAR PURPOSE. + */ +#include +#include +#include "utf.h" +#include "fmt.h" +#include "fmtdef.h" + +enum +{ + Maxfmt = 64 +}; + +typedef struct Convfmt Convfmt; +struct Convfmt +{ + int c; + volatile Fmts fmt; /* for spin lock in fmtfmt; avoids race due to write order */ +}; + +struct +{ + /* lock by calling __fmtlock, __fmtunlock */ + int nfmt; + Convfmt fmt[Maxfmt]; +} fmtalloc; + +static Convfmt knownfmt[] = { + ' ', __flagfmt, + '#', __flagfmt, + '%', __percentfmt, + '+', __flagfmt, + ',', __flagfmt, + '-', __flagfmt, + 'C', __runefmt, /* Plan 9 addition */ + 'E', __efgfmt, + 'F', __efgfmt, /* ANSI only */ + 'G', __efgfmt, + 'L', __flagfmt, /* ANSI only */ + 'S', __runesfmt, /* Plan 9 addition */ + 'X', __ifmt, + 'b', __ifmt, /* Plan 9 addition */ + 'c', __charfmt, + 'd', __ifmt, + 'e', __efgfmt, + 'f', __efgfmt, + 'g', __efgfmt, + 'h', __flagfmt, + 'i', __ifmt, /* ANSI only */ + 'l', __flagfmt, + 'n', __countfmt, + 'o', __ifmt, + 'p', __ifmt, + 'r', __errfmt, + 's', __strfmt, + 'u', __flagfmt, /* in Unix, __ifmt */ + 'x', __ifmt, + 0, nil, +}; + + +int (*fmtdoquote)(int); + +/* + * __fmtlock() must be set + */ +static int +__fmtinstall(int c, Fmts f) +{ + Convfmt *p, *ep; + + if(c<=0 || c>=65536) + return -1; + if(!f) + f = __badfmt; + + ep = &fmtalloc.fmt[fmtalloc.nfmt]; + for(p=fmtalloc.fmt; pc == c) + break; + + if(p == &fmtalloc.fmt[Maxfmt]) + return -1; + + p->fmt = f; + if(p == ep){ /* installing a new format character */ + fmtalloc.nfmt++; + p->c = c; + } + + return 0; +} + +int +fmtinstall(int c, Fmts f) +{ + int ret; + + __fmtlock(); + ret = __fmtinstall(c, f); + __fmtunlock(); + return ret; +} + +static Fmts +fmtfmt(int c) +{ + Convfmt *p, *ep; + + ep = &fmtalloc.fmt[fmtalloc.nfmt]; + for(p=fmtalloc.fmt; pc == c){ + while(p->fmt == nil) /* loop until value is updated */ + ; + return p->fmt; + } + + /* is this a predefined format char? */ + __fmtlock(); + for(p=knownfmt; p->c; p++) + if(p->c == c){ + __fmtinstall(p->c, p->fmt); + __fmtunlock(); + return p->fmt; + } + __fmtunlock(); + + return __badfmt; +} + +void* +__fmtdispatch(Fmt *f, void *fmt, int isrunes) +{ + Rune rune, r; + int i, n; + + f->flags = 0; + f->width = f->prec = 0; + + for(;;){ + if(isrunes){ + r = *(Rune*)fmt; + fmt = (Rune*)fmt + 1; + }else{ + fmt = (char*)fmt + chartorune(&rune, (char*)fmt); + r = rune; + } + f->r = r; + switch(r){ + case '\0': + return nil; + case '.': + f->flags |= FmtWidth|FmtPrec; + continue; + case '0': + if(!(f->flags & FmtWidth)){ + f->flags |= FmtZero; + continue; + } + /* fall through */ + case '1': case '2': case '3': case '4': + case '5': case '6': case '7': case '8': case '9': + i = 0; + while(r >= '0' && r <= '9'){ + i = i * 10 + r - '0'; + if(isrunes){ + r = *(Rune*)fmt; + fmt = (Rune*)fmt + 1; + }else{ + r = *(char*)fmt; + fmt = (char*)fmt + 1; + } + } + if(isrunes) + fmt = (Rune*)fmt - 1; + else + fmt = (char*)fmt - 1; + numflag: + if(f->flags & FmtWidth){ + f->flags |= FmtPrec; + f->prec = i; + }else{ + f->flags |= FmtWidth; + f->width = i; + } + continue; + case '*': + i = va_arg(f->args, int); + if(i < 0){ + /* + * negative precision => + * ignore the precision. + */ + if(f->flags & FmtPrec){ + f->flags &= ~FmtPrec; + f->prec = 0; + continue; + } + i = -i; + f->flags |= FmtLeft; + } + goto numflag; + } + n = (*fmtfmt(r))(f); + if(n < 0) + return nil; + if(n == 0) + return fmt; + } +} diff --git a/src/lib9/fmt/fmtdef.h b/src/lib9/fmt/fmtdef.h new file mode 100644 index 00000000..ca2010ab --- /dev/null +++ b/src/lib9/fmt/fmtdef.h @@ -0,0 +1,121 @@ +/* + * The authors of this software are Rob Pike and Ken Thompson. + * Copyright (c) 2002 by Lucent Technologies. + * Permission to use, copy, modify, and distribute this software for any + * purpose without fee is hereby granted, provided that this entire notice + * is included in all copies of any software which is or includes a copy + * or modification of this software and in all copies of the supporting + * documentation for such software. + * THIS SOFTWARE IS BEING PROVIDED "AS IS", WITHOUT ANY EXPRESS OR IMPLIED + * WARRANTY. IN PARTICULAR, NEITHER THE AUTHORS NOR LUCENT TECHNOLOGIES MAKE ANY + * REPRESENTATION OR WARRANTY OF ANY KIND CONCERNING THE MERCHANTABILITY + * OF THIS SOFTWARE OR ITS FITNESS FOR ANY PARTICULAR PURPOSE. + */ +/* + * dofmt -- format to a buffer + * the number of characters formatted is returned, + * or -1 if there was an error. + * if the buffer is ever filled, flush is called. + * it should reset the buffer and return whether formatting should continue. + */ +#define uchar _fmtuchar +#define ushort _fmtushort +#define uint _fmtuint +#define ulong _fmtulong +#define vlong _fmtvlong +#define uvlong _fmtuvlong + +#define USED(x) if(x);else + +typedef unsigned char uchar; +typedef unsigned short ushort; +typedef unsigned int uint; +typedef unsigned long ulong; + +#ifndef NOVLONGS +typedef unsigned long long uvlong; +typedef long long vlong; +#endif + +#define nil 0 /* cannot be ((void*)0) because used for function pointers */ + +typedef int (*Fmts)(Fmt*); + +typedef struct Quoteinfo Quoteinfo; +struct Quoteinfo +{ + int quoted; /* if set, string must be quoted */ + int nrunesin; /* number of input runes that can be accepted */ + int nbytesin; /* number of input bytes that can be accepted */ + int nrunesout; /* number of runes that will be generated */ + int nbytesout; /* number of bytes that will be generated */ +}; + +void *__fmtflush(Fmt*, void*, int); +void *__fmtdispatch(Fmt*, void*, int); +int __floatfmt(Fmt*, double); +int __fmtpad(Fmt*, int); +int __rfmtpad(Fmt*, int); +int __fmtFdFlush(Fmt*); + +int __efgfmt(Fmt*); +int __charfmt(Fmt*); +int __runefmt(Fmt*); +int __runesfmt(Fmt*); +int __countfmt(Fmt*); +int __flagfmt(Fmt*); +int __percentfmt(Fmt*); +int __ifmt(Fmt*); +int __strfmt(Fmt*); +int __badfmt(Fmt*); +int __fmtcpy(Fmt*, const void*, int, int); +int __fmtrcpy(Fmt*, const void*, int n); +int __errfmt(Fmt *f); + +double __fmtpow10(int); + +void __fmtlock(void); +void __fmtunlock(void); + +#define FMTCHAR(f, t, s, c)\ + do{\ + if(t + 1 > (char*)s){\ + t = __fmtflush(f, t, 1);\ + if(t != nil)\ + s = f->stop;\ + else\ + return -1;\ + }\ + *t++ = c;\ + }while(0) + +#define FMTRCHAR(f, t, s, c)\ + do{\ + if(t + 1 > (Rune*)s){\ + t = __fmtflush(f, t, sizeof(Rune));\ + if(t != nil)\ + s = f->stop;\ + else\ + return -1;\ + }\ + *t++ = c;\ + }while(0) + +#define FMTRUNE(f, t, s, r)\ + do{\ + Rune _rune;\ + int _runelen;\ + if(t + UTFmax > (char*)s && t + (_runelen = runelen(r)) > (char*)s){\ + t = __fmtflush(f, t, _runelen);\ + if(t != nil)\ + s = f->stop;\ + else\ + return -1;\ + }\ + if(r < Runeself)\ + *t++ = r;\ + else{\ + _rune = r;\ + t += runetochar(t, &_rune);\ + }\ + }while(0) diff --git a/src/lib9/fmt/fmtfd.c b/src/lib9/fmt/fmtfd.c new file mode 100644 index 00000000..d4251402 --- /dev/null +++ b/src/lib9/fmt/fmtfd.c @@ -0,0 +1,46 @@ +/* + * The authors of this software are Rob Pike and Ken Thompson. + * Copyright (c) 2002 by Lucent Technologies. + * Permission to use, copy, modify, and distribute this software for any + * purpose without fee is hereby granted, provided that this entire notice + * is included in all copies of any software which is or includes a copy + * or modification of this software and in all copies of the supporting + * documentation for such software. + * THIS SOFTWARE IS BEING PROVIDED "AS IS", WITHOUT ANY EXPRESS OR IMPLIED + * WARRANTY. IN PARTICULAR, NEITHER THE AUTHORS NOR LUCENT TECHNOLOGIES MAKE ANY + * REPRESENTATION OR WARRANTY OF ANY KIND CONCERNING THE MERCHANTABILITY + * OF THIS SOFTWARE OR ITS FITNESS FOR ANY PARTICULAR PURPOSE. + */ +#include +#include +#include "utf.h" +#include "fmt.h" +#include "fmtdef.h" + +/* + * public routine for final flush of a formatting buffer + * to a file descriptor; returns total char count. + */ +int +fmtfdflush(Fmt *f) +{ + if(__fmtFdFlush(f) <= 0) + return -1; + return f->nfmt; +} + +/* + * initialize an output buffer for buffered printing + */ +int +fmtfdinit(Fmt *f, int fd, char *buf, int size) +{ + f->runes = 0; + f->start = buf; + f->to = buf; + f->stop = buf + size; + f->flush = __fmtFdFlush; + f->farg = (void*)fd; + f->nfmt = 0; + return 0; +} diff --git a/src/lib9/fmt/fmtfdflush.c b/src/lib9/fmt/fmtfdflush.c new file mode 100644 index 00000000..796feab2 --- /dev/null +++ b/src/lib9/fmt/fmtfdflush.c @@ -0,0 +1,33 @@ +/* + * The authors of this software are Rob Pike and Ken Thompson. + * Copyright (c) 2002 by Lucent Technologies. + * Permission to use, copy, modify, and distribute this software for any + * purpose without fee is hereby granted, provided that this entire notice + * is included in all copies of any software which is or includes a copy + * or modification of this software and in all copies of the supporting + * documentation for such software. + * THIS SOFTWARE IS BEING PROVIDED "AS IS", WITHOUT ANY EXPRESS OR IMPLIED + * WARRANTY. IN PARTICULAR, NEITHER THE AUTHORS NOR LUCENT TECHNOLOGIES MAKE ANY + * REPRESENTATION OR WARRANTY OF ANY KIND CONCERNING THE MERCHANTABILITY + * OF THIS SOFTWARE OR ITS FITNESS FOR ANY PARTICULAR PURPOSE. + */ +#include +#include +#include "fmt.h" +#include "fmtdef.h" + +/* + * generic routine for flushing a formatting buffer + * to a file descriptor + */ +int +__fmtFdFlush(Fmt *f) +{ + int n; + + n = (char*)f->to - (char*)f->start; + if(n && write((int)f->farg, f->start, n) != n) + return 0; + f->to = f->start; + return 1; +} diff --git a/src/lib9/fmt/fmtlock.c b/src/lib9/fmt/fmtlock.c new file mode 100644 index 00000000..fffe81cf --- /dev/null +++ b/src/lib9/fmt/fmtlock.c @@ -0,0 +1,28 @@ +/* + * The authors of this software are Rob Pike and Ken Thompson. + * Copyright (c) 2002 by Lucent Technologies. + * Permission to use, copy, modify, and distribute this software for any + * purpose without fee is hereby granted, provided that this entire notice + * is included in all copies of any software which is or includes a copy + * or modification of this software and in all copies of the supporting + * documentation for such software. + * THIS SOFTWARE IS BEING PROVIDED "AS IS", WITHOUT ANY EXPRESS OR IMPLIED + * WARRANTY. IN PARTICULAR, NEITHER THE AUTHORS NOR LUCENT TECHNOLOGIES MAKE ANY + * REPRESENTATION OR WARRANTY OF ANY KIND CONCERNING THE MERCHANTABILITY + * OF THIS SOFTWARE OR ITS FITNESS FOR ANY PARTICULAR PURPOSE. + */ +#include +#include "fmt.h" +#include "fmtdef.h" + +void +__fmtlock(void) +{ + ; +} + +void +__fmtunlock(void) +{ + ; +} diff --git a/src/lib9/fmt/fmtprint.c b/src/lib9/fmt/fmtprint.c new file mode 100644 index 00000000..fe2ad3cc --- /dev/null +++ b/src/lib9/fmt/fmtprint.c @@ -0,0 +1,47 @@ +/* + * The authors of this software are Rob Pike and Ken Thompson. + * Copyright (c) 2002 by Lucent Technologies. + * Permission to use, copy, modify, and distribute this software for any + * purpose without fee is hereby granted, provided that this entire notice + * is included in all copies of any software which is or includes a copy + * or modification of this software and in all copies of the supporting + * documentation for such software. + * THIS SOFTWARE IS BEING PROVIDED "AS IS", WITHOUT ANY EXPRESS OR IMPLIED + * WARRANTY. IN PARTICULAR, NEITHER THE AUTHORS NOR LUCENT TECHNOLOGIES MAKE ANY + * REPRESENTATION OR WARRANTY OF ANY KIND CONCERNING THE MERCHANTABILITY + * OF THIS SOFTWARE OR ITS FITNESS FOR ANY PARTICULAR PURPOSE. + */ +#include +#include +#include "utf.h" +#include "fmt.h" +#include "fmtdef.h" + + +/* + * format a string into the output buffer + * designed for formats which themselves call fmt, + * but ignore any width flags + */ +int +fmtprint(Fmt *f, char *fmt, ...) +{ + va_list va; + int n; + + f->flags = 0; + f->width = 0; + f->prec = 0; + va = f->args; + va_start(f->args, fmt); + n = dofmt(f, fmt); + va_end(f->args); + f->flags = 0; + f->width = 0; + f->prec = 0; + f->args = va; + if(n >= 0) + return 0; + return n; +} + diff --git a/src/lib9/fmt/fmtquote.c b/src/lib9/fmt/fmtquote.c new file mode 100644 index 00000000..9d5633d6 --- /dev/null +++ b/src/lib9/fmt/fmtquote.c @@ -0,0 +1,262 @@ +/* + * The authors of this software are Rob Pike and Ken Thompson. + * Copyright (c) 2002 by Lucent Technologies. + * Permission to use, copy, modify, and distribute this software for any + * purpose without fee is hereby granted, provided that this entire notice + * is included in all copies of any software which is or includes a copy + * or modification of this software and in all copies of the supporting + * documentation for such software. + * THIS SOFTWARE IS BEING PROVIDED "AS IS", WITHOUT ANY EXPRESS OR IMPLIED + * WARRANTY. IN PARTICULAR, NEITHER THE AUTHORS NOR LUCENT TECHNOLOGIES MAKE ANY + * REPRESENTATION OR WARRANTY OF ANY KIND CONCERNING THE MERCHANTABILITY + * OF THIS SOFTWARE OR ITS FITNESS FOR ANY PARTICULAR PURPOSE. + */ +#include +#include +#include "utf.h" +#include "fmt.h" +#include "fmtdef.h" + +/* + * How many bytes of output UTF will be produced by quoting (if necessary) this string? + * How many runes? How much of the input will be consumed? + * The parameter q is filled in by __quotesetup. + * The string may be UTF or Runes (s or r). + * Return count does not include NUL. + * Terminate the scan at the first of: + * NUL in input + * count exceeded in input + * count exceeded on output + * *ninp is set to number of input bytes accepted. + * nin may be <0 initially, to avoid checking input by count. + */ +void +__quotesetup(char *s, Rune *r, int nin, int nout, Quoteinfo *q, int sharp, int runesout) +{ + int w; + Rune c; + + q->quoted = 0; + q->nbytesout = 0; + q->nrunesout = 0; + q->nbytesin = 0; + q->nrunesin = 0; + if(sharp || nin==0 || (s && *s=='\0') || (r && *r=='\0')){ + if(nout < 2) + return; + q->quoted = 1; + q->nbytesout = 2; + q->nrunesout = 2; + } + for(; nin!=0; nin-=w){ + if(s) + w = chartorune(&c, s); + else{ + c = *r; + w = runelen(c); + } + + if(c == '\0') + break; + if(runesout){ + if(q->nrunesout+1 > nout) + break; + }else{ + if(q->nbytesout+w > nout) + break; + } + + if((c <= L' ') || (c == L'\'') || (fmtdoquote!=nil && fmtdoquote(c))){ + if(!q->quoted){ + if(runesout){ + if(1+q->nrunesout+1+1 > nout) /* no room for quotes */ + break; + }else{ + if(1+q->nbytesout+w+1 > nout) /* no room for quotes */ + break; + } + q->nrunesout += 2; /* include quotes */ + q->nbytesout += 2; /* include quotes */ + q->quoted = 1; + } + if(c == '\'') { + if(runesout){ + if(1+q->nrunesout+1 > nout) /* no room for quotes */ + break; + }else{ + if(1+q->nbytesout+w > nout) /* no room for quotes */ + break; + } + q->nbytesout++; + q->nrunesout++; /* quotes reproduce as two characters */ + } + } + + /* advance input */ + if(s) + s += w; + else + r++; + q->nbytesin += w; + q->nrunesin++; + + /* advance output */ + q->nbytesout += w; + q->nrunesout++; + } +} + +static int +qstrfmt(char *sin, Rune *rin, Quoteinfo *q, Fmt *f) +{ + Rune r, *rm, *rme; + char *t, *s, *m, *me; + Rune *rt, *rs; + ulong fl; + int nc, w; + + m = sin; + me = m + q->nbytesin; + rm = rin; + rme = rm + q->nrunesin; + + w = f->width; + fl = f->flags; + if(f->runes){ + if(!(fl & FmtLeft) && __rfmtpad(f, w - q->nrunesout) < 0) + return -1; + }else{ + if(!(fl & FmtLeft) && __fmtpad(f, w - q->nbytesout) < 0) + return -1; + } + t = (char*)f->to; + s = (char*)f->stop; + rt = (Rune*)f->to; + rs = (Rune*)f->stop; + if(f->runes) + FMTRCHAR(f, rt, rs, '\''); + else + FMTRUNE(f, t, s, '\''); + for(nc = q->nrunesin; nc > 0; nc--){ + if(sin){ + r = *(uchar*)m; + if(r < Runeself) + m++; + else if((me - m) >= UTFmax || fullrune(m, me-m)) + m += chartorune(&r, m); + else + break; + }else{ + if(rm >= rme) + break; + r = *(uchar*)rm++; + } + if(f->runes){ + FMTRCHAR(f, rt, rs, r); + if(r == '\'') + FMTRCHAR(f, rt, rs, r); + }else{ + FMTRUNE(f, t, s, r); + if(r == '\'') + FMTRUNE(f, t, s, r); + } + } + + if(f->runes){ + FMTRCHAR(f, rt, rs, '\''); + USED(rs); + f->nfmt += rt - (Rune *)f->to; + f->to = rt; + if(fl & FmtLeft && __rfmtpad(f, w - q->nrunesout) < 0) + return -1; + }else{ + FMTRUNE(f, t, s, '\''); + USED(s); + f->nfmt += t - (char *)f->to; + f->to = t; + if(fl & FmtLeft && __fmtpad(f, w - q->nbytesout) < 0) + return -1; + } + return 0; +} + +int +__quotestrfmt(int runesin, Fmt *f) +{ + int outlen; + Rune *r; + char *s; + Quoteinfo q; + + f->flags &= ~FmtPrec; /* ignored for %q %Q, so disable for %s %S in easy case */ + if(runesin){ + r = va_arg(f->args, Rune *); + s = nil; + }else{ + s = va_arg(f->args, char *); + r = nil; + } + if(!s && !r) + return __fmtcpy(f, (void*)"", 5, 5); + + if(f->flush) + outlen = 0x7FFFFFFF; /* if we can flush, no output limit */ + else if(f->runes) + outlen = (Rune*)f->stop - (Rune*)f->to; + else + outlen = (char*)f->stop - (char*)f->to; + + __quotesetup(s, r, -1, outlen, &q, f->flags&FmtSharp, f->runes); +//print("bytes in %d bytes out %d runes in %d runesout %d\n", q.nbytesin, q.nbytesout, q.nrunesin, q.nrunesout); + + if(runesin){ + if(!q.quoted) + return __fmtrcpy(f, r, q.nrunesin); + return qstrfmt(nil, r, &q, f); + } + + if(!q.quoted) + return __fmtcpy(f, s, q.nrunesin, q.nbytesin); + return qstrfmt(s, nil, &q, f); +} + +int +quotestrfmt(Fmt *f) +{ + return __quotestrfmt(0, f); +} + +int +quoterunestrfmt(Fmt *f) +{ + return __quotestrfmt(1, f); +} + +void +quotefmtinstall(void) +{ + fmtinstall('q', quotestrfmt); + fmtinstall('Q', quoterunestrfmt); +} + +int +__needsquotes(char *s, int *quotelenp) +{ + Quoteinfo q; + + __quotesetup(s, nil, -1, 0x7FFFFFFF, &q, 0, 0); + *quotelenp = q.nbytesout; + + return q.quoted; +} + +int +__runeneedsquotes(Rune *r, int *quotelenp) +{ + Quoteinfo q; + + __quotesetup(nil, r, -1, 0x7FFFFFFF, &q, 0, 0); + *quotelenp = q.nrunesout; + + return q.quoted; +} diff --git a/src/lib9/fmt/fmtrune.c b/src/lib9/fmt/fmtrune.c new file mode 100644 index 00000000..a034546b --- /dev/null +++ b/src/lib9/fmt/fmtrune.c @@ -0,0 +1,40 @@ +/* + * The authors of this software are Rob Pike and Ken Thompson. + * Copyright (c) 2002 by Lucent Technologies. + * Permission to use, copy, modify, and distribute this software for any + * purpose without fee is hereby granted, provided that this entire notice + * is included in all copies of any software which is or includes a copy + * or modification of this software and in all copies of the supporting + * documentation for such software. + * THIS SOFTWARE IS BEING PROVIDED "AS IS", WITHOUT ANY EXPRESS OR IMPLIED + * WARRANTY. IN PARTICULAR, NEITHER THE AUTHORS NOR LUCENT TECHNOLOGIES MAKE ANY + * REPRESENTATION OR WARRANTY OF ANY KIND CONCERNING THE MERCHANTABILITY + * OF THIS SOFTWARE OR ITS FITNESS FOR ANY PARTICULAR PURPOSE. + */ +#include +#include +#include "utf.h" +#include "fmt.h" +#include "fmtdef.h" + +int +fmtrune(Fmt *f, int r) +{ + Rune *rt; + char *t; + int n; + + if(f->runes){ + rt = (Rune*)f->to; + FMTRCHAR(f, rt, f->stop, r); + f->to = rt; + n = 1; + }else{ + t = (char*)f->to; + FMTRUNE(f, t, f->stop, r); + n = t - (char*)f->to; + f->to = t; + } + f->nfmt += n; + return 0; +} diff --git a/src/lib9/fmt/fmtstr.c b/src/lib9/fmt/fmtstr.c new file mode 100644 index 00000000..7af1fa24 --- /dev/null +++ b/src/lib9/fmt/fmtstr.c @@ -0,0 +1,65 @@ +/* + * The authors of this software are Rob Pike and Ken Thompson. + * Copyright (c) 2002 by Lucent Technologies. + * Permission to use, copy, modify, and distribute this software for any + * purpose without fee is hereby granted, provided that this entire notice + * is included in all copies of any software which is or includes a copy + * or modification of this software and in all copies of the supporting + * documentation for such software. + * THIS SOFTWARE IS BEING PROVIDED "AS IS", WITHOUT ANY EXPRESS OR IMPLIED + * WARRANTY. IN PARTICULAR, NEITHER THE AUTHORS NOR LUCENT TECHNOLOGIES MAKE ANY + * REPRESENTATION OR WARRANTY OF ANY KIND CONCERNING THE MERCHANTABILITY + * OF THIS SOFTWARE OR ITS FITNESS FOR ANY PARTICULAR PURPOSE. + */ +#include +#include +#include +#include "utf.h" +#include "fmt.h" +#include "fmtdef.h" + +static int +fmtStrFlush(Fmt *f) +{ + char *s; + int n; + + n = (int)f->farg; + n += 256; + f->farg = (void*)n; + s = (char*)f->start; + f->start = realloc(s, n); + if(f->start == nil){ + f->start = s; + return 0; + } + f->to = (char*)f->start + ((char*)f->to - s); + f->stop = (char*)f->start + n - 1; + return 1; +} + +int +fmtstrinit(Fmt *f) +{ + int n; + + f->runes = 0; + n = 32; + f->start = malloc(n); + if(f->start == nil) + return -1; + f->to = f->start; + f->stop = (char*)f->start + n - 1; + f->flush = fmtStrFlush; + f->farg = (void*)n; + f->nfmt = 0; + return 0; +} + +char* +fmtstrflush(Fmt *f) +{ + *(char*)f->to = '\0'; + f->to = f->start; + return (char*)f->start; +} diff --git a/src/lib9/fmt/fmtvprint.c b/src/lib9/fmt/fmtvprint.c new file mode 100644 index 00000000..6aed013d --- /dev/null +++ b/src/lib9/fmt/fmtvprint.c @@ -0,0 +1,46 @@ +/* + * The authors of this software are Rob Pike and Ken Thompson. + * Copyright (c) 2002 by Lucent Technologies. + * Permission to use, copy, modify, and distribute this software for any + * purpose without fee is hereby granted, provided that this entire notice + * is included in all copies of any software which is or includes a copy + * or modification of this software and in all copies of the supporting + * documentation for such software. + * THIS SOFTWARE IS BEING PROVIDED "AS IS", WITHOUT ANY EXPRESS OR IMPLIED + * WARRANTY. IN PARTICULAR, NEITHER THE AUTHORS NOR LUCENT TECHNOLOGIES MAKE ANY + * REPRESENTATION OR WARRANTY OF ANY KIND CONCERNING THE MERCHANTABILITY + * OF THIS SOFTWARE OR ITS FITNESS FOR ANY PARTICULAR PURPOSE. + */ +#include +#include +#include "utf.h" +#include "fmt.h" +#include "fmtdef.h" + + +/* + * format a string into the output buffer + * designed for formats which themselves call fmt, + * but ignore any width flags + */ +int +fmtvprint(Fmt *f, char *fmt, va_list args) +{ + va_list va; + int n; + + f->flags = 0; + f->width = 0; + f->prec = 0; + va = f->args; + f->args = args; + n = dofmt(f, fmt); + f->flags = 0; + f->width = 0; + f->prec = 0; + f->args = va; + if(n >= 0) + return 0; + return n; +} + diff --git a/src/lib9/fmt/fprint.c b/src/lib9/fmt/fprint.c new file mode 100644 index 00000000..c8b889de --- /dev/null +++ b/src/lib9/fmt/fprint.c @@ -0,0 +1,28 @@ +/* + * The authors of this software are Rob Pike and Ken Thompson. + * Copyright (c) 2002 by Lucent Technologies. + * Permission to use, copy, modify, and distribute this software for any + * purpose without fee is hereby granted, provided that this entire notice + * is included in all copies of any software which is or includes a copy + * or modification of this software and in all copies of the supporting + * documentation for such software. + * THIS SOFTWARE IS BEING PROVIDED "AS IS", WITHOUT ANY EXPRESS OR IMPLIED + * WARRANTY. IN PARTICULAR, NEITHER THE AUTHORS NOR LUCENT TECHNOLOGIES MAKE ANY + * REPRESENTATION OR WARRANTY OF ANY KIND CONCERNING THE MERCHANTABILITY + * OF THIS SOFTWARE OR ITS FITNESS FOR ANY PARTICULAR PURPOSE. + */ +#include +#include "utf.h" +#include "fmt.h" + +int +fprint(int fd, char *fmt, ...) +{ + int n; + va_list args; + + va_start(args, fmt); + n = vfprint(fd, fmt, args); + va_end(args); + return n; +} diff --git a/src/lib9/fmt/mkfile b/src/lib9/fmt/mkfile new file mode 100644 index 00000000..0a5d3b67 --- /dev/null +++ b/src/lib9/fmt/mkfile @@ -0,0 +1,57 @@ +PLAN9=../.. +<$PLAN9/src/mkhdr + +LIB=libfmt.a + +NUM=\ + charstod.$O\ + pow10.$O\ + +# Could add errfmt, but we want to pick it up from lib9 instead. +OFILES=\ + dofmt.$O\ + errfmt.$O\ + fltfmt.$O\ + fmt.$O\ + fmtfd.$O\ + fmtfdflush.$O\ + fmtlock.$O\ + fmtprint.$O\ + fmtquote.$O\ + fmtrune.$O\ + fmtstr.$O\ + fmtvprint.$O\ + fprint.$O\ + nan64.$O\ + print.$O\ + runefmtstr.$O\ + runeseprint.$O\ + runesmprint.$O\ + runesnprint.$O\ + runesprint.$O\ + runevseprint.$O\ + runevsmprint.$O\ + runevsnprint.$O\ + seprint.$O\ + smprint.$O\ + snprint.$O\ + sprint.$O\ + strtod.$O\ + vfprint.$O\ + vseprint.$O\ + vsmprint.$O\ + vsnprint.$O\ + $NUM\ + +HFILES=\ + fmtdef.h\ + $PLAN9/include/fmt.h\ + +<$PLAN9/src/mksyslib + +$NAN.$O: nan.h +strtod.$O: nan.h + +test: $LIB test.$O + $CC -o test test.$O $LIB -L$PLAN9/lib -lutf + diff --git a/src/lib9/fmt/nan.h b/src/lib9/fmt/nan.h new file mode 100644 index 00000000..be78e14e --- /dev/null +++ b/src/lib9/fmt/nan.h @@ -0,0 +1,4 @@ +extern double __NaN(void); +extern double __Inf(int); +extern int __isNaN(double); +extern int __isInf(double, int); diff --git a/src/lib9/fmt/nan64.c b/src/lib9/fmt/nan64.c new file mode 100644 index 00000000..a02a3adb --- /dev/null +++ b/src/lib9/fmt/nan64.c @@ -0,0 +1,76 @@ +/* + * 64-bit IEEE not-a-number routines. + * This is big/little-endian portable assuming that + * the 64-bit doubles and 64-bit integers have the + * same byte ordering. + */ + +#include "nan.h" + +#ifdef __APPLE__ +#define _NEEDLL +#endif + +typedef unsigned long long uvlong; +typedef unsigned long ulong; + +#ifdef _NEEDLL +static uvlong uvnan = 0x7FF0000000000001LL; +static uvlong uvinf = 0x7FF0000000000000LL; +static uvlong uvneginf = 0xFFF0000000000000LL; +#else +static uvlong uvnan = 0x7FF0000000000001; +static uvlong uvinf = 0x7FF0000000000000; +static uvlong uvneginf = 0xFFF0000000000000; +#endif + +double +__NaN(void) +{ + uvlong *p; + + /* gcc complains about "return *(double*)&uvnan;" */ + p = &uvnan; + return *(double*)p; +} + +int +__isNaN(double d) +{ + uvlong x; + double *p; + + p = &d; + x = *(uvlong*)p; + return (ulong)(x>>32)==0x7FF00000 && !__isInf(d, 0); +} + +double +__Inf(int sign) +{ + uvlong *p; + + if(sign < 0) + p = &uvinf; + else + p = &uvneginf; + return *(double*)p; +} + +int +__isInf(double d, int sign) +{ + uvlong x; + double *p; + + p = &d; + x = *(uvlong*)p; + if(sign == 0) + return x==uvinf || x==uvneginf; + else if(sign > 0) + return x==uvinf; + else + return x==uvneginf; +} + + diff --git a/src/lib9/fmt/pow10.c b/src/lib9/fmt/pow10.c new file mode 100644 index 00000000..65a426c3 --- /dev/null +++ b/src/lib9/fmt/pow10.c @@ -0,0 +1,57 @@ +/* + * The authors of this software are Rob Pike and Ken Thompson. + * Copyright (c) 2002 by Lucent Technologies. + * Permission to use, copy, modify, and distribute this software for any + * purpose without fee is hereby granted, provided that this entire notice + * is included in all copies of any software which is or includes a copy + * or modification of this software and in all copies of the supporting + * documentation for such software. + * THIS SOFTWARE IS BEING PROVIDED "AS IS", WITHOUT ANY EXPRESS OR IMPLIED + * WARRANTY. IN PARTICULAR, NEITHER THE AUTHORS NOR LUCENT TECHNOLOGIES MAKE ANY + * REPRESENTATION OR WARRANTY OF ANY KIND CONCERNING THE MERCHANTABILITY + * OF THIS SOFTWARE OR ITS FITNESS FOR ANY PARTICULAR PURPOSE. + */ +#include +#include +#include "utf.h" +#include "fmt.h" +#include "fmtdef.h" + +/* + * this table might overflow 127-bit exponent representations. + * in that case, truncate it after 1.0e38. + * it is important to get all one can from this + * routine since it is used in atof to scale numbers. + * the presumption is that C converts fp numbers better + * than multipication of lower powers of 10. + */ + +static +double tab[] = +{ + 1.0e0, 1.0e1, 1.0e2, 1.0e3, 1.0e4, 1.0e5, 1.0e6, 1.0e7, 1.0e8, 1.0e9, + 1.0e10,1.0e11,1.0e12,1.0e13,1.0e14,1.0e15,1.0e16,1.0e17,1.0e18,1.0e19, + 1.0e20,1.0e21,1.0e22,1.0e23,1.0e24,1.0e25,1.0e26,1.0e27,1.0e28,1.0e29, + 1.0e30,1.0e31,1.0e32,1.0e33,1.0e34,1.0e35,1.0e36,1.0e37,1.0e38,1.0e39, + 1.0e40,1.0e41,1.0e42,1.0e43,1.0e44,1.0e45,1.0e46,1.0e47,1.0e48,1.0e49, + 1.0e50,1.0e51,1.0e52,1.0e53,1.0e54,1.0e55,1.0e56,1.0e57,1.0e58,1.0e59, + 1.0e60,1.0e61,1.0e62,1.0e63,1.0e64,1.0e65,1.0e66,1.0e67,1.0e68,1.0e69, +}; + +double +__fmtpow10(int n) +{ + int m; + + if(n < 0) { + n = -n; + if(n < (int)(sizeof(tab)/sizeof(tab[0]))) + return 1/tab[n]; + m = n/2; + return __fmtpow10(-m) * __fmtpow10(m-n); + } + if(n < (int)(sizeof(tab)/sizeof(tab[0]))) + return tab[n]; + m = n/2; + return __fmtpow10(m) * __fmtpow10(n-m); +} diff --git a/src/lib9/fmt/print.c b/src/lib9/fmt/print.c new file mode 100644 index 00000000..381a1da9 --- /dev/null +++ b/src/lib9/fmt/print.c @@ -0,0 +1,28 @@ +/* + * The authors of this software are Rob Pike and Ken Thompson. + * Copyright (c) 2002 by Lucent Technologies. + * Permission to use, copy, modify, and distribute this software for any + * purpose without fee is hereby granted, provided that this entire notice + * is included in all copies of any software which is or includes a copy + * or modification of this software and in all copies of the supporting + * documentation for such software. + * THIS SOFTWARE IS BEING PROVIDED "AS IS", WITHOUT ANY EXPRESS OR IMPLIED + * WARRANTY. IN PARTICULAR, NEITHER THE AUTHORS NOR LUCENT TECHNOLOGIES MAKE ANY + * REPRESENTATION OR WARRANTY OF ANY KIND CONCERNING THE MERCHANTABILITY + * OF THIS SOFTWARE OR ITS FITNESS FOR ANY PARTICULAR PURPOSE. + */ +#include +#include "utf.h" +#include "fmt.h" + +int +print(char *fmt, ...) +{ + int n; + va_list args; + + va_start(args, fmt); + n = vfprint(1, fmt, args); + va_end(args); + return n; +} diff --git a/src/lib9/fmt/runefmtstr.c b/src/lib9/fmt/runefmtstr.c new file mode 100644 index 00000000..a2ec6cb4 --- /dev/null +++ b/src/lib9/fmt/runefmtstr.c @@ -0,0 +1,65 @@ +/* + * The authors of this software are Rob Pike and Ken Thompson. + * Copyright (c) 2002 by Lucent Technologies. + * Permission to use, copy, modify, and distribute this software for any + * purpose without fee is hereby granted, provided that this entire notice + * is included in all copies of any software which is or includes a copy + * or modification of this software and in all copies of the supporting + * documentation for such software. + * THIS SOFTWARE IS BEING PROVIDED "AS IS", WITHOUT ANY EXPRESS OR IMPLIED + * WARRANTY. IN PARTICULAR, NEITHER THE AUTHORS NOR LUCENT TECHNOLOGIES MAKE ANY + * REPRESENTATION OR WARRANTY OF ANY KIND CONCERNING THE MERCHANTABILITY + * OF THIS SOFTWARE OR ITS FITNESS FOR ANY PARTICULAR PURPOSE. + */ +#include +#include +#include +#include "utf.h" +#include "fmt.h" +#include "fmtdef.h" + +static int +runeFmtStrFlush(Fmt *f) +{ + Rune *s; + int n; + + n = (int)f->farg; + n += 256; + f->farg = (void*)n; + s = (Rune*)f->start; + f->start = realloc(s, sizeof(Rune)*n); + if(f->start == nil){ + f->start = s; + return 0; + } + f->to = (Rune*)f->start + ((Rune*)f->to - s); + f->stop = (Rune*)f->start + n - 1; + return 1; +} + +int +runefmtstrinit(Fmt *f) +{ + int n; + + f->runes = 1; + n = 32; + f->start = malloc(sizeof(Rune)*n); + if(f->start == nil) + return -1; + f->to = f->start; + f->stop = (Rune*)f->start + n - 1; + f->flush = runeFmtStrFlush; + f->farg = (void*)n; + f->nfmt = 0; + return 0; +} + +Rune* +runefmtstrflush(Fmt *f) +{ + *(Rune*)f->to = '\0'; + f->to = f->start; + return f->start; +} diff --git a/src/lib9/fmt/runeseprint.c b/src/lib9/fmt/runeseprint.c new file mode 100644 index 00000000..7829a439 --- /dev/null +++ b/src/lib9/fmt/runeseprint.c @@ -0,0 +1,30 @@ +/* + * The authors of this software are Rob Pike and Ken Thompson. + * Copyright (c) 2002 by Lucent Technologies. + * Permission to use, copy, modify, and distribute this software for any + * purpose without fee is hereby granted, provided that this entire notice + * is included in all copies of any software which is or includes a copy + * or modification of this software and in all copies of the supporting + * documentation for such software. + * THIS SOFTWARE IS BEING PROVIDED "AS IS", WITHOUT ANY EXPRESS OR IMPLIED + * WARRANTY. IN PARTICULAR, NEITHER THE AUTHORS NOR LUCENT TECHNOLOGIES MAKE ANY + * REPRESENTATION OR WARRANTY OF ANY KIND CONCERNING THE MERCHANTABILITY + * OF THIS SOFTWARE OR ITS FITNESS FOR ANY PARTICULAR PURPOSE. + */ +#include +#include +#include "utf.h" +#include "fmt.h" +#include "fmtdef.h" + +Rune* +runeseprint(Rune *buf, Rune *e, char *fmt, ...) +{ + Rune *p; + va_list args; + + va_start(args, fmt); + p = runevseprint(buf, e, fmt, args); + va_end(args); + return p; +} diff --git a/src/lib9/fmt/runesmprint.c b/src/lib9/fmt/runesmprint.c new file mode 100644 index 00000000..538ae0ee --- /dev/null +++ b/src/lib9/fmt/runesmprint.c @@ -0,0 +1,30 @@ +/* + * The authors of this software are Rob Pike and Ken Thompson. + * Copyright (c) 2002 by Lucent Technologies. + * Permission to use, copy, modify, and distribute this software for any + * purpose without fee is hereby granted, provided that this entire notice + * is included in all copies of any software which is or includes a copy + * or modification of this software and in all copies of the supporting + * documentation for such software. + * THIS SOFTWARE IS BEING PROVIDED "AS IS", WITHOUT ANY EXPRESS OR IMPLIED + * WARRANTY. IN PARTICULAR, NEITHER THE AUTHORS NOR LUCENT TECHNOLOGIES MAKE ANY + * REPRESENTATION OR WARRANTY OF ANY KIND CONCERNING THE MERCHANTABILITY + * OF THIS SOFTWARE OR ITS FITNESS FOR ANY PARTICULAR PURPOSE. + */ +#include +#include +#include "utf.h" +#include "fmt.h" +#include "fmtdef.h" + +Rune* +runesmprint(char *fmt, ...) +{ + va_list args; + Rune *p; + + va_start(args, fmt); + p = runevsmprint(fmt, args); + va_end(args); + return p; +} diff --git a/src/lib9/fmt/runesnprint.c b/src/lib9/fmt/runesnprint.c new file mode 100644 index 00000000..5abb6dbf --- /dev/null +++ b/src/lib9/fmt/runesnprint.c @@ -0,0 +1,31 @@ +/* + * The authors of this software are Rob Pike and Ken Thompson. + * Copyright (c) 2002 by Lucent Technologies. + * Permission to use, copy, modify, and distribute this software for any + * purpose without fee is hereby granted, provided that this entire notice + * is included in all copies of any software which is or includes a copy + * or modification of this software and in all copies of the supporting + * documentation for such software. + * THIS SOFTWARE IS BEING PROVIDED "AS IS", WITHOUT ANY EXPRESS OR IMPLIED + * WARRANTY. IN PARTICULAR, NEITHER THE AUTHORS NOR LUCENT TECHNOLOGIES MAKE ANY + * REPRESENTATION OR WARRANTY OF ANY KIND CONCERNING THE MERCHANTABILITY + * OF THIS SOFTWARE OR ITS FITNESS FOR ANY PARTICULAR PURPOSE. + */ +#include +#include +#include "utf.h" +#include "fmt.h" +#include "fmtdef.h" + +int +runesnprint(Rune *buf, int len, char *fmt, ...) +{ + int n; + va_list args; + + va_start(args, fmt); + n = runevsnprint(buf, len, fmt, args); + va_end(args); + return n; +} + diff --git a/src/lib9/fmt/runesprint.c b/src/lib9/fmt/runesprint.c new file mode 100644 index 00000000..8abfeb65 --- /dev/null +++ b/src/lib9/fmt/runesprint.c @@ -0,0 +1,30 @@ +/* + * The authors of this software are Rob Pike and Ken Thompson. + * Copyright (c) 2002 by Lucent Technologies. + * Permission to use, copy, modify, and distribute this software for any + * purpose without fee is hereby granted, provided that this entire notice + * is included in all copies of any software which is or includes a copy + * or modification of this software and in all copies of the supporting + * documentation for such software. + * THIS SOFTWARE IS BEING PROVIDED "AS IS", WITHOUT ANY EXPRESS OR IMPLIED + * WARRANTY. IN PARTICULAR, NEITHER THE AUTHORS NOR LUCENT TECHNOLOGIES MAKE ANY + * REPRESENTATION OR WARRANTY OF ANY KIND CONCERNING THE MERCHANTABILITY + * OF THIS SOFTWARE OR ITS FITNESS FOR ANY PARTICULAR PURPOSE. + */ +#include +#include +#include "utf.h" +#include "fmt.h" +#include "fmtdef.h" + +int +runesprint(Rune *buf, char *fmt, ...) +{ + int n; + va_list args; + + va_start(args, fmt); + n = runevsnprint(buf, 256, fmt, args); + va_end(args); + return n; +} diff --git a/src/lib9/fmt/runevseprint.c b/src/lib9/fmt/runevseprint.c new file mode 100644 index 00000000..e4cc1ea8 --- /dev/null +++ b/src/lib9/fmt/runevseprint.c @@ -0,0 +1,39 @@ +/* + * The authors of this software are Rob Pike and Ken Thompson. + * Copyright (c) 2002 by Lucent Technologies. + * Permission to use, copy, modify, and distribute this software for any + * purpose without fee is hereby granted, provided that this entire notice + * is included in all copies of any software which is or includes a copy + * or modification of this software and in all copies of the supporting + * documentation for such software. + * THIS SOFTWARE IS BEING PROVIDED "AS IS", WITHOUT ANY EXPRESS OR IMPLIED + * WARRANTY. IN PARTICULAR, NEITHER THE AUTHORS NOR LUCENT TECHNOLOGIES MAKE ANY + * REPRESENTATION OR WARRANTY OF ANY KIND CONCERNING THE MERCHANTABILITY + * OF THIS SOFTWARE OR ITS FITNESS FOR ANY PARTICULAR PURPOSE. + */ +#include +#include +#include "utf.h" +#include "fmt.h" +#include "fmtdef.h" + +Rune* +runevseprint(Rune *buf, Rune *e, char *fmt, va_list args) +{ + Fmt f; + + if(e <= buf) + return nil; + f.runes = 1; + f.start = buf; + f.to = buf; + f.stop = e - 1; + f.flush = nil; + f.farg = nil; + f.nfmt = 0; + f.args = args; + dofmt(&f, fmt); + *(Rune*)f.to = '\0'; + return (Rune*)f.to; +} + diff --git a/src/lib9/fmt/runevsmprint.c b/src/lib9/fmt/runevsmprint.c new file mode 100644 index 00000000..71c23d84 --- /dev/null +++ b/src/lib9/fmt/runevsmprint.c @@ -0,0 +1,37 @@ +/* + * The authors of this software are Rob Pike and Ken Thompson. + * Copyright (c) 2002 by Lucent Technologies. + * Permission to use, copy, modify, and distribute this software for any + * purpose without fee is hereby granted, provided that this entire notice + * is included in all copies of any software which is or includes a copy + * or modification of this software and in all copies of the supporting + * documentation for such software. + * THIS SOFTWARE IS BEING PROVIDED "AS IS", WITHOUT ANY EXPRESS OR IMPLIED + * WARRANTY. IN PARTICULAR, NEITHER THE AUTHORS NOR LUCENT TECHNOLOGIES MAKE ANY + * REPRESENTATION OR WARRANTY OF ANY KIND CONCERNING THE MERCHANTABILITY + * OF THIS SOFTWARE OR ITS FITNESS FOR ANY PARTICULAR PURPOSE. + */ +#include +#include +#include "utf.h" +#include "fmt.h" +#include "fmtdef.h" + +/* + * print into an allocated string buffer + */ +Rune* +runevsmprint(char *fmt, va_list args) +{ + Fmt f; + int n; + + if(runefmtstrinit(&f) < 0) + return nil; + f.args = args; + n = dofmt(&f, fmt); + if(n < 0) + return nil; + *(Rune*)f.to = '\0'; + return (Rune*)f.start; +} diff --git a/src/lib9/fmt/runevsnprint.c b/src/lib9/fmt/runevsnprint.c new file mode 100644 index 00000000..933a04ca --- /dev/null +++ b/src/lib9/fmt/runevsnprint.c @@ -0,0 +1,38 @@ +/* + * The authors of this software are Rob Pike and Ken Thompson. + * Copyright (c) 2002 by Lucent Technologies. + * Permission to use, copy, modify, and distribute this software for any + * purpose without fee is hereby granted, provided that this entire notice + * is included in all copies of any software which is or includes a copy + * or modification of this software and in all copies of the supporting + * documentation for such software. + * THIS SOFTWARE IS BEING PROVIDED "AS IS", WITHOUT ANY EXPRESS OR IMPLIED + * WARRANTY. IN PARTICULAR, NEITHER THE AUTHORS NOR LUCENT TECHNOLOGIES MAKE ANY + * REPRESENTATION OR WARRANTY OF ANY KIND CONCERNING THE MERCHANTABILITY + * OF THIS SOFTWARE OR ITS FITNESS FOR ANY PARTICULAR PURPOSE. + */ +#include +#include +#include "utf.h" +#include "fmt.h" +#include "fmtdef.h" + +int +runevsnprint(Rune *buf, int len, char *fmt, va_list args) +{ + Fmt f; + + if(len <= 0) + return -1; + f.runes = 1; + f.start = buf; + f.to = buf; + f.stop = buf + len - 1; + f.flush = nil; + f.farg = nil; + f.nfmt = 0; + f.args = args; + dofmt(&f, fmt); + *(Rune*)f.to = '\0'; + return (Rune*)f.to - buf; +} diff --git a/src/lib9/fmt/seprint.c b/src/lib9/fmt/seprint.c new file mode 100644 index 00000000..d5031a2b --- /dev/null +++ b/src/lib9/fmt/seprint.c @@ -0,0 +1,27 @@ +/* + * The authors of this software are Rob Pike and Ken Thompson. + * Copyright (c) 2002 by Lucent Technologies. + * Permission to use, copy, modify, and distribute this software for any + * purpose without fee is hereby granted, provided that this entire notice + * is included in all copies of any software which is or includes a copy + * or modification of this software and in all copies of the supporting + * documentation for such software. + * THIS SOFTWARE IS BEING PROVIDED "AS IS", WITHOUT ANY EXPRESS OR IMPLIED + * WARRANTY. IN PARTICULAR, NEITHER THE AUTHORS NOR LUCENT TECHNOLOGIES MAKE ANY + * REPRESENTATION OR WARRANTY OF ANY KIND CONCERNING THE MERCHANTABILITY + * OF THIS SOFTWARE OR ITS FITNESS FOR ANY PARTICULAR PURPOSE. + */ +#include +#include "fmt.h" + +char* +seprint(char *buf, char *e, char *fmt, ...) +{ + char *p; + va_list args; + + va_start(args, fmt); + p = vseprint(buf, e, fmt, args); + va_end(args); + return p; +} diff --git a/src/lib9/fmt/smprint.c b/src/lib9/fmt/smprint.c new file mode 100644 index 00000000..543411a0 --- /dev/null +++ b/src/lib9/fmt/smprint.c @@ -0,0 +1,27 @@ +/* + * The authors of this software are Rob Pike and Ken Thompson. + * Copyright (c) 2002 by Lucent Technologies. + * Permission to use, copy, modify, and distribute this software for any + * purpose without fee is hereby granted, provided that this entire notice + * is included in all copies of any software which is or includes a copy + * or modification of this software and in all copies of the supporting + * documentation for such software. + * THIS SOFTWARE IS BEING PROVIDED "AS IS", WITHOUT ANY EXPRESS OR IMPLIED + * WARRANTY. IN PARTICULAR, NEITHER THE AUTHORS NOR LUCENT TECHNOLOGIES MAKE ANY + * REPRESENTATION OR WARRANTY OF ANY KIND CONCERNING THE MERCHANTABILITY + * OF THIS SOFTWARE OR ITS FITNESS FOR ANY PARTICULAR PURPOSE. + */ +#include +#include "fmt.h" + +char* +smprint(char *fmt, ...) +{ + va_list args; + char *p; + + va_start(args, fmt); + p = vsmprint(fmt, args); + va_end(args); + return p; +} diff --git a/src/lib9/fmt/snprint.c b/src/lib9/fmt/snprint.c new file mode 100644 index 00000000..594606d2 --- /dev/null +++ b/src/lib9/fmt/snprint.c @@ -0,0 +1,28 @@ +/* + * The authors of this software are Rob Pike and Ken Thompson. + * Copyright (c) 2002 by Lucent Technologies. + * Permission to use, copy, modify, and distribute this software for any + * purpose without fee is hereby granted, provided that this entire notice + * is included in all copies of any software which is or includes a copy + * or modification of this software and in all copies of the supporting + * documentation for such software. + * THIS SOFTWARE IS BEING PROVIDED "AS IS", WITHOUT ANY EXPRESS OR IMPLIED + * WARRANTY. IN PARTICULAR, NEITHER THE AUTHORS NOR LUCENT TECHNOLOGIES MAKE ANY + * REPRESENTATION OR WARRANTY OF ANY KIND CONCERNING THE MERCHANTABILITY + * OF THIS SOFTWARE OR ITS FITNESS FOR ANY PARTICULAR PURPOSE. + */ +#include +#include "fmt.h" + +int +snprint(char *buf, int len, char *fmt, ...) +{ + int n; + va_list args; + + va_start(args, fmt); + n = vsnprint(buf, len, fmt, args); + va_end(args); + return n; +} + diff --git a/src/lib9/fmt/sprint.c b/src/lib9/fmt/sprint.c new file mode 100644 index 00000000..57150c36 --- /dev/null +++ b/src/lib9/fmt/sprint.c @@ -0,0 +1,27 @@ +/* + * The authors of this software are Rob Pike and Ken Thompson. + * Copyright (c) 2002 by Lucent Technologies. + * Permission to use, copy, modify, and distribute this software for any + * purpose without fee is hereby granted, provided that this entire notice + * is included in all copies of any software which is or includes a copy + * or modification of this software and in all copies of the supporting + * documentation for such software. + * THIS SOFTWARE IS BEING PROVIDED "AS IS", WITHOUT ANY EXPRESS OR IMPLIED + * WARRANTY. IN PARTICULAR, NEITHER THE AUTHORS NOR LUCENT TECHNOLOGIES MAKE ANY + * REPRESENTATION OR WARRANTY OF ANY KIND CONCERNING THE MERCHANTABILITY + * OF THIS SOFTWARE OR ITS FITNESS FOR ANY PARTICULAR PURPOSE. + */ +#include +#include "fmt.h" + +int +sprint(char *buf, char *fmt, ...) +{ + int n; + va_list args; + + va_start(args, fmt); + n = vsnprint(buf, 65536, fmt, args); /* big number, but sprint is deprecated anyway */ + va_end(args); + return n; +} diff --git a/src/lib9/fmt/strtod.c b/src/lib9/fmt/strtod.c new file mode 100644 index 00000000..685b7fae --- /dev/null +++ b/src/lib9/fmt/strtod.c @@ -0,0 +1,539 @@ +/* + * The authors of this software are Rob Pike and Ken Thompson. + * Copyright (c) 2002 by Lucent Technologies. + * Permission to use, copy, modify, and distribute this software for any + * purpose without fee is hereby granted, provided that this entire notice + * is included in all copies of any software which is or includes a copy + * or modification of this software and in all copies of the supporting + * documentation for such software. + * THIS SOFTWARE IS BEING PROVIDED "AS IS", WITHOUT ANY EXPRESS OR IMPLIED + * WARRANTY. IN PARTICULAR, NEITHER THE AUTHORS NOR LUCENT TECHNOLOGIES MAKE ANY + * REPRESENTATION OR WARRANTY OF ANY KIND CONCERNING THE MERCHANTABILITY + * OF THIS SOFTWARE OR ITS FITNESS FOR ANY PARTICULAR PURPOSE. + */ +#include +#include +#include +#include +#include +#include +#include "fmt.h" +#include "nan.h" + +#ifndef nelem +#define nelem(x) (sizeof(x)/sizeof *(x)) +#endif +#define nil ((void*)0) +#define ulong _fmtulong +typedef unsigned long ulong; + +static ulong +umuldiv(ulong a, ulong b, ulong c) +{ + double d; + + d = ((double)a * (double)b) / (double)c; + if(d >= 4294967295.) + d = 4294967295.; + return (ulong)d; +} + +/* + * This routine will convert to arbitrary precision + * floating point entirely in multi-precision fixed. + * The answer is the closest floating point number to + * the given decimal number. Exactly half way are + * rounded ala ieee rules. + * Method is to scale input decimal between .500 and .999... + * with external power of 2, then binary search for the + * closest mantissa to this decimal number. + * Nmant is is the required precision. (53 for ieee dp) + * Nbits is the max number of bits/word. (must be <= 28) + * Prec is calculated - the number of words of fixed mantissa. + */ +enum +{ + Nbits = 28, /* bits safely represented in a ulong */ + Nmant = 53, /* bits of precision required */ + Prec = (Nmant+Nbits+1)/Nbits, /* words of Nbits each to represent mantissa */ + Sigbit = 1<<(Prec*Nbits-Nmant), /* first significant bit of Prec-th word */ + Ndig = 1500, + One = (ulong)(1<>1), + Maxe = 310, + + Fsign = 1<<0, /* found - */ + Fesign = 1<<1, /* found e- */ + Fdpoint = 1<<2, /* found . */ + + S0 = 0, /* _ _S0 +S1 #S2 .S3 */ + S1, /* _+ #S2 .S3 */ + S2, /* _+# #S2 .S4 eS5 */ + S3, /* _+. #S4 */ + S4, /* _+#.# #S4 eS5 */ + S5, /* _+#.#e +S6 #S7 */ + S6, /* _+#.#e+ #S7 */ + S7, /* _+#.#e+# #S7 */ +}; + +static int xcmp(char*, char*); +static int fpcmp(char*, ulong*); +static void frnorm(ulong*); +static void divascii(char*, int*, int*, int*); +static void mulascii(char*, int*, int*, int*); + +typedef struct Tab Tab; +struct Tab +{ + int bp; + int siz; + char* cmp; +}; + +double +fmtstrtod(const char *as, char **aas) +{ + int na, ex, dp, bp, c, i, flag, state; + ulong low[Prec], hig[Prec], mid[Prec]; + double d; + char *s, a[Ndig]; + + flag = 0; /* Fsign, Fesign, Fdpoint */ + na = 0; /* number of digits of a[] */ + dp = 0; /* na of decimal point */ + ex = 0; /* exonent */ + + state = S0; + for(s=(char*)as;; s++) { + c = *s; + if(c >= '0' && c <= '9') { + switch(state) { + case S0: + case S1: + case S2: + state = S2; + break; + case S3: + case S4: + state = S4; + break; + + case S5: + case S6: + case S7: + state = S7; + ex = ex*10 + (c-'0'); + continue; + } + if(na == 0 && c == '0') { + dp--; + continue; + } + if(na < Ndig-50) + a[na++] = c; + continue; + } + switch(c) { + case '\t': + case '\n': + case '\v': + case '\f': + case '\r': + case ' ': + if(state == S0) + continue; + break; + case '-': + if(state == S0) + flag |= Fsign; + else + flag |= Fesign; + case '+': + if(state == S0) + state = S1; + else + if(state == S5) + state = S6; + else + break; /* syntax */ + continue; + case '.': + flag |= Fdpoint; + dp = na; + if(state == S0 || state == S1) { + state = S3; + continue; + } + if(state == S2) { + state = S4; + continue; + } + break; + case 'e': + case 'E': + if(state == S2 || state == S4) { + state = S5; + continue; + } + break; + } + break; + } + + /* + * clean up return char-pointer + */ + switch(state) { + case S0: + if(xcmp(s, "nan") == 0) { + if(aas != nil) + *aas = s+3; + goto retnan; + } + case S1: + if(xcmp(s, "infinity") == 0) { + if(aas != nil) + *aas = s+8; + goto retinf; + } + if(xcmp(s, "inf") == 0) { + if(aas != nil) + *aas = s+3; + goto retinf; + } + case S3: + if(aas != nil) + *aas = (char*)as; + goto ret0; /* no digits found */ + case S6: + s--; /* back over +- */ + case S5: + s--; /* back over e */ + break; + } + if(aas != nil) + *aas = s; + + if(flag & Fdpoint) + while(na > 0 && a[na-1] == '0') + na--; + if(na == 0) + goto ret0; /* zero */ + a[na] = 0; + if(!(flag & Fdpoint)) + dp = na; + if(flag & Fesign) + ex = -ex; + dp += ex; + if(dp < -Maxe){ + errno = ERANGE; + goto ret0; /* underflow by exp */ + } else + if(dp > +Maxe) + goto retinf; /* overflow by exp */ + + /* + * normalize the decimal ascii number + * to range .[5-9][0-9]* e0 + */ + bp = 0; /* binary exponent */ + while(dp > 0) + divascii(a, &na, &dp, &bp); + while(dp < 0 || a[0] < '5') + mulascii(a, &na, &dp, &bp); + + /* close approx by naive conversion */ + mid[0] = 0; + mid[1] = 1; + for(i=0; c=a[i]; i++) { + mid[0] = mid[0]*10 + (c-'0'); + mid[1] = mid[1]*10; + if(i >= 8) + break; + } + low[0] = umuldiv(mid[0], One, mid[1]); + hig[0] = umuldiv(mid[0]+1, One, mid[1]); + for(i=1; i>= 1; + } + frnorm(mid); + + /* compare */ + c = fpcmp(a, mid); + if(c > 0) { + c = 1; + for(i=0; i= Sigbit/2) { + mid[Prec-1] += Sigbit; + frnorm(mid); + } + goto out; + +ret0: + return 0; + +retnan: + return __NaN(); + +retinf: + /* + * Unix strtod requires these. Plan 9 would return Inf(0) or Inf(-1). */ + errno = ERANGE; + if(flag & Fsign) + return -HUGE_VAL; + return HUGE_VAL; + +out: + d = 0; + for(i=0; i0; i--) { + f[i] += c; + c = f[i] >> Nbits; + f[i] &= One-1; + } + f[0] += c; +} + +static int +fpcmp(char *a, ulong* f) +{ + ulong tf[Prec]; + int i, d, c; + + for(i=0; i> Nbits) + '0'; + tf[0] &= One-1; + + /* compare next digit */ + c = *a; + if(c == 0) { + if('0' < d) + return -1; + if(tf[0] != 0) + goto cont; + for(i=1; i d) + return +1; + if(c < d) + return -1; + a++; + cont:; + } + return 0; +} + +static void +divby(char *a, int *na, int b) +{ + int n, c; + char *p; + + p = a; + n = 0; + while(n>>b == 0) { + c = *a++; + if(c == 0) { + while(n) { + c = n*10; + if(c>>b) + break; + n = c; + } + goto xx; + } + n = n*10 + c-'0'; + (*na)--; + } + for(;;) { + c = n>>b; + n -= c<>b; + n -= c<= (int)(nelem(tab1))) + d = (int)(nelem(tab1))-1; + t = tab1 + d; + b = t->bp; + if(memcmp(a, t->cmp, t->siz) > 0) + d--; + *dp -= d; + *bp += b; + divby(a, na, b); +} + +static void +mulby(char *a, char *p, char *q, int b) +{ + int n, c; + + n = 0; + *p = 0; + for(;;) { + q--; + if(q < a) + break; + c = *q - '0'; + c = (c<= (int)(nelem(tab2))) + d = (int)(nelem(tab2))-1; + t = tab2 + d; + b = t->bp; + if(memcmp(a, t->cmp, t->siz) < 0) + d--; + p = a + *na; + *bp -= b; + *dp += d; + *na += d; + mulby(a, p+d, p, b); +} + +static int +xcmp(char *a, char *b) +{ + int c1, c2; + + while(c1 = *b++) { + c2 = *a++; + if(isupper(c2)) + c2 = tolower(c2); + if(c1 != c2) + return 1; + } + return 0; +} diff --git a/src/lib9/fmt/strtod.h b/src/lib9/fmt/strtod.h new file mode 100644 index 00000000..82c3d46e --- /dev/null +++ b/src/lib9/fmt/strtod.h @@ -0,0 +1,4 @@ +extern double __NaN(void); +extern double __Inf(int); +extern double __isNaN(double); +extern double __isInf(double, int); diff --git a/src/lib9/fmt/test.c b/src/lib9/fmt/test.c new file mode 100644 index 00000000..a1a1d5ed --- /dev/null +++ b/src/lib9/fmt/test.c @@ -0,0 +1,39 @@ +/* + * The authors of this software are Rob Pike and Ken Thompson. + * Copyright (c) 2002 by Lucent Technologies. + * Permission to use, copy, modify, and distribute this software for any + * purpose without fee is hereby granted, provided that this entire notice + * is included in all copies of any software which is or includes a copy + * or modification of this software and in all copies of the supporting + * documentation for such software. + * THIS SOFTWARE IS BEING PROVIDED "AS IS", WITHOUT ANY EXPRESS OR IMPLIED + * WARRANTY. IN PARTICULAR, NEITHER THE AUTHORS NOR LUCENT TECHNOLOGIES MAKE ANY + * REPRESENTATION OR WARRANTY OF ANY KIND CONCERNING THE MERCHANTABILITY + * OF THIS SOFTWARE OR ITS FITNESS FOR ANY PARTICULAR PURPOSE. + */ +#include +#include +#include "fmt.h" + +int +main(int argc, char *argv[]) +{ + quotefmtinstall(); + print("hello world\n"); + print("x: %x\n", 0x87654321); + print("u: %u\n", 0x87654321); + print("d: %d\n", 0x87654321); + print("s: %s\n", "hi there"); + print("q: %q\n", "hi i'm here"); + print("c: %c\n", '!'); + print("g: %g %g %g\n", 3.14159, 3.14159e10, 3.14159e-10); + print("e: %e %e %e\n", 3.14159, 3.14159e10, 3.14159e-10); + print("f: %f %f %f\n", 3.14159, 3.14159e10, 3.14159e-10); + print("smiley: %C\n", (Rune)0x263a); + print("%g %.18\n", 2e25, 2e25); + print("%2.18g\n", 1.0); + print("%f\n", 3.1415927/4); + print("%d\n", 23); + print("%i\n", 23); + return 0; +} diff --git a/src/lib9/fmt/vfprint.c b/src/lib9/fmt/vfprint.c new file mode 100644 index 00000000..e4ab82ac --- /dev/null +++ b/src/lib9/fmt/vfprint.c @@ -0,0 +1,31 @@ +/* + * The authors of this software are Rob Pike and Ken Thompson. + * Copyright (c) 2002 by Lucent Technologies. + * Permission to use, copy, modify, and distribute this software for any + * purpose without fee is hereby granted, provided that this entire notice + * is included in all copies of any software which is or includes a copy + * or modification of this software and in all copies of the supporting + * documentation for such software. + * THIS SOFTWARE IS BEING PROVIDED "AS IS", WITHOUT ANY EXPRESS OR IMPLIED + * WARRANTY. IN PARTICULAR, NEITHER THE AUTHORS NOR LUCENT TECHNOLOGIES MAKE ANY + * REPRESENTATION OR WARRANTY OF ANY KIND CONCERNING THE MERCHANTABILITY + * OF THIS SOFTWARE OR ITS FITNESS FOR ANY PARTICULAR PURPOSE. + */ +#include +#include "fmt.h" +#include "fmtdef.h" + +int +vfprint(int fd, char *fmt, va_list args) +{ + Fmt f; + char buf[256]; + int n; + + fmtfdinit(&f, fd, buf, sizeof(buf)); + f.args = args; + n = dofmt(&f, fmt); + if(n > 0 && __fmtFdFlush(&f) == 0) + return -1; + return n; +} diff --git a/src/lib9/fmt/vseprint.c b/src/lib9/fmt/vseprint.c new file mode 100644 index 00000000..85ed810b --- /dev/null +++ b/src/lib9/fmt/vseprint.c @@ -0,0 +1,37 @@ +/* + * The authors of this software are Rob Pike and Ken Thompson. + * Copyright (c) 2002 by Lucent Technologies. + * Permission to use, copy, modify, and distribute this software for any + * purpose without fee is hereby granted, provided that this entire notice + * is included in all copies of any software which is or includes a copy + * or modification of this software and in all copies of the supporting + * documentation for such software. + * THIS SOFTWARE IS BEING PROVIDED "AS IS", WITHOUT ANY EXPRESS OR IMPLIED + * WARRANTY. IN PARTICULAR, NEITHER THE AUTHORS NOR LUCENT TECHNOLOGIES MAKE ANY + * REPRESENTATION OR WARRANTY OF ANY KIND CONCERNING THE MERCHANTABILITY + * OF THIS SOFTWARE OR ITS FITNESS FOR ANY PARTICULAR PURPOSE. + */ +#include +#include "fmt.h" +#include "fmtdef.h" + +char* +vseprint(char *buf, char *e, char *fmt, va_list args) +{ + Fmt f; + + if(e <= buf) + return nil; + f.runes = 0; + f.start = buf; + f.to = buf; + f.stop = e - 1; + f.flush = 0; + f.farg = nil; + f.nfmt = 0; + f.args = args; + dofmt(&f, fmt); + *(char*)f.to = '\0'; + return (char*)f.to; +} + diff --git a/src/lib9/fmt/vsmprint.c b/src/lib9/fmt/vsmprint.c new file mode 100644 index 00000000..38ace62d --- /dev/null +++ b/src/lib9/fmt/vsmprint.c @@ -0,0 +1,36 @@ +/* + * The authors of this software are Rob Pike and Ken Thompson. + * Copyright (c) 2002 by Lucent Technologies. + * Permission to use, copy, modify, and distribute this software for any + * purpose without fee is hereby granted, provided that this entire notice + * is included in all copies of any software which is or includes a copy + * or modification of this software and in all copies of the supporting + * documentation for such software. + * THIS SOFTWARE IS BEING PROVIDED "AS IS", WITHOUT ANY EXPRESS OR IMPLIED + * WARRANTY. IN PARTICULAR, NEITHER THE AUTHORS NOR LUCENT TECHNOLOGIES MAKE ANY + * REPRESENTATION OR WARRANTY OF ANY KIND CONCERNING THE MERCHANTABILITY + * OF THIS SOFTWARE OR ITS FITNESS FOR ANY PARTICULAR PURPOSE. + */ +#include +#include +#include "fmt.h" +#include "fmtdef.h" + +/* + * print into an allocated string buffer + */ +char* +vsmprint(char *fmt, va_list args) +{ + Fmt f; + int n; + + if(fmtstrinit(&f) < 0) + return nil; + f.args = args; + n = dofmt(&f, fmt); + if(n < 0) + return nil; + *(char*)f.to = '\0'; + return (char*)f.start; +} diff --git a/src/lib9/fmt/vsnprint.c b/src/lib9/fmt/vsnprint.c new file mode 100644 index 00000000..21662e6d --- /dev/null +++ b/src/lib9/fmt/vsnprint.c @@ -0,0 +1,37 @@ +/* + * The authors of this software are Rob Pike and Ken Thompson. + * Copyright (c) 2002 by Lucent Technologies. + * Permission to use, copy, modify, and distribute this software for any + * purpose without fee is hereby granted, provided that this entire notice + * is included in all copies of any software which is or includes a copy + * or modification of this software and in all copies of the supporting + * documentation for such software. + * THIS SOFTWARE IS BEING PROVIDED "AS IS", WITHOUT ANY EXPRESS OR IMPLIED + * WARRANTY. IN PARTICULAR, NEITHER THE AUTHORS NOR LUCENT TECHNOLOGIES MAKE ANY + * REPRESENTATION OR WARRANTY OF ANY KIND CONCERNING THE MERCHANTABILITY + * OF THIS SOFTWARE OR ITS FITNESS FOR ANY PARTICULAR PURPOSE. + */ +#include +#include +#include "fmt.h" +#include "fmtdef.h" + +int +vsnprint(char *buf, int len, char *fmt, va_list args) +{ + Fmt f; + + if(len <= 0) + return -1; + f.runes = 0; + f.start = buf; + f.to = buf; + f.stop = buf + len - 1; + f.flush = 0; + f.farg = nil; + f.nfmt = 0; + f.args = args; + dofmt(&f, fmt); + *(char*)f.to = '\0'; + return (char*)f.to - buf; +} diff --git a/src/lib9/utf/LICENSE b/src/lib9/utf/LICENSE new file mode 100644 index 00000000..ad76cd52 --- /dev/null +++ b/src/lib9/utf/LICENSE @@ -0,0 +1,13 @@ +/* + * The authors of this software are Rob Pike and Ken Thompson. + * Copyright (c) 1998-2002 by Lucent Technologies. + * Permission to use, copy, modify, and distribute this software for any + * purpose without fee is hereby granted, provided that this entire notice + * is included in all copies of any software which is or includes a copy + * or modification of this software and in all copies of the supporting + * documentation for such software. + * THIS SOFTWARE IS BEING PROVIDED "AS IS", WITHOUT ANY EXPRESS OR IMPLIED + * WARRANTY. IN PARTICULAR, NEITHER THE AUTHORS NOR LUCENT TECHNOLOGIES MAKE ANY + * REPRESENTATION OR WARRANTY OF ANY KIND CONCERNING THE MERCHANTABILITY + * OF THIS SOFTWARE OR ITS FITNESS FOR ANY PARTICULAR PURPOSE. + */ diff --git a/src/lib9/utf/NOTICE b/src/lib9/utf/NOTICE new file mode 100644 index 00000000..ad76cd52 --- /dev/null +++ b/src/lib9/utf/NOTICE @@ -0,0 +1,13 @@ +/* + * The authors of this software are Rob Pike and Ken Thompson. + * Copyright (c) 1998-2002 by Lucent Technologies. + * Permission to use, copy, modify, and distribute this software for any + * purpose without fee is hereby granted, provided that this entire notice + * is included in all copies of any software which is or includes a copy + * or modification of this software and in all copies of the supporting + * documentation for such software. + * THIS SOFTWARE IS BEING PROVIDED "AS IS", WITHOUT ANY EXPRESS OR IMPLIED + * WARRANTY. IN PARTICULAR, NEITHER THE AUTHORS NOR LUCENT TECHNOLOGIES MAKE ANY + * REPRESENTATION OR WARRANTY OF ANY KIND CONCERNING THE MERCHANTABILITY + * OF THIS SOFTWARE OR ITS FITNESS FOR ANY PARTICULAR PURPOSE. + */ diff --git a/src/lib9/utf/README b/src/lib9/utf/README new file mode 100644 index 00000000..ad76cd52 --- /dev/null +++ b/src/lib9/utf/README @@ -0,0 +1,13 @@ +/* + * The authors of this software are Rob Pike and Ken Thompson. + * Copyright (c) 1998-2002 by Lucent Technologies. + * Permission to use, copy, modify, and distribute this software for any + * purpose without fee is hereby granted, provided that this entire notice + * is included in all copies of any software which is or includes a copy + * or modification of this software and in all copies of the supporting + * documentation for such software. + * THIS SOFTWARE IS BEING PROVIDED "AS IS", WITHOUT ANY EXPRESS OR IMPLIED + * WARRANTY. IN PARTICULAR, NEITHER THE AUTHORS NOR LUCENT TECHNOLOGIES MAKE ANY + * REPRESENTATION OR WARRANTY OF ANY KIND CONCERNING THE MERCHANTABILITY + * OF THIS SOFTWARE OR ITS FITNESS FOR ANY PARTICULAR PURPOSE. + */ diff --git a/src/lib9/utf/lib9.h b/src/lib9/utf/lib9.h new file mode 100644 index 00000000..e6128ae4 --- /dev/null +++ b/src/lib9/utf/lib9.h @@ -0,0 +1,17 @@ +#include +#include "utf.h" + +#define nil ((void*)0) + +#define uchar _fmtuchar +#define ushort _fmtushort +#define uint _fmtuint +#define ulong _fmtulong +#define vlong _fmtvlong +#define uvlong _fmtuvlong + +typedef unsigned char uchar; +typedef unsigned short ushort; +typedef unsigned int uint; +typedef unsigned long ulong; + diff --git a/src/lib9/utf/mkfile b/src/lib9/utf/mkfile new file mode 100644 index 00000000..46c92fef --- /dev/null +++ b/src/lib9/utf/mkfile @@ -0,0 +1,31 @@ +PLAN9=../.. +<$PLAN9/src/mkhdr + +LIB=libutf.a + +OFILES=\ + rune.$O\ + runestrcat.$O\ + runestrchr.$O\ + runestrcmp.$O\ + runestrcpy.$O\ + runestrdup.$O\ + runestrlen.$O\ + runestrecpy.$O\ + runestrncat.$O\ + runestrncmp.$O\ + runestrncpy.$O\ + runestrrchr.$O\ + runestrstr.$O\ + runetype.$O\ + utfecpy.$O\ + utflen.$O\ + utfnlen.$O\ + utfrrune.$O\ + utfrune.$O\ + utfutf.$O\ + +HFILES=\ + $PLAN9/include/utf.h\ + +<$PLAN9/src/mksyslib diff --git a/src/lib9/utf/rune.c b/src/lib9/utf/rune.c new file mode 100644 index 00000000..e1aaa9be --- /dev/null +++ b/src/lib9/utf/rune.c @@ -0,0 +1,177 @@ +/* + * The authors of this software are Rob Pike and Ken Thompson. + * Copyright (c) 2002 by Lucent Technologies. + * Permission to use, copy, modify, and distribute this software for any + * purpose without fee is hereby granted, provided that this entire notice + * is included in all copies of any software which is or includes a copy + * or modification of this software and in all copies of the supporting + * documentation for such software. + * THIS SOFTWARE IS BEING PROVIDED "AS IS", WITHOUT ANY EXPRESS OR IMPLIED + * WARRANTY. IN PARTICULAR, NEITHER THE AUTHORS NOR LUCENT TECHNOLOGIES MAKE ANY + * REPRESENTATION OR WARRANTY OF ANY KIND CONCERNING THE MERCHANTABILITY + * OF THIS SOFTWARE OR ITS FITNESS FOR ANY PARTICULAR PURPOSE. + */ +#include +#include +#include "utf.h" +#include "utfdef.h" + +enum +{ + Bit1 = 7, + Bitx = 6, + Bit2 = 5, + Bit3 = 4, + Bit4 = 3, + + T1 = ((1<<(Bit1+1))-1) ^ 0xFF, /* 0000 0000 */ + Tx = ((1<<(Bitx+1))-1) ^ 0xFF, /* 1000 0000 */ + T2 = ((1<<(Bit2+1))-1) ^ 0xFF, /* 1100 0000 */ + T3 = ((1<<(Bit3+1))-1) ^ 0xFF, /* 1110 0000 */ + T4 = ((1<<(Bit4+1))-1) ^ 0xFF, /* 1111 0000 */ + + Rune1 = (1<<(Bit1+0*Bitx))-1, /* 0000 0000 0111 1111 */ + Rune2 = (1<<(Bit2+1*Bitx))-1, /* 0000 0111 1111 1111 */ + Rune3 = (1<<(Bit3+2*Bitx))-1, /* 1111 1111 1111 1111 */ + + Maskx = (1< T1 + */ + c = *(uchar*)str; + if(c < Tx) { + *rune = c; + return 1; + } + + /* + * two character sequence + * 0080-07FF => T2 Tx + */ + c1 = *(uchar*)(str+1) ^ Tx; + if(c1 & Testx) + goto bad; + if(c < T3) { + if(c < T2) + goto bad; + l = ((c << Bitx) | c1) & Rune2; + if(l <= Rune1) + goto bad; + *rune = l; + return 2; + } + + /* + * three character sequence + * 0800-FFFF => T3 Tx Tx + */ + c2 = *(uchar*)(str+2) ^ Tx; + if(c2 & Testx) + goto bad; + if(c < T4) { + l = ((((c << Bitx) | c1) << Bitx) | c2) & Rune3; + if(l <= Rune2) + goto bad; + *rune = l; + return 3; + } + + /* + * bad decoding + */ +bad: + *rune = Bad; + return 1; +} + +int +runetochar(char *str, Rune *rune) +{ + long c; + + /* + * one character sequence + * 00000-0007F => 00-7F + */ + c = *rune; + if(c <= Rune1) { + str[0] = c; + return 1; + } + + /* + * two character sequence + * 0080-07FF => T2 Tx + */ + if(c <= Rune2) { + str[0] = T2 | (c >> 1*Bitx); + str[1] = Tx | (c & Maskx); + return 2; + } + + /* + * three character sequence + * 0800-FFFF => T3 Tx Tx + */ + str[0] = T3 | (c >> 2*Bitx); + str[1] = Tx | ((c >> 1*Bitx) & Maskx); + str[2] = Tx | (c & Maskx); + return 3; +} + +int +runelen(long c) +{ + Rune rune; + char str[10]; + + rune = c; + return runetochar(str, &rune); +} + +int +runenlen(Rune *r, int nrune) +{ + int nb, c; + + nb = 0; + while(nrune--) { + c = *r++; + if(c <= Rune1) + nb++; + else + if(c <= Rune2) + nb += 2; + else + nb += 3; + } + return nb; +} + +int +fullrune(char *str, int n) +{ + int c; + + if(n > 0) { + c = *(uchar*)str; + if(c < Tx) + return 1; + if(n > 1) + if(c < T3 || n > 2) + return 1; + } + return 0; +} diff --git a/src/lib9/utf/runestrcat.c b/src/lib9/utf/runestrcat.c new file mode 100644 index 00000000..e1e8c9d5 --- /dev/null +++ b/src/lib9/utf/runestrcat.c @@ -0,0 +1,25 @@ +/* + * The authors of this software are Rob Pike and Ken Thompson. + * Copyright (c) 2002 by Lucent Technologies. + * Permission to use, copy, modify, and distribute this software for any + * purpose without fee is hereby granted, provided that this entire notice + * is included in all copies of any software which is or includes a copy + * or modification of this software and in all copies of the supporting + * documentation for such software. + * THIS SOFTWARE IS BEING PROVIDED "AS IS", WITHOUT ANY EXPRESS OR IMPLIED + * WARRANTY. IN PARTICULAR, NEITHER THE AUTHORS NOR LUCENT TECHNOLOGIES MAKE ANY + * REPRESENTATION OR WARRANTY OF ANY KIND CONCERNING THE MERCHANTABILITY + * OF THIS SOFTWARE OR ITS FITNESS FOR ANY PARTICULAR PURPOSE. + */ +#include +#include +#include "utf.h" +#include "utfdef.h" + +Rune* +runestrcat(Rune *s1, Rune *s2) +{ + + runestrcpy(runestrchr(s1, 0), s2); + return s1; +} diff --git a/src/lib9/utf/runestrchr.c b/src/lib9/utf/runestrchr.c new file mode 100644 index 00000000..45ee673a --- /dev/null +++ b/src/lib9/utf/runestrchr.c @@ -0,0 +1,35 @@ +/* + * The authors of this software are Rob Pike and Ken Thompson. + * Copyright (c) 2002 by Lucent Technologies. + * Permission to use, copy, modify, and distribute this software for any + * purpose without fee is hereby granted, provided that this entire notice + * is included in all copies of any software which is or includes a copy + * or modification of this software and in all copies of the supporting + * documentation for such software. + * THIS SOFTWARE IS BEING PROVIDED "AS IS", WITHOUT ANY EXPRESS OR IMPLIED + * WARRANTY. IN PARTICULAR, NEITHER THE AUTHORS NOR LUCENT TECHNOLOGIES MAKE ANY + * REPRESENTATION OR WARRANTY OF ANY KIND CONCERNING THE MERCHANTABILITY + * OF THIS SOFTWARE OR ITS FITNESS FOR ANY PARTICULAR PURPOSE. + */ +#include +#include +#include "utf.h" +#include "utfdef.h" + +Rune* +runestrchr(Rune *s, Rune c) +{ + Rune c0 = c; + Rune c1; + + if(c == 0) { + while(*s++) + ; + return s-1; + } + + while(c1 = *s++) + if(c1 == c0) + return s-1; + return 0; +} diff --git a/src/lib9/utf/runestrcmp.c b/src/lib9/utf/runestrcmp.c new file mode 100644 index 00000000..d14145f3 --- /dev/null +++ b/src/lib9/utf/runestrcmp.c @@ -0,0 +1,35 @@ +/* + * The authors of this software are Rob Pike and Ken Thompson. + * Copyright (c) 2002 by Lucent Technologies. + * Permission to use, copy, modify, and distribute this software for any + * purpose without fee is hereby granted, provided that this entire notice + * is included in all copies of any software which is or includes a copy + * or modification of this software and in all copies of the supporting + * documentation for such software. + * THIS SOFTWARE IS BEING PROVIDED "AS IS", WITHOUT ANY EXPRESS OR IMPLIED + * WARRANTY. IN PARTICULAR, NEITHER THE AUTHORS NOR LUCENT TECHNOLOGIES MAKE ANY + * REPRESENTATION OR WARRANTY OF ANY KIND CONCERNING THE MERCHANTABILITY + * OF THIS SOFTWARE OR ITS FITNESS FOR ANY PARTICULAR PURPOSE. + */ +#include +#include +#include "utf.h" +#include "utfdef.h" + +int +runestrcmp(Rune *s1, Rune *s2) +{ + Rune c1, c2; + + for(;;) { + c1 = *s1++; + c2 = *s2++; + if(c1 != c2) { + if(c1 > c2) + return 1; + return -1; + } + if(c1 == 0) + return 0; + } +} diff --git a/src/lib9/utf/runestrcpy.c b/src/lib9/utf/runestrcpy.c new file mode 100644 index 00000000..9d7aa026 --- /dev/null +++ b/src/lib9/utf/runestrcpy.c @@ -0,0 +1,28 @@ +/* + * The authors of this software are Rob Pike and Ken Thompson. + * Copyright (c) 2002 by Lucent Technologies. + * Permission to use, copy, modify, and distribute this software for any + * purpose without fee is hereby granted, provided that this entire notice + * is included in all copies of any software which is or includes a copy + * or modification of this software and in all copies of the supporting + * documentation for such software. + * THIS SOFTWARE IS BEING PROVIDED "AS IS", WITHOUT ANY EXPRESS OR IMPLIED + * WARRANTY. IN PARTICULAR, NEITHER THE AUTHORS NOR LUCENT TECHNOLOGIES MAKE ANY + * REPRESENTATION OR WARRANTY OF ANY KIND CONCERNING THE MERCHANTABILITY + * OF THIS SOFTWARE OR ITS FITNESS FOR ANY PARTICULAR PURPOSE. + */ +#include +#include +#include "utf.h" +#include "utfdef.h" + +Rune* +runestrcpy(Rune *s1, Rune *s2) +{ + Rune *os1; + + os1 = s1; + while(*s1++ = *s2++) + ; + return os1; +} diff --git a/src/lib9/utf/runestrdup.c b/src/lib9/utf/runestrdup.c new file mode 100644 index 00000000..97a889c8 --- /dev/null +++ b/src/lib9/utf/runestrdup.c @@ -0,0 +1,30 @@ +/* + * The authors of this software are Rob Pike and Ken Thompson. + * Copyright (c) 2002 by Lucent Technologies. + * Permission to use, copy, modify, and distribute this software for any + * purpose without fee is hereby granted, provided that this entire notice + * is included in all copies of any software which is or includes a copy + * or modification of this software and in all copies of the supporting + * documentation for such software. + * THIS SOFTWARE IS BEING PROVIDED "AS IS", WITHOUT ANY EXPRESS OR IMPLIED + * WARRANTY. IN PARTICULAR, NEITHER THE AUTHORS NOR LUCENT TECHNOLOGIES MAKE ANY + * REPRESENTATION OR WARRANTY OF ANY KIND CONCERNING THE MERCHANTABILITY + * OF THIS SOFTWARE OR ITS FITNESS FOR ANY PARTICULAR PURPOSE. + */ +#include +#include +#include +#include "utf.h" +#include "utfdef.h" + +Rune* +runestrdup(Rune *s) +{ + Rune *ns; + + ns = malloc(sizeof(Rune)*(runestrlen(s) + 1)); + if(ns == 0) + return 0; + + return runestrcpy(ns, s); +} diff --git a/src/lib9/utf/runestrecpy.c b/src/lib9/utf/runestrecpy.c new file mode 100644 index 00000000..f3e4a096 --- /dev/null +++ b/src/lib9/utf/runestrecpy.c @@ -0,0 +1,32 @@ +/* + * The authors of this software are Rob Pike and Ken Thompson. + * Copyright (c) 2002 by Lucent Technologies. + * Permission to use, copy, modify, and distribute this software for any + * purpose without fee is hereby granted, provided that this entire notice + * is included in all copies of any software which is or includes a copy + * or modification of this software and in all copies of the supporting + * documentation for such software. + * THIS SOFTWARE IS BEING PROVIDED "AS IS", WITHOUT ANY EXPRESS OR IMPLIED + * WARRANTY. IN PARTICULAR, NEITHER THE AUTHORS NOR LUCENT TECHNOLOGIES MAKE ANY + * REPRESENTATION OR WARRANTY OF ANY KIND CONCERNING THE MERCHANTABILITY + * OF THIS SOFTWARE OR ITS FITNESS FOR ANY PARTICULAR PURPOSE. + */ +#include +#include +#include "utf.h" +#include "utfdef.h" + +Rune* +runestrecpy(Rune *s1, Rune *es1, Rune *s2) +{ + if(s1 >= es1) + return s1; + + while(*s1++ = *s2++){ + if(s1 == es1){ + *--s1 = '\0'; + break; + } + } + return s1; +} diff --git a/src/lib9/utf/runestrlen.c b/src/lib9/utf/runestrlen.c new file mode 100644 index 00000000..bca3050c --- /dev/null +++ b/src/lib9/utf/runestrlen.c @@ -0,0 +1,24 @@ +/* + * The authors of this software are Rob Pike and Ken Thompson. + * Copyright (c) 2002 by Lucent Technologies. + * Permission to use, copy, modify, and distribute this software for any + * purpose without fee is hereby granted, provided that this entire notice + * is included in all copies of any software which is or includes a copy + * or modification of this software and in all copies of the supporting + * documentation for such software. + * THIS SOFTWARE IS BEING PROVIDED "AS IS", WITHOUT ANY EXPRESS OR IMPLIED + * WARRANTY. IN PARTICULAR, NEITHER THE AUTHORS NOR LUCENT TECHNOLOGIES MAKE ANY + * REPRESENTATION OR WARRANTY OF ANY KIND CONCERNING THE MERCHANTABILITY + * OF THIS SOFTWARE OR ITS FITNESS FOR ANY PARTICULAR PURPOSE. + */ +#include +#include +#include "utf.h" +#include "utfdef.h" + +long +runestrlen(Rune *s) +{ + + return runestrchr(s, 0) - s; +} diff --git a/src/lib9/utf/runestrncat.c b/src/lib9/utf/runestrncat.c new file mode 100644 index 00000000..74400c30 --- /dev/null +++ b/src/lib9/utf/runestrncat.c @@ -0,0 +1,32 @@ +/* + * The authors of this software are Rob Pike and Ken Thompson. + * Copyright (c) 2002 by Lucent Technologies. + * Permission to use, copy, modify, and distribute this software for any + * purpose without fee is hereby granted, provided that this entire notice + * is included in all copies of any software which is or includes a copy + * or modification of this software and in all copies of the supporting + * documentation for such software. + * THIS SOFTWARE IS BEING PROVIDED "AS IS", WITHOUT ANY EXPRESS OR IMPLIED + * WARRANTY. IN PARTICULAR, NEITHER THE AUTHORS NOR LUCENT TECHNOLOGIES MAKE ANY + * REPRESENTATION OR WARRANTY OF ANY KIND CONCERNING THE MERCHANTABILITY + * OF THIS SOFTWARE OR ITS FITNESS FOR ANY PARTICULAR PURPOSE. + */ +#include +#include +#include "utf.h" +#include "utfdef.h" + +Rune* +runestrncat(Rune *s1, Rune *s2, long n) +{ + Rune *os1; + + os1 = s1; + s1 = runestrchr(s1, 0); + while(*s1++ = *s2++) + if(--n < 0) { + s1[-1] = 0; + break; + } + return os1; +} diff --git a/src/lib9/utf/runestrncmp.c b/src/lib9/utf/runestrncmp.c new file mode 100644 index 00000000..dbd9da45 --- /dev/null +++ b/src/lib9/utf/runestrncmp.c @@ -0,0 +1,37 @@ +/* + * The authors of this software are Rob Pike and Ken Thompson. + * Copyright (c) 2002 by Lucent Technologies. + * Permission to use, copy, modify, and distribute this software for any + * purpose without fee is hereby granted, provided that this entire notice + * is included in all copies of any software which is or includes a copy + * or modification of this software and in all copies of the supporting + * documentation for such software. + * THIS SOFTWARE IS BEING PROVIDED "AS IS", WITHOUT ANY EXPRESS OR IMPLIED + * WARRANTY. IN PARTICULAR, NEITHER THE AUTHORS NOR LUCENT TECHNOLOGIES MAKE ANY + * REPRESENTATION OR WARRANTY OF ANY KIND CONCERNING THE MERCHANTABILITY + * OF THIS SOFTWARE OR ITS FITNESS FOR ANY PARTICULAR PURPOSE. + */ +#include +#include +#include "utf.h" +#include "utfdef.h" + +int +runestrncmp(Rune *s1, Rune *s2, long n) +{ + Rune c1, c2; + + while(n > 0) { + c1 = *s1++; + c2 = *s2++; + n--; + if(c1 != c2) { + if(c1 > c2) + return 1; + return -1; + } + if(c1 == 0) + break; + } + return 0; +} diff --git a/src/lib9/utf/runestrncpy.c b/src/lib9/utf/runestrncpy.c new file mode 100644 index 00000000..acba7bc2 --- /dev/null +++ b/src/lib9/utf/runestrncpy.c @@ -0,0 +1,33 @@ +/* + * The authors of this software are Rob Pike and Ken Thompson. + * Copyright (c) 2002 by Lucent Technologies. + * Permission to use, copy, modify, and distribute this software for any + * purpose without fee is hereby granted, provided that this entire notice + * is included in all copies of any software which is or includes a copy + * or modification of this software and in all copies of the supporting + * documentation for such software. + * THIS SOFTWARE IS BEING PROVIDED "AS IS", WITHOUT ANY EXPRESS OR IMPLIED + * WARRANTY. IN PARTICULAR, NEITHER THE AUTHORS NOR LUCENT TECHNOLOGIES MAKE ANY + * REPRESENTATION OR WARRANTY OF ANY KIND CONCERNING THE MERCHANTABILITY + * OF THIS SOFTWARE OR ITS FITNESS FOR ANY PARTICULAR PURPOSE. + */ +#include +#include +#include "utf.h" +#include "utfdef.h" + +Rune* +runestrncpy(Rune *s1, Rune *s2, long n) +{ + int i; + Rune *os1; + + os1 = s1; + for(i = 0; i < n; i++) + if((*s1++ = *s2++) == 0) { + while(++i < n) + *s1++ = 0; + return os1; + } + return os1; +} diff --git a/src/lib9/utf/runestrrchr.c b/src/lib9/utf/runestrrchr.c new file mode 100644 index 00000000..1f40c5a6 --- /dev/null +++ b/src/lib9/utf/runestrrchr.c @@ -0,0 +1,30 @@ +/* + * The authors of this software are Rob Pike and Ken Thompson. + * Copyright (c) 2002 by Lucent Technologies. + * Permission to use, copy, modify, and distribute this software for any + * purpose without fee is hereby granted, provided that this entire notice + * is included in all copies of any software which is or includes a copy + * or modification of this software and in all copies of the supporting + * documentation for such software. + * THIS SOFTWARE IS BEING PROVIDED "AS IS", WITHOUT ANY EXPRESS OR IMPLIED + * WARRANTY. IN PARTICULAR, NEITHER THE AUTHORS NOR LUCENT TECHNOLOGIES MAKE ANY + * REPRESENTATION OR WARRANTY OF ANY KIND CONCERNING THE MERCHANTABILITY + * OF THIS SOFTWARE OR ITS FITNESS FOR ANY PARTICULAR PURPOSE. + */ +#include +#include +#include "utf.h" +#include "utfdef.h" + +Rune* +runestrrchr(Rune *s, Rune c) +{ + Rune *r; + + if(c == 0) + return runestrchr(s, 0); + r = 0; + while(s = runestrchr(s, c)) + r = s++; + return r; +} diff --git a/src/lib9/utf/runestrstr.c b/src/lib9/utf/runestrstr.c new file mode 100644 index 00000000..48583a3d --- /dev/null +++ b/src/lib9/utf/runestrstr.c @@ -0,0 +1,44 @@ +/* + * The authors of this software are Rob Pike and Ken Thompson. + * Copyright (c) 2002 by Lucent Technologies. + * Permission to use, copy, modify, and distribute this software for any + * purpose without fee is hereby granted, provided that this entire notice + * is included in all copies of any software which is or includes a copy + * or modification of this software and in all copies of the supporting + * documentation for such software. + * THIS SOFTWARE IS BEING PROVIDED "AS IS", WITHOUT ANY EXPRESS OR IMPLIED + * WARRANTY. IN PARTICULAR, NEITHER THE AUTHORS NOR LUCENT TECHNOLOGIES MAKE ANY + * REPRESENTATION OR WARRANTY OF ANY KIND CONCERNING THE MERCHANTABILITY + * OF THIS SOFTWARE OR ITS FITNESS FOR ANY PARTICULAR PURPOSE. + */ +#include +#include +#include "utf.h" +#include "utfdef.h" + +/* + * Return pointer to first occurrence of s2 in s1, + * 0 if none + */ +Rune* +runestrstr(Rune *s1, Rune *s2) +{ + Rune *p, *pa, *pb; + int c0, c; + + c0 = *s2; + if(c0 == 0) + return s1; + s2++; + for(p=runestrchr(s1, c0); p; p=runestrchr(p+1, c0)) { + pa = p; + for(pb=s2;; pb++) { + c = *pb; + if(c == 0) + return p; + if(c != *++pa) + break; + } + } + return 0; +} diff --git a/src/lib9/utf/runetype.c b/src/lib9/utf/runetype.c new file mode 100644 index 00000000..ffae30e6 --- /dev/null +++ b/src/lib9/utf/runetype.c @@ -0,0 +1,1152 @@ +/* + * The authors of this software are Rob Pike and Ken Thompson. + * Copyright (c) 2002 by Lucent Technologies. + * Permission to use, copy, modify, and distribute this software for any + * purpose without fee is hereby granted, provided that this entire notice + * is included in all copies of any software which is or includes a copy + * or modification of this software and in all copies of the supporting + * documentation for such software. + * THIS SOFTWARE IS BEING PROVIDED "AS IS", WITHOUT ANY EXPRESS OR IMPLIED + * WARRANTY. IN PARTICULAR, NEITHER THE AUTHORS NOR LUCENT TECHNOLOGIES MAKE ANY + * REPRESENTATION OR WARRANTY OF ANY KIND CONCERNING THE MERCHANTABILITY + * OF THIS SOFTWARE OR ITS FITNESS FOR ANY PARTICULAR PURPOSE. + */ +#include +#include +#include "utf.h" +#include "utfdef.h" + +/* + * alpha ranges - + * only covers ranges not in lower||upper + */ +static +Rune __alpha2[] = +{ + 0x00d8, 0x00f6, /* Ø - ö */ + 0x00f8, 0x01f5, /* ø - ǵ */ + 0x0250, 0x02a8, /* ɐ - ʨ */ + 0x038e, 0x03a1, /* Ύ - Ρ */ + 0x03a3, 0x03ce, /* Σ - ώ */ + 0x03d0, 0x03d6, /* ϐ - ϖ */ + 0x03e2, 0x03f3, /* Ϣ - ϳ */ + 0x0490, 0x04c4, /* Ґ - ӄ */ + 0x0561, 0x0587, /* ա - և */ + 0x05d0, 0x05ea, /* א - ת */ + 0x05f0, 0x05f2, /* װ - ײ */ + 0x0621, 0x063a, /* ء - غ */ + 0x0640, 0x064a, /* ـ - ي */ + 0x0671, 0x06b7, /* ٱ - ڷ */ + 0x06ba, 0x06be, /* ں - ھ */ + 0x06c0, 0x06ce, /* ۀ - ێ */ + 0x06d0, 0x06d3, /* ې - ۓ */ + 0x0905, 0x0939, /* अ - ह */ + 0x0958, 0x0961, /* क़ - ॡ */ + 0x0985, 0x098c, /* অ - ঌ */ + 0x098f, 0x0990, /* এ - ঐ */ + 0x0993, 0x09a8, /* ও - ন */ + 0x09aa, 0x09b0, /* প - র */ + 0x09b6, 0x09b9, /* শ - হ */ + 0x09dc, 0x09dd, /* ড় - ঢ় */ + 0x09df, 0x09e1, /* য় - ৡ */ + 0x09f0, 0x09f1, /* ৰ - ৱ */ + 0x0a05, 0x0a0a, /* ਅ - ਊ */ + 0x0a0f, 0x0a10, /* ਏ - ਐ */ + 0x0a13, 0x0a28, /* ਓ - ਨ */ + 0x0a2a, 0x0a30, /* ਪ - ਰ */ + 0x0a32, 0x0a33, /* ਲ - ਲ਼ */ + 0x0a35, 0x0a36, /* ਵ - ਸ਼ */ + 0x0a38, 0x0a39, /* ਸ - ਹ */ + 0x0a59, 0x0a5c, /* ਖ਼ - ੜ */ + 0x0a85, 0x0a8b, /* અ - ઋ */ + 0x0a8f, 0x0a91, /* એ - ઑ */ + 0x0a93, 0x0aa8, /* ઓ - ન */ + 0x0aaa, 0x0ab0, /* પ - ર */ + 0x0ab2, 0x0ab3, /* લ - ળ */ + 0x0ab5, 0x0ab9, /* વ - હ */ + 0x0b05, 0x0b0c, /* ଅ - ଌ */ + 0x0b0f, 0x0b10, /* ଏ - ଐ */ + 0x0b13, 0x0b28, /* ଓ - ନ */ + 0x0b2a, 0x0b30, /* ପ - ର */ + 0x0b32, 0x0b33, /* ଲ - ଳ */ + 0x0b36, 0x0b39, /* ଶ - ହ */ + 0x0b5c, 0x0b5d, /* ଡ଼ - ଢ଼ */ + 0x0b5f, 0x0b61, /* ୟ - ୡ */ + 0x0b85, 0x0b8a, /* அ - ஊ */ + 0x0b8e, 0x0b90, /* எ - ஐ */ + 0x0b92, 0x0b95, /* ஒ - க */ + 0x0b99, 0x0b9a, /* ங - ச */ + 0x0b9e, 0x0b9f, /* ஞ - ட */ + 0x0ba3, 0x0ba4, /* ண - த */ + 0x0ba8, 0x0baa, /* ந - ப */ + 0x0bae, 0x0bb5, /* ம - வ */ + 0x0bb7, 0x0bb9, /* ஷ - ஹ */ + 0x0c05, 0x0c0c, /* అ - ఌ */ + 0x0c0e, 0x0c10, /* ఎ - ఐ */ + 0x0c12, 0x0c28, /* ఒ - న */ + 0x0c2a, 0x0c33, /* ప - ళ */ + 0x0c35, 0x0c39, /* వ - హ */ + 0x0c60, 0x0c61, /* ౠ - ౡ */ + 0x0c85, 0x0c8c, /* ಅ - ಌ */ + 0x0c8e, 0x0c90, /* ಎ - ಐ */ + 0x0c92, 0x0ca8, /* ಒ - ನ */ + 0x0caa, 0x0cb3, /* ಪ - ಳ */ + 0x0cb5, 0x0cb9, /* ವ - ಹ */ + 0x0ce0, 0x0ce1, /* ೠ - ೡ */ + 0x0d05, 0x0d0c, /* അ - ഌ */ + 0x0d0e, 0x0d10, /* എ - ഐ */ + 0x0d12, 0x0d28, /* ഒ - ന */ + 0x0d2a, 0x0d39, /* പ - ഹ */ + 0x0d60, 0x0d61, /* ൠ - ൡ */ + 0x0e01, 0x0e30, /* ก - ะ */ + 0x0e32, 0x0e33, /* า - ำ */ + 0x0e40, 0x0e46, /* เ - ๆ */ + 0x0e5a, 0x0e5b, /* ๚ - ๛ */ + 0x0e81, 0x0e82, /* ກ - ຂ */ + 0x0e87, 0x0e88, /* ງ - ຈ */ + 0x0e94, 0x0e97, /* ດ - ທ */ + 0x0e99, 0x0e9f, /* ນ - ຟ */ + 0x0ea1, 0x0ea3, /* ມ - ຣ */ + 0x0eaa, 0x0eab, /* ສ - ຫ */ + 0x0ead, 0x0eae, /* ອ - ຮ */ + 0x0eb2, 0x0eb3, /* າ - ຳ */ + 0x0ec0, 0x0ec4, /* ເ - ໄ */ + 0x0edc, 0x0edd, /* ໜ - ໝ */ + 0x0f18, 0x0f19, /* ༘ - ༙ */ + 0x0f40, 0x0f47, /* ཀ - ཇ */ + 0x0f49, 0x0f69, /* ཉ - ཀྵ */ + 0x10d0, 0x10f6, /* ა - ჶ */ + 0x1100, 0x1159, /* ᄀ - ᅙ */ + 0x115f, 0x11a2, /* ᅟ - ᆢ */ + 0x11a8, 0x11f9, /* ᆨ - ᇹ */ + 0x1e00, 0x1e9b, /* Ḁ - ẛ */ + 0x1f50, 0x1f57, /* ὐ - ὗ */ + 0x1f80, 0x1fb4, /* ᾀ - ᾴ */ + 0x1fb6, 0x1fbc, /* ᾶ - ᾼ */ + 0x1fc2, 0x1fc4, /* ῂ - ῄ */ + 0x1fc6, 0x1fcc, /* ῆ - ῌ */ + 0x1fd0, 0x1fd3, /* ῐ - ΐ */ + 0x1fd6, 0x1fdb, /* ῖ - Ί */ + 0x1fe0, 0x1fec, /* ῠ - Ῥ */ + 0x1ff2, 0x1ff4, /* ῲ - ῴ */ + 0x1ff6, 0x1ffc, /* ῶ - ῼ */ + 0x210a, 0x2113, /* ℊ - ℓ */ + 0x2115, 0x211d, /* ℕ - ℝ */ + 0x2120, 0x2122, /* ℠ - ™ */ + 0x212a, 0x2131, /* K - ℱ */ + 0x2133, 0x2138, /* ℳ - ℸ */ + 0x3041, 0x3094, /* ぁ - ゔ */ + 0x30a1, 0x30fa, /* ァ - ヺ */ + 0x3105, 0x312c, /* ㄅ - ㄬ */ + 0x3131, 0x318e, /* ㄱ - ㆎ */ + 0x3192, 0x319f, /* ㆒ - ㆟ */ + 0x3260, 0x327b, /* ㉠ - ㉻ */ + 0x328a, 0x32b0, /* ㊊ - ㊰ */ + 0x32d0, 0x32fe, /* ㋐ - ㋾ */ + 0x3300, 0x3357, /* ㌀ - ㍗ */ + 0x3371, 0x3376, /* ㍱ - ㍶ */ + 0x337b, 0x3394, /* ㍻ - ㎔ */ + 0x3399, 0x339e, /* ㎙ - ㎞ */ + 0x33a9, 0x33ad, /* ㎩ - ㎭ */ + 0x33b0, 0x33c1, /* ㎰ - ㏁ */ + 0x33c3, 0x33c5, /* ㏃ - ㏅ */ + 0x33c7, 0x33d7, /* ㏇ - ㏗ */ + 0x33d9, 0x33dd, /* ㏙ - ㏝ */ + 0x4e00, 0x9fff, /* 一 - 鿿 */ + 0xac00, 0xd7a3, /* 가 - 힣 */ + 0xf900, 0xfb06, /* 豈 - st */ + 0xfb13, 0xfb17, /* ﬓ - ﬗ */ + 0xfb1f, 0xfb28, /* ײַ - ﬨ */ + 0xfb2a, 0xfb36, /* שׁ - זּ */ + 0xfb38, 0xfb3c, /* טּ - לּ */ + 0xfb40, 0xfb41, /* נּ - סּ */ + 0xfb43, 0xfb44, /* ףּ - פּ */ + 0xfb46, 0xfbb1, /* צּ - ﮱ */ + 0xfbd3, 0xfd3d, /* ﯓ - ﴽ */ + 0xfd50, 0xfd8f, /* ﵐ - ﶏ */ + 0xfd92, 0xfdc7, /* ﶒ - ﷇ */ + 0xfdf0, 0xfdf9, /* ﷰ - ﷹ */ + 0xfe70, 0xfe72, /* ﹰ - ﹲ */ + 0xfe76, 0xfefc, /* ﹶ - ﻼ */ + 0xff66, 0xff6f, /* ｦ - ｯ */ + 0xff71, 0xff9d, /* ｱ - ﾝ */ + 0xffa0, 0xffbe, /* ﾠ - ﾾ */ + 0xffc2, 0xffc7, /* ￂ - ￇ */ + 0xffca, 0xffcf, /* ￊ - ￏ */ + 0xffd2, 0xffd7, /* ￒ - ￗ */ + 0xffda, 0xffdc, /* ￚ - ￜ */ +}; + +/* + * alpha singlets - + * only covers ranges not in lower||upper + */ +static +Rune __alpha1[] = +{ + 0x00aa, /* ª */ + 0x00b5, /* µ */ + 0x00ba, /* º */ + 0x03da, /* Ϛ */ + 0x03dc, /* Ϝ */ + 0x03de, /* Ϟ */ + 0x03e0, /* Ϡ */ + 0x06d5, /* ە */ + 0x09b2, /* ল */ + 0x0a5e, /* ਫ਼ */ + 0x0a8d, /* ઍ */ + 0x0ae0, /* ૠ */ + 0x0b9c, /* ஜ */ + 0x0cde, /* ೞ */ + 0x0e4f, /* ๏ */ + 0x0e84, /* ຄ */ + 0x0e8a, /* ຊ */ + 0x0e8d, /* ຍ */ + 0x0ea5, /* ລ */ + 0x0ea7, /* ວ */ + 0x0eb0, /* ະ */ + 0x0ebd, /* ຽ */ + 0x1fbe, /* ι */ + 0x207f, /* ⁿ */ + 0x20a8, /* ₨ */ + 0x2102, /* ℂ */ + 0x2107, /* ℇ */ + 0x2124, /* ℤ */ + 0x2126, /* Ω */ + 0x2128, /* ℨ */ + 0xfb3e, /* מּ */ + 0xfe74, /* ﹴ */ +}; + +/* + * space ranges + */ +static +Rune __space2[] = +{ + 0x0009, 0x000a, /* tab and newline */ + 0x0020, 0x0020, /* space */ + 0x00a0, 0x00a0, /*   */ + 0x2000, 0x200b, /*   - ​ */ + 0x2028, 0x2029, /* 
 - 
 */ + 0x3000, 0x3000, /* 　 */ + 0xfeff, 0xfeff, /*  */ +}; + +/* + * lower case ranges + * 3rd col is conversion excess 500 + */ +static +Rune __toupper2[] = +{ + 0x0061, 0x007a, 468, /* a-z A-Z */ + 0x00e0, 0x00f6, 468, /* à-ö À-Ö */ + 0x00f8, 0x00fe, 468, /* ø-þ Ø-Þ */ + 0x0256, 0x0257, 295, /* ɖ-ɗ Ɖ-Ɗ */ + 0x0258, 0x0259, 298, /* ɘ-ə Ǝ-Ə */ + 0x028a, 0x028b, 283, /* ʊ-ʋ Ʊ-Ʋ */ + 0x03ad, 0x03af, 463, /* έ-ί Έ-Ί */ + 0x03b1, 0x03c1, 468, /* α-ρ Α-Ρ */ + 0x03c3, 0x03cb, 468, /* σ-ϋ Σ-Ϋ */ + 0x03cd, 0x03ce, 437, /* ύ-ώ Ύ-Ώ */ + 0x0430, 0x044f, 468, /* а-я А-Я */ + 0x0451, 0x045c, 420, /* ё-ќ Ё-Ќ */ + 0x045e, 0x045f, 420, /* ў-џ Ў-Џ */ + 0x0561, 0x0586, 452, /* ա-ֆ Ա-Ֆ */ + 0x1f00, 0x1f07, 508, /* ἀ-ἇ Ἀ-Ἇ */ + 0x1f10, 0x1f15, 508, /* ἐ-ἕ Ἐ-Ἕ */ + 0x1f20, 0x1f27, 508, /* ἠ-ἧ Ἠ-Ἧ */ + 0x1f30, 0x1f37, 508, /* ἰ-ἷ Ἰ-Ἷ */ + 0x1f40, 0x1f45, 508, /* ὀ-ὅ Ὀ-Ὅ */ + 0x1f60, 0x1f67, 508, /* ὠ-ὧ Ὠ-Ὧ */ + 0x1f70, 0x1f71, 574, /* ὰ-ά Ὰ-Ά */ + 0x1f72, 0x1f75, 586, /* ὲ-ή Ὲ-Ή */ + 0x1f76, 0x1f77, 600, /* ὶ-ί Ὶ-Ί */ + 0x1f78, 0x1f79, 628, /* ὸ-ό Ὸ-Ό */ + 0x1f7a, 0x1f7b, 612, /* ὺ-ύ Ὺ-Ύ */ + 0x1f7c, 0x1f7d, 626, /* ὼ-ώ Ὼ-Ώ */ + 0x1f80, 0x1f87, 508, /* ᾀ-ᾇ ᾈ-ᾏ */ + 0x1f90, 0x1f97, 508, /* ᾐ-ᾗ ᾘ-ᾟ */ + 0x1fa0, 0x1fa7, 508, /* ᾠ-ᾧ ᾨ-ᾯ */ + 0x1fb0, 0x1fb1, 508, /* ᾰ-ᾱ Ᾰ-Ᾱ */ + 0x1fd0, 0x1fd1, 508, /* ῐ-ῑ Ῐ-Ῑ */ + 0x1fe0, 0x1fe1, 508, /* ῠ-ῡ Ῠ-Ῡ */ + 0x2170, 0x217f, 484, /* ⅰ-ⅿ Ⅰ-Ⅿ */ + 0x24d0, 0x24e9, 474, /* ⓐ-ⓩ Ⓐ-Ⓩ */ + 0xff41, 0xff5a, 468, /* ａ-ｚ Ａ-Ｚ */ +}; + +/* + * lower case singlets + * 2nd col is conversion excess 500 + */ +static +Rune __toupper1[] = +{ + 0x00ff, 621, /* ÿ Ÿ */ + 0x0101, 499, /* ā Ā */ + 0x0103, 499, /* ă Ă */ + 0x0105, 499, /* ą Ą */ + 0x0107, 499, /* ć Ć */ + 0x0109, 499, /* ĉ Ĉ */ + 0x010b, 499, /* ċ Ċ */ + 0x010d, 499, /* č Č */ + 0x010f, 499, /* ď Ď */ + 0x0111, 499, /* đ Đ */ + 0x0113, 499, /* ē Ē */ + 0x0115, 499, /* ĕ Ĕ */ + 0x0117, 499, /* ė Ė */ + 0x0119, 499, /* ę Ę */ + 0x011b, 499, /* ě Ě */ + 0x011d, 499, /* ĝ Ĝ */ + 0x011f, 499, /* ğ Ğ */ + 0x0121, 499, /* ġ Ġ */ + 0x0123, 499, /* ģ Ģ */ + 0x0125, 499, /* ĥ Ĥ */ + 0x0127, 499, /* ħ Ħ */ + 0x0129, 499, /* ĩ Ĩ */ + 0x012b, 499, /* ī Ī */ + 0x012d, 499, /* ĭ Ĭ */ + 0x012f, 499, /* į Į */ + 0x0131, 268, /* ı I */ + 0x0133, 499, /* ij IJ */ + 0x0135, 499, /* ĵ Ĵ */ + 0x0137, 499, /* ķ Ķ */ + 0x013a, 499, /* ĺ Ĺ */ + 0x013c, 499, /* ļ Ļ */ + 0x013e, 499, /* ľ Ľ */ + 0x0140, 499, /* ŀ Ŀ */ + 0x0142, 499, /* ł Ł */ + 0x0144, 499, /* ń Ń */ + 0x0146, 499, /* ņ Ņ */ + 0x0148, 499, /* ň Ň */ + 0x014b, 499, /* ŋ Ŋ */ + 0x014d, 499, /* ō Ō */ + 0x014f, 499, /* ŏ Ŏ */ + 0x0151, 499, /* ő Ő */ + 0x0153, 499, /* œ Œ */ + 0x0155, 499, /* ŕ Ŕ */ + 0x0157, 499, /* ŗ Ŗ */ + 0x0159, 499, /* ř Ř */ + 0x015b, 499, /* ś Ś */ + 0x015d, 499, /* ŝ Ŝ */ + 0x015f, 499, /* ş Ş */ + 0x0161, 499, /* š Š */ + 0x0163, 499, /* ţ Ţ */ + 0x0165, 499, /* ť Ť */ + 0x0167, 499, /* ŧ Ŧ */ + 0x0169, 499, /* ũ Ũ */ + 0x016b, 499, /* ū Ū */ + 0x016d, 499, /* ŭ Ŭ */ + 0x016f, 499, /* ů Ů */ + 0x0171, 499, /* ű Ű */ + 0x0173, 499, /* ų Ų */ + 0x0175, 499, /* ŵ Ŵ */ + 0x0177, 499, /* ŷ Ŷ */ + 0x017a, 499, /* ź Ź */ + 0x017c, 499, /* ż Ż */ + 0x017e, 499, /* ž Ž */ + 0x017f, 200, /* ſ S */ + 0x0183, 499, /* ƃ Ƃ */ + 0x0185, 499, /* ƅ Ƅ */ + 0x0188, 499, /* ƈ Ƈ */ + 0x018c, 499, /* ƌ Ƌ */ + 0x0192, 499, /* ƒ Ƒ */ + 0x0199, 499, /* ƙ Ƙ */ + 0x01a1, 499, /* ơ Ơ */ + 0x01a3, 499, /* ƣ Ƣ */ + 0x01a5, 499, /* ƥ Ƥ */ + 0x01a8, 499, /* ƨ Ƨ */ + 0x01ad, 499, /* ƭ Ƭ */ + 0x01b0, 499, /* ư Ư */ + 0x01b4, 499, /* ƴ Ƴ */ + 0x01b6, 499, /* ƶ Ƶ */ + 0x01b9, 499, /* ƹ Ƹ */ + 0x01bd, 499, /* ƽ Ƽ */ + 0x01c5, 499, /* Dž DŽ */ + 0x01c6, 498, /* dž DŽ */ + 0x01c8, 499, /* Lj LJ */ + 0x01c9, 498, /* lj LJ */ + 0x01cb, 499, /* Nj NJ */ + 0x01cc, 498, /* nj NJ */ + 0x01ce, 499, /* ǎ Ǎ */ + 0x01d0, 499, /* ǐ Ǐ */ + 0x01d2, 499, /* ǒ Ǒ */ + 0x01d4, 499, /* ǔ Ǔ */ + 0x01d6, 499, /* ǖ Ǖ */ + 0x01d8, 499, /* ǘ Ǘ */ + 0x01da, 499, /* ǚ Ǚ */ + 0x01dc, 499, /* ǜ Ǜ */ + 0x01df, 499, /* ǟ Ǟ */ + 0x01e1, 499, /* ǡ Ǡ */ + 0x01e3, 499, /* ǣ Ǣ */ + 0x01e5, 499, /* ǥ Ǥ */ + 0x01e7, 499, /* ǧ Ǧ */ + 0x01e9, 499, /* ǩ Ǩ */ + 0x01eb, 499, /* ǫ Ǫ */ + 0x01ed, 499, /* ǭ Ǭ */ + 0x01ef, 499, /* ǯ Ǯ */ + 0x01f2, 499, /* Dz DZ */ + 0x01f3, 498, /* dz DZ */ + 0x01f5, 499, /* ǵ Ǵ */ + 0x01fb, 499, /* ǻ Ǻ */ + 0x01fd, 499, /* ǽ Ǽ */ + 0x01ff, 499, /* ǿ Ǿ */ + 0x0201, 499, /* ȁ Ȁ */ + 0x0203, 499, /* ȃ Ȃ */ + 0x0205, 499, /* ȅ Ȅ */ + 0x0207, 499, /* ȇ Ȇ */ + 0x0209, 499, /* ȉ Ȉ */ + 0x020b, 499, /* ȋ Ȋ */ + 0x020d, 499, /* ȍ Ȍ */ + 0x020f, 499, /* ȏ Ȏ */ + 0x0211, 499, /* ȑ Ȑ */ + 0x0213, 499, /* ȓ Ȓ */ + 0x0215, 499, /* ȕ Ȕ */ + 0x0217, 499, /* ȗ Ȗ */ + 0x0253, 290, /* ɓ Ɓ */ + 0x0254, 294, /* ɔ Ɔ */ + 0x025b, 297, /* ɛ Ɛ */ + 0x0260, 295, /* ɠ Ɠ */ + 0x0263, 293, /* ɣ Ɣ */ + 0x0268, 291, /* ɨ Ɨ */ + 0x0269, 289, /* ɩ Ɩ */ + 0x026f, 289, /* ɯ Ɯ */ + 0x0272, 287, /* ɲ Ɲ */ + 0x0283, 282, /* ʃ Ʃ */ + 0x0288, 282, /* ʈ Ʈ */ + 0x0292, 281, /* ʒ Ʒ */ + 0x03ac, 462, /* ά Ά */ + 0x03cc, 436, /* ό Ό */ + 0x03d0, 438, /* ϐ Β */ + 0x03d1, 443, /* ϑ Θ */ + 0x03d5, 453, /* ϕ Φ */ + 0x03d6, 446, /* ϖ Π */ + 0x03e3, 499, /* ϣ Ϣ */ + 0x03e5, 499, /* ϥ Ϥ */ + 0x03e7, 499, /* ϧ Ϧ */ + 0x03e9, 499, /* ϩ Ϩ */ + 0x03eb, 499, /* ϫ Ϫ */ + 0x03ed, 499, /* ϭ Ϭ */ + 0x03ef, 499, /* ϯ Ϯ */ + 0x03f0, 414, /* ϰ Κ */ + 0x03f1, 420, /* ϱ Ρ */ + 0x0461, 499, /* ѡ Ѡ */ + 0x0463, 499, /* ѣ Ѣ */ + 0x0465, 499, /* ѥ Ѥ */ + 0x0467, 499, /* ѧ Ѧ */ + 0x0469, 499, /* ѩ Ѩ */ + 0x046b, 499, /* ѫ Ѫ */ + 0x046d, 499, /* ѭ Ѭ */ + 0x046f, 499, /* ѯ Ѯ */ + 0x0471, 499, /* ѱ Ѱ */ + 0x0473, 499, /* ѳ Ѳ */ + 0x0475, 499, /* ѵ Ѵ */ + 0x0477, 499, /* ѷ Ѷ */ + 0x0479, 499, /* ѹ Ѹ */ + 0x047b, 499, /* ѻ Ѻ */ + 0x047d, 499, /* ѽ Ѽ */ + 0x047f, 499, /* ѿ Ѿ */ + 0x0481, 499, /* ҁ Ҁ */ + 0x0491, 499, /* ґ Ґ */ + 0x0493, 499, /* ғ Ғ */ + 0x0495, 499, /* ҕ Ҕ */ + 0x0497, 499, /* җ Җ */ + 0x0499, 499, /* ҙ Ҙ */ + 0x049b, 499, /* қ Қ */ + 0x049d, 499, /* ҝ Ҝ */ + 0x049f, 499, /* ҟ Ҟ */ + 0x04a1, 499, /* ҡ Ҡ */ + 0x04a3, 499, /* ң Ң */ + 0x04a5, 499, /* ҥ Ҥ */ + 0x04a7, 499, /* ҧ Ҧ */ + 0x04a9, 499, /* ҩ Ҩ */ + 0x04ab, 499, /* ҫ Ҫ */ + 0x04ad, 499, /* ҭ Ҭ */ + 0x04af, 499, /* ү Ү */ + 0x04b1, 499, /* ұ Ұ */ + 0x04b3, 499, /* ҳ Ҳ */ + 0x04b5, 499, /* ҵ Ҵ */ + 0x04b7, 499, /* ҷ Ҷ */ + 0x04b9, 499, /* ҹ Ҹ */ + 0x04bb, 499, /* һ Һ */ + 0x04bd, 499, /* ҽ Ҽ */ + 0x04bf, 499, /* ҿ Ҿ */ + 0x04c2, 499, /* ӂ Ӂ */ + 0x04c4, 499, /* ӄ Ӄ */ + 0x04c8, 499, /* ӈ Ӈ */ + 0x04cc, 499, /* ӌ Ӌ */ + 0x04d1, 499, /* ӑ Ӑ */ + 0x04d3, 499, /* ӓ Ӓ */ + 0x04d5, 499, /* ӕ Ӕ */ + 0x04d7, 499, /* ӗ Ӗ */ + 0x04d9, 499, /* ә Ә */ + 0x04db, 499, /* ӛ Ӛ */ + 0x04dd, 499, /* ӝ Ӝ */ + 0x04df, 499, /* ӟ Ӟ */ + 0x04e1, 499, /* ӡ Ӡ */ + 0x04e3, 499, /* ӣ Ӣ */ + 0x04e5, 499, /* ӥ Ӥ */ + 0x04e7, 499, /* ӧ Ӧ */ + 0x04e9, 499, /* ө Ө */ + 0x04eb, 499, /* ӫ Ӫ */ + 0x04ef, 499, /* ӯ Ӯ */ + 0x04f1, 499, /* ӱ Ӱ */ + 0x04f3, 499, /* ӳ Ӳ */ + 0x04f5, 499, /* ӵ Ӵ */ + 0x04f9, 499, /* ӹ Ӹ */ + 0x1e01, 499, /* ḁ Ḁ */ + 0x1e03, 499, /* ḃ Ḃ */ + 0x1e05, 499, /* ḅ Ḅ */ + 0x1e07, 499, /* ḇ Ḇ */ + 0x1e09, 499, /* ḉ Ḉ */ + 0x1e0b, 499, /* ḋ Ḋ */ + 0x1e0d, 499, /* ḍ Ḍ */ + 0x1e0f, 499, /* ḏ Ḏ */ + 0x1e11, 499, /* ḑ Ḑ */ + 0x1e13, 499, /* ḓ Ḓ */ + 0x1e15, 499, /* ḕ Ḕ */ + 0x1e17, 499, /* ḗ Ḗ */ + 0x1e19, 499, /* ḙ Ḙ */ + 0x1e1b, 499, /* ḛ Ḛ */ + 0x1e1d, 499, /* ḝ Ḝ */ + 0x1e1f, 499, /* ḟ Ḟ */ + 0x1e21, 499, /* ḡ Ḡ */ + 0x1e23, 499, /* ḣ Ḣ */ + 0x1e25, 499, /* ḥ Ḥ */ + 0x1e27, 499, /* ḧ Ḧ */ + 0x1e29, 499, /* ḩ Ḩ */ + 0x1e2b, 499, /* ḫ Ḫ */ + 0x1e2d, 499, /* ḭ Ḭ */ + 0x1e2f, 499, /* ḯ Ḯ */ + 0x1e31, 499, /* ḱ Ḱ */ + 0x1e33, 499, /* ḳ Ḳ */ + 0x1e35, 499, /* ḵ Ḵ */ + 0x1e37, 499, /* ḷ Ḷ */ + 0x1e39, 499, /* ḹ Ḹ */ + 0x1e3b, 499, /* ḻ Ḻ */ + 0x1e3d, 499, /* ḽ Ḽ */ + 0x1e3f, 499, /* ḿ Ḿ */ + 0x1e41, 499, /* ṁ Ṁ */ + 0x1e43, 499, /* ṃ Ṃ */ + 0x1e45, 499, /* ṅ Ṅ */ + 0x1e47, 499, /* ṇ Ṇ */ + 0x1e49, 499, /* ṉ Ṉ */ + 0x1e4b, 499, /* ṋ Ṋ */ + 0x1e4d, 499, /* ṍ Ṍ */ + 0x1e4f, 499, /* ṏ Ṏ */ + 0x1e51, 499, /* ṑ Ṑ */ + 0x1e53, 499, /* ṓ Ṓ */ + 0x1e55, 499, /* ṕ Ṕ */ + 0x1e57, 499, /* ṗ Ṗ */ + 0x1e59, 499, /* ṙ Ṙ */ + 0x1e5b, 499, /* ṛ Ṛ */ + 0x1e5d, 499, /* ṝ Ṝ */ + 0x1e5f, 499, /* ṟ Ṟ */ + 0x1e61, 499, /* ṡ Ṡ */ + 0x1e63, 499, /* ṣ Ṣ */ + 0x1e65, 499, /* ṥ Ṥ */ + 0x1e67, 499, /* ṧ Ṧ */ + 0x1e69, 499, /* ṩ Ṩ */ + 0x1e6b, 499, /* ṫ Ṫ */ + 0x1e6d, 499, /* ṭ Ṭ */ + 0x1e6f, 499, /* ṯ Ṯ */ + 0x1e71, 499, /* ṱ Ṱ */ + 0x1e73, 499, /* ṳ Ṳ */ + 0x1e75, 499, /* ṵ Ṵ */ + 0x1e77, 499, /* ṷ Ṷ */ + 0x1e79, 499, /* ṹ Ṹ */ + 0x1e7b, 499, /* ṻ Ṻ */ + 0x1e7d, 499, /* ṽ Ṽ */ + 0x1e7f, 499, /* ṿ Ṿ */ + 0x1e81, 499, /* ẁ Ẁ */ + 0x1e83, 499, /* ẃ Ẃ */ + 0x1e85, 499, /* ẅ Ẅ */ + 0x1e87, 499, /* ẇ Ẇ */ + 0x1e89, 499, /* ẉ Ẉ */ + 0x1e8b, 499, /* ẋ Ẋ */ + 0x1e8d, 499, /* ẍ Ẍ */ + 0x1e8f, 499, /* ẏ Ẏ */ + 0x1e91, 499, /* ẑ Ẑ */ + 0x1e93, 499, /* ẓ Ẓ */ + 0x1e95, 499, /* ẕ Ẕ */ + 0x1ea1, 499, /* ạ Ạ */ + 0x1ea3, 499, /* ả Ả */ + 0x1ea5, 499, /* ấ Ấ */ + 0x1ea7, 499, /* ầ Ầ */ + 0x1ea9, 499, /* ẩ Ẩ */ + 0x1eab, 499, /* ẫ Ẫ */ + 0x1ead, 499, /* ậ Ậ */ + 0x1eaf, 499, /* ắ Ắ */ + 0x1eb1, 499, /* ằ Ằ */ + 0x1eb3, 499, /* ẳ Ẳ */ + 0x1eb5, 499, /* ẵ Ẵ */ + 0x1eb7, 499, /* ặ Ặ */ + 0x1eb9, 499, /* ẹ Ẹ */ + 0x1ebb, 499, /* ẻ Ẻ */ + 0x1ebd, 499, /* ẽ Ẽ */ + 0x1ebf, 499, /* ế Ế */ + 0x1ec1, 499, /* ề Ề */ + 0x1ec3, 499, /* ể Ể */ + 0x1ec5, 499, /* ễ Ễ */ + 0x1ec7, 499, /* ệ Ệ */ + 0x1ec9, 499, /* ỉ Ỉ */ + 0x1ecb, 499, /* ị Ị */ + 0x1ecd, 499, /* ọ Ọ */ + 0x1ecf, 499, /* ỏ Ỏ */ + 0x1ed1, 499, /* ố Ố */ + 0x1ed3, 499, /* ồ Ồ */ + 0x1ed5, 499, /* ổ Ổ */ + 0x1ed7, 499, /* ỗ Ỗ */ + 0x1ed9, 499, /* ộ Ộ */ + 0x1edb, 499, /* ớ Ớ */ + 0x1edd, 499, /* ờ Ờ */ + 0x1edf, 499, /* ở Ở */ + 0x1ee1, 499, /* ỡ Ỡ */ + 0x1ee3, 499, /* ợ Ợ */ + 0x1ee5, 499, /* ụ Ụ */ + 0x1ee7, 499, /* ủ Ủ */ + 0x1ee9, 499, /* ứ Ứ */ + 0x1eeb, 499, /* ừ Ừ */ + 0x1eed, 499, /* ử Ử */ + 0x1eef, 499, /* ữ Ữ */ + 0x1ef1, 499, /* ự Ự */ + 0x1ef3, 499, /* ỳ Ỳ */ + 0x1ef5, 499, /* ỵ Ỵ */ + 0x1ef7, 499, /* ỷ Ỷ */ + 0x1ef9, 499, /* ỹ Ỹ */ + 0x1f51, 508, /* ὑ Ὑ */ + 0x1f53, 508, /* ὓ Ὓ */ + 0x1f55, 508, /* ὕ Ὕ */ + 0x1f57, 508, /* ὗ Ὗ */ + 0x1fb3, 509, /* ᾳ ᾼ */ + 0x1fc3, 509, /* ῃ ῌ */ + 0x1fe5, 507, /* ῥ Ῥ */ + 0x1ff3, 509, /* ῳ ῼ */ +}; + +/* + * upper case ranges + * 3rd col is conversion excess 500 + */ +static +Rune __tolower2[] = +{ + 0x0041, 0x005a, 532, /* A-Z a-z */ + 0x00c0, 0x00d6, 532, /* À-Ö à-ö */ + 0x00d8, 0x00de, 532, /* Ø-Þ ø-þ */ + 0x0189, 0x018a, 705, /* Ɖ-Ɗ ɖ-ɗ */ + 0x018e, 0x018f, 702, /* Ǝ-Ə ɘ-ə */ + 0x01b1, 0x01b2, 717, /* Ʊ-Ʋ ʊ-ʋ */ + 0x0388, 0x038a, 537, /* Έ-Ί έ-ί */ + 0x038e, 0x038f, 563, /* Ύ-Ώ ύ-ώ */ + 0x0391, 0x03a1, 532, /* Α-Ρ α-ρ */ + 0x03a3, 0x03ab, 532, /* Σ-Ϋ σ-ϋ */ + 0x0401, 0x040c, 580, /* Ё-Ќ ё-ќ */ + 0x040e, 0x040f, 580, /* Ў-Џ ў-џ */ + 0x0410, 0x042f, 532, /* А-Я а-я */ + 0x0531, 0x0556, 548, /* Ա-Ֆ ա-ֆ */ + 0x10a0, 0x10c5, 548, /* Ⴀ-Ⴥ ა-ჵ */ + 0x1f08, 0x1f0f, 492, /* Ἀ-Ἇ ἀ-ἇ */ + 0x1f18, 0x1f1d, 492, /* Ἐ-Ἕ ἐ-ἕ */ + 0x1f28, 0x1f2f, 492, /* Ἠ-Ἧ ἠ-ἧ */ + 0x1f38, 0x1f3f, 492, /* Ἰ-Ἷ ἰ-ἷ */ + 0x1f48, 0x1f4d, 492, /* Ὀ-Ὅ ὀ-ὅ */ + 0x1f68, 0x1f6f, 492, /* Ὠ-Ὧ ὠ-ὧ */ + 0x1f88, 0x1f8f, 492, /* ᾈ-ᾏ ᾀ-ᾇ */ + 0x1f98, 0x1f9f, 492, /* ᾘ-ᾟ ᾐ-ᾗ */ + 0x1fa8, 0x1faf, 492, /* ᾨ-ᾯ ᾠ-ᾧ */ + 0x1fb8, 0x1fb9, 492, /* Ᾰ-Ᾱ ᾰ-ᾱ */ + 0x1fba, 0x1fbb, 426, /* Ὰ-Ά ὰ-ά */ + 0x1fc8, 0x1fcb, 414, /* Ὲ-Ή ὲ-ή */ + 0x1fd8, 0x1fd9, 492, /* Ῐ-Ῑ ῐ-ῑ */ + 0x1fda, 0x1fdb, 400, /* Ὶ-Ί ὶ-ί */ + 0x1fe8, 0x1fe9, 492, /* Ῠ-Ῡ ῠ-ῡ */ + 0x1fea, 0x1feb, 388, /* Ὺ-Ύ ὺ-ύ */ + 0x1ff8, 0x1ff9, 372, /* Ὸ-Ό ὸ-ό */ + 0x1ffa, 0x1ffb, 374, /* Ὼ-Ώ ὼ-ώ */ + 0x2160, 0x216f, 516, /* Ⅰ-Ⅿ ⅰ-ⅿ */ + 0x24b6, 0x24cf, 526, /* Ⓐ-Ⓩ ⓐ-ⓩ */ + 0xff21, 0xff3a, 532, /* Ａ-Ｚ ａ-ｚ */ +}; + +/* + * upper case singlets + * 2nd col is conversion excess 500 + */ +static +Rune __tolower1[] = +{ + 0x0100, 501, /* Ā ā */ + 0x0102, 501, /* Ă ă */ + 0x0104, 501, /* Ą ą */ + 0x0106, 501, /* Ć ć */ + 0x0108, 501, /* Ĉ ĉ */ + 0x010a, 501, /* Ċ ċ */ + 0x010c, 501, /* Č č */ + 0x010e, 501, /* Ď ď */ + 0x0110, 501, /* Đ đ */ + 0x0112, 501, /* Ē ē */ + 0x0114, 501, /* Ĕ ĕ */ + 0x0116, 501, /* Ė ė */ + 0x0118, 501, /* Ę ę */ + 0x011a, 501, /* Ě ě */ + 0x011c, 501, /* Ĝ ĝ */ + 0x011e, 501, /* Ğ ğ */ + 0x0120, 501, /* Ġ ġ */ + 0x0122, 501, /* Ģ ģ */ + 0x0124, 501, /* Ĥ ĥ */ + 0x0126, 501, /* Ħ ħ */ + 0x0128, 501, /* Ĩ ĩ */ + 0x012a, 501, /* Ī ī */ + 0x012c, 501, /* Ĭ ĭ */ + 0x012e, 501, /* Į į */ + 0x0130, 301, /* İ i */ + 0x0132, 501, /* IJ ij */ + 0x0134, 501, /* Ĵ ĵ */ + 0x0136, 501, /* Ķ ķ */ + 0x0139, 501, /* Ĺ ĺ */ + 0x013b, 501, /* Ļ ļ */ + 0x013d, 501, /* Ľ ľ */ + 0x013f, 501, /* Ŀ ŀ */ + 0x0141, 501, /* Ł ł */ + 0x0143, 501, /* Ń ń */ + 0x0145, 501, /* Ņ ņ */ + 0x0147, 501, /* Ň ň */ + 0x014a, 501, /* Ŋ ŋ */ + 0x014c, 501, /* Ō ō */ + 0x014e, 501, /* Ŏ ŏ */ + 0x0150, 501, /* Ő ő */ + 0x0152, 501, /* Œ œ */ + 0x0154, 501, /* Ŕ ŕ */ + 0x0156, 501, /* Ŗ ŗ */ + 0x0158, 501, /* Ř ř */ + 0x015a, 501, /* Ś ś */ + 0x015c, 501, /* Ŝ ŝ */ + 0x015e, 501, /* Ş ş */ + 0x0160, 501, /* Š š */ + 0x0162, 501, /* Ţ ţ */ + 0x0164, 501, /* Ť ť */ + 0x0166, 501, /* Ŧ ŧ */ + 0x0168, 501, /* Ũ ũ */ + 0x016a, 501, /* Ū ū */ + 0x016c, 501, /* Ŭ ŭ */ + 0x016e, 501, /* Ů ů */ + 0x0170, 501, /* Ű ű */ + 0x0172, 501, /* Ų ų */ + 0x0174, 501, /* Ŵ ŵ */ + 0x0176, 501, /* Ŷ ŷ */ + 0x0178, 379, /* Ÿ ÿ */ + 0x0179, 501, /* Ź ź */ + 0x017b, 501, /* Ż ż */ + 0x017d, 501, /* Ž ž */ + 0x0181, 710, /* Ɓ ɓ */ + 0x0182, 501, /* Ƃ ƃ */ + 0x0184, 501, /* Ƅ ƅ */ + 0x0186, 706, /* Ɔ ɔ */ + 0x0187, 501, /* Ƈ ƈ */ + 0x018b, 501, /* Ƌ ƌ */ + 0x0190, 703, /* Ɛ ɛ */ + 0x0191, 501, /* Ƒ ƒ */ + 0x0193, 705, /* Ɠ ɠ */ + 0x0194, 707, /* Ɣ ɣ */ + 0x0196, 711, /* Ɩ ɩ */ + 0x0197, 709, /* Ɨ ɨ */ + 0x0198, 501, /* Ƙ ƙ */ + 0x019c, 711, /* Ɯ ɯ */ + 0x019d, 713, /* Ɲ ɲ */ + 0x01a0, 501, /* Ơ ơ */ + 0x01a2, 501, /* Ƣ ƣ */ + 0x01a4, 501, /* Ƥ ƥ */ + 0x01a7, 501, /* Ƨ ƨ */ + 0x01a9, 718, /* Ʃ ʃ */ + 0x01ac, 501, /* Ƭ ƭ */ + 0x01ae, 718, /* Ʈ ʈ */ + 0x01af, 501, /* Ư ư */ + 0x01b3, 501, /* Ƴ ƴ */ + 0x01b5, 501, /* Ƶ ƶ */ + 0x01b7, 719, /* Ʒ ʒ */ + 0x01b8, 501, /* Ƹ ƹ */ + 0x01bc, 501, /* Ƽ ƽ */ + 0x01c4, 502, /* DŽ dž */ + 0x01c5, 501, /* Dž dž */ + 0x01c7, 502, /* LJ lj */ + 0x01c8, 501, /* Lj lj */ + 0x01ca, 502, /* NJ nj */ + 0x01cb, 501, /* Nj nj */ + 0x01cd, 501, /* Ǎ ǎ */ + 0x01cf, 501, /* Ǐ ǐ */ + 0x01d1, 501, /* Ǒ ǒ */ + 0x01d3, 501, /* Ǔ ǔ */ + 0x01d5, 501, /* Ǖ ǖ */ + 0x01d7, 501, /* Ǘ ǘ */ + 0x01d9, 501, /* Ǚ ǚ */ + 0x01db, 501, /* Ǜ ǜ */ + 0x01de, 501, /* Ǟ ǟ */ + 0x01e0, 501, /* Ǡ ǡ */ + 0x01e2, 501, /* Ǣ ǣ */ + 0x01e4, 501, /* Ǥ ǥ */ + 0x01e6, 501, /* Ǧ ǧ */ + 0x01e8, 501, /* Ǩ ǩ */ + 0x01ea, 501, /* Ǫ ǫ */ + 0x01ec, 501, /* Ǭ ǭ */ + 0x01ee, 501, /* Ǯ ǯ */ + 0x01f1, 502, /* DZ dz */ + 0x01f2, 501, /* Dz dz */ + 0x01f4, 501, /* Ǵ ǵ */ + 0x01fa, 501, /* Ǻ ǻ */ + 0x01fc, 501, /* Ǽ ǽ */ + 0x01fe, 501, /* Ǿ ǿ */ + 0x0200, 501, /* Ȁ ȁ */ + 0x0202, 501, /* Ȃ ȃ */ + 0x0204, 501, /* Ȅ ȅ */ + 0x0206, 501, /* Ȇ ȇ */ + 0x0208, 501, /* Ȉ ȉ */ + 0x020a, 501, /* Ȋ ȋ */ + 0x020c, 501, /* Ȍ ȍ */ + 0x020e, 501, /* Ȏ ȏ */ + 0x0210, 501, /* Ȑ ȑ */ + 0x0212, 501, /* Ȓ ȓ */ + 0x0214, 501, /* Ȕ ȕ */ + 0x0216, 501, /* Ȗ ȗ */ + 0x0386, 538, /* Ά ά */ + 0x038c, 564, /* Ό ό */ + 0x03e2, 501, /* Ϣ ϣ */ + 0x03e4, 501, /* Ϥ ϥ */ + 0x03e6, 501, /* Ϧ ϧ */ + 0x03e8, 501, /* Ϩ ϩ */ + 0x03ea, 501, /* Ϫ ϫ */ + 0x03ec, 501, /* Ϭ ϭ */ + 0x03ee, 501, /* Ϯ ϯ */ + 0x0460, 501, /* Ѡ ѡ */ + 0x0462, 501, /* Ѣ ѣ */ + 0x0464, 501, /* Ѥ ѥ */ + 0x0466, 501, /* Ѧ ѧ */ + 0x0468, 501, /* Ѩ ѩ */ + 0x046a, 501, /* Ѫ ѫ */ + 0x046c, 501, /* Ѭ ѭ */ + 0x046e, 501, /* Ѯ ѯ */ + 0x0470, 501, /* Ѱ ѱ */ + 0x0472, 501, /* Ѳ ѳ */ + 0x0474, 501, /* Ѵ ѵ */ + 0x0476, 501, /* Ѷ ѷ */ + 0x0478, 501, /* Ѹ ѹ */ + 0x047a, 501, /* Ѻ ѻ */ + 0x047c, 501, /* Ѽ ѽ */ + 0x047e, 501, /* Ѿ ѿ */ + 0x0480, 501, /* Ҁ ҁ */ + 0x0490, 501, /* Ґ ґ */ + 0x0492, 501, /* Ғ ғ */ + 0x0494, 501, /* Ҕ ҕ */ + 0x0496, 501, /* Җ җ */ + 0x0498, 501, /* Ҙ ҙ */ + 0x049a, 501, /* Қ қ */ + 0x049c, 501, /* Ҝ ҝ */ + 0x049e, 501, /* Ҟ ҟ */ + 0x04a0, 501, /* Ҡ ҡ */ + 0x04a2, 501, /* Ң ң */ + 0x04a4, 501, /* Ҥ ҥ */ + 0x04a6, 501, /* Ҧ ҧ */ + 0x04a8, 501, /* Ҩ ҩ */ + 0x04aa, 501, /* Ҫ ҫ */ + 0x04ac, 501, /* Ҭ ҭ */ + 0x04ae, 501, /* Ү ү */ + 0x04b0, 501, /* Ұ ұ */ + 0x04b2, 501, /* Ҳ ҳ */ + 0x04b4, 501, /* Ҵ ҵ */ + 0x04b6, 501, /* Ҷ ҷ */ + 0x04b8, 501, /* Ҹ ҹ */ + 0x04ba, 501, /* Һ һ */ + 0x04bc, 501, /* Ҽ ҽ */ + 0x04be, 501, /* Ҿ ҿ */ + 0x04c1, 501, /* Ӂ ӂ */ + 0x04c3, 501, /* Ӄ ӄ */ + 0x04c7, 501, /* Ӈ ӈ */ + 0x04cb, 501, /* Ӌ ӌ */ + 0x04d0, 501, /* Ӑ ӑ */ + 0x04d2, 501, /* Ӓ ӓ */ + 0x04d4, 501, /* Ӕ ӕ */ + 0x04d6, 501, /* Ӗ ӗ */ + 0x04d8, 501, /* Ә ә */ + 0x04da, 501, /* Ӛ ӛ */ + 0x04dc, 501, /* Ӝ ӝ */ + 0x04de, 501, /* Ӟ ӟ */ + 0x04e0, 501, /* Ӡ ӡ */ + 0x04e2, 501, /* Ӣ ӣ */ + 0x04e4, 501, /* Ӥ ӥ */ + 0x04e6, 501, /* Ӧ ӧ */ + 0x04e8, 501, /* Ө ө */ + 0x04ea, 501, /* Ӫ ӫ */ + 0x04ee, 501, /* Ӯ ӯ */ + 0x04f0, 501, /* Ӱ ӱ */ + 0x04f2, 501, /* Ӳ ӳ */ + 0x04f4, 501, /* Ӵ ӵ */ + 0x04f8, 501, /* Ӹ ӹ */ + 0x1e00, 501, /* Ḁ ḁ */ + 0x1e02, 501, /* Ḃ ḃ */ + 0x1e04, 501, /* Ḅ ḅ */ + 0x1e06, 501, /* Ḇ ḇ */ + 0x1e08, 501, /* Ḉ ḉ */ + 0x1e0a, 501, /* Ḋ ḋ */ + 0x1e0c, 501, /* Ḍ ḍ */ + 0x1e0e, 501, /* Ḏ ḏ */ + 0x1e10, 501, /* Ḑ ḑ */ + 0x1e12, 501, /* Ḓ ḓ */ + 0x1e14, 501, /* Ḕ ḕ */ + 0x1e16, 501, /* Ḗ ḗ */ + 0x1e18, 501, /* Ḙ ḙ */ + 0x1e1a, 501, /* Ḛ ḛ */ + 0x1e1c, 501, /* Ḝ ḝ */ + 0x1e1e, 501, /* Ḟ ḟ */ + 0x1e20, 501, /* Ḡ ḡ */ + 0x1e22, 501, /* Ḣ ḣ */ + 0x1e24, 501, /* Ḥ ḥ */ + 0x1e26, 501, /* Ḧ ḧ */ + 0x1e28, 501, /* Ḩ ḩ */ + 0x1e2a, 501, /* Ḫ ḫ */ + 0x1e2c, 501, /* Ḭ ḭ */ + 0x1e2e, 501, /* Ḯ ḯ */ + 0x1e30, 501, /* Ḱ ḱ */ + 0x1e32, 501, /* Ḳ ḳ */ + 0x1e34, 501, /* Ḵ ḵ */ + 0x1e36, 501, /* Ḷ ḷ */ + 0x1e38, 501, /* Ḹ ḹ */ + 0x1e3a, 501, /* Ḻ ḻ */ + 0x1e3c, 501, /* Ḽ ḽ */ + 0x1e3e, 501, /* Ḿ ḿ */ + 0x1e40, 501, /* Ṁ ṁ */ + 0x1e42, 501, /* Ṃ ṃ */ + 0x1e44, 501, /* Ṅ ṅ */ + 0x1e46, 501, /* Ṇ ṇ */ + 0x1e48, 501, /* Ṉ ṉ */ + 0x1e4a, 501, /* Ṋ ṋ */ + 0x1e4c, 501, /* Ṍ ṍ */ + 0x1e4e, 501, /* Ṏ ṏ */ + 0x1e50, 501, /* Ṑ ṑ */ + 0x1e52, 501, /* Ṓ ṓ */ + 0x1e54, 501, /* Ṕ ṕ */ + 0x1e56, 501, /* Ṗ ṗ */ + 0x1e58, 501, /* Ṙ ṙ */ + 0x1e5a, 501, /* Ṛ ṛ */ + 0x1e5c, 501, /* Ṝ ṝ */ + 0x1e5e, 501, /* Ṟ ṟ */ + 0x1e60, 501, /* Ṡ ṡ */ + 0x1e62, 501, /* Ṣ ṣ */ + 0x1e64, 501, /* Ṥ ṥ */ + 0x1e66, 501, /* Ṧ ṧ */ + 0x1e68, 501, /* Ṩ ṩ */ + 0x1e6a, 501, /* Ṫ ṫ */ + 0x1e6c, 501, /* Ṭ ṭ */ + 0x1e6e, 501, /* Ṯ ṯ */ + 0x1e70, 501, /* Ṱ ṱ */ + 0x1e72, 501, /* Ṳ ṳ */ + 0x1e74, 501, /* Ṵ ṵ */ + 0x1e76, 501, /* Ṷ ṷ */ + 0x1e78, 501, /* Ṹ ṹ */ + 0x1e7a, 501, /* Ṻ ṻ */ + 0x1e7c, 501, /* Ṽ ṽ */ + 0x1e7e, 501, /* Ṿ ṿ */ + 0x1e80, 501, /* Ẁ ẁ */ + 0x1e82, 501, /* Ẃ ẃ */ + 0x1e84, 501, /* Ẅ ẅ */ + 0x1e86, 501, /* Ẇ ẇ */ + 0x1e88, 501, /* Ẉ ẉ */ + 0x1e8a, 501, /* Ẋ ẋ */ + 0x1e8c, 501, /* Ẍ ẍ */ + 0x1e8e, 501, /* Ẏ ẏ */ + 0x1e90, 501, /* Ẑ ẑ */ + 0x1e92, 501, /* Ẓ ẓ */ + 0x1e94, 501, /* Ẕ ẕ */ + 0x1ea0, 501, /* Ạ ạ */ + 0x1ea2, 501, /* Ả ả */ + 0x1ea4, 501, /* Ấ ấ */ + 0x1ea6, 501, /* Ầ ầ */ + 0x1ea8, 501, /* Ẩ ẩ */ + 0x1eaa, 501, /* Ẫ ẫ */ + 0x1eac, 501, /* Ậ ậ */ + 0x1eae, 501, /* Ắ ắ */ + 0x1eb0, 501, /* Ằ ằ */ + 0x1eb2, 501, /* Ẳ ẳ */ + 0x1eb4, 501, /* Ẵ ẵ */ + 0x1eb6, 501, /* Ặ ặ */ + 0x1eb8, 501, /* Ẹ ẹ */ + 0x1eba, 501, /* Ẻ ẻ */ + 0x1ebc, 501, /* Ẽ ẽ */ + 0x1ebe, 501, /* Ế ế */ + 0x1ec0, 501, /* Ề ề */ + 0x1ec2, 501, /* Ể ể */ + 0x1ec4, 501, /* Ễ ễ */ + 0x1ec6, 501, /* Ệ ệ */ + 0x1ec8, 501, /* Ỉ ỉ */ + 0x1eca, 501, /* Ị ị */ + 0x1ecc, 501, /* Ọ ọ */ + 0x1ece, 501, /* Ỏ ỏ */ + 0x1ed0, 501, /* Ố ố */ + 0x1ed2, 501, /* Ồ ồ */ + 0x1ed4, 501, /* Ổ ổ */ + 0x1ed6, 501, /* Ỗ ỗ */ + 0x1ed8, 501, /* Ộ ộ */ + 0x1eda, 501, /* Ớ ớ */ + 0x1edc, 501, /* Ờ ờ */ + 0x1ede, 501, /* Ở ở */ + 0x1ee0, 501, /* Ỡ ỡ */ + 0x1ee2, 501, /* Ợ ợ */ + 0x1ee4, 501, /* Ụ ụ */ + 0x1ee6, 501, /* Ủ ủ */ + 0x1ee8, 501, /* Ứ ứ */ + 0x1eea, 501, /* Ừ ừ */ + 0x1eec, 501, /* Ử ử */ + 0x1eee, 501, /* Ữ ữ */ + 0x1ef0, 501, /* Ự ự */ + 0x1ef2, 501, /* Ỳ ỳ */ + 0x1ef4, 501, /* Ỵ ỵ */ + 0x1ef6, 501, /* Ỷ ỷ */ + 0x1ef8, 501, /* Ỹ ỹ */ + 0x1f59, 492, /* Ὑ ὑ */ + 0x1f5b, 492, /* Ὓ ὓ */ + 0x1f5d, 492, /* Ὕ ὕ */ + 0x1f5f, 492, /* Ὗ ὗ */ + 0x1fbc, 491, /* ᾼ ᾳ */ + 0x1fcc, 491, /* ῌ ῃ */ + 0x1fec, 493, /* Ῥ ῥ */ + 0x1ffc, 491, /* ῼ ῳ */ +}; + +/* + * title characters are those between + * upper and lower case. ie DZ Dz dz + */ +static +Rune __totitle1[] = +{ + 0x01c4, 501, /* DŽ Dž */ + 0x01c6, 499, /* dž Dž */ + 0x01c7, 501, /* LJ Lj */ + 0x01c9, 499, /* lj Lj */ + 0x01ca, 501, /* NJ Nj */ + 0x01cc, 499, /* nj Nj */ + 0x01f1, 501, /* DZ Dz */ + 0x01f3, 499, /* dz Dz */ +}; + +static +Rune* +bsearch(Rune c, Rune *t, int n, int ne) +{ + Rune *p; + int m; + + while(n > 1) { + m = n/2; + p = t + m*ne; + if(c >= p[0]) { + t = p; + n = n-m; + } else + n = m; + } + if(n && c >= t[0]) + return t; + return 0; +} + +Rune +tolowerrune(Rune c) +{ + Rune *p; + + p = bsearch(c, __tolower2, nelem(__tolower2)/3, 3); + if(p && c >= p[0] && c <= p[1]) + return c + p[2] - 500; + p = bsearch(c, __tolower1, nelem(__tolower1)/2, 2); + if(p && c == p[0]) + return c + p[1] - 500; + return c; +} + +Rune +toupperrune(Rune c) +{ + Rune *p; + + p = bsearch(c, __toupper2, nelem(__toupper2)/3, 3); + if(p && c >= p[0] && c <= p[1]) + return c + p[2] - 500; + p = bsearch(c, __toupper1, nelem(__toupper1)/2, 2); + if(p && c == p[0]) + return c + p[1] - 500; + return c; +} + +Rune +totitlerune(Rune c) +{ + Rune *p; + + p = bsearch(c, __totitle1, nelem(__totitle1)/2, 2); + if(p && c == p[0]) + return c + p[1] - 500; + return c; +} + +int +islowerrune(Rune c) +{ + Rune *p; + + p = bsearch(c, __toupper2, nelem(__toupper2)/3, 3); + if(p && c >= p[0] && c <= p[1]) + return 1; + p = bsearch(c, __toupper1, nelem(__toupper1)/2, 2); + if(p && c == p[0]) + return 1; + return 0; +} + +int +isupperrune(Rune c) +{ + Rune *p; + + p = bsearch(c, __tolower2, nelem(__tolower2)/3, 3); + if(p && c >= p[0] && c <= p[1]) + return 1; + p = bsearch(c, __tolower1, nelem(__tolower1)/2, 2); + if(p && c == p[0]) + return 1; + return 0; +} + +int +isalpharune(Rune c) +{ + Rune *p; + + if(isupperrune(c) || islowerrune(c)) + return 1; + p = bsearch(c, __alpha2, nelem(__alpha2)/2, 2); + if(p && c >= p[0] && c <= p[1]) + return 1; + p = bsearch(c, __alpha1, nelem(__alpha1), 1); + if(p && c == p[0]) + return 1; + return 0; +} + +int +istitlerune(Rune c) +{ + return isupperrune(c) && islowerrune(c); +} + +int +isspacerune(Rune c) +{ + Rune *p; + + p = bsearch(c, __space2, nelem(__space2)/2, 2); + if(p && c >= p[0] && c <= p[1]) + return 1; + return 0; +} diff --git a/src/lib9/utf/utfdef.h b/src/lib9/utf/utfdef.h new file mode 100644 index 00000000..4b58ae87 --- /dev/null +++ b/src/lib9/utf/utfdef.h @@ -0,0 +1,14 @@ +#define uchar _utfuchar +#define ushort _utfushort +#define uint _utfuint +#define ulong _utfulong +#define vlong _utfvlong +#define uvlong _utfuvlong + +typedef unsigned char uchar; +typedef unsigned short ushort; +typedef unsigned int uint; +typedef unsigned long ulong; + +#define nelem(x) (sizeof(x)/sizeof((x)[0])) +#define nil ((void*)0) diff --git a/src/lib9/utf/utfecpy.c b/src/lib9/utf/utfecpy.c new file mode 100644 index 00000000..57159350 --- /dev/null +++ b/src/lib9/utf/utfecpy.c @@ -0,0 +1,36 @@ +/* + * The authors of this software are Rob Pike and Ken Thompson. + * Copyright (c) 2002 by Lucent Technologies. + * Permission to use, copy, modify, and distribute this software for any + * purpose without fee is hereby granted, provided that this entire notice + * is included in all copies of any software which is or includes a copy + * or modification of this software and in all copies of the supporting + * documentation for such software. + * THIS SOFTWARE IS BEING PROVIDED "AS IS", WITHOUT ANY EXPRESS OR IMPLIED + * WARRANTY. IN PARTICULAR, NEITHER THE AUTHORS NOR LUCENT TECHNOLOGIES MAKE ANY + * REPRESENTATION OR WARRANTY OF ANY KIND CONCERNING THE MERCHANTABILITY + * OF THIS SOFTWARE OR ITS FITNESS FOR ANY PARTICULAR PURPOSE. + */ +#include +#include +#include "utf.h" +#include "utfdef.h" + +char* +utfecpy(char *to, char *e, char *from) +{ + char *end; + + if(to >= e) + return to; + end = memccpy(to, from, '\0', e - to); + if(end == nil){ + end = e-1; + while(end>to && (*--end&0xC0)==0x80) + ; + *end = '\0'; + }else{ + end--; + } + return end; +} diff --git a/src/lib9/utf/utflen.c b/src/lib9/utf/utflen.c new file mode 100644 index 00000000..a10a5318 --- /dev/null +++ b/src/lib9/utf/utflen.c @@ -0,0 +1,38 @@ +/* + * The authors of this software are Rob Pike and Ken Thompson. + * Copyright (c) 2002 by Lucent Technologies. + * Permission to use, copy, modify, and distribute this software for any + * purpose without fee is hereby granted, provided that this entire notice + * is included in all copies of any software which is or includes a copy + * or modification of this software and in all copies of the supporting + * documentation for such software. + * THIS SOFTWARE IS BEING PROVIDED "AS IS", WITHOUT ANY EXPRESS OR IMPLIED + * WARRANTY. IN PARTICULAR, NEITHER THE AUTHORS NOR LUCENT TECHNOLOGIES MAKE ANY + * REPRESENTATION OR WARRANTY OF ANY KIND CONCERNING THE MERCHANTABILITY + * OF THIS SOFTWARE OR ITS FITNESS FOR ANY PARTICULAR PURPOSE. + */ +#include +#include +#include "utf.h" +#include "utfdef.h" + +int +utflen(char *s) +{ + int c; + long n; + Rune rune; + + n = 0; + for(;;) { + c = *(uchar*)s; + if(c < Runeself) { + if(c == 0) + return n; + s++; + } else + s += chartorune(&rune, s); + n++; + } + return 0; +} diff --git a/src/lib9/utf/utfnlen.c b/src/lib9/utf/utfnlen.c new file mode 100644 index 00000000..66d674fd --- /dev/null +++ b/src/lib9/utf/utfnlen.c @@ -0,0 +1,41 @@ +/* + * The authors of this software are Rob Pike and Ken Thompson. + * Copyright (c) 2002 by Lucent Technologies. + * Permission to use, copy, modify, and distribute this software for any + * purpose without fee is hereby granted, provided that this entire notice + * is included in all copies of any software which is or includes a copy + * or modification of this software and in all copies of the supporting + * documentation for such software. + * THIS SOFTWARE IS BEING PROVIDED "AS IS", WITHOUT ANY EXPRESS OR IMPLIED + * WARRANTY. IN PARTICULAR, NEITHER THE AUTHORS NOR LUCENT TECHNOLOGIES MAKE ANY + * REPRESENTATION OR WARRANTY OF ANY KIND CONCERNING THE MERCHANTABILITY + * OF THIS SOFTWARE OR ITS FITNESS FOR ANY PARTICULAR PURPOSE. + */ +#include +#include +#include "utf.h" +#include "utfdef.h" + +int +utfnlen(char *s, long m) +{ + int c; + long n; + Rune rune; + char *es; + + es = s + m; + for(n = 0; s < es; n++) { + c = *(uchar*)s; + if(c < Runeself){ + if(c == '\0') + break; + s++; + continue; + } + if(!fullrune(s, es-s)) + break; + s += chartorune(&rune, s); + } + return n; +} diff --git a/src/lib9/utf/utfrrune.c b/src/lib9/utf/utfrrune.c new file mode 100644 index 00000000..37d2a725 --- /dev/null +++ b/src/lib9/utf/utfrrune.c @@ -0,0 +1,46 @@ +/* + * The authors of this software are Rob Pike and Ken Thompson. + * Copyright (c) 2002 by Lucent Technologies. + * Permission to use, copy, modify, and distribute this software for any + * purpose without fee is hereby granted, provided that this entire notice + * is included in all copies of any software which is or includes a copy + * or modification of this software and in all copies of the supporting + * documentation for such software. + * THIS SOFTWARE IS BEING PROVIDED "AS IS", WITHOUT ANY EXPRESS OR IMPLIED + * WARRANTY. IN PARTICULAR, NEITHER THE AUTHORS NOR LUCENT TECHNOLOGIES MAKE ANY + * REPRESENTATION OR WARRANTY OF ANY KIND CONCERNING THE MERCHANTABILITY + * OF THIS SOFTWARE OR ITS FITNESS FOR ANY PARTICULAR PURPOSE. + */ +#include +#include +#include "utf.h" +#include "utfdef.h" + +char* +utfrrune(char *s, long c) +{ + long c1; + Rune r; + char *s1; + + if(c < Runesync) /* not part of utf sequence */ + return strrchr(s, c); + + s1 = 0; + for(;;) { + c1 = *(uchar*)s; + if(c1 < Runeself) { /* one byte rune */ + if(c1 == 0) + return s1; + if(c1 == c) + s1 = s; + s++; + continue; + } + c1 = chartorune(&r, s); + if(r == c) + s1 = s; + s += c1; + } + return 0; +} diff --git a/src/lib9/utf/utfrune.c b/src/lib9/utf/utfrune.c new file mode 100644 index 00000000..1a2e04c9 --- /dev/null +++ b/src/lib9/utf/utfrune.c @@ -0,0 +1,45 @@ +/* + * The authors of this software are Rob Pike and Ken Thompson. + * Copyright (c) 2002 by Lucent Technologies. + * Permission to use, copy, modify, and distribute this software for any + * purpose without fee is hereby granted, provided that this entire notice + * is included in all copies of any software which is or includes a copy + * or modification of this software and in all copies of the supporting + * documentation for such software. + * THIS SOFTWARE IS BEING PROVIDED "AS IS", WITHOUT ANY EXPRESS OR IMPLIED + * WARRANTY. IN PARTICULAR, NEITHER THE AUTHORS NOR LUCENT TECHNOLOGIES MAKE ANY + * REPRESENTATION OR WARRANTY OF ANY KIND CONCERNING THE MERCHANTABILITY + * OF THIS SOFTWARE OR ITS FITNESS FOR ANY PARTICULAR PURPOSE. + */ +#include +#include +#include "utf.h" +#include "utfdef.h" + +char* +utfrune(char *s, long c) +{ + long c1; + Rune r; + int n; + + if(c < Runesync) /* not part of utf sequence */ + return strchr(s, c); + + for(;;) { + c1 = *(uchar*)s; + if(c1 < Runeself) { /* one byte rune */ + if(c1 == 0) + return 0; + if(c1 == c) + return s; + s++; + continue; + } + n = chartorune(&r, s); + if(r == c) + return s; + s += n; + } + return 0; +} diff --git a/src/lib9/utf/utfutf.c b/src/lib9/utf/utfutf.c new file mode 100644 index 00000000..ea360361 --- /dev/null +++ b/src/lib9/utf/utfutf.c @@ -0,0 +1,41 @@ +/* + * The authors of this software are Rob Pike and Ken Thompson. + * Copyright (c) 2002 by Lucent Technologies. + * Permission to use, copy, modify, and distribute this software for any + * purpose without fee is hereby granted, provided that this entire notice + * is included in all copies of any software which is or includes a copy + * or modification of this software and in all copies of the supporting + * documentation for such software. + * THIS SOFTWARE IS BEING PROVIDED "AS IS", WITHOUT ANY EXPRESS OR IMPLIED + * WARRANTY. IN PARTICULAR, NEITHER THE AUTHORS NOR LUCENT TECHNOLOGIES MAKE ANY + * REPRESENTATION OR WARRANTY OF ANY KIND CONCERNING THE MERCHANTABILITY + * OF THIS SOFTWARE OR ITS FITNESS FOR ANY PARTICULAR PURPOSE. + */ +#include +#include +#include "utf.h" +#include "utfdef.h" + + +/* + * Return pointer to first occurrence of s2 in s1, + * 0 if none + */ +char* +utfutf(char *s1, char *s2) +{ + char *p; + long f, n1, n2; + Rune r; + + n1 = chartorune(&r, s2); + f = r; + if(f <= Runesync) /* represents self */ + return strstr(s1, s2); + + n2 = strlen(s2); + for(p=s1; p=utfrune(p, f); p+=n1) + if(strncmp(p, s2, n2) == 0) + return p; + return 0; +} -- cgit v1.2.3