X-Git-Url: https://git.distorted.org.uk/~mdw/sgt/halibut/blobdiff_plain/f0e51ce1f8e8c03966509153e9ff360684d4dd6f..7c8c4239f226d2dabdd84dabae9f5869d6d64f2f:/bk_pdf.c diff --git a/bk_pdf.c b/bk_pdf.c index fd1424a..f6babe5 100644 --- a/bk_pdf.c +++ b/bk_pdf.c @@ -56,12 +56,15 @@ struct objlist_Tag { static object *new_object(objlist *list); static void objtext(object *o, char const *text); static void objstream(object *o, char const *text); +static void pdf_string(void (*add)(object *, char const *), + object *, char const *); static void objref(object *o, object *dest); static void make_pages_node(object *node, object *parent, page_data *first, page_data *last, object *resources); static int make_outline(object *parent, outline_element *start, int n, int open); +static int pdf_versionid(FILE *fp, word *words); void pdf_backend(paragraph *sourceform, keywordlist *keywords, indexdata *idx, void *vdoc) { @@ -194,8 +197,9 @@ void pdf_backend(paragraph *sourceform, keywordlist *keywords, for (page = doc->pages; page; page = page->next) { object *opage, *cstr; rect *r; - text_fragment *frag; + text_fragment *frag, *frag_end; char buf[256]; + int x, y, lx, ly; opage = (object *)page->spare; /* @@ -235,24 +239,89 @@ void pdf_backend(paragraph *sourceform, keywordlist *keywords, } objstream(cstr, "BT\n"); - for (frag = page->first_text; frag; frag = frag->next) { - char *c; + /* + * PDF tracks two separate current positions: the position + * given in the `line matrix' and the position given in the + * `text matrix'. We must therefore track both as well. + * They start off at -1 (unset). + */ + lx = ly = -1; + x = y = -1; + + frag = page->first_text; + while (frag) { + /* + * For compactness, I'm going to group text fragments + * into subsequences that use the same font+size. So + * first find the end of this subsequence. + */ + for (frag_end = frag; + (frag_end && + frag_end->fe == frag->fe && + frag_end->fontsize == frag->fontsize); + frag_end = frag_end->next); + + /* + * Now select the text fragment, and prepare to display + * the text. + */ objstream(cstr, "/"); objstream(cstr, frag->fe->name); - sprintf(buf, " %d Tf 1 0 0 1 %g %g Tm (", frag->fontsize, - frag->x/4096.0, frag->y/4096.0); + sprintf(buf, " %d Tf ", frag->fontsize); objstream(cstr, buf); - for (c = frag->text; *c; c++) { - if (*c == '(' || *c == ')' || *c == '\\') - objstream(cstr, "\\"); - buf[0] = *c; - buf[1] = '\0'; + while (frag && frag != frag_end) { + /* + * Place the text position for the first piece of + * text. + */ + if (lx < 0) { + sprintf(buf, "1 0 0 1 %g %g Tm ", + frag->x/4096.0, frag->y/4096.0); + } else { + sprintf(buf, "%g %g Td ", + (frag->x - lx)/4096.0, (frag->y - ly)/4096.0); + } objstream(cstr, buf); + lx = x = frag->x; + ly = y = frag->y; + + /* + * See if we're going to use Tj (show a single + * string) or TJ (show an array of strings with + * x-spacings between them). We determine this by + * seeing if there's more than one text fragment in + * sequence with the same y-coordinate. + */ + if (frag->next && frag->next != frag_end && + frag->next->y == y) { + /* + * The TJ strategy. + */ + objstream(cstr, "["); + while (frag && frag != frag_end && frag->y == y) { + if (frag->x != x) { + sprintf(buf, "%g", + (x - frag->x) * 1000.0 / + (4096.0 * frag->fontsize)); + objstream(cstr, buf); + } + pdf_string(objstream, cstr, frag->text); + x = frag->x + frag->width; + frag = frag->next; + } + objstream(cstr, "]TJ\n"); + } else + { + /* + * The Tj strategy. + */ + pdf_string(objstream, cstr, frag->text); + objstream(cstr, "Tj\n"); + frag = frag->next; + } } - - objstream(cstr, ") Tj\n"); } objstream(cstr, "ET"); @@ -284,18 +353,9 @@ void pdf_backend(paragraph *sourceform, keywordlist *keywords, objref(annot, (object *)xr->dest.page->spare); objtext(annot, " /XYZ null null null]\n"); } else { - char *p; - - objtext(annot, "/A <<\n/Type /Action\n/S /URI\n/URI ("); - for (p = xr->dest.url; *p; p++) { - char c[2]; - c[0] = *p; - c[1] = '\0'; - if (*p == '(' || *p == ')' || *p == '\\') - objtext(annot, "\\"); - objtext(annot, c); - } - objtext(annot, ")\n>>\n"); + objtext(annot, "/A <<\n/Type /Action\n/S /URI\n/URI "); + pdf_string(objtext, annot, xr->dest.url); + objtext(annot, "\n>>\n"); } objtext(annot, ">>\n"); @@ -372,9 +432,13 @@ void pdf_backend(paragraph *sourceform, keywordlist *keywords, } /* - * Header + * Header. I'm going to put the version IDs in the header as + * well, simply in PDF comments. */ fileoff = fprintf(fp, "%%PDF-1.3\n"); + for (p = sourceform; p; p = p->next) + if (p->type == para_VersionID) + fileoff += pdf_versionid(fp, p->words); /* * Body @@ -579,78 +643,19 @@ static int pdf_convert(wchar_t *s, char **result) { return ok; } -static void pdf_rdaddwc(rdstringc *rs, word *text) { - char *c; - - for (; text; text = text->next) switch (text->type) { - case word_HyperLink: - case word_HyperEnd: - case word_UpperXref: - case word_LowerXref: - case word_XrefEnd: - case word_IndexRef: - break; - - case word_Normal: - case word_Emph: - case word_Code: - case word_WeakCode: - case word_WhiteSpace: - case word_EmphSpace: - case word_CodeSpace: - case word_WkCodeSpace: - case word_Quote: - case word_EmphQuote: - case word_CodeQuote: - case word_WkCodeQuote: - assert(text->type != word_CodeQuote && - text->type != word_WkCodeQuote); - if (towordstyle(text->type) == word_Emph && - (attraux(text->aux) == attr_First || - attraux(text->aux) == attr_Only)) - rdaddc(rs, '_'); /* FIXME: configurability */ - else if (towordstyle(text->type) == word_Code && - (attraux(text->aux) == attr_First || - attraux(text->aux) == attr_Only)) - rdaddc(rs, '\''); /* FIXME: configurability */ - if (removeattr(text->type) == word_Normal) { - if (pdf_convert(text->text, &c)) - rdaddsc(rs, c); - else - pdf_rdaddwc(rs, text->alt); - sfree(c); - } else if (removeattr(text->type) == word_WhiteSpace) { - rdaddc(rs, ' '); - } else if (removeattr(text->type) == word_Quote) { - rdaddc(rs, '\''); /* FIXME: configurability */ - } - if (towordstyle(text->type) == word_Emph && - (attraux(text->aux) == attr_Last || - attraux(text->aux) == attr_Only)) - rdaddc(rs, '_'); /* FIXME: configurability */ - else if (towordstyle(text->type) == word_Code && - (attraux(text->aux) == attr_Last || - attraux(text->aux) == attr_Only)) - rdaddc(rs, '\''); /* FIXME: configurability */ - break; - } -} - static int make_outline(object *parent, outline_element *items, int n, int open) { int level, totalcount = 0; outline_element *itemp; object *curr, *prev = NULL, *first = NULL, *last = NULL; - para_data *pdata; assert(n > 0); level = items->level; while (n > 0) { - rdstringc rs = {0, 0, NULL}; - char *p; + char *title; /* * Here we expect to be sitting on an item at the given @@ -659,33 +664,18 @@ static int make_outline(object *parent, outline_element *items, int n, */ assert(items->level == level); - if (level == 1 && items->para->kwtext) { - pdf_rdaddwc(&rs, items->para->kwtext); - rdaddsc(&rs, ": "); - } else if (level > 1 && items->para->kwtext2) { - pdf_rdaddwc(&rs, items->para->kwtext2); - rdaddsc(&rs, " "); - } - pdf_rdaddwc(&rs, items->para->words); + pdf_convert(items->pdata->outline_title, &title); totalcount++; curr = new_object(parent->list); if (!first) first = curr; last = curr; - objtext(curr, "<<\n/Title ("); - for (p = rs.text; p < rs.text+rs.pos; p++) { - char c[2]; - if (*p == '\\' || *p == '(' || *p == ')') - objtext(curr, "\\"); - c[0] = *p; - c[1] = '\0'; - objtext(curr, c); - } - objtext(curr, ")\n/Parent "); + objtext(curr, "<<\n/Title "); + pdf_string(objtext, curr, title); + objtext(curr, "\n/Parent "); objref(curr, parent); objtext(curr, "\n/Dest ["); - pdata = (para_data *)items->para->private_data; - objref(curr, (object *)pdata->first->page->spare); + objref(curr, (object *)items->pdata->first->page->spare); objtext(curr, " /XYZ null null null]\n"); if (prev) { objtext(curr, "/Prev "); @@ -727,3 +717,64 @@ static int make_outline(object *parent, outline_element *items, int n, return totalcount; } + +static int pdf_versionid(FILE *fp, word *words) +{ + int ret; + + ret = fprintf(fp, "%% "); + + for (; words; words = words->next) { + char *text; + int type; + + switch (words->type) { + case word_HyperLink: + case word_HyperEnd: + case word_UpperXref: + case word_LowerXref: + case word_XrefEnd: + case word_IndexRef: + continue; + } + + type = removeattr(words->type); + + switch (type) { + case word_Normal: + text = utoa_dup(words->text); + break; + case word_WhiteSpace: + text = dupstr(" "); + break; + case word_Quote: + text = dupstr("'"); + break; + } + + fputs(text, fp); + ret += strlen(text); + sfree(text); + } + + ret += fprintf(fp, "\n"); + + return ret; +} + +static void pdf_string(void (*add)(object *, char const *), + object *o, char const *str) +{ + char const *p; + + add(o, "("); + for (p = str; *p; p++) { + char c[2]; + if (*p == '\\' || *p == '(' || *p == ')') + add(o, "\\"); + c[0] = *p; + c[1] = '\0'; + add(o, c); + } + add(o, ")"); +}