=================================================================== RCS file: /cvs/mandoc/html.c,v retrieving revision 1.231 retrieving revision 1.252 diff -u -p -r1.231 -r1.252 --- mandoc/html.c 2018/05/29 02:10:10 1.231 +++ mandoc/html.c 2019/01/18 14:36:21 1.252 @@ -1,7 +1,7 @@ -/* $Id: html.c,v 1.231 2018/05/29 02:10:10 schwarze Exp $ */ +/* $Id: html.c,v 1.252 2019/01/18 14:36:21 schwarze Exp $ */ /* * Copyright (c) 2008-2011, 2014 Kristaps Dzonsons - * Copyright (c) 2011-2015, 2017, 2018 Ingo Schwarze + * Copyright (c) 2011-2015, 2017-2019 Ingo Schwarze * * Permission to use, copy, modify, and distribute this software for any * purpose with or without fee is hereby granted, provided that the above @@ -18,6 +18,7 @@ #include "config.h" #include +#include #include #include @@ -69,8 +70,6 @@ static const struct htmldata htmltags[TAG_MAX] = { {"br", HTML_NOSTACK | HTML_AUTOCLOSE | HTML_NLALL}, {"a", 0}, {"table", HTML_NLALL | HTML_INDENT}, - {"colgroup", HTML_NLALL | HTML_INDENT}, - {"col", HTML_NOSTACK | HTML_AUTOCLOSE | HTML_NLALL}, {"tr", HTML_NLALL | HTML_INDENT}, {"td", HTML_NLAROUND}, {"li", HTML_NLAROUND | HTML_INDENT}, @@ -79,6 +78,7 @@ static const struct htmldata htmltags[TAG_MAX] = { {"dl", HTML_NLALL | HTML_INDENT}, {"dt", HTML_NLAROUND}, {"dd", HTML_NLAROUND | HTML_INDENT}, + {"p", HTML_NLAROUND | HTML_INDENT}, {"pre", HTML_NLALL | HTML_NOINDENT}, {"var", 0}, {"cite", 0}, @@ -106,23 +106,9 @@ static const struct htmldata htmltags[TAG_MAX] = { {"mover", 0}, }; -static const char *const roffscales[SCALE_MAX] = { - "cm", /* SCALE_CM */ - "in", /* SCALE_IN */ - "pc", /* SCALE_PC */ - "pt", /* SCALE_PT */ - "em", /* SCALE_EM */ - "em", /* SCALE_MM */ - "ex", /* SCALE_EN */ - "ex", /* SCALE_BU */ - "em", /* SCALE_VS */ - "ex", /* SCALE_FS */ -}; - /* Avoid duplicate HTML id= attributes. */ static struct ohash id_unique; -static void a2width(const char *, struct roffsu *); static void print_byte(struct html *, char); static void print_endword(struct html *); static void print_indent(struct html *); @@ -132,7 +118,6 @@ static void print_ctag(struct html *, struct tag *); static int print_escape(struct html *, char); static int print_encode(struct html *, const char *, const char *, int); static void print_href(struct html *, const char *, const char *, int); -static void print_metaf(struct html *, enum mandoc_esc); void * @@ -144,10 +129,15 @@ html_alloc(const struct manoutput *outopts) h->tag = NULL; h->style = outopts->style; - h->base_man = outopts->man; + if ((h->base_man1 = outopts->man) == NULL) + h->base_man2 = NULL; + else if ((h->base_man2 = strchr(h->base_man1, ';')) != NULL) + *h->base_man2++ = '\0'; h->base_includes = outopts->includes; if (outopts->fragment) h->oflags |= HTML_FRAGMENT; + if (outopts->toc) + h->oflags |= HTML_TOC; mandoc_ohash_init(&id_unique, 4, 0); @@ -220,7 +210,7 @@ print_gen_head(struct html *h) print_tagq(h, t); } -static void +void print_metaf(struct html *h, enum mandoc_esc deco) { enum htmlfont font; @@ -238,12 +228,15 @@ print_metaf(struct html *h, enum mandoc_esc deco) case ESCAPE_FONTBI: font = HTMLFONT_BI; break; + case ESCAPE_FONTCW: + font = HTMLFONT_CW; + break; case ESCAPE_FONT: case ESCAPE_FONTROMAN: font = HTMLFONT_NONE; break; default: - abort(); + return; } if (h->metaf) { @@ -265,11 +258,69 @@ print_metaf(struct html *h, enum mandoc_esc deco) h->metaf = print_otag(h, TAG_B, ""); print_otag(h, TAG_I, ""); break; + case HTMLFONT_CW: + h->metaf = print_otag(h, TAG_SPAN, "c", "Li"); + break; default: break; } } +void +html_close_paragraph(struct html *h) +{ + struct tag *t; + + for (t = h->tag; t != NULL && t->closed == 0; t = t->next) { + switch(t->tag) { + case TAG_P: + case TAG_PRE: + print_tagq(h, t); + break; + case TAG_A: + print_tagq(h, t); + continue; + default: + continue; + } + break; + } +} + +/* + * ROFF_nf switches to no-fill mode, ROFF_fi to fill mode. + * TOKEN_NONE does not switch. The old mode is returned. + */ +enum roff_tok +html_fillmode(struct html *h, enum roff_tok want) +{ + struct tag *t; + enum roff_tok had; + + for (t = h->tag; t != NULL; t = t->next) + if (t->tag == TAG_PRE) + break; + + had = t == NULL ? ROFF_fi : ROFF_nf; + + if (want != had) { + switch (want) { + case ROFF_fi: + print_tagq(h, t); + break; + case ROFF_nf: + html_close_paragraph(h); + print_otag(h, TAG_PRE, ""); + break; + case TOKEN_NONE: + break; + default: + abort(); + } + } + return had; +} + char * html_make_id(const struct roff_node *n, int unique) { @@ -326,57 +377,6 @@ html_make_id(const struct roff_node *n, int unique) return buf; } -int -html_strlen(const char *cp) -{ - size_t rsz; - int skip, sz; - - /* - * Account for escaped sequences within string length - * calculations. This follows the logic in term_strlen() as we - * must calculate the width of produced strings. - * Assume that characters are always width of "1". This is - * hacky, but it gets the job done for approximation of widths. - */ - - sz = 0; - skip = 0; - while (1) { - rsz = strcspn(cp, "\\"); - if (rsz) { - cp += rsz; - if (skip) { - skip = 0; - rsz--; - } - sz += rsz; - } - if ('\0' == *cp) - break; - cp++; - switch (mandoc_escape(&cp, NULL, NULL)) { - case ESCAPE_ERROR: - return sz; - case ESCAPE_UNICODE: - case ESCAPE_NUMBERED: - case ESCAPE_SPECIAL: - case ESCAPE_OVERSTRIKE: - if (skip) - skip = 0; - else - sz++; - break; - case ESCAPE_SKIPCHAR: - skip = 1; - break; - default: - break; - } - } - return sz; -} - static int print_escape(struct html *h, char c) { @@ -412,7 +412,6 @@ static int print_encode(struct html *h, const char *p, const char *pend, int norecurse) { char numbuf[16]; - struct tag *t; const char *seq; size_t sz; int c, len, breakline, nospace; @@ -438,9 +437,7 @@ print_encode(struct html *h, const char *p, const char if (breakline && (p >= pend || *p == ' ' || *p == ASCII_NBRSP)) { - t = print_otag(h, TAG_DIV, ""); - print_text(h, "\\~"); - print_tagq(h, t); + print_otag(h, TAG_BR, ""); breakline = 0; while (p < pend && (*p == ' ' || *p == ASCII_NBRSP)) p++; @@ -460,22 +457,25 @@ print_encode(struct html *h, const char *p, const char continue; esc = mandoc_escape(&p, &seq, &len); - if (ESCAPE_ERROR == esc) - break; - switch (esc) { case ESCAPE_FONT: case ESCAPE_FONTPREV: case ESCAPE_FONTBOLD: case ESCAPE_FONTITALIC: case ESCAPE_FONTBI: + case ESCAPE_FONTCW: case ESCAPE_FONTROMAN: - if (0 == norecurse) + if (0 == norecurse) { + h->flags |= HTML_NOSPACE; print_metaf(h, esc); + h->flags &= ~HTML_NOSPACE; + } continue; case ESCAPE_SKIPCHAR: h->flags |= HTML_SKIPCHAR; continue; + case ESCAPE_ERROR: + continue; default: break; } @@ -500,6 +500,12 @@ print_encode(struct html *h, const char *p, const char if (c <= 0) continue; break; + case ESCAPE_UNDEF: + c = *seq; + break; + case ESCAPE_DEVICE: + print_word(h, "html"); + continue; case ESCAPE_BREAK: breakline = 1; continue; @@ -531,9 +537,21 @@ print_encode(struct html *h, const char *p, const char static void print_href(struct html *h, const char *name, const char *sec, int man) { + struct stat sb; const char *p, *pp; + char *filename; - pp = man ? h->base_man : h->base_includes; + if (man) { + pp = h->base_man1; + if (h->base_man2 != NULL) { + mandoc_asprintf(&filename, "%s.%s", name, sec); + if (stat(filename, &sb) == -1) + pp = h->base_man2; + free(filename); + } + } else + pp = h->base_includes; + while ((p = strchr(pp, '%')) != NULL) { print_encode(h, pp, p, 1); if (man && p[1] == 'S') { @@ -556,13 +574,10 @@ struct tag * print_otag(struct html *h, enum htmltag tag, const char *fmt, ...) { va_list ap; - struct roffsu mysu, *su; - char numbuf[16]; struct tag *t; const char *attr; char *arg1, *arg2; - double v; - int i, have_style, tflags; + int style_written, tflags; tflags = htmltags[tag].flags; @@ -572,6 +587,8 @@ print_otag(struct html *h, enum htmltag tag, const cha t = mandoc_malloc(sizeof(struct tag)); t->tag = tag; t->next = h->tag; + t->refcnt = 0; + t->closed = 0; h->tag = t; } else t = NULL; @@ -602,17 +619,12 @@ print_otag(struct html *h, enum htmltag tag, const cha va_start(ap, fmt); - have_style = 0; - while (*fmt != '\0') { - if (*fmt == 's') { - have_style = 1; - fmt++; - break; - } + while (*fmt != '\0' && *fmt != 's') { - /* Parse a non-style attribute and its arguments. */ + /* Parse attributes and arguments. */ arg1 = va_arg(ap, char *); + arg2 = NULL; switch (*fmt++) { case 'c': attr = "class"; @@ -630,13 +642,12 @@ print_otag(struct html *h, enum htmltag tag, const cha default: abort(); } - arg2 = NULL; if (*fmt == 'M') arg2 = va_arg(ap, char *); if (arg1 == NULL) continue; - /* Print the non-style attributes. */ + /* Print the attributes. */ print_byte(h, ' '); print_word(h, attr); @@ -656,12 +667,6 @@ print_otag(struct html *h, enum htmltag tag, const cha print_encode(h, arg1, NULL, 1); fmt++; break; - case 'T': - print_encode(h, arg1, NULL, 1); - print_word(h, "\" title=\""); - print_encode(h, arg1, NULL, 1); - fmt++; - break; default: print_encode(h, arg1, NULL, 1); break; @@ -669,95 +674,24 @@ print_otag(struct html *h, enum htmltag tag, const cha print_byte(h, '"'); } - /* Print out styles. */ - - while (*fmt != '\0') { - arg1 = NULL; - su = NULL; - - /* First letter: input argument type. */ - - switch (*fmt++) { - case 'h': - i = va_arg(ap, int); - su = &mysu; - SCALE_HS_INIT(su, i); - break; - case 's': - arg1 = va_arg(ap, char *); - break; - case 'u': - su = va_arg(ap, struct roffsu *); - break; - case 'w': - if ((arg2 = va_arg(ap, char *)) != NULL) { - su = &mysu; - a2width(arg2, su); - } - if (*fmt == '+') { - if (su != NULL) { - /* Make even bold text fit. */ - su->scale *= 1.2; - /* Add padding. */ - su->scale += 3.0; - } - fmt++; - } - break; - default: - abort(); - } - - /* Second letter: style name. */ - - switch (*fmt++) { - case 'h': - attr = "height"; - break; - case 'i': - attr = "text-indent"; - break; - case 'l': - attr = "margin-left"; - break; - case 'w': - attr = "width"; - break; - case 'W': - attr = "min-width"; - break; - case '?': - attr = arg1; - arg1 = va_arg(ap, char *); - break; - default: - abort(); - } - if (su == NULL && arg1 == NULL) + style_written = 0; + while (*fmt++ == 's') { + arg1 = va_arg(ap, char *); + arg2 = va_arg(ap, char *); + if (arg2 == NULL) continue; - - if (have_style == 1) - print_word(h, " style=\""); - else - print_byte(h, ' '); - print_word(h, attr); + print_byte(h, ' '); + if (style_written == 0) { + print_word(h, "style=\""); + style_written = 1; + } + print_word(h, arg1); print_byte(h, ':'); print_byte(h, ' '); - if (su != NULL) { - v = su->scale; - if (su->unit == SCALE_MM && (v /= 100.0) == 0.0) - v = 1.0; - else if (su->unit == SCALE_BU) - v /= 24.0; - (void)snprintf(numbuf, sizeof(numbuf), "%.2f", v); - print_word(h, numbuf); - print_word(h, roffscales[su->unit]); - } else - print_word(h, arg1); + print_word(h, arg2); print_byte(h, ';'); - have_style = 2; } - if (have_style == 2) + if (style_written) print_byte(h, '"'); va_end(ap); @@ -787,33 +721,32 @@ print_ctag(struct html *h, struct tag *tag) { int tflags; - /* - * Remember to close out and nullify the current - * meta-font and table, if applicable. - */ - if (tag == h->metaf) - h->metaf = NULL; - if (tag == h->tblt) - h->tblt = NULL; + if (tag->closed == 0) { + tag->closed = 1; + if (tag == h->metaf) + h->metaf = NULL; + if (tag == h->tblt) + h->tblt = NULL; - tflags = htmltags[tag->tag].flags; - - if (tflags & HTML_INDENT) - h->indent--; - if (tflags & HTML_NOINDENT) - h->noindent--; - if (tflags & HTML_NLEND) - print_endline(h); - print_indent(h); - print_byte(h, '<'); - print_byte(h, '/'); - print_word(h, htmltags[tag->tag].name); - print_byte(h, '>'); - if (tflags & HTML_NLAFTER) - print_endline(h); - - h->tag = tag->next; - free(tag); + tflags = htmltags[tag->tag].flags; + if (tflags & HTML_INDENT) + h->indent--; + if (tflags & HTML_NOINDENT) + h->noindent--; + if (tflags & HTML_NLEND) + print_endline(h); + print_indent(h); + print_byte(h, '<'); + print_byte(h, '/'); + print_word(h, htmltags[tag->tag].name); + print_byte(h, '>'); + if (tflags & HTML_NLAFTER) + print_endline(h); + } + if (tag->refcnt == 0) { + h->tag = tag->next; + free(tag); + } } void @@ -873,6 +806,9 @@ print_text(struct html *h, const char *word) h->metaf = print_otag(h, TAG_B, ""); print_otag(h, TAG_I, ""); break; + case HTMLFONT_CW: + h->metaf = print_otag(h, TAG_SPAN, "c", "Li"); + break; default: print_indent(h); break; @@ -897,37 +833,34 @@ print_text(struct html *h, const char *word) void print_tagq(struct html *h, const struct tag *until) { - struct tag *tag; + struct tag *this, *next; - while ((tag = h->tag) != NULL) { - print_ctag(h, tag); - if (until && tag == until) - return; + for (this = h->tag; this != NULL; this = next) { + next = this == until ? NULL : this->next; + print_ctag(h, this); } } +/* + * Close out all open elements up to but excluding suntil. + * Note that a paragraph just inside stays open together with it + * because paragraphs include subsequent phrasing content. + */ void print_stagq(struct html *h, const struct tag *suntil) { - struct tag *tag; + struct tag *this, *next; - while ((tag = h->tag) != NULL) { - if (suntil && tag == suntil) - return; - print_ctag(h, tag); + for (this = h->tag; this != NULL; this = next) { + next = this->next; + if (this == suntil || (next == suntil && + (this->tag == TAG_P || this->tag == TAG_PRE))) + break; + print_ctag(h, this); } } -void -print_paragraph(struct html *h) -{ - struct tag *t; - t = print_otag(h, TAG_DIV, "c", "Pp"); - print_tagq(h, t); -} - - /*********************************************************************** * Low level output functions. * They implement line breaking using a short static buffer. @@ -1039,22 +972,4 @@ print_word(struct html *h, const char *cp) { while (*cp != '\0') print_byte(h, *cp++); -} - -/* - * Calculate the scaling unit passed in a `-width' argument. This uses - * either a native scaling unit (e.g., 1i, 2m) or the string length of - * the value. - */ -static void -a2width(const char *p, struct roffsu *su) -{ - const char *end; - - end = a2roffsu(p, su, SCALE_MAX); - if (end == NULL || *end != '\0') { - su->unit = SCALE_EN; - su->scale = html_strlen(p); - } else if (su->scale < 0.0) - su->scale = 0.0; }