=================================================================== RCS file: /cvs/mandoc/html.c,v retrieving revision 1.231 retrieving revision 1.269 diff -u -p -r1.231 -r1.269 --- mandoc/html.c 2018/05/29 02:10:10 1.231 +++ mandoc/html.c 2020/04/19 15:16:56 1.269 @@ -1,7 +1,7 @@ -/* $Id: html.c,v 1.231 2018/05/29 02:10:10 schwarze Exp $ */ +/* $Id: html.c,v 1.269 2020/04/19 15:16:56 schwarze Exp $ */ /* + * Copyright (c) 2011-2015, 2017-2020 Ingo Schwarze * Copyright (c) 2008-2011, 2014 Kristaps Dzonsons - * Copyright (c) 2011-2015, 2017, 2018 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 @@ -14,10 +14,14 @@ * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE. + * + * Common functions for mandoc(1) HTML formatters. + * For use by individual formatters and by the main program. */ #include "config.h" #include +#include #include #include @@ -41,36 +45,31 @@ struct htmldata { const char *name; int flags; -#define HTML_NOSTACK (1 << 0) -#define HTML_AUTOCLOSE (1 << 1) -#define HTML_NLBEFORE (1 << 2) -#define HTML_NLBEGIN (1 << 3) -#define HTML_NLEND (1 << 4) -#define HTML_NLAFTER (1 << 5) +#define HTML_INPHRASE (1 << 0) /* Can appear in phrasing context. */ +#define HTML_TOPHRASE (1 << 1) /* Establishes phrasing context. */ +#define HTML_NOSTACK (1 << 2) /* Does not have an end tag. */ +#define HTML_NLBEFORE (1 << 3) /* Output line break before opening. */ +#define HTML_NLBEGIN (1 << 4) /* Output line break after opening. */ +#define HTML_NLEND (1 << 5) /* Output line break before closing. */ +#define HTML_NLAFTER (1 << 6) /* Output line break after closing. */ #define HTML_NLAROUND (HTML_NLBEFORE | HTML_NLAFTER) #define HTML_NLINSIDE (HTML_NLBEGIN | HTML_NLEND) #define HTML_NLALL (HTML_NLAROUND | HTML_NLINSIDE) -#define HTML_INDENT (1 << 6) -#define HTML_NOINDENT (1 << 7) +#define HTML_INDENT (1 << 7) /* Indent content by two spaces. */ +#define HTML_NOINDENT (1 << 8) /* Exception: never indent content. */ }; static const struct htmldata htmltags[TAG_MAX] = { {"html", HTML_NLALL}, {"head", HTML_NLALL | HTML_INDENT}, - {"body", HTML_NLALL}, - {"meta", HTML_NOSTACK | HTML_AUTOCLOSE | HTML_NLALL}, + {"meta", HTML_NOSTACK | HTML_NLALL}, + {"link", HTML_NOSTACK | HTML_NLALL}, + {"style", HTML_NLALL | HTML_INDENT}, {"title", HTML_NLAROUND}, + {"body", HTML_NLALL}, {"div", HTML_NLAROUND}, - {"div", 0}, - {"h1", HTML_NLAROUND}, - {"h2", HTML_NLAROUND}, - {"span", 0}, - {"link", HTML_NOSTACK | HTML_AUTOCLOSE | HTML_NLALL}, - {"br", HTML_NOSTACK | HTML_AUTOCLOSE | HTML_NLALL}, - {"a", 0}, + {"section", HTML_NLALL}, {"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}, @@ -79,15 +78,21 @@ static const struct htmldata htmltags[TAG_MAX] = { {"dl", HTML_NLALL | HTML_INDENT}, {"dt", HTML_NLAROUND}, {"dd", HTML_NLAROUND | HTML_INDENT}, - {"pre", HTML_NLALL | HTML_NOINDENT}, - {"var", 0}, - {"cite", 0}, - {"b", 0}, - {"i", 0}, - {"code", 0}, - {"small", 0}, - {"style", HTML_NLALL | HTML_INDENT}, - {"math", HTML_NLALL | HTML_INDENT}, + {"h1", HTML_TOPHRASE | HTML_NLAROUND}, + {"h2", HTML_TOPHRASE | HTML_NLAROUND}, + {"p", HTML_TOPHRASE | HTML_NLAROUND | HTML_INDENT}, + {"pre", HTML_TOPHRASE | HTML_NLALL | HTML_NOINDENT}, + {"a", HTML_INPHRASE | HTML_TOPHRASE}, + {"b", HTML_INPHRASE | HTML_TOPHRASE}, + {"cite", HTML_INPHRASE | HTML_TOPHRASE}, + {"code", HTML_INPHRASE | HTML_TOPHRASE}, + {"i", HTML_INPHRASE | HTML_TOPHRASE}, + {"small", HTML_INPHRASE | HTML_TOPHRASE}, + {"span", HTML_INPHRASE | HTML_TOPHRASE}, + {"var", HTML_INPHRASE | HTML_TOPHRASE}, + {"br", HTML_INPHRASE | HTML_NOSTACK | HTML_NLALL}, + {"mark", HTML_INPHRASE }, + {"math", HTML_INPHRASE | HTML_NLALL | HTML_INDENT}, {"mrow", 0}, {"mi", 0}, {"mn", 0}, @@ -106,23 +111,15 @@ 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. */ + +struct id_entry { + int ord; /* Ordinal number of the latest occurrence. */ + char id[]; /* The id= attribute without any ordinal suffix. */ +}; static struct ohash id_unique; -static void a2width(const char *, struct roffsu *); +static void html_reset_internal(struct html *); static void print_byte(struct html *, char); static void print_endword(struct html *); static void print_indent(struct html *); @@ -132,7 +129,7 @@ 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); +static void print_metaf(struct html *); void * @@ -144,40 +141,55 @@ 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); + mandoc_ohash_init(&id_unique, 4, offsetof(struct id_entry, id)); return h; } -void -html_free(void *p) +static void +html_reset_internal(struct html *h) { struct tag *tag; - struct html *h; - char *cp; + struct id_entry *entry; unsigned int slot; - h = (struct html *)p; while ((tag = h->tag) != NULL) { h->tag = tag->next; free(tag); } - free(h); - - cp = ohash_first(&id_unique, &slot); - while (cp != NULL) { - free(cp); - cp = ohash_next(&id_unique, &slot); + entry = ohash_first(&id_unique, &slot); + while (entry != NULL) { + free(entry); + entry = ohash_next(&id_unique, &slot); } ohash_delete(&id_unique); } void +html_reset(void *p) +{ + html_reset_internal(p); + mandoc_ohash_init(&id_unique, 4, offsetof(struct id_entry, id)); +} + +void +html_free(void *p) +{ + html_reset_internal(p); + free(p); +} + +void print_gen_head(struct html *h) { struct tag *t; @@ -200,93 +212,172 @@ print_gen_head(struct html *h) print_endline(h); print_text(h, "td.head-vol { text-align: center; }"); print_endline(h); - print_text(h, "div.Pp { margin: 1ex 0ex; }"); + print_text(h, ".Nd, .Bf, .Op { display: inline; }"); print_endline(h); - print_text(h, "div.Nd, div.Bf, div.Op { display: inline; }"); + print_text(h, ".Pa, .Ad { font-style: italic; }"); print_endline(h); - print_text(h, "span.Pa, span.Ad { font-style: italic; }"); + print_text(h, ".Ms { font-weight: bold; }"); print_endline(h); - print_text(h, "span.Ms { font-weight: bold; }"); - print_endline(h); - print_text(h, "dl.Bl-diag "); + print_text(h, ".Bl-diag "); print_byte(h, '>'); print_text(h, " dt { font-weight: bold; }"); print_endline(h); - print_text(h, "code.Nm, code.Fl, code.Cm, code.Ic, " - "code.In, code.Fd, code.Fn,"); - print_endline(h); - print_text(h, "code.Cd { font-weight: bold; " - "font-family: inherit; }"); + print_text(h, "code.Nm, .Fl, .Cm, .Ic, code.In, .Fd, .Fn, .Cd " + "{ font-weight: bold; font-family: inherit; }"); print_tagq(h, t); } -static void -print_metaf(struct html *h, enum mandoc_esc deco) +int +html_setfont(struct html *h, enum mandoc_esc font) { - enum htmlfont font; - - switch (deco) { + switch (font) { case ESCAPE_FONTPREV: font = h->metal; break; case ESCAPE_FONTITALIC: - font = HTMLFONT_ITALIC; - break; case ESCAPE_FONTBOLD: - font = HTMLFONT_BOLD; - break; case ESCAPE_FONTBI: - font = HTMLFONT_BI; + case ESCAPE_FONTCW: + case ESCAPE_FONTROMAN: break; case ESCAPE_FONT: - case ESCAPE_FONTROMAN: - font = HTMLFONT_NONE; + font = ESCAPE_FONTROMAN; break; default: - abort(); + return 0; } + h->metal = h->metac; + h->metac = font; + return 1; +} +static void +print_metaf(struct html *h) +{ if (h->metaf) { print_tagq(h, h->metaf); h->metaf = NULL; } - - h->metal = h->metac; - h->metac = font; - - switch (font) { - case HTMLFONT_ITALIC: + switch (h->metac) { + case ESCAPE_FONTITALIC: h->metaf = print_otag(h, TAG_I, ""); break; - case HTMLFONT_BOLD: + case ESCAPE_FONTBOLD: h->metaf = print_otag(h, TAG_B, ""); break; - case HTMLFONT_BI: + case ESCAPE_FONTBI: h->metaf = print_otag(h, TAG_B, ""); print_otag(h, TAG_I, ""); break; + case ESCAPE_FONTCW: + h->metaf = print_otag(h, TAG_SPAN, "c", "Li"); + break; default: break; } } +void +html_close_paragraph(struct html *h) +{ + struct tag *this, *next; + int flags; + + this = h->tag; + for (;;) { + next = this->next; + flags = htmltags[this->tag].flags; + if (flags & (HTML_INPHRASE | HTML_TOPHRASE)) + print_ctag(h, this); + if ((flags & HTML_INPHRASE) == 0) + break; + this = next; + } +} + +/* + * 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; +} + +/* + * Allocate a string to be used for the "id=" attribute of an HTML + * element and/or as a segment identifier for a URI in an element. + * The function may fail and return NULL if the node lacks text data + * to create the attribute from. + * The caller is responsible for free(3)ing the returned string. + * + * If the "unique" argument is non-zero, the "id_unique" ohash table + * is used for de-duplication. If the "unique" argument is 1, + * it is the first time the function is called for this tag and + * location, so if an ordinal suffix is needed, it is incremented. + * If the "unique" argument is 2, it is the second time the function + * is called for this tag and location, so the ordinal suffix + * remains unchanged. + */ char * html_make_id(const struct roff_node *n, int unique) { const struct roff_node *nch; - char *buf, *bufs, *cp; + struct id_entry *entry; + char *buf, *cp; + size_t len; unsigned int slot; - int suffix; - for (nch = n->child; nch != NULL; nch = nch->next) - if (nch->type != ROFFT_TEXT) - return NULL; + if (n->tag != NULL) + buf = mandoc_strdup(n->tag); + else { + switch (n->tok) { + case MDOC_Sh: + case MDOC_Ss: + case MDOC_Sx: + case MAN_SH: + case MAN_SS: + for (nch = n->child; nch != NULL; nch = nch->next) + if (nch->type != ROFFT_TEXT) + return NULL; + buf = NULL; + deroff(&buf, n); + if (buf == NULL) + return NULL; + break; + default: + if (n->child == NULL || n->child->type != ROFFT_TEXT) + return NULL; + buf = mandoc_strdup(n->child->string); + break; + } + } - buf = NULL; - deroff(&buf, n); - if (buf == NULL) - return NULL; - /* * In ID attributes, only use ASCII characters that are * permitted in URL-fragment strings according to the @@ -304,79 +395,24 @@ html_make_id(const struct roff_node *n, int unique) /* Avoid duplicate HTML id= attributes. */ - bufs = NULL; - suffix = 1; slot = ohash_qlookup(&id_unique, buf); - cp = ohash_find(&id_unique, slot); - if (cp != NULL) { - while (cp != NULL) { - free(bufs); - if (++suffix > 127) { - free(buf); - return NULL; - } - mandoc_asprintf(&bufs, "%s_%d", buf, suffix); - slot = ohash_qlookup(&id_unique, bufs); - cp = ohash_find(&id_unique, slot); - } - free(buf); - buf = bufs; + if ((entry = ohash_find(&id_unique, slot)) == NULL) { + len = strlen(buf) + 1; + entry = mandoc_malloc(sizeof(*entry) + len); + entry->ord = 1; + memcpy(entry->id, buf, len); + ohash_insert(&id_unique, slot, entry); + } else if (unique == 1) + entry->ord++; + + if (entry->ord > 1) { + cp = buf; + mandoc_asprintf(&buf, "%s_%d", cp, entry->ord); + free(cp); } - ohash_insert(&id_unique, slot, buf); 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 +448,6 @@ 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 +473,7 @@ 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 +493,26 @@ 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) - print_metaf(h, esc); + if (0 == norecurse) { + h->flags |= HTML_NOSPACE; + if (html_setfont(h, esc)) + print_metaf(h); + h->flags &= ~HTML_NOSPACE; + } continue; case ESCAPE_SKIPCHAR: h->flags |= HTML_SKIPCHAR; continue; + case ESCAPE_ERROR: + continue; default: break; } @@ -500,6 +537,12 @@ 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 +574,21 @@ 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') { @@ -556,22 +611,40 @@ 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; + /* Flow content is not allowed in phrasing context. */ + + if ((tflags & HTML_INPHRASE) == 0) { + for (t = h->tag; t != NULL; t = t->next) { + if (t->closed) + continue; + assert((htmltags[t->tag].flags & HTML_TOPHRASE) == 0); + break; + } + + /* + * Always wrap phrasing elements in a paragraph + * unless already contained in some flow container; + * never put them directly into a section. + */ + + } else if (tflags & HTML_TOPHRASE && h->tag->tag == TAG_SECTION) + print_otag(h, TAG_P, "c", "Pp"); + /* Push this tag onto the stack of open scopes. */ if ((tflags & HTML_NOSTACK) == 0) { 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 +675,12 @@ 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 +698,12 @@ 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 +723,6 @@ 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; @@ -669,102 +730,31 @@ 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); /* Accommodate for "well-formed" singleton escaping. */ - if (HTML_AUTOCLOSE & htmltags[tag].flags) + if (htmltags[tag].flags & HTML_NOSTACK) print_byte(h, '/'); print_byte(h, '>'); @@ -782,38 +772,80 @@ print_otag(struct html *h, enum htmltag tag, const cha return t; } +/* + * Print an element with an optional "id=" attribute. + * If the element has phrasing content and an "id=" attribute, + * also add a permalink: outside if it can be in phrasing context, + * inside otherwise. + */ +struct tag * +print_otag_id(struct html *h, enum htmltag elemtype, const char *cattr, + struct roff_node *n) +{ + struct roff_node *nch; + struct tag *ret, *t; + char *id, *href; + + ret = NULL; + id = href = NULL; + if (n->flags & NODE_ID) + id = html_make_id(n, 1); + if (n->flags & NODE_HREF) + href = id == NULL ? html_make_id(n, 2) : id; + if (href != NULL && htmltags[elemtype].flags & HTML_INPHRASE) + ret = print_otag(h, TAG_A, "chR", "permalink", href); + t = print_otag(h, elemtype, "ci", cattr, id); + if (ret == NULL) { + ret = t; + if (href != NULL && (nch = n->child) != NULL) { + /* man(7) is safe, it tags phrasing content only. */ + if (n->tok > MDOC_MAX || + htmltags[elemtype].flags & HTML_TOPHRASE) + nch = NULL; + else /* For mdoc(7), beware of nested blocks. */ + while (nch != NULL && nch->type == ROFFT_TEXT) + nch = nch->next; + if (nch == NULL) + print_otag(h, TAG_A, "chR", "permalink", href); + } + } + free(id); + if (id == NULL) + free(href); + return ret; +} + static void 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 @@ -852,6 +884,25 @@ print_gen_comment(struct html *h, struct roff_node *n) void print_text(struct html *h, const char *word) { + print_tagged_text(h, word, NULL); +} + +void +print_tagged_text(struct html *h, const char *word, struct roff_node *n) +{ + struct tag *t; + char *href; + + /* + * Always wrap text in a paragraph unless already contained in + * some flow container; never put it directly into a section. + */ + + if (h->tag->tag == TAG_SECTION) + print_otag(h, TAG_P, "c", "Pp"); + + /* Output whitespace before this text? */ + if (h->col && (h->flags & HTML_NOSPACE) == 0) { if ( ! (HTML_KEEP & h->flags)) { if (HTML_PREKEEP & h->flags) @@ -861,24 +912,21 @@ print_text(struct html *h, const char *word) print_word(h, " "); } - assert(NULL == h->metaf); - switch (h->metac) { - case HTMLFONT_ITALIC: - h->metaf = print_otag(h, TAG_I, ""); - break; - case HTMLFONT_BOLD: - h->metaf = print_otag(h, TAG_B, ""); - break; - case HTMLFONT_BI: - h->metaf = print_otag(h, TAG_B, ""); - print_otag(h, TAG_I, ""); - break; - default: - print_indent(h); - break; - } + /* + * Optionally switch fonts, optionally write a permalink, then + * print the text, optionally surrounded by HTML whitespace. + */ - assert(word); + assert(h->metaf == NULL); + print_metaf(h); + print_indent(h); + + if (n != NULL && (href = html_make_id(n, 2)) != NULL) { + t = print_otag(h, TAG_A, "chR", "permalink", href); + free(href); + } else + t = NULL; + if ( ! print_encode(h, word, NULL, 0)) { if ( ! (h->flags & HTML_NONOSPACE)) h->flags &= ~HTML_NOSPACE; @@ -886,10 +934,11 @@ print_text(struct html *h, const char *word) } else h->flags |= HTML_NOSPACE | HTML_NONEWLINE; - if (h->metaf) { + if (h->metaf != NULL) { print_tagq(h, h->metaf); h->metaf = NULL; - } + } else if (t != NULL) + print_tagq(h, t); h->flags &= ~HTML_IGNDELIM; } @@ -897,37 +946,34 @@ 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. @@ -1019,15 +1065,12 @@ print_indent(struct html *h) { size_t i; - if (h->col) + if (h->col || h->noindent) return; - if (h->noindent == 0) { - h->col = h->indent * 2; - for (i = 0; i < h->col; i++) - putchar(' '); - } - h->flags &= ~HTML_NOSPACE; + h->col = h->indent * 2; + for (i = 0; i < h->col; i++) + putchar(' '); } /* @@ -1039,22 +1082,4 @@ 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; }