=================================================================== RCS file: /cvs/texi2mdoc/main.c,v retrieving revision 1.45 retrieving revision 1.46 diff -u -p -r1.45 -r1.46 --- texi2mdoc/main.c 2015/02/25 10:01:54 1.45 +++ texi2mdoc/main.c 2015/02/25 14:37:17 1.46 @@ -1,4 +1,4 @@ -/* $Id: main.c,v 1.45 2015/02/25 10:01:54 kristaps Exp $ */ +/* $Id: main.c,v 1.46 2015/02/25 14:37:17 kristaps Exp $ */ /* * Copyright (c) 2015 Kristaps Dzonsons * @@ -38,43 +38,43 @@ static const char *const sects[SECTSZ] = { "No", }; -static void doaccent(struct texi *, enum texicmd, const char *, size_t, size_t *); -static void doblock(struct texi *, enum texicmd, const char *, size_t, size_t *); -static void dobracket(struct texi *, enum texicmd, const char *, size_t, size_t *); -static void dobye(struct texi *, enum texicmd, const char *, size_t, size_t *); -static void dodefindex(struct texi *, enum texicmd, const char *, size_t, size_t *); -static void dodefn(struct texi *, enum texicmd, const char *, size_t, size_t *); -static void dodisplay(struct texi *, enum texicmd, const char *, size_t, size_t *); -static void doend(struct texi *, enum texicmd, const char *, size_t, size_t *); -static void doenumerate(struct texi *, enum texicmd, const char *, size_t, size_t *); -static void doexample(struct texi *, enum texicmd, const char *, size_t, size_t *); -static void doignargn(struct texi *, enum texicmd, const char *, size_t, size_t *); -static void doignblock(struct texi *, enum texicmd, const char *, size_t, size_t *); -static void doignbracket(struct texi *, enum texicmd, const char *, size_t, size_t *); -static void doignline(struct texi *, enum texicmd, const char *, size_t, size_t *); -static void doinline(struct texi *, enum texicmd, const char *, size_t, size_t *); -static void doinclude(struct texi *, enum texicmd, const char *, size_t, size_t *); -static void doitem(struct texi *, enum texicmd, const char *, size_t, size_t *); -static void doitemize(struct texi *, enum texicmd, const char *, size_t, size_t *); -static void dolink(struct texi *, enum texicmd, const char *, size_t, size_t *); -static void domacro(struct texi *, enum texicmd, const char *, size_t, size_t *); -static void domath(struct texi *, enum texicmd, const char *, size_t, size_t *); -static void domultitable(struct texi *, enum texicmd, const char *, size_t, size_t *); -static void doquotation(struct texi *, enum texicmd, const char *, size_t, size_t *); -static void dotable(struct texi *, enum texicmd, const char *, size_t, size_t *); -static void dotop(struct texi *, enum texicmd, const char *, size_t, size_t *); -static void dosecoffs(struct texi *, enum texicmd, const char *, size_t, size_t *); -static void dosection(struct texi *, enum texicmd, const char *, size_t, size_t *); -static void dosp(struct texi *, enum texicmd, const char *, size_t, size_t *); -static void dosubsection(struct texi *, enum texicmd, const char *, size_t, size_t *); -static void dosubsubsection(struct texi *, enum texicmd, const char *, size_t, size_t *); -static void dosymbol(struct texi *, enum texicmd, const char *, size_t, size_t *); -static void dotab(struct texi *, enum texicmd, const char *, size_t, size_t *); -static void dotitle(struct texi *, enum texicmd, const char *, size_t, size_t *); -static void dovalue(struct texi *, enum texicmd, const char *, size_t, size_t *); -static void doverb(struct texi *, enum texicmd, const char *, size_t, size_t *); -static void doverbatim(struct texi *, enum texicmd, const char *, size_t, size_t *); -static void doverbinclude(struct texi *, enum texicmd, const char *, size_t, size_t *); +static void doaccent(struct texi *, enum texicmd, size_t *); +static void doblock(struct texi *, enum texicmd, size_t *); +static void dobracket(struct texi *, enum texicmd, size_t *); +static void dobye(struct texi *, enum texicmd, size_t *); +static void dodefindex(struct texi *, enum texicmd, size_t *); +static void dodefn(struct texi *, enum texicmd, size_t *); +static void dodisplay(struct texi *, enum texicmd, size_t *); +static void doend(struct texi *, enum texicmd, size_t *); +static void doenumerate(struct texi *, enum texicmd, size_t *); +static void doexample(struct texi *, enum texicmd, size_t *); +static void doignargn(struct texi *, enum texicmd, size_t *); +static void doignblock(struct texi *, enum texicmd, size_t *); +static void doignbracket(struct texi *, enum texicmd, size_t *); +static void doignline(struct texi *, enum texicmd, size_t *); +static void doinline(struct texi *, enum texicmd, size_t *); +static void doinclude(struct texi *, enum texicmd, size_t *); +static void doitem(struct texi *, enum texicmd, size_t *); +static void doitemize(struct texi *, enum texicmd, size_t *); +static void dolink(struct texi *, enum texicmd, size_t *); +static void domacro(struct texi *, enum texicmd, size_t *); +static void domath(struct texi *, enum texicmd, size_t *); +static void domultitable(struct texi *, enum texicmd, size_t *); +static void doquotation(struct texi *, enum texicmd, size_t *); +static void dotable(struct texi *, enum texicmd, size_t *); +static void dotop(struct texi *, enum texicmd, size_t *); +static void dosecoffs(struct texi *, enum texicmd, size_t *); +static void dosection(struct texi *, enum texicmd, size_t *); +static void dosp(struct texi *, enum texicmd, size_t *); +static void dosubsection(struct texi *, enum texicmd, size_t *); +static void dosubsubsection(struct texi *, enum texicmd, size_t *); +static void dosymbol(struct texi *, enum texicmd, size_t *); +static void dotab(struct texi *, enum texicmd, size_t *); +static void dotitle(struct texi *, enum texicmd, size_t *); +static void dovalue(struct texi *, enum texicmd, size_t *); +static void doverb(struct texi *, enum texicmd, size_t *); +static void doverbatim(struct texi *, enum texicmd, size_t *); +static void doverbinclude(struct texi *, enum texicmd, size_t *); static const struct texitok __texitoks[TEXICMD__MAX] = { /* TEXICMD__BEGIN */ @@ -342,26 +342,25 @@ static const struct texitok __texitoks[TEXICMD__MAX] = const struct texitok *const texitoks = __texitoks; static void -dodefindex(struct texi *p, enum texicmd cmd, - const char *buf, size_t sz, size_t *pos) +dodefindex(struct texi *p, enum texicmd cmd, size_t *pos) { size_t start, end; char *cp; - while (*pos < sz && isws(buf[*pos])) - advance(p, buf, pos); + while (*pos < BUFSZ(p) && isws(BUF(p)[*pos])) + advance(p, pos); start = end = *pos; - while (end < sz && ! ismspace(buf[end])) + while (end < BUFSZ(p) && ! ismspace(BUF(p)[end])) end++; if (start == end) { - advanceeoln(p, buf, sz, pos, 1); + advanceeoln(p, pos, 1); return; } else if (NULL == (cp = malloc(end - start + 1))) texiabort(p, NULL); - memcpy(cp, &buf[start], end - start); + memcpy(cp, &BUF(p)[start], end - start); cp[end - start] = '\0'; p->indexs = realloc(p->indexs, @@ -373,8 +372,7 @@ dodefindex(struct texi *p, enum texicmd cmd, } static void -dodefn(struct texi *p, enum texicmd cmd, - const char *buf, size_t sz, size_t *pos) +dodefn(struct texi *p, enum texicmd cmd, size_t *pos) { const char *blk; @@ -399,8 +397,8 @@ dodefn(struct texi *p, enum texicmd cmd, if (p->ign) { NULL != blk ? - parseto(p, buf, sz, pos, blk) : - parseeoln(p, buf, sz, pos); + parseto(p, pos, blk) : + parseeoln(p, pos); return; } @@ -429,7 +427,7 @@ dodefn(struct texi *p, enum texicmd cmd, texiputchars(p, "Function"); break; default: - parselinearg(p, buf, sz, pos); + parselinearg(p, pos); break; } @@ -439,7 +437,7 @@ dodefn(struct texi *p, enum texicmd cmd, case (TEXICMD_DEFMAC): case (TEXICMD_DEFMACX): teximacroopen(p, "Dv"); - while (parselinearg(p, buf, sz, pos)) + while (parselinearg(p, pos)) /* Spin. */ ; teximacroclose(p); break; @@ -448,10 +446,10 @@ dodefn(struct texi *p, enum texicmd cmd, case (TEXICMD_DEFUN): case (TEXICMD_DEFUNX): teximacroopen(p, "Fo"); - parselinearg(p, buf, sz, pos); + parselinearg(p, pos); teximacroclose(p); teximacroopen(p, "Fa"); - while (parselinearg(p, buf, sz, pos)) + while (parselinearg(p, pos)) /* Spin. */ ; teximacroclose(p); teximacro(p, "Fc"); @@ -463,13 +461,13 @@ dodefn(struct texi *p, enum texicmd cmd, case (TEXICMD_DEFTYPEMETHOD): case (TEXICMD_DEFTYPEMETHODX): teximacroopen(p, "Ft"); - parselinearg(p, buf, sz, pos); + parselinearg(p, pos); teximacroclose(p); teximacroopen(p, "Fo"); - parselinearg(p, buf, sz, pos); + parselinearg(p, pos); teximacroclose(p); teximacroopen(p, "Fa"); - while (parselinearg(p, buf, sz, pos)) + while (parselinearg(p, pos)) /* Spin. */ ; teximacroclose(p); teximacro(p, "Fc"); @@ -481,7 +479,7 @@ dodefn(struct texi *p, enum texicmd cmd, case (TEXICMD_DEFTYPEVR): case (TEXICMD_DEFTYPEVRX): teximacroopen(p, "Vt"); - while (parselinearg(p, buf, sz, pos)) + while (parselinearg(p, pos)) /* Spin. */ ; teximacroclose(p); break; @@ -490,7 +488,7 @@ dodefn(struct texi *p, enum texicmd cmd, case (TEXICMD_DEFVR): case (TEXICMD_DEFVRX): teximacroopen(p, "Va"); - while (parselinearg(p, buf, sz, pos)) + while (parselinearg(p, pos)) /* Spin. */ ; teximacroclose(p); break; @@ -500,12 +498,11 @@ dodefn(struct texi *p, enum texicmd cmd, texivspace(p); if (NULL != blk) - parseto(p, buf, sz, pos, blk); + parseto(p, pos, blk); } static void -domacro(struct texi *p, enum texicmd cmd, - const char *buf, size_t sz, size_t *pos) +domacro(struct texi *p, enum texicmd cmd, size_t *pos) { size_t start, end, endtoksz, len; struct teximacro m; @@ -513,42 +510,58 @@ domacro(struct texi *p, enum texicmd cmd, memset(&m, 0, sizeof(struct teximacro)); - while (*pos < sz && isws(buf[*pos])) - advance(p, buf, pos); + while (*pos < BUFSZ(p) && isws(BUF(p)[*pos])) + advance(p, pos); - for (start = end = *pos; end < sz; end++) - if (ismspace(buf[end]) || '{' == buf[end]) + for (start = end = *pos; end < BUFSZ(p); end++) + if (ismspace(BUF(p)[end]) || '{' == BUF(p)[end]) break; if (start == end) texierr(p, "zero-length macro name"); - advanceto(p, buf, pos, end); + advanceto(p, pos, end); m.key = malloc(end - start + 1); if (NULL == m.key) texiabort(p, NULL); - memcpy(m.key, &buf[start], end - start); + memcpy(m.key, &BUF(p)[start], end - start); m.key[end - start] = '\0'; - m.args = argparse(p, buf, sz, pos, &m.argsz, 0); - advanceeoln(p, buf, sz, pos, 0); + m.args = argparse(p, pos, &m.argsz, 0); + /* Note: we advance to the beginning of the macro. */ + advanceeoln(p, pos, 1); + + /* + * According to the Texinfo manual, the macro ends on the + * newline subsequent the @end macro. + * That's COMPLETELY FUCKING WRONG. + * It ends inclusive the newline, which is why so many macros + * say things like @r{hello}@c, where the subsequent @c swallows + * the newline. + * However, it does swallow the leading newline, so look for the + * @end macro without the leading newline else we might look + * past empty macros. + */ start = *pos; - endtok = "\n@end macro\n"; + endtok = "@end macro\n"; endtoksz = strlen(endtok); - blk = memmem(&buf[start], sz - start, endtok, endtoksz); + blk = memmem(&BUF(p)[start], BUFSZ(p) - start, endtok, endtoksz); if (NULL == blk) texierr(p, "unterminated macro body"); - while (&buf[*pos] != blk) - advance(p, buf, pos); - assert('\n' == buf[*pos]); - advance(p, buf, pos); - len = blk - &buf[start]; + /* Roll us back one character. */ + while (&BUF(p)[*pos] != blk) + advance(p, pos); + assert('@' == BUF(p)[*pos]); + if ('\n' != BUF(p)[*pos - 1]) + texierr(p, "cannot handle @end macro in-line"); + + len = blk - &BUF(p)[start]; m.value = malloc(len + 1); if (NULL == m.value) texiabort(p, NULL); - memcpy(m.value, &buf[start], len); + memcpy(m.value, &BUF(p)[start], len); m.value[len] = '\0'; p->macros = realloc @@ -559,12 +572,11 @@ domacro(struct texi *p, enum texicmd cmd, texiabort(p, NULL); p->macros[p->macrosz++] = m; - advanceeoln(p, buf, sz, pos, 1); + advanceeoln(p, pos, 1); } static void -doignblock(struct texi *p, enum texicmd cmd, - const char *buf, size_t sz, size_t *pos) +doignblock(struct texi *p, enum texicmd cmd, size_t *pos) { char end[32], start[32]; const char *endt, *startt; @@ -599,44 +611,42 @@ doignblock(struct texi *p, enum texicmd cmd, * then there'll be an "end" token of the same kind between us. * Thus, we keep track of scopes for matching "end" blocks. */ - while (stack > 0 && *pos < sz) { + while (stack > 0 && *pos < BUFSZ(p)) { if (stack > 10) abort(); - endt = memmem(&buf[*pos], sz - *pos, end, esz); - startt = memmem(&buf[*pos], sz - *pos, start, ssz); + endt = memmem(&BUF(p)[*pos], BUFSZ(p) - *pos, end, esz); + startt = memmem(&BUF(p)[*pos], BUFSZ(p) - *pos, start, ssz); if (NULL == endt) { texiwarn(p, "unterminated \"%s\" " "block", texitoks[cmd].tok); - *pos = sz; + *pos = BUFSZ(p); break; } newpos = *pos; if (NULL == startt || startt > endt) { - newpos += esz + (size_t)(endt - &buf[*pos]); + newpos += esz + (size_t)(endt - &BUF(p)[*pos]); stack--; } else { - newpos += ssz + (size_t)(startt - &buf[*pos]); + newpos += ssz + (size_t)(startt - &BUF(p)[*pos]); stack++; } - assert(newpos <= sz); + assert(newpos <= BUFSZ(p)); while (*pos < newpos) - advance(p, buf, pos); + advance(p, pos); } } static void -doblock(struct texi *p, enum texicmd cmd, - const char *buf, size_t sz, size_t *pos) +doblock(struct texi *p, enum texicmd cmd, size_t *pos) { - parseto(p, buf, sz, pos, texitoks[cmd].tok); + parseto(p, pos, texitoks[cmd].tok); } static void -doinline(struct texi *p, enum texicmd cmd, - const char *buf, size_t sz, size_t *pos) +doinline(struct texi *p, enum texicmd cmd, size_t *pos) { const char *macro = NULL; @@ -678,59 +688,57 @@ doinline(struct texi *p, enum texicmd cmd, } if (NULL == macro || p->literal || TEXILIST_TABLE == p->list) { - parsebracket(p, buf, sz, pos); + parsebracket(p, pos); return; } teximacroopen(p, macro); p->seenws = 0; - parsebracket(p, buf, sz, pos); - texipunctuate(p, buf, sz, pos); + parsebracket(p, pos); + texipunctuate(p, pos); teximacroclose(p); } static void -doverb(struct texi *p, enum texicmd cmd, - const char *buf, size_t sz, size_t *pos) +doverb(struct texi *p, enum texicmd cmd, size_t *pos) { char delim; size_t start; - while (*pos < sz && isws(buf[*pos])) - advance(p, buf, pos); - if (*pos == sz || '{' != buf[*pos]) + while (*pos < BUFSZ(p) && isws(BUF(p)[*pos])) + advance(p, pos); + if (*pos == BUFSZ(p) || '{' != BUF(p)[*pos]) return; - advance(p, buf, pos); - if (*pos == sz) + advance(p, pos); + if (*pos == BUFSZ(p)) return; - delim = buf[*pos]; - advance(p, buf, pos); + delim = BUF(p)[*pos]; + advance(p, pos); /* Make sure we flush out our initial whitespace... */ if (p->seenws && p->outcol && 0 == p->literal) texiputchar(p, ' '); p->seenws = 0; start = *pos; /* Read until we see the delimiter then end-brace. */ - while (*pos < sz - 1) { - if (buf[*pos] == delim && buf[*pos + 1] == '}') + while (*pos < BUFSZ(p) - 1) { + if (BUF(p)[*pos] == delim && BUF(p)[*pos + 1] == '}') break; - advance(p, buf, pos); + advance(p, pos); } - if (*pos == sz - 1) + if (*pos == BUFSZ(p) - 1) return; - texiputbuf(p, buf, start, *pos); + texiputbuf(p, start, *pos); /* Make sure we read after the end-brace. */ - assert(delim == buf[*pos]); - advance(p, buf, pos); - assert('}' == buf[*pos]); - advance(p, buf, pos); + assert(delim == BUF(p)[*pos]); + advance(p, pos); + assert('}' == BUF(p)[*pos]); + advance(p, pos); } static void -doverbatim(struct texi *p, enum texicmd cmd, - const char *buf, size_t sz, size_t *pos) +doverbatim(struct texi *p, enum texicmd cmd, size_t *pos) { const char *end, *term; size_t endsz, endpos; @@ -738,35 +746,34 @@ doverbatim(struct texi *p, enum texicmd cmd, /* We read from the @verbatim\n newline inclusive! */ end = "\n@end verbatim\n"; endsz = strlen(end); - advanceeoln(p, buf, sz, pos, 0); - if (*pos == sz) { + advanceeoln(p, pos, 0); + if (*pos == BUFSZ(p)) { texiwarn(p, "unexpected end of file"); return; } - term = memmem(&buf[*pos], sz - *pos, end, endsz); + term = memmem(&BUF(p)[*pos], BUFSZ(p) - *pos, end, endsz); if (NULL == term) { texiwarn(p, "unterminated verbatim block"); - endpos = sz; + endpos = BUFSZ(p); } else - endpos = *pos + (size_t)(term - &buf[*pos]); + endpos = *pos + (size_t)(term - &BUF(p)[*pos]); - assert(endpos <= sz); - assert('\n' == buf[*pos]); - advance(p, buf, pos); + assert(endpos <= BUFSZ(p)); + assert('\n' == BUF(p)[*pos]); + advance(p, pos); teximacro(p, "Bd -literal -offset indent"); while (*pos < endpos) { - texiputchar(p, buf[*pos]); - advance(p, buf, pos); + texiputchar(p, BUF(p)[*pos]); + advance(p, pos); } teximacro(p, "Ed"); - if (*pos < sz) - advanceto(p, buf, pos, endpos + endsz); + if (*pos < BUFSZ(p)) + advanceto(p, pos, endpos + endsz); } static void -doverbinclude(struct texi *p, enum texicmd cmd, - const char *buf, size_t sz, size_t *pos) +doverbinclude(struct texi *p, enum texicmd cmd, size_t *pos) { char fname[PATH_MAX], path[PATH_MAX]; int rc; @@ -774,22 +781,22 @@ doverbinclude(struct texi *p, enum texicmd cmd, const char *v; enum texicmd type; - while (*pos < sz && ' ' == buf[*pos]) - advance(p, buf, pos); + while (*pos < BUFSZ(p) && ' ' == BUF(p)[*pos]) + advance(p, pos); - for (i = 0; *pos < sz && '\n' != buf[*pos]; ) { + for (i = 0; *pos < BUFSZ(p) && '\n' != BUF(p)[*pos]; ) { if (i == sizeof(fname) - 1) break; - if ('@' != buf[*pos]) { - fname[i++] = buf[*pos]; - advance(p, buf, pos); + if ('@' != BUF(p)[*pos]) { + fname[i++] = BUF(p)[*pos]; + advance(p, pos); continue; } - type = texicmd(p, buf, *pos, sz, &end, NULL); - advanceto(p, buf, pos, end); + type = texicmd(p, *pos, &end, NULL); + advanceto(p, pos, end); if (TEXICMD_VALUE != type) texierr(p, "unknown verbatiminclude command"); - v = valueblookup(p, buf, sz, pos); + v = valueblookup(p, pos); if (NULL == v) continue; while ('\0' != *v) { @@ -803,7 +810,7 @@ doverbinclude(struct texi *p, enum texicmd cmd, if (i == 0) texierr(p, "path too short"); - else if ('\n' != buf[*pos]) + else if ('\n' != BUF(p)[*pos]) texierr(p, "path too long"); else if ('/' == fname[0]) texierr(p, "no absolute paths"); @@ -823,8 +830,7 @@ doverbinclude(struct texi *p, enum texicmd cmd, } static void -doinclude(struct texi *p, enum texicmd cmd, - const char *buf, size_t sz, size_t *pos) +doinclude(struct texi *p, enum texicmd cmd, size_t *pos) { char fname[PATH_MAX], path[PATH_MAX]; size_t i, end; @@ -832,23 +838,23 @@ doinclude(struct texi *p, enum texicmd cmd, const char *v; enum texicmd type; - while (*pos < sz && ' ' == buf[*pos]) - advance(p, buf, pos); + while (*pos < BUFSZ(p) && ' ' == BUF(p)[*pos]) + advance(p, pos); /* Read in the filename. */ - for (i = 0; *pos < sz && '\n' != buf[*pos]; ) { + for (i = 0; *pos < BUFSZ(p) && '\n' != BUF(p)[*pos]; ) { if (i == sizeof(fname) - 1) break; - if ('@' != buf[*pos]) { - fname[i++] = buf[*pos]; - advance(p, buf, pos); + if ('@' != BUF(p)[*pos]) { + fname[i++] = BUF(p)[*pos]; + advance(p, pos); continue; } - type = texicmd(p, buf, *pos, sz, &end, NULL); - advanceto(p, buf, pos, end); + type = texicmd(p, *pos, &end, NULL); + advanceto(p, pos, end); if (TEXICMD_VALUE != type) texierr(p, "unknown include command"); - v = valueblookup(p, buf, sz, pos); + v = valueblookup(p, pos); if (NULL == v) continue; while ('\0' != *v) { @@ -862,7 +868,7 @@ doinclude(struct texi *p, enum texicmd cmd, if (i == 0) texierr(p, "path too short"); - else if ('\n' != buf[*pos]) + else if ('\n' != BUF(p)[*pos]) texierr(p, "path too long"); else if ('/' == fname[0]) texierr(p, "no absolute paths"); @@ -889,16 +895,14 @@ doinclude(struct texi *p, enum texicmd cmd, } static void -dobracket(struct texi *p, enum texicmd cmd, - const char *buf, size_t sz, size_t *pos) +dobracket(struct texi *p, enum texicmd cmd, size_t *pos) { - parsebracket(p, buf, sz, pos); + parsebracket(p, pos); } static void -dodisplay(struct texi *p, enum texicmd cmd, - const char *buf, size_t sz, size_t *pos) +dodisplay(struct texi *p, enum texicmd cmd, size_t *pos) { switch (cmd) { @@ -913,28 +917,26 @@ dodisplay(struct texi *p, enum texicmd cmd, p->seenvs = 1; /* FIXME: ignore and parseeoln. */ - advanceeoln(p, buf, sz, pos, 1); - parseto(p, buf, sz, pos, texitoks[cmd].tok); + advanceeoln(p, pos, 1); + parseto(p, pos, texitoks[cmd].tok); teximacro(p, "Ed"); } static void -doexample(struct texi *p, enum texicmd cmd, - const char *buf, size_t sz, size_t *pos) +doexample(struct texi *p, enum texicmd cmd, size_t *pos) { teximacro(p, "Bd -literal -offset indent"); /* FIXME: ignore and parseeoln. */ - advanceeoln(p, buf, sz, pos, 1); + advanceeoln(p, pos, 1); p->literal++; - parseto(p, buf, sz, pos, texitoks[cmd].tok); + parseto(p, pos, texitoks[cmd].tok); p->literal--; teximacro(p, "Ed"); } static void -dobye(struct texi *p, enum texicmd cmd, - const char *buf, size_t sz, size_t *pos) +dobye(struct texi *p, enum texicmd cmd, size_t *pos) { texiexit(p); @@ -942,32 +944,30 @@ dobye(struct texi *p, enum texicmd cmd, } static void -dotitle(struct texi *p, enum texicmd cmd, - const char *buf, size_t sz, size_t *pos) +dotitle(struct texi *p, enum texicmd cmd, size_t *pos) { size_t start, end; - while (*pos < sz && isws(buf[*pos])) - advance(p, buf, pos); + while (*pos < BUFSZ(p) && isws(BUF(p)[*pos])) + advance(p, pos); start = end = *pos; - while (end < sz && '\n' != buf[end]) + while (end < BUFSZ(p) && '\n' != BUF(p)[end]) end++; - advanceeoln(p, buf, sz, pos, 1); + advanceeoln(p, pos, 1); free(p->subtitle); p->subtitle = malloc(end - start + 1); if (NULL == p->subtitle) texiabort(p, NULL); - memcpy(p->subtitle, &buf[start], end - start); + memcpy(p->subtitle, &BUF(p)[start], end - start); p->subtitle[end - start] = '\0'; } static void -doaccent(struct texi *p, enum texicmd cmd, - const char *buf, size_t sz, size_t *pos) +doaccent(struct texi *p, enum texicmd cmd, size_t *pos) { int brace = 0; - if (*pos == sz) { + if (*pos == BUFSZ(p)) { texiwarn(p, "truncated: @%s", texitoks[cmd].tok); return; } @@ -985,21 +985,21 @@ doaccent(struct texi *p, enum texicmd cmd, * the critical one. * Otherwise, space can sit between us and our argument. */ - if ('{' == buf[*pos]) { + if ('{' == BUF(p)[*pos]) { brace = 1; - advance(p, buf, pos); + advance(p, pos); } else if (isalpha(texitoks[cmd].tok[0])) - while (*pos < sz && isws(buf[*pos])) - advance(p, buf, pos); + while (*pos < BUFSZ(p) && isws(BUF(p)[*pos])) + advance(p, pos); - if (*pos == sz) { + if (*pos == BUFSZ(p)) { texiwarn(p, "truncated: @%s", texitoks[cmd].tok); return; } switch (cmd) { case (TEXICMD_ACUTE): - switch (buf[*pos]) { + switch (BUF(p)[*pos]) { case ('a'): case ('A'): case ('e'): case ('E'): case ('i'): case ('I'): @@ -1011,19 +1011,19 @@ doaccent(struct texi *p, enum texicmd cmd, texiwarn(p, "ignoring accent"); break; } - texiputchar(p, buf[*pos]); - advance(p, buf, pos); + texiputchar(p, BUF(p)[*pos]); + advance(p, pos); break; case (TEXICMD_CEDILLA): - if ('c' == buf[*pos] || 'C' == buf[*pos]) + if ('c' == BUF(p)[*pos] || 'C' == BUF(p)[*pos]) texiputchars(p, "\\(,"); else texiwarn(p, "ignoring accent"); - texiputchar(p, buf[*pos]); - advance(p, buf, pos); + texiputchar(p, BUF(p)[*pos]); + advance(p, pos); break; case (TEXICMD_CIRCUMFLEX): - switch (buf[*pos]) { + switch (BUF(p)[*pos]) { case ('a'): case ('A'): case ('e'): case ('E'): case ('i'): case ('I'): @@ -1035,19 +1035,19 @@ doaccent(struct texi *p, enum texicmd cmd, texiwarn(p, "ignoring accent"); break; } - texiputchar(p, buf[*pos]); - advance(p, buf, pos); + texiputchar(p, BUF(p)[*pos]); + advance(p, pos); break; case (TEXICMD_DOTLESS): - if ('i' == buf[*pos] || 'j' == buf[*pos]) + if ('i' == BUF(p)[*pos] || 'j' == BUF(p)[*pos]) texiputchars(p, "\\(."); else texiwarn(p, "ignoring accent"); - texiputchar(p, buf[*pos]); - advance(p, buf, pos); + texiputchar(p, BUF(p)[*pos]); + advance(p, pos); break; case (TEXICMD_GRAVE): - switch (buf[*pos]) { + switch (BUF(p)[*pos]) { case ('a'): case ('A'): case ('e'): case ('E'): case ('i'): case ('I'): @@ -1059,11 +1059,11 @@ doaccent(struct texi *p, enum texicmd cmd, texiwarn(p, "ignoring accent"); break; } - texiputchar(p, buf[*pos]); - advance(p, buf, pos); + texiputchar(p, BUF(p)[*pos]); + advance(p, pos); break; case (TEXICMD_TILDE): - switch (buf[*pos]) { + switch (BUF(p)[*pos]) { case ('a'): case ('A'): case ('n'): case ('N'): case ('o'): case ('O'): @@ -1073,11 +1073,11 @@ doaccent(struct texi *p, enum texicmd cmd, texiwarn(p, "ignoring accent"); break; } - texiputchar(p, buf[*pos]); - advance(p, buf, pos); + texiputchar(p, BUF(p)[*pos]); + advance(p, pos); break; case (TEXICMD_UMLAUT): - switch (buf[*pos]) { + switch (BUF(p)[*pos]) { case ('a'): case ('A'): case ('e'): case ('E'): case ('i'): case ('I'): @@ -1090,22 +1090,22 @@ doaccent(struct texi *p, enum texicmd cmd, texiwarn(p, "ignoring accent"); break; } - texiputchar(p, buf[*pos]); - advance(p, buf, pos); + texiputchar(p, BUF(p)[*pos]); + advance(p, pos); break; default: - texiputchar(p, buf[*pos]); - advance(p, buf, pos); + texiputchar(p, BUF(p)[*pos]); + advance(p, pos); break; } if (brace) { - while (*pos < sz && '}' != buf[*pos]) { - texiputchar(p, buf[*pos]); - advance(p, buf, pos); + while (*pos < BUFSZ(p) && '}' != BUF(p)[*pos]) { + texiputchar(p, BUF(p)[*pos]); + advance(p, pos); } - if (*pos < sz) - advance(p, buf, pos); + if (*pos < BUFSZ(p)) + advance(p, pos); } switch (cmd) { @@ -1121,8 +1121,7 @@ doaccent(struct texi *p, enum texicmd cmd, } static void -dosymbol(struct texi *p, enum texicmd cmd, - const char *buf, size_t sz, size_t *pos) +dosymbol(struct texi *p, enum texicmd cmd, size_t *pos) { /* Remember to pad us. */ @@ -1312,22 +1311,20 @@ dosymbol(struct texi *p, enum texicmd cmd, /* Alphabetic commands have braces we ignore. */ if (isalpha(texitoks[cmd].tok[0])) - doignbracket(p, cmd, buf, sz, pos); + doignbracket(p, cmd, pos); } static void -doquotation(struct texi *p, enum texicmd cmd, - const char *buf, size_t sz, size_t *pos) +doquotation(struct texi *p, enum texicmd cmd, size_t *pos) { teximacro(p, "Qo"); - parseto(p, buf, sz, pos, "quotation"); + parseto(p, pos, "quotation"); teximacro(p, "Qc"); } static void -domath(struct texi *p, enum texicmd cmd, - const char *buf, size_t sz, size_t *pos) +domath(struct texi *p, enum texicmd cmd, size_t *pos) { size_t nest, start; @@ -1337,102 +1334,91 @@ domath(struct texi *p, enum texicmd cmd, * terms of @-commands. * This departs from GNU's rules, but whatever. */ - while (*pos < sz && isws(buf[*pos])) - advance(p, buf, pos); - if (*pos == sz || '{' != buf[*pos]) + while (*pos < BUFSZ(p) && isws(BUF(p)[*pos])) + advance(p, pos); + if (*pos == BUFSZ(p) || '{' != BUF(p)[*pos]) return; - advance(p, buf, pos); + advance(p, pos); if (p->seenws && p->outcol && 0 == p->literal) texiputchar(p, ' '); p->seenws = 0; - for (nest = 1, start = *pos; *pos < sz && nest > 0; ) { - if ('{' == buf[*pos]) + for (nest = 1, start = *pos; *pos < BUFSZ(p) && nest > 0; ) { + if ('{' == BUF(p)[*pos]) nest++; - else if ('}' == buf[*pos]) + else if ('}' == BUF(p)[*pos]) if (0 == --nest) continue; - advance(p, buf, pos); + advance(p, pos); } - if (*pos == sz) + if (*pos == BUFSZ(p)) return; - assert('}' == buf[*pos]); - texiputbuf(p, buf, start, *pos); - advance(p, buf, pos); + assert('}' == BUF(p)[*pos]); + texiputbuf(p, start, *pos); + advance(p, pos); } static void -dovalue(struct texi *p, enum texicmd cmd, - const char *buf, size_t sz, size_t *pos) +dovalue(struct texi *p, enum texicmd cmd, size_t *pos) { - size_t start, end, i; + size_t start, end; char *key, *val; const char *cp; if (TEXICMD_SET == cmd) { - while (*pos < sz && isws(buf[*pos])) - advance(p, buf, pos); - for (start = end = *pos; end < sz; end++) - if (ismspace(buf[end])) + while (*pos < BUFSZ(p) && isws(BUF(p)[*pos])) + advance(p, pos); + for (start = end = *pos; end < BUFSZ(p); end++) + if (ismspace(BUF(p)[end])) break; /* We don't allow empty keys. */ if (start == end) return; - advanceto(p, buf, pos, end); + advanceto(p, pos, end); key = malloc(end - start + 1); if (NULL == key) texiabort(p, NULL); - memcpy(key, &buf[start], end - start); + memcpy(key, &BUF(p)[start], end - start); key[end - start] = '\0'; - while (*pos < sz && isws(buf[*pos])) - advance(p, buf, pos); - for (start = end = *pos; end < sz; end++) - if ('\n' == buf[end]) + while (*pos < BUFSZ(p) && isws(BUF(p)[*pos])) + advance(p, pos); + for (start = end = *pos; end < BUFSZ(p); end++) + if ('\n' == BUF(p)[end]) break; /* We do allow empty values. */ - advanceeoln(p, buf, sz, pos, 1); + advanceeoln(p, pos, 1); val = malloc(end - start + 1); if (NULL == val) texiabort(p, NULL); - memcpy(val, &buf[start], end - start); + memcpy(val, &BUF(p)[start], end - start); val[end - start] = '\0'; valueadd(p, key, val); } else if (TEXICMD_VALUE == cmd) { if (p->seenws) texiputchar(p, ' '); p->seenws = 0; - if (NULL != (cp = valueblookup(p, buf, sz, pos))) { - for (i = 0; i < p->valstackpos; i++) - if (cp == p->valstack[i]) - break; - if (i < p->valstackpos) - texierr(p, "recursive value"); - if (64 == p->valstackpos) - texierr(p, "too many nested values"); - p->valstack[p->valstackpos++] = cp; - parsemembuf(p, cp, strlen(cp)); - p->valstackpos--; - } else + if (NULL != (cp = valueblookup(p, pos))) + texisplice(p, cp, strlen(cp), pos); + else texiputchars(p, "{No value}"); } else if (TEXICMD_IFCLEAR == cmd) { - if (NULL != valuellookup(p, buf, sz, pos)) - doignblock(p, cmd, buf, sz, pos); + if (NULL != valuellookup(p, pos)) + doignblock(p, cmd, pos); else - parseto(p, buf, sz, pos, texitoks[cmd].tok); + parseto(p, pos, texitoks[cmd].tok); } else if (TEXICMD_IFSET == cmd) { - if (NULL == valuellookup(p, buf, sz, pos)) - doignblock(p, cmd, buf, sz, pos); + if (NULL == valuellookup(p, pos)) + doignblock(p, cmd, pos); else - parseto(p, buf, sz, pos, texitoks[cmd].tok); + parseto(p, pos, texitoks[cmd].tok); } else if (TEXICMD_CLEAR == cmd) - valuelclear(p, buf, sz, pos); + valuelclear(p, pos); } static void -dolink(struct texi *p, enum texicmd cmd, - const char *buf, size_t sz, size_t *pos) +dolink(struct texi *p, enum texicmd cmd, size_t *pos) { int c; @@ -1461,26 +1447,25 @@ dolink(struct texi *p, enum texicmd cmd, abort(); } - c = parsearg(p, buf, sz, pos, 0); + c = parsearg(p, pos, 0); p->ign++; while (c > 0) - c = parsearg(p, buf, sz, pos, 1); + c = parsearg(p, pos, 1); p->ign--; - texipunctuate(p, buf, sz, pos); + texipunctuate(p, pos); teximacroclose(p); } static void -doignargn(struct texi *p, enum texicmd cmd, - const char *buf, size_t sz, size_t *pos) +doignargn(struct texi *p, enum texicmd cmd, size_t *pos) { int c; - c = parsearg(p, buf, sz, pos, 0); + c = parsearg(p, pos, 0); p->ign++; while (c > 0) - c = parsearg(p, buf, sz, pos, 1); + c = parsearg(p, pos, 1); p->ign--; } @@ -1505,8 +1490,7 @@ sectioner(struct texi *p, int sec) } static void -dosubsubsection(struct texi *p, enum texicmd cmd, - const char *buf, size_t sz, size_t *pos) +dosubsubsection(struct texi *p, enum texicmd cmd, size_t *pos) { int sec; @@ -1515,14 +1499,13 @@ dosubsubsection(struct texi *p, enum texicmd cmd, /* We don't have a subsubsubsection, so make one up. */ texivspace(p); teximacroopen(p, sects[sec]); - parseeoln(p, buf, sz, pos); + parseeoln(p, pos); teximacroclose(p); texivspace(p); } static void -dosubsection(struct texi *p, enum texicmd cmd, - const char *buf, size_t sz, size_t *pos) +dosubsection(struct texi *p, enum texicmd cmd, size_t *pos) { int sec; @@ -1537,15 +1520,14 @@ dosubsection(struct texi *p, enum texicmd cmd, if (sec > 1) texivspace(p); teximacroopen(p, sects[sec]); - parseeoln(p, buf, sz, pos); + parseeoln(p, pos); teximacroclose(p); if (sec > 1) texivspace(p); } static void -dosecoffs(struct texi *p, enum texicmd cmd, - const char *buf, size_t sz, size_t *pos) +dosecoffs(struct texi *p, enum texicmd cmd, size_t *pos) { if (TEXICMD_RAISESECTIONS == cmd) @@ -1555,8 +1537,7 @@ dosecoffs(struct texi *p, enum texicmd cmd, } static void -dosection(struct texi *p, enum texicmd cmd, - const char *buf, size_t sz, size_t *pos) +dosection(struct texi *p, enum texicmd cmd, size_t *pos) { int sec; @@ -1583,14 +1564,13 @@ dosection(struct texi *p, enum texicmd cmd, texierr(p, "\"%s\" in a literal scope!?", sects[sec]); teximacroopen(p, sects[sec]); - parseeoln(p, buf, sz, pos); + parseeoln(p, pos); teximacroclose(p); p->seenvs = 1; } static void -dosp(struct texi *p, enum texicmd cmd, - const char *buf, size_t sz, size_t *pos) +dosp(struct texi *p, enum texicmd cmd, size_t *pos) { if (p->literal) @@ -1598,19 +1578,18 @@ dosp(struct texi *p, enum texicmd cmd, else texivspace(p); /* FIXME: ignore and parseeoln. */ - advanceeoln(p, buf, sz, pos, 1); + advanceeoln(p, pos, 1); } static void -dotop(struct texi *p, enum texicmd cmd, - const char *buf, size_t sz, size_t *pos) +dotop(struct texi *p, enum texicmd cmd, size_t *pos) { const char *cp; time_t t; char date[32]; if (--p->ign) - texierr(p, "@top command while ignoring (%d)", p->ign); + texierr(p, "@top command while ignoring"); /* * Here we print our standard mdoc(7) prologue. @@ -1644,12 +1623,11 @@ dotop(struct texi *p, enum texicmd cmd, texiputchars(p, "Unknown description"); teximacroclose(p); p->seenvs = 1; - dosection(p, cmd, buf, sz, pos); + dosection(p, cmd, pos); } static void -doitem(struct texi *p, enum texicmd cmd, - const char *buf, size_t sz, size_t *pos) +doitem(struct texi *p, enum texicmd cmd, size_t *pos) { /* Multitable is using raw tbl(7). */ @@ -1677,7 +1655,7 @@ doitem(struct texi *p, enum texicmd cmd, /* Trick so we don't start with Pp. */ p->seenvs = 1; - parseeoln(p, buf, sz, pos); + parseeoln(p, pos); if (TEXILIST_ITEM == p->list) teximacroclose(p); @@ -1686,8 +1664,7 @@ doitem(struct texi *p, enum texicmd cmd, } static void -dotab(struct texi *p, enum texicmd cmd, - const char *buf, size_t sz, size_t *pos) +dotab(struct texi *p, enum texicmd cmd, size_t *pos) { /* This command is only useful in @multitable. */ @@ -1696,8 +1673,7 @@ dotab(struct texi *p, enum texicmd cmd, } static void -domultitable(struct texi *p, enum texicmd cmd, - const char *buf, size_t sz, size_t *pos) +domultitable(struct texi *p, enum texicmd cmd, size_t *pos) { enum texilist sv = p->list; int svliteral = p->literal; @@ -1714,29 +1690,29 @@ domultitable(struct texi *p, enum texicmd cmd, columns = 0; /* Advance to the first argument... */ - while (*pos < sz && isws(buf[*pos])) - advance(p, buf, pos); + while (*pos < BUFSZ(p) && isws(BUF(p)[*pos])) + advance(p, pos); /* Make sure we don't print anything when scanning. */ p->ign++; - if ('@' == buf[*pos]) { + if ('@' == BUF(p)[*pos]) { /* * Look for @columnfractions. * We ignore these, but we do use the number of * arguments to set the number of columns that we'll * have. */ - type = texicmd(p, buf, *pos, sz, &end, NULL); - advanceto(p, buf, pos, end); + type = texicmd(p, *pos, &end, NULL); + advanceto(p, pos, end); if (TEXICMD_COLUMNFRACTIONS != type) texierr(p, "unknown multitable command"); - while (*pos < sz && '\n' != buf[*pos]) { - while (*pos < sz && isws(buf[*pos])) - advance(p, buf, pos); - while (*pos < sz && ! isws(buf[*pos])) { - if ('\n' == buf[*pos]) + while (*pos < BUFSZ(p) && '\n' != BUF(p)[*pos]) { + while (*pos < BUFSZ(p) && isws(BUF(p)[*pos])) + advance(p, pos); + while (*pos < BUFSZ(p) && ! isws(BUF(p)[*pos])) { + if ('\n' == BUF(p)[*pos]) break; - advance(p, buf, pos); + advance(p, pos); } columns++; } @@ -1747,7 +1723,7 @@ domultitable(struct texi *p, enum texicmd cmd, * tbl(7) figure it out. * So use this only to count arguments. */ - while (parselinearg(p, buf, sz, pos) > 0) + while (parselinearg(p, pos) > 0) columns++; p->ign--; @@ -1759,7 +1735,7 @@ domultitable(struct texi *p, enum texicmd cmd, } texiputchars(p, ".\n"); p->outmacro++; - parseto(p, buf, sz, pos, texitoks[cmd].tok); + parseto(p, pos, texitoks[cmd].tok); p->outmacro--; teximacro(p, "TE"); p->literal = svliteral; @@ -1767,40 +1743,37 @@ domultitable(struct texi *p, enum texicmd cmd, } static void -dotable(struct texi *p, enum texicmd cmd, - const char *buf, size_t sz, size_t *pos) +dotable(struct texi *p, enum texicmd cmd, size_t *pos) { enum texilist sv = p->list; p->list = TEXILIST_ITEM; teximacro(p, "Bl -tag -width Ds"); /* FIXME: ignore and parseeoln. */ - advanceeoln(p, buf, sz, pos, 1); + advanceeoln(p, pos, 1); p->seenvs = 1; - parseto(p, buf, sz, pos, texitoks[cmd].tok); + parseto(p, pos, texitoks[cmd].tok); teximacro(p, "El"); p->list = sv; } static void -doend(struct texi *p, enum texicmd cmd, - const char *buf, size_t sz, size_t *pos) +doend(struct texi *p, enum texicmd cmd, size_t *pos) { size_t start; - while (*pos < sz && isws(buf[*pos])) - advance(p, buf, pos); + while (*pos < BUFSZ(p) && isws(BUF(p)[*pos])) + advance(p, pos); start = *pos; - while (*pos < sz && '\n' != buf[*pos]) - advance(p, buf, pos); + while (*pos < BUFSZ(p) && '\n' != BUF(p)[*pos]) + advance(p, pos); - texiwarn(p, "unexpected \"end\": %.*s", (int)(*pos - start), &buf[start]); - advanceeoln(p, buf, sz, pos, 1); + texiwarn(p, "unexpected \"end\": %.*s", (int)(*pos - start), &BUF(p)[start]); + advanceeoln(p, pos, 1); } static void -doenumerate(struct texi *p, enum texicmd cmd, - const char *buf, size_t sz, size_t *pos) +doenumerate(struct texi *p, enum texicmd cmd, size_t *pos) { enum texilist sv = p->list; @@ -1808,15 +1781,14 @@ doenumerate(struct texi *p, enum texicmd cmd, teximacro(p, "Bl -enum"); p->seenvs = 1; /* FIXME: ignore and parseeoln. */ - advanceeoln(p, buf, sz, pos, 1); - parseto(p, buf, sz, pos, "enumerate"); + advanceeoln(p, pos, 1); + parseto(p, pos, "enumerate"); teximacro(p, "El"); p->list = sv; } static void -doitemize(struct texi *p, enum texicmd cmd, - const char *buf, size_t sz, size_t *pos) +doitemize(struct texi *p, enum texicmd cmd, size_t *pos) { enum texilist sv = p->list; @@ -1824,29 +1796,27 @@ doitemize(struct texi *p, enum texicmd cmd, teximacro(p, "Bl -bullet"); p->seenvs = 1; /* FIXME: ignore and parseeoln. */ - advanceeoln(p, buf, sz, pos, 1); - parseto(p, buf, sz, pos, "itemize"); + advanceeoln(p, pos, 1); + parseto(p, pos, "itemize"); teximacro(p, "El"); p->list = sv; } static void -doignbracket(struct texi *p, enum texicmd cmd, - const char *buf, size_t sz, size_t *pos) +doignbracket(struct texi *p, enum texicmd cmd, size_t *pos) { p->ign++; - parsebracket(p, buf, sz, pos); + parsebracket(p, pos); p->ign--; } static void -doignline(struct texi *p, enum texicmd cmd, - const char *buf, size_t sz, size_t *pos) +doignline(struct texi *p, enum texicmd cmd, size_t *pos) { /* FIXME: ignore and parseeoln. */ - advanceeoln(p, buf, sz, pos, 1); + advanceeoln(p, pos, 1); } /*