X-Git-Url: https://git.distorted.org.uk/~mdw/sgt/halibut/blobdiff_plain/44407fea05657f91da34fd83428b67e7100ab62b..7e2417cca16da830a1d0898380b256a77ec04de7:/bk_pdf.c

diff --git a/bk_pdf.c b/bk_pdf.c
index 0a36ad9..71307d2 100644
--- a/bk_pdf.c
+++ b/bk_pdf.c
@@ -5,6 +5,7 @@
 #include <assert.h>
 #include "halibut.h"
 #include "paper.h"
+#include "deflate.h"
 
 #define TREE_BRANCH 2		       /* max branching factor in page tree */
 
@@ -112,7 +113,7 @@ void pdf_backend(paragraph *sourceform, keywordlist *keywords,
 
     /*
      * The catalogue just contains references to the outlines and
-     * pages objects.
+     * pages objects, and the pagelabels dictionary.
      */
     objtext(cat, "<<\n/Type /Catalog");
     if (outlines) {
@@ -121,6 +122,8 @@ void pdf_backend(paragraph *sourceform, keywordlist *keywords,
     }
     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");
@@ -130,10 +133,10 @@ void pdf_backend(paragraph *sourceform, keywordlist *keywords,
      * 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;
+	int i, prev;
 	object *font;
 
 	sprintf(fname, "f%d", font_index++);
@@ -157,9 +160,13 @@ void pdf_backend(paragraph *sourceform, keywordlist *keywords,
 	    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");
+	    if (i != prev + 1) {
+		sprintf(buf, "\n%d", i);
+		objtext(font, buf);
+	    }
+	    objtext(font, i % 8 ? "/" : "\n/");
+	    objtext(font, fe->vector[i]);
+	    prev = i;
 	}
 
 	objtext(font, "\n]\n>>\n");
@@ -418,31 +425,26 @@ void pdf_backend(paragraph *sourceform, keywordlist *keywords,
 	    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[");
+		    objref(opage, (object *)xr->dest.page->spare);
+		    objtext(opage, "/XYZ null null null]");
 		} 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");
@@ -471,12 +473,19 @@ void pdf_backend(paragraph *sourceform, keywordlist *keywords,
     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);
+	    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, "<<\n/Filter/FlateDecode\n/Length %d\n>>\n", zlen);
 	    rdaddsc(&o->main, text);
 	}
 
@@ -488,15 +497,11 @@ void pdf_backend(paragraph *sourceform, keywordlist *keywords,
 	    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");
@@ -517,9 +522,11 @@ void pdf_backend(paragraph *sourceform, keywordlist *keywords,
 
     /*
      * 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);