=================================================================== RCS file: /cvs/texi2mdoc/util.c,v retrieving revision 1.25 retrieving revision 1.29 diff -u -p -r1.25 -r1.29 --- texi2mdoc/util.c 2015/03/03 22:37:24 1.25 +++ texi2mdoc/util.c 2015/03/07 11:53:21 1.29 @@ -1,4 +1,4 @@ -/* $Id: util.c,v 1.25 2015/03/03 22:37:24 kristaps Exp $ */ +/* $Id: util.c,v 1.29 2015/03/07 11:53:21 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()). */ @@ -100,6 +138,7 @@ texiexit(struct texi *p) free(p->dirs); free(p->subtitle); free(p->title); + free(p->copying); } /* @@ -190,7 +229,6 @@ texiputchar(struct texi *p, char c) fputc(c, p->outfile); if ('\\' == c) fputc('e', p->outfile); - p->seenvs = 0; if ('\n' == c) { p->outcol = 0; p->seenws = 0; @@ -219,7 +257,6 @@ texiputchars(struct texi *p, const char *s) ((unsigned int)*s), p->outfile); else p->outcol += fputs(s, p->outfile); - p->seenvs = 0; } /* @@ -250,6 +287,7 @@ teximacroclose(struct texi *p) fputc('\n', p->outfile); p->outcol = p->seenws = 0; } + p->seenvs = 0; } /* @@ -283,6 +321,7 @@ teximacroopen(struct texi *p, const char *s) p->outcol++; p->outmacro++; p->seenws = 0; + p->seenvs = 0; } /* @@ -307,6 +346,7 @@ teximacro(struct texi *p, const char *s) fputs(s, p->outfile); fputc('\n', p->outfile); p->outcol = p->seenws = 0; + p->seenvs = 0; } /* @@ -316,10 +356,8 @@ void texivspace(struct texi *p) { - if (p->seenvs || TEXILIST_TABLE == p->list) - return; - teximacro(p, "Pp"); - p->seenvs = 1; + if (TEXILIST_TABLE != p->list) + teximacro(p, "Pp"); } /* @@ -410,13 +448,6 @@ 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); advance(p, pos); } return(*pos); @@ -568,8 +599,21 @@ 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 == p->literal && TEXILIST_TABLE != p->list) + teximacro(p, "Pp"); + + p->seenvs = 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. @@ -584,6 +628,46 @@ parseword(struct texi *p, size_t *pos, char extra) 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 ('@'): @@ -593,7 +677,26 @@ 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 (*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"); @@ -605,9 +708,15 @@ parseword(struct texi *p, size_t *pos, char extra) advance(p, pos); } else texiputchar(p, BUF(p)[*pos]); + advance(p, pos); } + if (*pos + 1 < BUFSZ(p) && + '\n' == BUF(p)[*pos] && + '\n' == BUF(p)[*pos + 1]) + p->seenvs = 1; + /* * New sentence, new line:if we (non-macro, non-literal) see a * period at the end of the last printed word, then open a @@ -1493,6 +1602,5 @@ teximdocopen(struct texi *p, size_t *pos) } else texiputchars(p, "Unknown description"); teximacroclose(p); - p->seenvs = 1; }