Refactored xref opening functions, command line tools, and rewrote pdfdraw to be more illustrative.

1 files changed, 189 insertions, 153 deletions

diff --git a/apps/pdfclean.c b/apps/pdfclean.c
index ee6ca722..0ecbb98e 100644
--- a/apps/pdfclean.c
+++ b/apps/pdfclean.c
@@ -4,10 +4,13 @@
  * Rewrite PDF with pretty printed objects.
  * Garbage collect unreachable objects.
  * Inflate compressed streams.
- * Encrypt output.
+ * Create subset documents.
+ *
+ * TODO: linearize document for fast web view
  */
 
-#include "pdftool.h"
+#include "fitz.h"
+#include "mupdf.h"
 
 static FILE *out = NULL;
 
@@ -20,6 +23,29 @@ static pdf_xrefentry *oldxreflist = NULL;
 static int dogarbage = 0;
 static int doexpand = 0;
 
+static pdf_xref *xref = NULL;
+
+void die(fz_error error)
+{
+	fz_catch(error, "aborting");
+	if (xref)
+		pdf_freexref(xref);
+	exit(1);
+}
+
+static void usage(void)
+{
+	fprintf(stderr,
+		"usage: pdfclean [options] input.pdf [output.pdf] [pages]\n"
+		"\t-p -\tpassword\n"
+		"\t-g\tgarbage collect unused objects\n"
+		"\t-gg\tin addition to -g compact xref table\n"
+		"\t-ggg\tin addition to -gg merge duplicate objects\n"
+		"\t-x\texpand compressed streams\n"
+		"\tpages\tcomma separated list of ranges\n");
+	exit(1);
+}
+
 /*
  * Garbage collect objects not reachable from the trailer.
  */
@@ -67,6 +93,10 @@ static void sweepref(fz_obj *obj)
 	sweepobj(fz_resolveindirect(obj));
 }
 
+/*
+ * Renumber objects to compact the xref table
+ */
+
 static void renumberobj(fz_obj *obj)
 {
 	int i;
@@ -109,6 +139,53 @@ static void renumberobj(fz_obj *obj)
 	}
 }
 
+static void renumberxref(void)
+{
+	int num, newnum;
+
+	newnumlist = fz_malloc(xref->len * sizeof(int));
+	oldxreflist = fz_malloc(xref->len * sizeof(pdf_xrefentry));
+	for (num = 0; num < xref->len; num++)
+	{
+		newnumlist[num] = -1;
+		oldxreflist[num] = xref->table[num];
+	}
+
+	newnum = 1;
+	for (num = 0; num < xref->len; num++)
+	{
+		if (xref->table[num].type == 'f')
+			uselist[num] = 0;
+		if (uselist[num])
+			newnumlist[num] = newnum++;
+	}
+
+	renumberobj(xref->trailer);
+	for (num = 0; num < xref->len; num++)
+		renumberobj(xref->table[num].obj);
+
+	for (num = 0; num < xref->len; num++)
+		uselist[num] = 0;
+
+	for (num = 0; num < xref->len; num++)
+	{
+		if (newnumlist[num] >= 0)
+		{
+			xref->table[newnumlist[num]] = oldxreflist[num];
+			uselist[newnumlist[num]] = 1;
+		}
+	}
+
+	fz_free(oldxreflist);
+	fz_free(newnumlist);
+
+	xref->len = newnum;
+}
+
+/*
+ * Scan and remove duplicate objects (slow)
+ */
+
 static void removeduplicateobjs(void)
 {
 	int num, other;
@@ -142,6 +219,99 @@ static void removeduplicateobjs(void)
 	fz_free(newnumlist);
 }
 
+/*
+ * Recreate page tree to only retain specified pages.
+ */
+
+static void retainpages(int argc, char **argv)
+{
+	fz_obj *root, *pages, *kids;
+	int count;
+
+	/* Snatch pages entry from root dict */
+	root = fz_dictgets(xref->trailer, "Root");
+	pages = fz_keepobj(fz_dictgets(root, "Pages"));
+
+	/* Then empty the root dict */
+	while (fz_dictlen(root) > 0)
+	{
+		fz_obj *key = fz_dictgetkey(root, 0);
+		fz_dictdel(root, key);
+	}
+
+	/* And only retain pages and type entries */
+	fz_dictputs(root, "Pages", pages);
+	fz_dictputs(root, "Type", fz_newname("Catalog"));
+	fz_dropobj(pages);
+
+	/* Create a new kids array too add into pages dict
+	 * since each element must be replaced to point to
+	 * a retained page */
+	kids = fz_newarray(1);
+	count = 0;
+
+	/* Retain pages specified */
+	while (argc - fz_optind)
+	{
+		int page, spage, epage;
+		char *spec, *dash;
+		char *pagelist = argv[fz_optind];
+
+		spec = fz_strsep(&pagelist, ",");
+		while (spec)
+		{
+			dash = strchr(spec, '-');
+
+			if (dash == spec)
+				spage = epage = 1;
+			else
+				spage = epage = atoi(spec);
+
+			if (dash)
+			{
+				if (strlen(dash) > 1)
+					epage = atoi(dash + 1);
+				else
+					epage = pdf_getpagecount(xref);
+			}
+
+			if (spage > epage)
+				page = spage, spage = epage, epage = page;
+
+			if (spage < 1)
+				spage = 1;
+			if (epage > pdf_getpagecount(xref))
+				epage = pdf_getpagecount(xref);
+
+			for (page = spage; page <= epage; page++)
+			{
+				fz_obj *pageobj = pdf_getpageobject(xref, page);
+				fz_obj *pageref = pdf_getpageref(xref, page);
+
+				/* Update parent reference */
+				fz_dictputs(pageobj, "Parent", pages);
+
+				/* Store page object in new kids array */
+				fz_arraypush(kids, pageref);
+				count++;
+
+				fz_dropobj(pageref);
+			}
+
+			spec = fz_strsep(&pagelist, ",");
+		}
+
+		fz_optind++;
+	}
+
+	/* Update page count and kids array */
+	fz_dictputs(pages, "Count", fz_newint(count));
+	fz_dictputs(pages, "Kids", kids);
+}
+
+/*
+ * Make sure we have loaded objects from object streams.
+ */
 
 static void preloadobjstms(void)
 {
@@ -161,6 +331,10 @@ static void preloadobjstms(void)
 	}
 }
 
+/*
+ * Save streams and objects to the output
+ */
+
 static void copystream(fz_obj *obj, int num, int gen)
 {
 	fz_error error;
@@ -208,7 +382,7 @@ static void expandstream(fz_obj *obj, int num, int gen)
 	fz_dropbuffer(buf);
 }
 
-static void saveobject(int num, int gen)
+static void writeobject(int num, int gen)
 {
 	fz_error error;
 	fz_obj *obj;
@@ -253,7 +427,7 @@ static void saveobject(int num, int gen)
 	fz_dropobj(obj);
 }
 
-static void savexref(void)
+static void writexref(void)
 {
 	fz_obj *trailer;
 	fz_obj *obj;
@@ -297,148 +471,7 @@ static void savexref(void)
 	fprintf(out, "startxref\n%d\n%%%%EOF\n", startxref);
 }
 
-static void cleanusage(void)
-{
-	fprintf(stderr,
-		"usage: pdfclean [options] input.pdf [outfile.pdf] [pages]\n"
-		"\t-p -\tpassword for decryption\n"
-		"\t-g\tgarbage collect unused objects\n"
-		"\t-gg\tin addition to -g xref is compacted\n"
-		"\t-ggg\tin addition to -gg identical objects are garbage collected\n"
-		"\t-x\texpand compressed streams\n");
-	exit(1);
-}
-
-static void retainpages(int argc, char **argv)
-{
-	fz_obj *root, *pages, *kids;
-	int count;
-
-	/* Snatch pages entry from root dict */
-	root = fz_dictgets(xref->trailer, "Root");
-	pages = fz_keepobj(fz_dictgets(root, "Pages"));
-
-	/* Then empty the root dict */
-	while (fz_dictlen(root) > 0)
-	{
-		fz_obj *key = fz_dictgetkey(root, 0);
-		fz_dictdel(root, key);
-	}
-
-	/* And only retain pages and type entries */
-	fz_dictputs(root, "Pages", pages);
-	fz_dictputs(root, "Type", fz_newname("Catalog"));
-	fz_dropobj(pages);
-
-	/* Create a new kids array too add into pages dict
-	 * since each element must be replaced to point to
-	 * a retained page */
-	kids = fz_newarray(1);
-	count = 0;
-
-	/* Retain pages specified */
-	while (argc - fz_optind)
-	{
-		int page, spage, epage;
-		char *spec, *dash;
-		char *pagelist = argv[fz_optind];
-
-		spec = fz_strsep(&pagelist, ",");
-		while (spec)
-		{
-			dash = strchr(spec, '-');
-
-			if (dash == spec)
-				spage = epage = 1;
-			else
-				spage = epage = atoi(spec);
-
-			if (dash)
-			{
-				if (strlen(dash) > 1)
-					epage = atoi(dash + 1);
-				else
-					epage = pagecount;
-			}
-
-			if (spage > epage)
-				page = spage, spage = epage, epage = page;
-
-			if (spage < 1)
-				spage = 1;
-			if (epage > pagecount)
-				epage = pagecount;
-
-			for (page = spage; page <= epage; page++)
-			{
-				fz_obj *pageobj = pdf_getpageobject(xref, page);
-				fz_obj *pageref = pdf_getpageref(xref, page);
-
-				/* Update parent reference */
-				fz_dictputs(pageobj, "Parent", pages);
-
-				/* Store page object in new kids array */
-				fz_arraypush(kids, pageref);
-				count++;
-
-				fz_dropobj(pageref);
-			}
-
-			spec = fz_strsep(&pagelist, ",");
-		}
-
-		fz_optind++;
-	}
-
-	/* Update page count and kids array */
-	fz_dictputs(pages, "Count", fz_newint(count));
-	fz_dictputs(pages, "Kids", kids);
-}
-
-static void renumberxref(void)
-{
-	int num, newnum;
-
-	newnumlist = fz_malloc(xref->len * sizeof(int));
-	oldxreflist = fz_malloc(xref->len * sizeof(pdf_xrefentry));
-	for (num = 0; num < xref->len; num++)
-	{
-		newnumlist[num] = -1;
-		oldxreflist[num] = xref->table[num];
-	}
-
-	newnum = 1;
-	for (num = 0; num < xref->len; num++)
-	{
-		if (xref->table[num].type == 'f')
-			uselist[num] = 0;
-		if (uselist[num])
-			newnumlist[num] = newnum++;
-	}
-
-	renumberobj(xref->trailer);
-	for (num = 0; num < xref->len; num++)
-		renumberobj(xref->table[num].obj);
-
-	for (num = 0; num < xref->len; num++)
-		uselist[num] = 0;
-
-	for (num = 0; num < xref->len; num++)
-	{
-		if (newnumlist[num] >= 0)
-		{
-			xref->table[newnumlist[num]] = oldxreflist[num];
-			uselist[newnumlist[num]] = 1;
-		}
-	}
-
-	fz_free(oldxreflist);
-	fz_free(newnumlist);
-
-	xref->len = newnum;
-}
-
-static void outputpdf(void)
+static void writepdf(void)
 {
 	int lastfree;
 	int num;
@@ -461,7 +494,7 @@ static void outputpdf(void)
 		if (xref->table[num].type == 'n' || xref->table[num].type == 'o')
 		{
 			ofslist[num] = ftell(out);
-			saveobject(num, genlist[num]);
+			writeobject(num, genlist[num]);
 		}
 	}
 
@@ -477,11 +510,12 @@ static void outputpdf(void)
 		}
 	}
 
-	savexref();
+	writexref();
 }
 
 int main(int argc, char **argv)
 {
+	fz_error error;
 	char *infile;
 	char *outfile = "out.pdf";
 	char *password = "";
@@ -495,12 +529,12 @@ int main(int argc, char **argv)
 		case 'p': password = fz_optarg; break;
 		case 'g': dogarbage ++; break;
 		case 'x': doexpand ++; break;
-		default: cleanusage(); break;
+		default: usage(); break;
 		}
 	}
 
 	if (argc - fz_optind < 1)
-		cleanusage();
+		usage();
 
 	infile = argv[fz_optind++];
 
@@ -514,7 +548,9 @@ int main(int argc, char **argv)
 	if (argc - fz_optind > 0)
 		subset = 1;
 
-	openxref(infile, password, 0, subset);
+	error = pdf_openxref(&xref, infile, password);
+	if (error)
+		die(fz_rethrow(error, "cannot open input file '%s'", infile));
 
 	out = fopen(outfile, "wb");
 	if (!out)
@@ -552,9 +588,9 @@ int main(int argc, char **argv)
 	if (dogarbage >= 2)
 		renumberxref();
 
-	outputpdf();
+	writepdf();
 
-	closexref();
+	pdf_freexref(xref);
 
 	return 0;
 }