From 7a4700f5f5afcea1580ec4fcfb628afe6beb3e6b Mon Sep 17 00:00:00 2001 From: Simon Tatham Date: Mon, 11 Dec 2006 19:43:10 +0000 Subject: Support for the MS HTML Help system in the HTML back end. As yet I don't know how to write out a .CHM directly, but I am at least able to have the HTML back end write out the three auxiliary files which enable a .CHM to be generated using the MS HTML Help compiler. [originally from svn r6991] --- bk_html.c | 517 +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++--- 1 file changed, 494 insertions(+), 23 deletions(-) (limited to 'bk_html.c') diff --git a/bk_html.c b/bk_html.c index a1b46f6..11d37d3 100644 --- a/bk_html.c +++ b/bk_html.c @@ -10,6 +10,13 @@ * sensible. Perhaps for the topmost section in the file, no * fragment should be used? (Though it should probably still be * _there_ even if unused.) + * + * - In HHK index mode: subsidiary hhk entries (as in replacing + * `foo, bar' with `foo\n\tbar') can be done by embedding + * sub-

@@ -41,10 +48,12 @@ typedef struct { int ncdepths; int address_section, visible_version_id; int leaf_contains_contents, leaf_smallest_contents; + int navlinks; char *contents_filename; char *index_filename; char *template_filename; char *single_filename; + char *chm_filename, *hhp_filename, *hhc_filename, *hhk_filename; char **template_fragments; int ntfragments; char *head_end, *body_start, *body_end, *addr_start, *addr_end; @@ -77,6 +86,13 @@ struct htmlfile { int last_fragment_number; int min_heading_depth; htmlsect *first, *last; /* first/last highest-level sections */ + /* + * The `temp' field is available for use in individual passes + * over the file list. For example, the HHK index generation + * uses it to ensure no index term references the same file + * more than once. + */ + int temp; }; struct htmlsect { @@ -92,6 +108,7 @@ typedef struct { htmlfile *head, *tail; htmlfile *single, *index; tree234 *frags; + tree234 *files; } htmlfilelist; typedef struct { @@ -127,6 +144,8 @@ typedef struct { enum { HO_NEUTRAL, HO_IN_TAG, HO_IN_EMPTY_TAG, HO_IN_TEXT } state; + int hackflags; /* used for icky .HH* stuff */ + int hacklimit; /* text size limit, again for .HH* */ /* * Stuff beyond here deals with the higher syntactic level: it * tracks how many levels of

\n"); + currdepth++; + } + while (currdepth > depth) { + fprintf(ho.fp, "

\n", + f->filename, leaf ? 11 : 1); + } + + while (currdepth > 0) { + fprintf(ho.fp, "

\n"); + + cleanup(&ho); + } + if (hhk_filename) { + htmlfile *f; + htmloutput ho; + indexentry *entry; + int i; + + /* + * First make a pass over all HTML files and set their + * `temp' fields to zero, because we're about to use them. + */ + for (f = files.head; f; f = f->next) + f->temp = 0; + + ho.fp = fopen(hhk_filename, "w"); + if (!ho.fp) + error(err_cantopenw, hhk_filename); + + ho.charset = CS_CP1252; /* as far as I know, HHK files are */ + ho.restrict_charset = CS_CP1252; /* hardwired to this charset */ + ho.cstate = charset_init_state; + ho.ver = HTML_4; /* *shrug* */ + ho.state = HO_NEUTRAL; + ho.contents_level = 0; + ho.hackflags = HO_HACK_QUOTEQUOTES; + + /* + * Magic DOCTYPE which seems to work for .HHK files. I'm + * wary of trying to change it! + */ + fprintf(ho.fp, "\n" + "\n" + "\n" + "

\n"); + + /* + * Now go through those files and re-clear the temp + * fields ready for the _next_ index term. + */ + for (j = 0; j < hi->nrefs; j++) { + htmlindexref *hr = + (htmlindexref *)hi->refs[j]->private_data; + hr->section->file->temp = 0; + } + } + } + + fprintf(ho.fp, "

\n"); + cleanup(&ho); + } + + /* * Go through and check that no index fragments were referenced * without being generated, or indeed vice versa. * @@ -1491,6 +1875,11 @@ void html_backend(paragraph *sourceform, keywordlist *keywords, } freetree234(files.frags); } + /* + * The strings in files.files are all owned by their containing + * htmlfile structures, so there's no need to free them here. + */ + freetree234(files.files); { htmlsect *sect, *tmp; sect = sects.head; @@ -1584,8 +1973,12 @@ static void html_file_section(htmlconfig *cfg, htmlfilelist *files, * we invent a fresh file and put this section at its head. * Otherwise, we put it in the same file as its parent * section. + * + * Another special value of cfg->leaf_level is -1, which + * means infinity (i.e. it's considered to always be + * greater than depth). */ - if (ldepth > cfg->leaf_level) { + if (cfg->leaf_level > 0 && ldepth > cfg->leaf_level) { /* * We know that sect->parent cannot be NULL. The only * circumstance in which it can be is if sect is at @@ -1641,7 +2034,8 @@ static htmlfile *html_new_file(htmlfilelist *list, char *filename) list->head = ret; list->tail = ret; - ret->filename = dupstr(filename); + ret->filename = html_sanitise_filename(list, dupstr(filename)); + add234(list->files, ret->filename); ret->last_fragment_number = 0; ret->min_heading_depth = INT_MAX; ret->first = ret->last = NULL; @@ -1943,8 +2337,16 @@ static void html_text_limit_internal(htmloutput *ho, wchar_t const *text, char outbuf[256]; int bytes, err; + if (ho->hackflags & (HO_HACK_QUOTEQUOTES | HO_HACK_OMITQUOTES)) + quote_quotes = TRUE; /* override the input value */ + if (maxlen > 0 && textlen > maxlen) textlen = maxlen; + if (ho->hacklimit >= 0) { + if (textlen > ho->hacklimit) + textlen = ho->hacklimit; + ho->hacklimit -= textlen; + } while (textlen > 0) { /* Scan ahead for characters we really can't display in HTML. */ @@ -1978,19 +2380,25 @@ static void html_text_limit_internal(htmloutput *ho, wchar_t const *text, * HTML. */ if (ho->fp) { - if (*text == L'<') - fprintf(ho->fp, "<"); - else if (*text == L'>') - fprintf(ho->fp, ">"); - else if (*text == L'&') - fprintf(ho->fp, "&"); - else if (*text == L'"') - fprintf(ho->fp, """); - else if (*text == L' ') { - assert(nbsp); - fprintf(ho->fp, " "); - } else - assert(!"Can't happen"); + if (*text == L'"' && (ho->hackflags & HO_HACK_OMITQUOTES)) { + fputc('\'', ho->fp); + } else if (ho->hackflags & HO_HACK_QUOTENOTHING) { + fputc(*text, ho->fp); + } else { + if (*text == L'<') + fprintf(ho->fp, "<"); + else if (*text == L'>') + fprintf(ho->fp, ">"); + else if (*text == L'&') + fprintf(ho->fp, "&"); + else if (*text == L'"') + fprintf(ho->fp, """); + else if (*text == L' ') { + assert(nbsp); + fprintf(ho->fp, " "); + } else + assert(!"Can't happen"); + } } text++, textlen--; } @@ -2162,6 +2570,69 @@ static char *html_sanitise_fragment(htmlfilelist *files, htmlfile *file, return text; } +static char *html_sanitise_filename(htmlfilelist *files, char *text) +{ + /* + * Unceremoniously rip out any character that might cause + * difficulty in some filesystem or another, or be otherwise + * inconvenient. + * + * That doesn't leave much punctuation. I permit alphanumerics + * and +-.=_ only. + */ + char *p = text, *q = text; + + while (*p) { + if ((*p>='A' && *p<='Z') || + (*p>='a' && *p<='z') || + (*p>='0' && *p<='9') || + *p=='-' || *p=='_' || *p=='+' || *p=='.' || *p=='=') + *q++ = *p; + p++; + } + *q = '\0'; + + /* If there's nothing left, make something valid up */ + if (!*text) { + static const char anonfrag[] = "anon.html"; + text = sresize(text, lenof(anonfrag), char); + strcpy(text, anonfrag); + } + + /* + * Now we check for clashes with other filenames, and adjust + * this one if necessary by appending a hyphen followed by a + * number just before the file extension (if any). + */ + { + int len, extpos; + int suffix = 1; + + p = NULL; + + while (find234(files->files, text, NULL)) { + if (!p) { + len = strlen(text); + p = text; + text = snewn(len+20, char); + + for (extpos = len; extpos > 0 && p[extpos-1] != '.'; extpos--); + if (extpos > 0) + extpos--; + else + extpos = len; + } + + sprintf(text, "%.*s-%d%s", extpos, p, ++suffix, p+extpos); + } + + if (p) + sfree(p); + } + + return text; +} + static void html_contents_entry(htmloutput *ho, int depth, htmlsect *s, htmlfile *thisfile, keywordlist *keywords, htmlconfig *cfg) -- cgit v1.1