=================================================================== RCS file: /cvs/texi2mdoc/util.c,v retrieving revision 1.20 retrieving revision 1.31 diff -u -p -r1.20 -r1.31 --- texi2mdoc/util.c 2015/03/01 00:25:08 1.20 +++ texi2mdoc/util.c 2015/03/12 04:24:19 1.31 @@ -1,4 +1,4 @@ -/* $Id: util.c,v 1.20 2015/03/01 00:25:08 kristaps Exp $ */ +/* $Id: util.c,v 1.31 2015/03/12 04:24:19 kristaps Exp $ */ /* * Copyright (c) 2015 Kristaps Dzonsons * @@ -31,6 +31,44 @@ #include "extern.h" /* + * Table of macros. + * These ABSOLUTELY MUST BE 2 or three characters long. + */ +static const char *const mdocs[] = { + "Ap", "Dd", "Dt", "Os", + "Sh", "Ss", "Pp", "D1", + "Dl", "Bd", "Ed", "Bl", + "El", "It", "Ad", "An", + "Ar", "Cd", "Cm", "Dv", + "Er", "Ev", "Ex", "Fa", + "Fd", "Fl", "Fn", "Ft", + "Ic", "In", "Li", "Nd", + "Nm", "Op", "Ot", "Pa", + "Rv", "St", "Va", "Vt", + "Xr", "%A", "%B", "%D", + "%I", "%J", "%N", "%O", + "%P", "%R", "%T", "%V", + "Ac", "Ao", "Aq", "At", + "Bc", "Bf", "Bo", "Bq", + "Bsx", "Bx", "Db", "Dc", + "Do", "Dq", "Ec", "Ef", + "Em", "Eo", "Fx", "Ms", + "No", "Ns", "Nx", "Ox", + "Pc", "Pf", "Po", "Pq", + "Qc", "Ql", "Qo", "Qq", + "Re", "Rs", "Sc", "So", + "Sq", "Sm", "Sx", "Sy", + "Tn", "Ux", "Xc", "Xo", + "Fo", "Fc", "Oo", "Oc", + "Bk", "Ek", "Bt", "Hf", + "Fr", "Ud", "Lb", "Lp", + "Lk", "Mt", "Brq", "Bro", + "Brc", "%C", "Es", "En", + "Dx", "%Q", "br", "sp", + "%U", "Ta", "ll", NULL, + }; + +/* * Unmap the top-most file in the stack of files currently opened (that * is, nested calls to parsefile()). */ @@ -65,7 +103,121 @@ texivaluefree(struct texivalue *p) free(p->value); } +static void +texidex_free(struct texidex *p) +{ + size_t i; + + for (i = 0; i < p->indexsz; i++) + free(p->index[i].term); + + free(p->index); + free(p->name); + p->index = NULL; + p->indexsz = 0; +} + /* + * Add the text beginning at "index" and of "sz" bytes to the index + * named "tok" with name size "toksz". + * This will also output the necessary mdoc(7) to construct the index. + */ +void +texindex(struct texi *p, const char *tok, + size_t toksz, const char *index, size_t sz) +{ + size_t i, isz; +#ifdef HAVE_INDEX + char *cp; +#endif + + if (0 == sz) { + texiwarn(p, "zero-length index entry"); + return; + } + + /* Look for the index. (Must be found.) */ + for (i = 0; i < p->indexsz; i++) { + if (strlen(p->indexs[i].name) != toksz) + continue; + if (strncmp(p->indexs[i].name, tok, toksz)) + continue; + break; + } + + assert(i < p->indexsz); + isz = p->indexs[i].indexsz; + /* Reallocate index's terms. */ + p->indexs[i].index = realloc + (p->indexs[i].index, + (isz + 1) * sizeof(struct texiterm)); + if (NULL == p->indexs[i].index) + texiabort(p, NULL); + + /* Add term to term array. */ + p->indexs[i].index[isz].chapter = p->nodesz - 1; + p->indexs[i].index[isz].term = malloc(sz + 1); + if (NULL == p->indexs[i].index[isz].term) + texiabort(p, NULL); + memcpy(p->indexs[i].index[isz].term, index, sz); + p->indexs[i].index[isz].term[sz] = '\0'; + + /* Output mdoc(7) for index. */ +#ifdef HAVE_INDEX + p->seenvs = -1; + teximacroopen(p, "Ix"); + texiputchars(p, "idx"); + texiputchars(p, p->indexs[i].name); + cp = p->indexs[i].index[isz].term; + while ('\n' != *cp) { + assert('\0' != *cp); + texiputchar(p, *cp++); + } + teximacroclose(p); +#endif + p->indexs[i].indexsz++; +} + +/* + * Add an index entry named "tok" of length "sz". + * This usually consists of two letters, e.g., "cp" or "vr". + * This does nothing if the index exists or is zero-sized. + */ +void +texindex_add(struct texi *p, const char *tok, size_t sz) +{ + size_t i; + char *cp; + + if (0 == sz) + return; + + /* Make sure we don't have a duplicate. */ + for (i = 0; i < p->indexsz; i++) { + if (strlen(p->indexs[i].name) != sz) + continue; + if (strncmp(p->indexs[i].name, tok, sz)) + continue; + return; + } + + /* Reallocate indices. */ + p->indexs = realloc(p->indexs, + sizeof(struct texidex) * + (p->indexsz + 1)); + if (NULL == p->indexs) + texiabort(p, NULL); + if (NULL == (cp = malloc(sz + 1))) + texiabort(p, NULL); + memcpy(cp, tok, sz); + cp[sz] = '\0'; + p->indexs[p->indexsz].name = cp; + p->indexs[p->indexsz].index = NULL; + p->indexs[p->indexsz].indexsz = 0; + p->indexsz++; +} + +/* * Unmap all files that we're currently using and free all resources * that we've allocated during the parse. * The utility should exit(...) after this is called. @@ -90,16 +242,18 @@ texiexit(struct texi *p) for (i = 0; i < p->dirsz; i++) free(p->dirs[i]); for (i = 0; i < p->indexsz; i++) - free(p->indexs[i]); + texidex_free(&p->indexs[i]); for (i = 0; i < p->valsz; i++) texivaluefree(&p->vals[i]); + free(p->nodes); free(p->macros); free(p->vals); free(p->indexs); free(p->dirs); free(p->subtitle); free(p->title); + free(p->copying); } /* @@ -184,10 +338,12 @@ texiputchar(struct texi *p, char c) if ('\'' == c && 0 == p->outcol) fputs("\\&", p->outfile); - fputc(c, p->outfile); + if (p->uppercase) + fputc(toupper((unsigned int)c), p->outfile); + else + fputc(c, p->outfile); if ('\\' == c) fputc('e', p->outfile); - p->seenvs = 0; if ('\n' == c) { p->outcol = 0; p->seenws = 0; @@ -210,8 +366,12 @@ texiputchars(struct texi *p, const char *s) fputs("\\&", p->outfile); if ('\'' == *s && 0 == p->outcol) fputs("\\&", p->outfile); - p->outcol += fputs(s, p->outfile); - p->seenvs = 0; + if (p->uppercase) + for ( ; '\0' != *s; s++) + p->outcol += fputc(toupper + ((unsigned int)*s), p->outfile); + else + p->outcol += fputs(s, p->outfile); } /* @@ -235,7 +395,7 @@ void teximacroclose(struct texi *p) { - if (p->ign) + if (p->ign || p->literal|| TEXILIST_TABLE == p->list) return; if (0 == --p->outmacro) { @@ -255,7 +415,7 @@ teximacroopen(struct texi *p, const char *s) { int rc; - if (p->ign) + if (p->ign || p->literal|| TEXILIST_TABLE == p->list) return; if (p->outcol && 0 == p->outmacro) { @@ -263,6 +423,9 @@ teximacroopen(struct texi *p, const char *s) p->outcol = 0; } + if (p->seenvs > 0 && 0 == p->outmacro) + fputs(".Pp\n", p->outfile); + if (0 == p->outmacro) fputc('.', p->outfile); else @@ -274,7 +437,7 @@ teximacroopen(struct texi *p, const char *s) fputc(' ', p->outfile); p->outcol++; p->outmacro++; - p->seenws = 0; + p->seenws = p->seenvs = 0; } /* @@ -291,9 +454,10 @@ teximacro(struct texi *p, const char *s) texierr(p, "\"%s\" in open line scope!?", s); if (p->literal) texierr(p, "\"%s\" in a literal scope!?", s); - if (p->outcol) fputc('\n', p->outfile); + if (p->seenvs > 0) + fputs(".Pp\n", p->outfile); fputc('.', p->outfile); fputs(s, p->outfile); @@ -308,10 +472,8 @@ void texivspace(struct texi *p) { - if (p->seenvs || TEXILIST_TABLE == p->list) - return; - teximacro(p, "Pp"); - p->seenvs = 1; + if (TEXILIST_TABLE != p->list && p->seenvs >= 0) + p->seenvs = 1; } /* @@ -362,6 +524,7 @@ texipunctuate(struct texi *p, size_t *pos) case ('.'): case ('"'): case (':'): + case (';'): case ('!'): case ('?'): continue; @@ -373,7 +536,7 @@ texipunctuate(struct texi *p, size_t *pos) if (end == *pos) return; if (end + 1 == BUFSZ(p) || ' ' == BUF(p)[end] || - '\n' == BUF(p)[end]) { + '@' == BUF(p)[end] || '\n' == BUF(p)[end]) { for ( ; start < end; start++) { texiputchar(p, ' '); texiputchar(p, BUF(p)[start]); @@ -401,13 +564,9 @@ advancenext(struct texi *p, size_t *pos) while (*pos < BUFSZ(p) && ismspace(BUF(p)[*pos])) { p->seenws = 1; - /* - * If it looks like we've printed a double-line, then - * output a paragraph. - * FIXME: this is stupid. - */ - if (*pos && '\n' == BUF(p)[*pos] && '\n' == BUF(p)[*pos - 1]) - texivspace(p); + if (0 == p->seenvs && '\n' == BUF(p)[*pos]) + if (*pos + 1 < BUFSZ(p) && '\n' == BUF(p)[*pos + 1]) + p->seenvs = 1; advance(p, pos); } return(*pos); @@ -415,15 +574,18 @@ advancenext(struct texi *p, size_t *pos) /* * Advance to the EOLN in the input stream. - * NOTE: THIS SHOULD NOT BE CALLED ON BLANK TEXT, as it will read up to - * the @\n. + * This will skip over '@' markers in an effort to ignore escaped + * newlines. */ size_t advanceeoln(struct texi *p, size_t *pos, int consumenl) { - while (*pos < BUFSZ(p) && '\n' != BUF(p)[*pos]) + while (*pos < BUFSZ(p) && '\n' != BUF(p)[*pos]) { + if ('@' == BUF(p)[*pos]) + advance(p, pos); advance(p, pos); + } if (*pos < BUFSZ(p) && consumenl) advance(p, pos); return(*pos); @@ -452,8 +614,13 @@ texiexecmacro(struct texi *p, struct teximacro *m, siz const char *cp; /* Disregard empty macros. */ - if (0 == (valsz = realsz = strlen(m->value))) + if (0 == (valsz = realsz = strlen(m->value))) { + args = argparse(p, pos, &asz, m->argsz); + for (i = 0; i < asz; i++) + free(args[i]); + free(args); return; + } /* * This is important: it protect us from macros that invoke more @@ -463,7 +630,7 @@ texiexecmacro(struct texi *p, struct teximacro *m, siz * The "sv" value was initialised at the start of the macro. */ if (sv > 0) - if (++p->files[p->filepos].depth > 64) + if (++p->files[p->filepos - 1].depth > 64) texierr(p, "maximium recursive depth"); args = argparse(p, pos, &asz, m->argsz); @@ -472,7 +639,7 @@ texiexecmacro(struct texi *p, struct teximacro *m, siz aasz = asz < m->argsz ? asz : m->argsz; if (0 == aasz) { - texisplice(p, m->value, valsz, pos); + texisplice(p, m->value, valsz, *pos); return; } @@ -535,7 +702,7 @@ texiexecmacro(struct texi *p, struct teximacro *m, siz i = end; } - texisplice(p, val, strlen(val), pos); + texisplice(p, val, strlen(val), *pos); for (i = 0; i < asz; i++) free(args[i]); @@ -551,16 +718,77 @@ texiexecmacro(struct texi *p, struct teximacro *m, siz static void parseword(struct texi *p, size_t *pos, char extra) { + size_t i, end, len; + int c; + /* + * If a prior word had a terminating double-newline, then begin + * this text block with a `Pp'. + * We don't do this if we're in a literal context (we'll print + * out the newlines themselves) nor in a `TS' table. + */ + if (p->seenvs > 0 && 0 == p->literal && TEXILIST_TABLE != p->list) { + if (p->outcol > 0) + fputc('\n', p->outfile); + fputs(".Pp\n", p->outfile); + p->outcol = 0; + } + + /* + * Some line control: if we (non-macro, non-literal) already + * have more than 72 characters written to the screen, then + * output a newline before getting started. + */ if (p->seenws && 0 == p->outmacro && p->outcol > 72 && 0 == p->literal) texiputchar(p, '\n'); - /* FIXME: abstract this: we use it elsewhere. */ + + /* Usual padding in the case of seen whitespace. */ if (p->seenws && p->outcol && 0 == p->literal) texiputchar(p, ' '); p->seenws = 0; + /* + * If we're in a macro line, we might want to print text that + * happens to be the same as an mdoc(7) macro. + * Obviously, we need to escape these words. + */ + if (p->outmacro) { + end = *pos; + /* Read ahead to get the word length. */ + while (end < BUFSZ(p) && ! ismspace(BUF(p)[end])) { + switch ((c = BUF(p)[end])) { + case ('@'): + case ('}'): + case ('{'): + break; + default: + if ('\0' != extra && extra == c) + break; + end++; + continue; + } + break; + } + len = end - *pos; + /* See if we have a match. */ + for (i = 0; NULL != mdocs[i]; i++) { + /* All macros are 2 or three letters. */ + if (len < 2 || len > 3) + continue; + /* Check the macro word length. */ + if ('\0' == mdocs[i][2] && 2 != len) + continue; + else if ('\0' == mdocs[i][3] && 3 != len) + continue; + if (strncmp(mdocs[i], &BUF(p)[*pos], len)) + continue; + texiputchars(p, "\\&"); + break; + } + } + while (*pos < BUFSZ(p) && ! ismspace(BUF(p)[*pos])) { switch (BUF(p)[*pos]) { case ('@'): @@ -570,7 +798,28 @@ parseword(struct texi *p, size_t *pos, char extra) } if ('\0' != extra && BUF(p)[*pos] == extra) return; - if (*pos < BUFSZ(p) - 1 && + + if (p->literal) { + texiputchar(p, BUF(p)[*pos]); + advance(p, pos); + continue; + } + + if ('"' == BUF(p)[*pos]) { + texiputchars(p, "\\(dq"); + } else if (*pos < BUFSZ(p) - 2 && + '-' == BUF(p)[*pos] && + '-' == BUF(p)[*pos + 1] && + '-' == BUF(p)[*pos + 2]) { + texiputchars(p, "\\(em"); + advance(p, pos); + advance(p, pos); + } else if (*pos < BUFSZ(p) - 1 && + '-' == BUF(p)[*pos] && + '-' == BUF(p)[*pos + 1]) { + texiputchars(p, "\\(en"); + advance(p, pos); + } else if (*pos < BUFSZ(p) - 1 && '`' == BUF(p)[*pos] && '`' == BUF(p)[*pos + 1]) { texiputchars(p, "\\(lq"); @@ -582,8 +831,27 @@ parseword(struct texi *p, size_t *pos, char extra) advance(p, pos); } else texiputchar(p, BUF(p)[*pos]); + advance(p, pos); } + + /* + * New sentence, new line:if we (non-macro, non-literal) see a + * period at the end of the last printed word, then open a + * newline. + */ + if (0 == p->literal && 0 == p->outmacro && *pos < BUFSZ(p)) + switch (BUF(p)[*pos - 1]) { + case ('.'): + case ('!'): + case ('?'): + texiputchar(p, '\n'); + break; + default: + break; + } + + p->seenvs = 0; } /* @@ -607,7 +875,7 @@ texicmd(const struct texi *p, size_t pos, size_t *end, return(TEXICMD__MAX); /* Alphabetic commands are special. */ - if ( ! isalpha((unsigned char)BUF(p)[pos])) { + if ( ! isalpha((unsigned int)BUF(p)[pos])) { if ((*end = pos + 1) == BUFSZ(p)) return(TEXICMD__MAX); for (i = 0; i < TEXICMD__MAX; i++) { @@ -637,10 +905,10 @@ texicmd(const struct texi *p, size_t pos, size_t *end, /* Look for it in our indices. */ for (i = 0; i < p->indexsz; i++) { - toksz = strlen(p->indexs[i]); + toksz = strlen(p->indexs[i].name); if (len != 5 + toksz) continue; - if (strncmp(&BUF(p)[pos], p->indexs[i], toksz)) + if (strncmp(&BUF(p)[pos], p->indexs[i].name, toksz)) continue; if (0 == strncmp(&BUF(p)[pos + toksz], "index", 5)) return(TEXICMD_USER_INDEX); @@ -805,6 +1073,8 @@ parseeoln(struct texi *p, size_t *pos) texiwarn(p, "unexpected \"{\""); advance(p, pos); continue; + case ('\n'): + continue; case ('@'): break; default: @@ -827,6 +1097,18 @@ parseeoln(struct texi *p, size_t *pos) advance(p, pos); } +enum texicmd +peeklinecmd(const struct texi *p, size_t pos) +{ + size_t end; + + while (pos < BUFSZ(p) && isws(BUF(p)[pos])) + pos++; + if (pos == BUFSZ(p) || '@' != BUF(p)[pos]) + return(TEXICMD__MAX); + return(texicmd(p, pos, &end, NULL)); +} + /* * Peek to see if there's a command after subsequent whitespace. * If so, return the macro identifier. @@ -927,7 +1209,7 @@ parseeof(struct texi *p) } void -texisplice(struct texi *p, const char *buf, size_t sz, size_t *pos) +texisplice(struct texi *p, const char *buf, size_t sz, size_t pos) { char *cp; struct texifile *f; @@ -944,8 +1226,8 @@ texisplice(struct texi *p, const char *buf, size_t sz, } f->insplice += sz; - memmove(f->map + *pos + sz, f->map + *pos, f->mapsz - *pos); - memcpy(f->map + *pos, buf, sz); + memmove(f->map + pos + sz, f->map + pos, f->mapsz - pos); + memcpy(f->map + pos, buf, sz); f->mapsz += sz; } @@ -1011,6 +1293,9 @@ parseto(struct texi *p, size_t *pos, const char *endto if (NULL != texitoks[cmd].fp) (*texitoks[cmd].fp)(p, cmd, pos); } + + if (*pos == BUFSZ(p)) + texiwarn(p, "EOF expecting \"%s\" end\n", endtoken); } /* @@ -1369,28 +1654,37 @@ teximdocclose(struct texi *p, int last) /* Print a reference to the "top" node. */ if (p->chapnum > 1) { - snprintf(buf, sizeof(buf), "node1 7"); + texiputchars(p, "Top node,"); + snprintf(buf, sizeof(buf), "%s-1 7", p->chapters); + p->seenvs = 0; teximacroopen(p, "Xr "); texiputchars(p, buf); - texiputchars(p, " ,"); + texiputchars(p, " ;"); teximacroclose(p); } /* Print a reference to the previous node. */ if (p->chapnum > 2) { + texiputchars(p, "previous node,"); snprintf(buf, sizeof(buf), - "node%zu 7", p->chapnum - 1); + "%s-%zu 7", p->chapters, p->chapnum - 1); + p->seenvs = 0; teximacroopen(p, "Xr "); texiputchars(p, buf); if ( ! last) - texiputchars(p, " ,"); + texiputchars(p, " ;"); teximacroclose(p); } /* Print a reference to the next node. */ if ( ! last) { + if (1 == p->chapnum) + texiputchars(p, "Next node,"); + else + texiputchars(p, "next node,"); snprintf(buf, sizeof(buf), - "node%zu 7", p->chapnum + 1); + "%s-%zu 7", p->chapters, p->chapnum + 1); + p->seenvs = 0; teximacroopen(p, "Xr "); texiputchars(p, buf); teximacroclose(p); @@ -1405,7 +1699,7 @@ teximdocclose(struct texi *p, int last) * Otherwise just print the mdoc(7) prologue. */ void -teximdocopen(struct texi *p) +teximdocopen(struct texi *p, size_t *pos) { const char *cp; time_t t; @@ -1413,8 +1707,19 @@ teximdocopen(struct texi *p) char fname[PATH_MAX]; if (NULL != p->chapters) { - snprintf(fname, sizeof(fname), "%s/node%zu.7", - p->chapters, ++p->chapnum); + p->chapnum++; + snprintf(fname, sizeof(fname), "%s-%zu", + p->chapters, p->chapnum); + p->nodes = realloc(p->nodes, + (p->nodesz + 1) * sizeof(char *)); + if (NULL == p->nodes) + texiabort(p, NULL); + p->nodes[p->nodesz] = strdup(fname); + if (NULL == p->nodes[p->nodesz]) + texiabort(p, NULL); + p->nodesz++; + snprintf(fname, sizeof(fname), "%s-%zu.7", + p->chapters, p->chapnum); p->outfile = fopen(fname, "w"); if (NULL == p->outfile) texiabort(p, fname); @@ -1430,6 +1735,7 @@ teximdocopen(struct texi *p) t = time(NULL); strftime(date, sizeof(date), "%F", localtime(&t)); + p->seenvs = -1; teximacroopen(p, "Dd"); texiputchars(p, date); teximacroclose(p); @@ -1445,12 +1751,15 @@ teximdocopen(struct texi *p) texiputchar(p, *cp); teximacroclose(p); teximacroopen(p, "Nd"); - if (NULL != p->subtitle) - for (cp = p->subtitle; '\0' != *cp; cp++) - texiputchar(p, *cp); - else + /* + * The subtitle `Nd' can consist of arbitrary macros, so paste + * it and parse to the end of the line. + */ + if (NULL != p->subtitle) { + texisplice(p, p->subtitle, strlen(p->subtitle), *pos); + parseeoln(p, pos); + } else texiputchars(p, "Unknown description"); teximacroclose(p); - p->seenvs = 1; }