#include <assert.h>
#include "halibut.h"
#include "paper.h"
+#include "deflate.h"
-#define TREE_BRANCH 2 /* max branching factor in page tree */
+#define TREE_BRANCH 8 /* max branching factor in page tree */
paragraph *pdf_config_filename(char *filename)
{
return cmdline_cfg_simple("pdf-filename", filename, NULL);
}
-typedef struct object_Tag object;
-typedef struct objlist_Tag objlist;
-
struct object_Tag {
objlist *list;
object *next;
object *head, *tail;
};
-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 pdf_string_len(void (*add)(object *, char const *),
object *, char const *, int);
static void objref(object *o, object *dest);
-static char *pdf_outline_convert(wchar_t *s, int *len);
+static void objdest(object *o, page_data *p);
static int is_std_font(char const *name);
int font_index;
font_encoding *fe;
page_data *page;
- int pageno;
FILE *fp;
char *filename;
paragraph *p;
/*
* The catalogue just contains references to the outlines and
- * pages objects.
+ * pages objects, and the pagelabels dictionary.
*/
objtext(cat, "<<\n/Type /Catalog");
if (outlines) {
}
objtext(cat, "\n/Pages ");
objref(cat, pages);
+ /* Halibut just numbers pages 1, 2, 3, ... */
+ objtext(cat, "\n/PageLabels<</Nums[0<</S/D>>]>>");
if (outlines)
objtext(cat, "\n/PageMode /UseOutlines");
objtext(cat, "\n>>\n");
* providing all the font objects and names to call them by.
*/
font_index = 0;
- objtext(resources, "<<\n/Font <<\n");
+ objtext(resources, "<<\n/ProcSet [/PDF/Text]\n/Font <<\n");
for (fe = doc->fonts->head; fe; fe = fe->next) {
char fname[40];
- int i;
- object *font;
+ char buf[80];
+ int i, prev;
+ object *font, *fontdesc;
+ int flags;
+ font_info const *fi = fe->font->info;
sprintf(fname, "f%d", font_index++);
fe->name = dupstr(fname);
objref(resources, font);
objtext(resources, "\n");
- objtext(font, "<<\n/Type /Font\n/Subtype /Type1\n/Name /");
- objtext(font, fe->name);
- objtext(font, "\n/BaseFont /");
- objtext(font, fe->font->info->name);
- objtext(font, "\n/Encoding <<\n/Type /Encoding\n/Differences [");
-
- for (i = 0; i < 256; i++) {
- char buf[20];
- if (!fe->vector[i])
- continue;
- sprintf(buf, "\n%d /", i);
- objtext(font, buf);
- objtext(font, fe->vector[i] ? fe->vector[i] : ".notdef");
- }
- objtext(font, "\n]\n>>\n");
+ /*
+ * Construct those parts of the font descriptor that don't dependd
+ * on the file format.
+ */
+ if (!is_std_font(fe->font->info->name)) {
+ fontdesc = new_object(&olist);
+
+#define FF_FIXEDPITCH 0x00000001
+#define FF_SERIF 0x00000002
+#define FF_SYMBOLIC 0x00000004
+#define FF_SCRIPT 0x00000008
+#define FF_NONSYMBOLIC 0x00000020
+#define FF_ITALIC 0x00000040
+#define FF_ALLCAP 0x00010000
+#define FF_SMALLCAP 0x00020000
+#define FF_FORCEBOLD 0x00040000
+
+ objtext(fontdesc, "<<\n/Type /FontDescriptor\n/Name /");
+ objtext(fontdesc, fi->name);
+ flags = 0;
+ if (fi->italicangle) flags |= FF_ITALIC;
+ flags |= FF_NONSYMBOLIC;
+ sprintf(buf, "\n/Flags %d\n", flags);
+ objtext(fontdesc, buf);
+ sprintf(buf, "/FontBBox [%g %g %g %g]\n", fi->fontbbox[0],
+ fi->fontbbox[1], fi->fontbbox[2], fi->fontbbox[3]);
+ objtext(fontdesc, buf);
+ sprintf(buf, "/ItalicAngle %g\n", fi->italicangle);
+ objtext(fontdesc, buf);
+ sprintf(buf, "/Ascent %g\n", fi->ascent);
+ objtext(fontdesc, buf);
+ sprintf(buf, "/Descent %g\n", fi->descent);
+ objtext(fontdesc, buf);
+ sprintf(buf, "/CapHeight %g\n", fi->capheight);
+ objtext(fontdesc, buf);
+ sprintf(buf, "/XHeight %g\n", fi->xheight);
+ objtext(fontdesc, buf);
+ sprintf(buf, "/StemH %g\n", fi->stemh);
+ objtext(fontdesc, buf);
+ sprintf(buf, "/StemV %g\n", fi->stemv);
+ objtext(fontdesc, buf);
+ }
- if (!is_std_font(fe->font->info->name)){
- object *widths = new_object(&olist);
- int firstchar = -1, lastchar = -1;
- char buf[80];
- for (i = 0; i < 256; i++)
- if (fe->indices[i] >= 0) {
- if (firstchar < 0) firstchar = i;
- lastchar = i;
+ objtext(font, "<<\n/Type /Font\n/BaseFont /");
+ objtext(font, fe->font->info->name);
+ if (fe->font->info->filetype == TRUETYPE) {
+ object *cidfont = new_object(&olist);
+ object *cmap = new_object(&olist);
+ unsigned short ranges[256];
+ unsigned startidx, nranges, nchars;
+ int start;
+
+ objtext(font, "/Subtype/Type0\n/Encoding ");
+ objtext(cmap, "<</Type/CMap\n/CMapName/");
+ objtext(cmap, fe->name);
+ objtext(cmap, "\n/CIDSystemInfo<</Registry(Adobe)"
+ "/Ordering(Identity)/Supplement 0>>\n");
+ objstream(cmap, "%!PS-Adobe-3.0 Resource-CMap\n"
+ "%%DocumentNeededResources: procset CIDInit\n"
+ "%%IncludeResource: procset CIDInit\n"
+ "%%BeginResource: CMap ");
+ objstream(cmap, fe->name);
+ objstream(cmap, "\n%%Title (");
+ objstream(cmap, fe->name);
+ objstream(cmap, " Adobe Identity 0)\n%%Version: 1\n%%EndComments\n");
+ objstream(cmap, "/CIDInit/ProcSet findresource begin\n");
+ objstream(cmap, "12 dict begin begincmap\n");
+ objstream(cmap, "/CIDSystemInfo 3 dict dup begin\n"
+ "/Registry(Adobe)def/Ordering(Identity)def"
+ "/Supplement 0 def end def\n");
+ objstream(cmap, "/CMapName/");
+ objstream(cmap, fe->name);
+ objstream(cmap, " def/CMapType 0 def/WMode 0 def\n");
+ objstream(cmap, "1 begincodespacerange<00><FF>"
+ "endcodespacerange\n");
+ start = -1; nranges = nchars = 0;
+ for (i = 0; i < 256; i++) {
+ unsigned idx;
+
+ ranges[i] = 0;
+ if (fe->vector[i] == NOGLYPH)
+ continue;
+ idx = sfnt_glyphtoindex(fe->font->info->fontfile,
+ fe->vector[i]);
+ if (start >= 0 && idx - startidx == (unsigned)(i - start)) {
+ if (ranges[start] == 1) {
+ nranges++; nchars--;
+ }
+ ranges[start] = i - start + 1;
+ } else {
+ ranges[i] = 1;
+ start = i;
+ startidx = idx;
+ nchars++;
+ }
+ }
+ i = 0;
+ while (nranges) {
+ unsigned blk = nranges > 100 ? 100 : nranges;
+ nranges -= blk;
+ sprintf(buf, "%u ", blk);
+ objstream(cmap, buf);
+ objstream(cmap, "begincidrange\n");
+ while (blk) {
+ if (ranges[i] > 1) {
+ sprintf(buf, "<%02X>", i);
+ objstream(cmap, buf);
+ sprintf(buf, "<%02X>", i + ranges[i] - 1);
+ objstream(cmap, buf);
+ sprintf(buf, "%hu\n",
+ sfnt_glyphtoindex(fe->font->info->fontfile,
+ fe->vector[i]));
+ objstream(cmap, buf);
+ blk--;
+ }
+ i++;
}
- sprintf(buf, "/FirstChar %d\n/LastChar %d\n/Widths ",
- firstchar, lastchar);
- objtext(font, buf);
- objref(font, widths);
- objtext(font, "\n");
- objtext(widths, "[\n");
- for (i = firstchar; i <= lastchar; i++) {
- char buf[80];
+ objstream(cmap, "endcidrange\n");
+ }
+ i = 0;
+ while (nchars) {
+ unsigned blk = nchars > 100 ? 100 : nchars;
+ nchars -= blk;
+ sprintf(buf, "%u ", blk);
+ objstream(cmap, buf);
+ objstream(cmap, "begincidchar\n");
+ while (blk) {
+ if (ranges[i] == 1) {
+ sprintf(buf, "<%02X>", i);
+ objstream(cmap, buf);
+ sprintf(buf, "%hu\n",
+ sfnt_glyphtoindex(fe->font->info->fontfile,
+ fe->vector[i]));
+ objstream(cmap, buf);
+ blk--;
+ }
+ i++;
+ }
+ objstream(cmap, "endcidchar\n");
+ }
+ objstream(cmap, "endcmap CMapName currentdict /CMap "
+ "defineresource pop end end\n%%EndResource\n%%EOF\n");
+
+ objref(font, cmap);
+ objtext(font, "\n/DescendantFonts[");
+ objref(font, cidfont);
+ objtext(font, "]\n");
+ objtext(cidfont, "<<\n/Type/Font\n/Subtype/CIDFontType2\n"
+ "/BaseFont/");
+ objtext(cidfont, fe->font->info->name);
+ objtext(cidfont, "\n/CIDSystemInfo<</Registry(Adobe)"
+ "/Ordering(Identity)/Supplement 0>>\n");
+ objtext(cidfont, "/FontDescriptor ");
+ objref(cidfont, fontdesc);
+ objtext(cidfont, "\n/W[0[");
+ for (i = 0; i < (int)sfnt_nglyphs(fe->font->info->fontfile); i++) {
+ char buf[20];
double width;
- if (fe->indices[i] < 0)
- width = 0.0;
- else
- width = fe->font->info->widths[fe->indices[i]];
- sprintf(buf, "%g\n", 1000.0 * width / FUNITS_PER_PT);
- objtext(widths, buf);
+ width = find_width(fe->font,
+ sfnt_indextoglyph(fe->font->info->fontfile, i));
+ sprintf(buf, "%g ", 1000.0 * width / FUNITS_PER_PT);
+ objtext(cidfont, buf);
+ }
+ objtext(cidfont, "]]>>\n");
+ } else {
+ objtext(font, "/Subtype /Type1\n");
+ objtext(font, "\n/Encoding <<\n/Type /Encoding\n/Differences [");
+
+ for (i = 0; i < 256; i++) {
+ char buf[20];
+ if (fe->vector[i] == NOGLYPH)
+ continue;
+ if (i != prev + 1) {
+ sprintf(buf, "\n%d", i);
+ objtext(font, buf);
+ }
+ objtext(font, i % 8 ? "/" : "\n/");
+ objtext(font, glyph_extern(fe->vector[i]));
+ prev = i;
+ }
+
+ objtext(font, "\n]\n>>\n");
+ if (!is_std_font(fe->font->info->name)){
+ object *widths = new_object(&olist);
+ int firstchar = -1, lastchar = -1;
+ for (i = 0; i < 256; i++)
+ if (fe->vector[i] != NOGLYPH) {
+ if (firstchar < 0) firstchar = i;
+ lastchar = i;
+ }
+ sprintf(buf, "/FirstChar %d\n/LastChar %d\n/Widths ",
+ firstchar, lastchar);
+ objtext(font, buf);
+ objref(font, widths);
+ objtext(font, "\n");
+ objtext(widths, "[\n");
+ for (i = firstchar; i <= lastchar; i++) {
+ double width;
+ if (fe->vector[i] == NOGLYPH)
+ width = 0.0;
+ else
+ width = find_width(fe->font, fe->vector[i]);
+ sprintf(buf, "%g\n", 1000.0 * width / FUNITS_PER_PT);
+ objtext(widths, buf);
+ }
+ objtext(widths, "]\n");
+ objtext(font, "/FontDescriptor ");
+ objref(font, fontdesc);
+ }
+
+ }
+
+ if (!is_std_font(fe->font->info->name)) {
+ if (fi->fontfile && fi->filetype == TYPE1) {
+ object *fontfile = new_object(&olist);
+ size_t len;
+ char *ffbuf;
+
+ pf_part1((font_info *)fi, &ffbuf, &len);
+ objstream_len(fontfile, ffbuf, len);
+ sfree(ffbuf);
+ sprintf(buf, "<<\n/Length1 %lu\n", (unsigned long)len);
+ objtext(fontfile, buf);
+ pf_part2((font_info *)fi, &ffbuf, &len);
+ objstream_len(fontfile, ffbuf, len);
+ sfree(ffbuf);
+ sprintf(buf, "/Length2 %lu\n", (unsigned long)len);
+ objtext(fontfile, buf);
+ objtext(fontfile, "/Length3 0\n");
+ objtext(fontdesc, "/FontFile ");
+ objref(fontdesc, fontfile);
+ } else if (fi->fontfile && fi->filetype == TRUETYPE) {
+ object *fontfile = new_object(&olist);
+ size_t len;
+ char *ffbuf;
+
+ sfnt_data((font_info *)fi, &ffbuf, &len);
+ objstream_len(fontfile, ffbuf, len);
+ sprintf(buf, "<<\n/Length1 %lu\n", (unsigned long)len);
+ objtext(fontfile, buf);
+ objtext(fontdesc, "/FontFile2 ");
+ objref(fontdesc, fontfile);
}
- objtext(widths, "]\n");
+ objtext(fontdesc, "\n>>\n");
}
- objtext(font, ">>\n");
+ objtext(font, "\n>>\n");
}
objtext(resources, ">>\n>>\n");
/*
* Create and render the individual pages.
*/
- pageno = 0;
for (page = doc->pages; page; page = page->next) {
object *opage, *cstr;
rect *r;
objtext(opage, "/Annots [\n");
for (xr = page->first_xref; xr; xr = xr->next) {
- object *annot;
char buf[256];
- annot = new_object(&olist);
- objref(opage, annot);
- objtext(opage, "\n");
-
- objtext(annot, "<<\n/Type /Annot\n/Subtype /Link\n/Rect [");
+ objtext(opage, "<</Subtype/Link\n/Rect[");
sprintf(buf, "%g %g %g %g",
xr->lx / FUNITS_PER_PT, xr->by / FUNITS_PER_PT,
xr->rx / FUNITS_PER_PT, xr->ty / FUNITS_PER_PT);
- objtext(annot, buf);
- objtext(annot, "]\n/Border [0 0 0]\n");
+ objtext(opage, buf);
+ objtext(opage, "]/Border[0 0 0]\n");
if (xr->dest.type == PAGE) {
- objtext(annot, "/Dest [");
- objref(annot, (object *)xr->dest.page->spare);
- objtext(annot, " /XYZ null null null]\n");
+ objtext(opage, "/Dest");
+ objdest(opage, xr->dest.page);
} else {
- objtext(annot, "/A <<\n/Type /Action\n/S /URI\n/URI ");
- pdf_string(objtext, annot, xr->dest.url);
- objtext(annot, "\n>>\n");
+ objtext(opage, "/A<</S/URI/URI");
+ pdf_string(objtext, opage, xr->dest.url);
+ objtext(opage, ">>");
}
- objtext(annot, ">>\n");
+ objtext(opage, ">>\n");
}
objtext(opage, "]\n");
for (o = olist.head; o; o = o->next) {
rdstringc rs = {0, 0, NULL};
char text[80];
+ deflate_compress_ctx *zcontext;
+ void *zbuf;
+ int zlen;
sprintf(text, "%d 0 obj\n", o->number);
rdaddsc(&rs, text);
- if (!o->main.text && o->stream.text) {
- sprintf(text, "<<\n/Length %d\n>>\n", o->stream.pos);
+ if (o->stream.text) {
+ if (!o->main.text)
+ rdaddsc(&o->main, "<<\n");
+#ifdef PDF_NOCOMPRESS
+ zlen = o->stream.pos;
+ zbuf = snewn(zlen, char);
+ memcpy(zbuf, o->stream.text, zlen);
+ sprintf(text, "/Length %d\n>>\n", zlen);
+#else
+ zcontext = deflate_compress_new(DEFLATE_TYPE_ZLIB);
+ deflate_compress_data(zcontext, o->stream.text, o->stream.pos,
+ DEFLATE_END_OF_DATA, &zbuf, &zlen);
+ deflate_compress_free(zcontext);
+ sprintf(text, "/Filter/FlateDecode\n/Length %d\n>>\n", zlen);
+#endif
rdaddsc(&o->main, text);
}
rdaddc(&rs, '\n');
if (o->stream.text) {
- /*
- * FIXME: If we ever start compressing stream data then
- * it will have zero bytes in it, so we'll have to be
- * more careful than this.
- */
rdaddsc(&rs, "stream\n");
- rdaddsc(&rs, o->stream.text);
+ rdaddsn(&rs, zbuf, zlen);
rdaddsc(&rs, "\nendstream\n");
sfree(o->stream.text);
+ sfree(zbuf);
}
rdaddsc(&rs, "endobj\n");
* Write out the PDF file.
*/
- fp = fopen(filename, "wb");
+ if (!strcmp(filename, "-"))
+ fp = stdout;
+ else
+ fp = fopen(filename, "wb");
if (!fp) {
error(err_cantopenw, filename);
return;
/*
* Header. I'm going to put the version IDs in the header as
- * well, simply in PDF comments.
+ * well, simply in PDF comments. The PDF Reference also suggests
+ * that binary PDF files contain four top-bit-set characters in
+ * the second line.
*/
- fileoff = fprintf(fp, "%%PDF-1.3\n");
+ fileoff = fprintf(fp, "%%PDF-1.3\n%% L\xc3\xba\xc3\xb0""a\n");
for (p = sourceform; p; p = p->next)
if (p->type == para_VersionID)
fileoff += pdf_versionid(fp, p->words);
olist.tail->number + 1, cat->number, info->number);
fprintf(fp, "startxref\n%d\n%%%%EOF\n", fileoff);
- fclose(fp);
+ if (fp != stdout)
+ fclose(fp);
sfree(filename);
}
-static object *new_object(objlist *list)
+object *new_object(objlist *list)
{
object *obj = snew(object);
return obj;
}
-static void objtext(object *o, char const *text)
+void objtext(object *o, char const *text)
{
rdaddsc(&o->main, text);
}
-static void objstream(object *o, char const *text)
+void objstream_len(object *o, char const *text, size_t len)
+{
+ rdaddsn(&o->stream, text, len);
+}
+
+void objstream(object *o, char const *text)
{
rdaddsc(&o->stream, text);
}
rdaddsc(&o->main, buf);
}
+static void objdest(object *o, page_data *p) {
+ objtext(o, "[");
+ objref(o, (object *)p->spare);
+ objtext(o, "/XYZ null null null]");
+}
+
static char const * const stdfonts[] = {
"Times-Roman", "Times-Bold", "Times-Italic", "Times-BoldItalic",
"Helvetica", "Helvetica-Bold", "Helvetica-Oblique","Helvetica-BoldOblique",
* encodes in either PDFDocEncoding (a custom superset of
* ISO-8859-1) or UTF-16BE.
*/
-static char *pdf_outline_convert(wchar_t *s, int *len) {
+char *pdf_outline_convert(wchar_t *s, int *len) {
char *ret;
ret = utoa_careful_dup(s, CS_PDF);
sfree(title);
objtext(curr, "\n/Parent ");
objref(curr, parent);
- objtext(curr, "\n/Dest [");
- objref(curr, (object *)items->pdata->first->page->spare);
- objtext(curr, " /XYZ null null null]\n");
+ objtext(curr, "\n/Dest");
+ objdest(curr, items->pdata->first->page);
+ objtext(curr, "\n");
if (prev) {
objtext(curr, "/Prev ");
objref(curr, prev);
~mdw / sgt / halibut / blobdiff