improved reading to buffers and writing files

13 files changed, 239 insertions, 681 deletions

diff --git a/filter/buffer.c b/filter/buffer.c
index 25e30a55..aae27474 100644
--- a/filter/buffer.c
+++ b/filter/buffer.c
@@ -32,7 +32,7 @@ fz_newbufferwithdata(fz_buffer **bp, unsigned char *data, int size)
 	b->bp = data;
 
 	b->rp = b->bp;
-	b->wp = b->bp;
+	b->wp = b->bp + size;
 	b->ep = b->bp + size;
 	b->eof = 0;
 
diff --git a/filter/filec.c b/filter/filec.c
index f62d1265..fc5599e1 100644
--- a/filter/filec.c
+++ b/filter/filec.c
@@ -13,15 +13,20 @@ fz_openfile(fz_file **filep, char *path, int mode)
 {
 	fz_error *error;
 	fz_file *file;
+	int realmode;
 	int fd;
 
-	assert(mode == O_RDONLY || mode == O_WRONLY);
+	assert(mode == O_RDONLY || mode == O_WRONLY || mode == O_APPEND);
+
+	realmode = mode;
+	if (mode == O_WRONLY)
+		realmode = O_WRONLY | O_CREAT | O_TRUNC;
 
 	file = *filep = fz_malloc(sizeof(fz_file));
 	if (!file)
 		return fz_outofmem;
 
-	fd = open(path, mode, 0);
+	fd = open(path, realmode, 0644);
 	if (fd == -1)
 		return fz_throw("ioerror: open '%s': %s", path, strerror(errno));
 
diff --git a/filter/filer.c b/filter/filer.c
index 804d7f7e..368f9dbe 100644
--- a/filter/filer.c
+++ b/filter/filer.c
@@ -175,11 +175,12 @@ fz_readline(fz_file *f, char *buf, int n)
  * a freshly allocated buffer; realloced and trimmed to size.
  */
 
-enum { CHUNKSIZE = 4096 };
+enum { CHUNKSIZE = 1024 * 32 };
 
 fz_error *
-fz_readfile(unsigned char **bufp, int *lenp, fz_file *file)
+fz_readfile(fz_buffer **bufp, fz_file *file)
 {
+	fz_buffer *real;
 	unsigned char *newbuf;
 	unsigned char *buf;
 	int len;
@@ -187,7 +188,6 @@ fz_readfile(unsigned char **bufp, int *lenp, fz_file *file)
 	int n;
 
 	*bufp = nil;
-	*lenp = 0;
 
 	len = 0;
 	pos = 0;
@@ -209,8 +209,6 @@ fz_readfile(unsigned char **bufp, int *lenp, fz_file *file)
 
 		n = fz_read(file, buf + pos, len - pos);
 
-printf("fz_read %d bytes\n", n);
-
 		if (n < 0)
 		{
 			fz_free(buf);
@@ -228,8 +226,19 @@ printf("fz_read %d bytes\n", n);
 				return fz_outofmem;
 			}
 
-			*bufp = newbuf;
-			*lenp = pos;
+			real = *bufp = fz_malloc(sizeof(fz_buffer));
+			if (!real)
+			{
+				fz_free(newbuf);
+				return fz_outofmem;
+			}
+
+			real->ownsdata = 1;
+			real->bp = buf;
+			real->rp = buf;
+			real->wp = buf + pos;
+			real->ep = buf + pos;
+
 			return nil;
 		}
 	}
diff --git a/filter/filew.c b/filter/filew.c
index 0422e60e..a7b15664 100644
--- a/filter/filew.c
+++ b/filter/filew.c
@@ -99,7 +99,12 @@ fz_write(fz_file *f, char *buf, int n)
 					return -1;
 				}
 
-				fz_rewindbuffer(f->in);
+				if (f->in->rp > f->in->bp)
+					f->error = fz_rewindbuffer(f->in);
+				else
+					f->error = fz_growbuffer(f->in);
+				if (f->error)
+					return -1;
 			}
 		}
 
diff --git a/include/fitz/file.h b/include/fitz/file.h
index 7807570c..2d6dc977 100644
--- a/include/fitz/file.h
+++ b/include/fitz/file.h
@@ -26,7 +26,7 @@ int fz_peekbyte(fz_file *f);
 int fz_readline(fz_file *f, char *buf, int n);
 int fz_read(fz_file *f, char *buf, int n);
 
-fz_error *fz_readfile(unsigned char **bufp, int *lenp, fz_file *file);
+fz_error *fz_readfile(fz_buffer **bufp, fz_file *file);
 
 int fz_printstring(fz_file *f, char *s);
 int fz_printobj(fz_file *f, fz_obj *o, int tight);
diff --git a/include/mupdf.h b/include/mupdf.h
index c5dcc133..98f40c62 100644
--- a/include/mupdf.h
+++ b/include/mupdf.h
@@ -81,7 +81,8 @@ fz_error *pdf_buildfilter(fz_filter**, pdf_xref*, fz_obj *stm, int oid, int gid)
 fz_error *pdf_openstream0(pdf_xref*, fz_obj *stmobj, int oid, int gid, int ofs);
 fz_error *pdf_openstream(pdf_xref*, fz_obj *stmref);
 void pdf_closestream(pdf_xref*);
-fz_error *pdf_readstream(unsigned char **bufp, int *lenp, pdf_xref*, fz_obj *stmref);
+fz_error *pdf_readstream0(fz_buffer **bufp, pdf_xref*, fz_obj *stmobj, int oid, int gid, int ofs);
+fz_error *pdf_readstream(fz_buffer **bufp, pdf_xref*, fz_obj *stmref);
 
 /* crypt.c */
 fz_error *pdf_newdecrypt(pdf_crypt **cp, fz_obj *enc, fz_obj *id);
@@ -118,6 +119,9 @@ fz_error *pdf_loadobject0(fz_obj **, pdf_xref*, int oid, int gid, int *stmofs);
 fz_error *pdf_loadobject(fz_obj **, pdf_xref*, fz_obj *ref, int *stmofs);
 fz_error *pdf_resolve(fz_obj **, pdf_xref*);
 
+fz_error *pdf_savestream(pdf_xref *xref, int oid, int gid, fz_buffer *buf);
+fz_error *pdf_deletestream(pdf_xref *xref, int oid, int gid);
+
 /* save.c */
 fz_error *pdf_saveincrementalpdf(pdf_xref *xref, char *path);
 fz_error *pdf_savepdf(pdf_xref *xref, char *path);
diff --git a/mupdf/font.c b/mupdf/font.c
index 6d499c60..7c75f585 100644
--- a/mupdf/font.c
+++ b/mupdf/font.c
@@ -562,17 +562,19 @@ printf("  embedded CMap\n");
 		cidtogidmap = fz_dictgets(dict, "CIDToGIDMap");
 		if (fz_isindirect(cidtogidmap))
 		{
-			unsigned char *buf;
+			fz_buffer *buf;
 			int len;
 
-			error = pdf_readstream(&buf, &len, xref, cidtogidmap);
+			error = pdf_readstream(&buf, xref, cidtogidmap);
 			if (error)
 				goto cleanup;
 
+			len = buf->wp - buf->rp;
+
 			font->cidtogidlen = len / 2;
 			font->cidtogidmap = fz_malloc((len / 2) * sizeof(int));
 			if (!font->cidtogidmap) {
-				fz_free(buf);
+				fz_freebuffer(buf);
 				error = fz_outofmem;
 				goto cleanup;
 			}
@@ -580,7 +582,9 @@ printf("  embedded CMap\n");
 printf("  cidtogidmap %d\n", len / 2);
 
 			for (i = 0; i < len / 2; i++)
-				font->cidtogidmap[i] = (buf[i * 2] << 8) + buf[i * 2 + 1];
+				font->cidtogidmap[i] = (buf->rp[i * 2] << 8) + buf->rp[i * 2 + 1];
+
+			fz_freebuffer(buf);
 		}
 
 		/* TODO: if truetype font is external, cidtogidmap should not be identity */
diff --git a/mupdf/fontfile.c b/mupdf/fontfile.c
index 3c0470f7..21a379b2 100644
--- a/mupdf/fontfile.c
+++ b/mupdf/fontfile.c
@@ -194,20 +194,17 @@ pdf_loadembeddedfont(void **fontp, pdf_xref *xref, fz_obj *stmref)
 	fz_error *error;
 	int fterr;
 	FT_Face face;
-	unsigned char *buf;
-	int len;
+	fz_buffer *buf;
 
 	error = initfontlibs();
 	if (error)
 		return error;
 
-	error = pdf_readstream(&buf, &len, xref, stmref);
+	error = pdf_readstream(&buf, xref, stmref);
 	if (error)
 		return error;
 
-printf("readstream $%p %d\n", buf, len);
-
-	fterr = FT_New_Memory_Face(ftlib, buf, len, 0, &face);
+	fterr = FT_New_Memory_Face(ftlib, buf->rp, buf->wp - buf->rp, 0, &face);
 
 	if (fterr) {
 		fz_free(buf);
diff --git a/mupdf/save.c b/mupdf/save.c
index 1698fc23..32dda709 100644
--- a/mupdf/save.c
+++ b/mupdf/save.c
@@ -1,9 +1,12 @@
 #include <fitz.h>
 #include <mupdf.h>
 
-static fz_error *writestored(FILE *out, pdf_xref *xref, int oid)
+#define TIGHT 0
+
+static fz_error *writestored(fz_file *out, pdf_xref *xref, int oid)
 {
 	pdf_xrefentry *x = xref->table + oid;
+	fz_error *error;
 	fz_obj *obj;
 	fz_buffer *stm;
 
@@ -16,26 +19,27 @@ static fz_error *writestored(FILE *out, pdf_xref *xref, int oid)
 	if (xref->crypt)
 		pdf_cryptobj(xref->crypt, obj, oid, x->gen);
 
-	fprintf(out, "%d %d obj\n", oid, x->gen);
-	fz_debugobj(obj);
-	fprintf(out, "\n");
+	fz_print(out, "%d %d obj\n", oid, x->gen);
+	fz_printobj(out, obj, TIGHT);
+	fz_print(out, "\n");
 
 	if (stm)
 	{
-		fprintf(out, "stream\n");
-		fwrite(stm->rp, 1, stm->wp - stm->rp, out);
-		fprintf(out, "endstream\n");
+		fz_print(out, "stream\n");
+		fz_write(out, stm->rp, stm->wp - stm->rp);
+		fz_print(out, "endstream\n");
 	}
 
-	fprintf(out, "endobj\n\n");
+	fz_print(out, "endobj\n\n");
 
-	if (ferror(out))
-		return fz_throw("ioerror: write failed");
+	error = fz_ferror(out);
+	if (error)
+		return error;
 
 	return nil;
 }
 
-static fz_error *writecopy(FILE *out, pdf_xref *xref, int oid)
+static fz_error *writecopy(fz_file *out, pdf_xref *xref, int oid)
 {
 	pdf_xrefentry *x = xref->table + oid;
 	fz_error *error;
@@ -52,13 +56,13 @@ static fz_error *writecopy(FILE *out, pdf_xref *xref, int oid)
 	if (error)
 		return error;
 
-	fprintf(out, "%d %d obj\n", oid, x->gen);
-    fz_debugobj(obj);
-    fprintf(out, "\n");
+	fz_print(out, "%d %d obj\n", oid, x->gen);
+	fz_printobj(out, obj, TIGHT);
+	fz_print(out, "\n");
 
 	if (stmofs != -1)
 	{
-		fprintf(out, "stream\n");
+		fz_print(out, "stream\n");
 
 		length = fz_dictgets(obj, "Length");
 		error = pdf_resolve(&length, xref);
@@ -98,20 +102,21 @@ static fz_error *writecopy(FILE *out, pdf_xref *xref, int oid)
 				fz_popfilter(xref->file);
 				goto cleanup;
 			}
-			fwrite(buf, 1, n, out);
+			fz_write(out, buf, n);
 		}
 
 		fz_popfilter(xref->file);
 
-		fprintf(out, "endstream\n");
+		fz_print(out, "endstream\n");
 	}
 
-	fprintf(out, "endobj\n\n");
+	fz_print(out, "endobj\n\n");
 
-	fz_dropobj(obj);
+	error = fz_ferror(out);
+	if (error)
+		goto cleanup;
 
-	if (ferror(out))
-        return fz_throw("ioerror: write failed");
+	fz_dropobj(obj);
 
 	return nil;
 
@@ -133,22 +138,23 @@ fz_error *
 pdf_saveincrementalpdf(pdf_xref *xref, char *path)
 {
 	fz_error *error;
-	FILE *out;
+	fz_file *out;
 	int oid;
 	int i, n;
 	int startxref;
 	fz_obj *obj;
 
-	out = fopen(path, "ab");
-	if (!out)
-		return fz_throw("ioerror: could not open '%s': %s", path, strerror(errno));
-	fprintf(out, "\n");
+	error = fz_openfile(&out, path, O_APPEND);
+	if (error)
+		return error;
+
+	fz_print(out, "\n");
 
 	for (oid = 0; oid < xref->size; oid++)
 	{
 		if (xref->table[oid].type == 'a')
 		{
-			xref->table[oid].ofs = ftell(out);
+			xref->table[oid].ofs = fz_tell(out);
 			error = writestored(out, xref, oid);
 			if (error)
 				goto cleanup;
@@ -158,15 +164,15 @@ pdf_saveincrementalpdf(pdf_xref *xref, char *path)
 	/* always write out entry 0 in appended xref sections */
 	xref->table[0].type = 'd';
 
-	startxref = ftell(out);
-	fprintf(out, "xref\n");
+	startxref = fz_tell(out);
+	fz_print(out, "xref\n");
 
 	oid = 0;
 	while (oid < xref->size)
 	{
 		n = countmodified(xref, oid);
 
-		fprintf(out, "%d %d\n", oid, n);
+		fz_print(out, "%d %d\n", oid, n);
 
 		for (i = 0; i < n; i++)
 		{
@@ -175,7 +181,7 @@ pdf_saveincrementalpdf(pdf_xref *xref, char *path)
 			if (xref->table[oid + i].type == 'a')
 				xref->table[oid + i].type = 'n';
 
-			fprintf(out, "%010d %05d %c \n",
+			fz_print(out, "%010d %05d %c \n",
 				xref->table[oid + i].ofs,
 				xref->table[oid + i].gen,
 				xref->table[oid + i].type);
@@ -188,42 +194,42 @@ pdf_saveincrementalpdf(pdf_xref *xref, char *path)
 			oid ++;
 	}
 
-	fprintf(out, "\n");
+	fz_print(out, "\n");
 
-	fprintf(out, "trailer\n<<\n  /Size %d\n  /Prev %d", xref->size, xref->startxref);
+	fz_print(out, "trailer\n<<\n  /Size %d\n  /Prev %d", xref->size, xref->startxref);
 
 	obj = fz_dictgets(xref->trailer, "Root");
-	fprintf(out,"\n  /Root %d %d R", fz_toobjid(obj), fz_togenid(obj));
+	fz_print(out,"\n  /Root %d %d R", fz_toobjid(obj), fz_togenid(obj));
 
 	obj = fz_dictgets(xref->trailer, "Info");
 	if (obj)
-		fprintf(out,"\n  /Info %d %d R", fz_toobjid(obj), fz_togenid(obj));
+		fz_print(out,"\n  /Info %d %d R", fz_toobjid(obj), fz_togenid(obj));
 
 	obj = fz_dictgets(xref->trailer, "Encrypt");
 	if (obj) {
-		fprintf(out,"\n  /Encrypt ");
-		fz_debugobj(obj);
+		fz_print(out,"\n  /Encrypt ");
+		fz_printobj(out, obj, TIGHT);
 	}
 
 	obj = fz_dictgets(xref->trailer, "ID");
 	if (obj) {
-		fprintf(out,"\n  /ID ");
-		fz_debugobj(obj);
+		fz_print(out,"\n  /ID ");
+		fz_printobj(out, obj, TIGHT);
 	}
 
-	fprintf(out, "\n>>\n\n");
+	fz_print(out, "\n>>\n\n");
 
-	fprintf(out, "startxref\n");
-	fprintf(out, "%d\n", startxref);
-	fprintf(out, "%%%%EOF\n");
+	fz_print(out, "startxref\n");
+	fz_print(out, "%d\n", startxref);
+	fz_print(out, "%%%%EOF\n");
 
 	xref->startxref = startxref;
 
-	fclose(out);
+	fz_closefile(out);
 	return nil;
 
 cleanup:
-	fclose(out);
+	fz_closefile(out);
 	return error;
 }
 
@@ -231,7 +237,7 @@ fz_error *
 pdf_savepdf(pdf_xref *xref, char *path)
 {
 	fz_error *error;
-	FILE *out;
+	fz_file *out;
 	int oid;
 	int startxref;
 	int *ofsbuf;
@@ -241,28 +247,28 @@ pdf_savepdf(pdf_xref *xref, char *path)
 	if (!ofsbuf)
 		return fz_outofmem;
 
-	out = fopen(path, "wb");
-	if (!out)
+	error = fz_openfile(&out, path, O_WRONLY);
+	if (error)
 	{
 		fz_free(ofsbuf);
-		return fz_throw("ioerror: could not open '%s': %s", path, strerror(errno));
+		return error;
 	}
 
-	fprintf(out, "%%PDF-%1.1f\n", xref->version);
-	fprintf(out, "%%\342\343\317\323\n\n");
+	fz_print(out, "%%PDF-%1.1f\n", xref->version);
+	fz_print(out, "%%\342\343\317\323\n\n");
 
 	for (oid = 0; oid < xref->size; oid++)
 	{
 		if (xref->table[oid].type == 'n' || xref->table[oid].type == 'o')
 		{
-			ofsbuf[oid] = ftell(out);
+			ofsbuf[oid] = fz_tell(out);
 			error = writecopy(out, xref, oid);
 			if (error)
 				goto cleanup;
 		}
 		else if (xref->table[oid].type == 'a')
 		{
-			ofsbuf[oid] = ftell(out);
+			ofsbuf[oid] = fz_tell(out);
 			error = writestored(out, xref, oid);
 			if (error)
 				goto cleanup;
@@ -273,9 +279,9 @@ pdf_savepdf(pdf_xref *xref, char *path)
 		}
 	}
 
-	startxref = ftell(out);
-	fprintf(out, "xref\n");
-	fprintf(out, "0 %d\n", xref->size);
+	startxref = fz_tell(out);
+	fz_print(out, "xref\n");
+	fz_print(out, "0 %d\n", xref->size);
 
 	for (oid = 0; oid < xref->size; oid++)
 	{
@@ -284,30 +290,30 @@ pdf_savepdf(pdf_xref *xref, char *path)
 			type = 'n';
 		if (type == 'd')
 			type = 'f';
-		fprintf(out, "%010d %05d %c \n", ofsbuf[oid], xref->table[oid].gen, type);
+		fz_print(out, "%010d %05d %c \n", ofsbuf[oid], xref->table[oid].gen, type);
 	}
 
-	fprintf(out, "\n");
+	fz_print(out, "\n");
 
-	fprintf(out, "trailer\n<<\n  /Size %d", xref->size);
+	fz_print(out, "trailer\n<<\n  /Size %d", xref->size);
 	obj = fz_dictgets(xref->trailer, "Root");
-	fprintf(out,"\n  /Root %d %d R", fz_toobjid(obj), fz_togenid(obj));
+	fz_print(out,"\n  /Root %d %d R", fz_toobjid(obj), fz_togenid(obj));
 	obj = fz_dictgets(xref->trailer, "Info");
 	if (obj)
-		fprintf(out,"\n  /Info %d %d R", fz_toobjid(obj), fz_togenid(obj));
-	fprintf(out, "\n>>\n\n");
+		fz_print(out,"\n  /Info %d %d R", fz_toobjid(obj), fz_togenid(obj));
+	fz_print(out, "\n>>\n\n");
 
-	fprintf(out, "startxref\n");
-	fprintf(out, "%d\n", startxref);
-	fprintf(out, "%%%%EOF\n");
+	fz_print(out, "startxref\n");
+	fz_print(out, "%d\n", startxref);
+	fz_print(out, "%%%%EOF\n");
 
 	xref->startxref = startxref;
 
-	fclose(out);
+	fz_closefile(out);
 	return nil;
 
 cleanup:
-	fclose(out);
+	fz_closefile(out);
 	return error;
 }
 
diff --git a/mupdf/stream.c b/mupdf/stream.c
index f3f40e53..ab051865 100644
--- a/mupdf/stream.c
+++ b/mupdf/stream.c
@@ -245,7 +245,23 @@ pdf_closestream(pdf_xref *xref)
 }
 
 fz_error *
-pdf_readstream(unsigned char **bufp, int *lenp, pdf_xref *xref, fz_obj *stmref)
+pdf_readstream0(fz_buffer **bufp, pdf_xref *xref, fz_obj *stmobj, int oid, int gid, int ofs)
+{
+	fz_error *error;
+
+	error = pdf_openstream0(xref, stmobj, oid, gid, ofs);
+	if (error)
+		return error;
+
+	error = fz_readfile(bufp, xref->file);
+
+	pdf_closestream(xref);
+
+	return error;
+}
+
+fz_error *
+pdf_readstream(fz_buffer **bufp, pdf_xref *xref, fz_obj *stmref)
 {
 	fz_error *error;
 
@@ -253,7 +269,7 @@ pdf_readstream(unsigned char **bufp, int *lenp, pdf_xref *xref, fz_obj *stmref)
 	if (error)
 		return error;
 
-	error = fz_readfile(bufp, lenp, xref->file);
+	error = fz_readfile(bufp, xref->file);
 
 	pdf_closestream(xref);
 
diff --git a/mupdf/xref.c b/mupdf/xref.c
index 53b9f239..08ab8621 100644
--- a/mupdf/xref.c
+++ b/mupdf/xref.c
@@ -325,6 +325,18 @@ pdf_saveobject(pdf_xref *xref, int oid, int gid, fz_obj *obj)
 	return nil;
 }
 
+fz_error *
+pdf_savestream(pdf_xref *xref, int oid, int gid, fz_buffer *buf)
+{
+	return pdf_storestream(xref->store, oid, gid, buf);
+}
+
+fz_error *
+pdf_deletestream(pdf_xref *xref, int oid, int gid)
+{
+	return pdf_deletestoredstream(xref->store, oid, gid);
+}
+
 /*
  * object loading
  */
diff --git a/test/pdfclean.c b/test/pdfclean.c
index 00721842..7eb72cda 100644
--- a/test/pdfclean.c
+++ b/test/pdfclean.c
@@ -1,30 +1,6 @@
 #include <fitz.h>
 #include <mupdf.h>
 
-#define encrypt encrypt3am
-
-int (*printobj)(FILE*,fz_obj*) = fz_fprintobj;
-
-/*
- * Rewrite PDF with cleaned up syntax, and consolidate the xref table.
- * Remove encryption while we're at it :)
- */
-
-static FILE *out;
-
-static pdf_xref *xt;
-
-static int *ofslist;
-static int *genlist;
-static int *uselist;
-
-static int dorebuild = 0;
-static int doexpand = 0;
-static pdf_crypt *encrypt = nil;
-static fz_obj *encryptdict = nil;
-static fz_obj *id = nil;
-static int dogc = 0;
-
 void usage()
 {
 	fprintf(stderr,
@@ -32,469 +8,129 @@ void usage()
 		"  -r\treconstruct broken xref table\n"
 		"  -g\tgarbage collect unused objects\n"
 		"  -x\texpand compressed streams\n"
-		"  -c\twrite compact objects\n"
 		"  -d -\tset user password for decryption\n"
 		"  -e\tencrypt outfile\n"
-		"  -u -\tset user password for encryption\n"
-		"  -o -\tset owner password\n"
-		"  -p -\tset permissions\n"
-		"  -n -\tkey length in bits: 40 <= n <= 128\n"
+		"    -u -\tset user password for encryption\n"
+		"    -o -\tset owner password\n"
+		"    -p -\tset permissions\n"
+		"    -n -\tkey length in bits: 40 <= n <= 128\n"
 		);
 	exit(1);
 }
 
-void garbagecollect(fz_obj *ref);
-
-void gc0(fz_obj *obj)
-{
-	int i;
-
-	if (fz_isdict(obj))
-		for (i = 0; i < fz_dictlen(obj); i++)
-			gc0(fz_dictgetval(obj, i));
-
-	if (fz_isarray(obj))
-		for (i = 0; i < fz_arraylen(obj); i++)
-			gc0(fz_arrayget(obj, i));
-
-	if (fz_isindirect(obj))
-		garbagecollect(obj);
-}
-
-void garbagecollect(fz_obj *ref)
-{
-	fz_obj *obj;
-	int stmofs;
-
-	if (uselist[fz_toobjid(ref)])
-		return;
-
-	uselist[fz_toobjid(ref)] = 1;
-
-	pdf_loadindirect(&obj, xt, ref, &stmofs);
-
-	if (stmofs != -1) {
-		fz_obj *len = fz_dictgets(obj, "Length");
-		if (fz_isindirect(len)) {
-			pdf_loadindirect(&len, xt, len, nil);
-			fz_dictputs(obj, "Length", len);
-		}
-	}
-
-	gc0(obj);
-}
-
-void decodestream(fz_obj *obj, int ofs, int oid, int gid)
-{
-	fz_error *error;
-	unsigned char buf[512];
-	fz_filter *filter;
-	fz_file *sf;
-	int n;
-	int len;
-	fz_obj *lenobj;
-	fz_obj *newdict;
-
-	/* count length of decoded data */
-	len = 0;
-
-	error = pdf_buildstream(&filter, xt, obj, oid, gid);
-	if (error) fz_abort(error);
-
-	n = fz_seek(xt->file, ofs);
-
-	fz_chainfile(&sf, xt->file, filter);
-
-	while (1) {
-		n = fz_read(sf, buf, sizeof buf);
-		if (n < 0)
-			fz_abort(fz_ferror(sf));
-		if (n == 0)
-			break;
-		len += n;
-	}
-
-	fz_unchainfile(sf);
-
-	/* change dictionary of object... */
-	fz_copydict(&newdict, obj);
-
-	fz_newint(&lenobj, len);
-	fz_dictputs(newdict, "Length", lenobj);
-	fz_dropobj(lenobj);
-
-	fz_dictdels(newdict, "Filter");
-	fz_dictdels(newdict, "DecodeParms");
-
-	/* save object */
-	fprintf(out, "%d %d obj\n", oid, gid);
-	printobj(out, newdict);
-	fprintf(out, "\n");
-	fprintf(out, "stream\n");
-
-	fz_dropobj(newdict);
-
-	/* now decode stream for real */
-	error = pdf_buildstream(&filter, xt, obj, oid, gid);
-	if (error) fz_abort(error);
-
-	fz_seek(xt->file, ofs);
-
-	if (encrypt) {
-		fz_filter *cf;
-		pdf_cryptstm(&cf, encrypt, oid, gid);
-		fz_newpipeline(&filter, filter, cf);
-	}
-
-	fz_chainfile(&sf, xt->file, filter);
-	while (1) {
-		n = fz_read(sf, buf, sizeof buf);
-		if (n < 0)
-			fz_abort(fz_ferror(sf));
-		if (n == 0)
-			break;
-		fwrite(buf, 1, n, out);
-	}
-	fz_unchainfile(sf);
-
-	/* the end */
-	fprintf(out, "endstream\nendobj\n\n");
-
-	return;
-}
-
-void savestream(fz_obj *obj, int ofs, int oid, int gid)
-{
-	unsigned char buf[512];
-	fz_filter *filter;
-	fz_file *sf;
-	int len;
-	int n;
-
-	/* save object */
-	fprintf(out, "%d %d obj\n", oid, gid);
-	printobj(out, obj);
-	fprintf(out, "\n");
-	fprintf(out, "stream\n");
-
-	/* copy stream */
-	obj = fz_dictgets(obj, "Length");
-	if (fz_isindirect(obj)) {
-		pdf_loadindirect(&obj, xt, obj, nil);
-		len = fz_toint(obj);
-		fz_dropobj(obj);
-	}
-	else {
-		len = fz_toint(obj);
-	}
-
-	fz_newnullfilter(&filter, len);
-
-	if (xt->crypt) {
-		fz_filter *cf;
-		pdf_cryptstm(&cf, xt->crypt, oid, gid);
-		fz_newpipeline(&filter, cf, filter);
-	}
-
-	if (encrypt) {
-		fz_filter *cf;
-		pdf_cryptstm(&cf, encrypt, oid, gid);
-		fz_newpipeline(&filter, filter, cf);
-	}
-
-	fz_seek(xt->file, ofs);
-	fz_chainfile(&sf, xt->file, filter);
-	while (1)
-	{
-		n = fz_read(sf, buf, sizeof buf);
-		if (n == 0)
-			break;
-		if (n < 0)
-			fz_abort(fz_ferror(sf));
-		fwrite(buf, 1, n, out);
-	}
-	fz_unchainfile(sf);
-
-	/* the end */
-	fprintf(out, "endstream\nendobj\n\n");
-}
-
-void deleteobject(int oid, int gid)
-{
-	uselist[oid] = 0;
-}
-
-void saveobject(int oid, int gid)
+void expandstreams(pdf_xref *xref)
 {
 	fz_error *error;
-	fz_obj *obj;
-	fz_obj *t;
+	fz_obj *stmobj;
 	int stmofs;
+	fz_buffer *buf;
+	fz_obj *stmlen;
+	int i, gen;
 
-	error = pdf_loadobj(&obj, xt, oid, gid, &stmofs);
-	if (error) fz_abort(error);
-
-	/* trash ObjStm and XrefStm objects */
-	if (fz_isdict(obj)) {
-		t = fz_dictgets(obj, "Type");
-		if (fz_isname(t) && strcmp(fz_toname(t), "ObjStm") == 0) {
-			deleteobject(oid, gid);
-			fz_dropobj(obj);
-			return;
-		}
-		if (fz_isname(t) && strcmp(fz_toname(t), "XRef") == 0) {
-			deleteobject(oid, gid);
-			fz_dropobj(obj);
-			return;
+	for (i = 0; i < xref->size; i++)
+	{
+		if (xref->table[i].type == 'n')
+		{
+			gen = xref->table[i].gen;
+
+			error = pdf_loadobject0(&stmobj, xref, i, gen, &stmofs);
+			if (error) fz_abort(error);
+
+			if (stmofs != -1)
+			{
+				error = pdf_readstream0(&buf, xref, stmobj, i, gen, stmofs);
+				if (error) fz_abort(error);
+
+				fz_dictdels(stmobj, "Filter");
+				fz_dictdels(stmobj, "DecodeParms");
+
+				error = fz_newint(&stmlen, buf->wp - buf->rp);
+				if (error) fz_abort(error);
+				error = fz_dictputs(stmobj, "Length", stmlen);
+				if (error) fz_abort(error);
+				fz_dropobj(stmlen);
+
+				error = pdf_saveobject(xref, i, gen, stmobj);
+				if (error) fz_abort(error);
+				error = pdf_savestream(xref, i, gen, buf);
+				if (error) fz_abort(error);
+			}
 		}
 	}
-
-	if (encrypt)
-		pdf_cryptobj(encrypt, obj, oid, gid);
-
-	if (stmofs == -1) {
-		fprintf(out, "%d %d obj\n", oid, gid);
-		printobj(out, obj);
-		fprintf(out, "\nendobj\n\n");
-	}
-	else if (doexpand) {
-		decodestream(obj, stmofs, oid, gid);
-	}
-	else {
-		savestream(obj, stmofs, oid, gid);
-	}
-
-	fz_dropobj(obj);
-}
-
-void savexref(void)
-{
-	fz_obj *newtrailer;
-	fz_obj *obj;
-	int startxref;
-	int i;
-
-	startxref = ftell(out);
-
-	fprintf(out, "xref\n0 %d\n", xt->size);
-	for (i = 0; i < xt->size; i++) {
-		if (uselist[i])
-			fprintf(out, "%010d %05d n \n", ofslist[i], genlist[i]);
-		else
-			fprintf(out, "%010d %05d f \n", ofslist[i], genlist[i]);
-	}
-	fprintf(out, "\n");
-
-	fz_newdict(&newtrailer, 5);
-
-	fz_newint(&obj, xt->size);
-	fz_dictputs(newtrailer, "Size", obj);
-	fz_dropobj(obj);
-
-	obj = fz_dictgets(xt->trailer, "Info");
-	if (obj) fz_dictputs(newtrailer, "Info", obj);
-
-	obj = fz_dictgets(xt->trailer, "Root");
-	if (obj) fz_dictputs(newtrailer, "Root", obj);
-
-	fz_dictputs(newtrailer, "ID", id);
-
-	if (encryptdict) {
-		fz_newindirect(&obj, xt->size - 1, 0);
-		fz_dictputs(newtrailer, "Encrypt", obj);
-		fz_dropobj(obj);
-	}
-
-	fprintf(out, "trailer\n");
-	printobj(out, newtrailer);
-	fprintf(out, "\n\n");
-
-	fprintf(out, "startxref\n%d\n%%%%EOF\n", startxref);
 }
 
 int main(int argc, char **argv)
 {
 	fz_error *error;
-	fz_obj *obj;
-	int lastfree;
-	char *filename;
-	int i;
+	char *infile;
+	char *outfile;
+	pdf_xref *xref;
 	int c;
 
 	int doencrypt = 0;
-	char *password = "";
+	int dorepair = 0;
+	int doexpand = 0;
+	int dogc = 0;
+
 	char *userpw = "";
 	char *ownerpw = "";
 	int perms = -4; /* 0xfffffffc */
 	int keylen = 40;
+	char *password = "";
 
-	while (1)
+	while ((c = getopt(argc, argv, "rgxd:eu:o:p:n:")) != -1)
 	{
-		c = getopt(argc, argv, "rcxgeu:o:p:n:d:");
-
-		if (c == -1)
-			break;
-
 		switch (c)
 		{
-		case 'r':
-			dorebuild ++;
-			break;
-		case 'x':
-			doexpand ++;
-			break;
-		case 'g':
-			dogc ++;
-			break;
-		case 'c':
-			printobj = fz_fprintcobj;
-			break;
-		case 'd':
-			password = optarg;
-			break;
-		case 'e':
-			doencrypt ++;
-			break;
-		case 'u':
-			userpw = optarg;
-			break;
-		case 'o':
-			ownerpw = optarg;
-			break;
-		case 'p':
-			perms = atoi(optarg);
-			break;
-		case 'n':
-			keylen = atoi(optarg);
-			break;
-		default:
-			usage();
+		case 'r': ++ dorepair; break;
+		case 'x': ++ doexpand; break;
+		case 'g': ++ dogc; break;
+		case 'e': ++ doencrypt; break;
+		case 'u': userpw = optarg; break;
+		case 'o': ownerpw = optarg; break;
+		case 'p': perms = atoi(optarg); break;
+		case 'n': keylen = atoi(optarg); break;
+		case 'd': password = optarg; break;
+		default: usage();
 		}
 	}
 
-	if (argc - optind != 2)
+	if (argc - optind < 2)
 		usage();
 
-	filename = argv[optind];
+	infile = argv[optind++];
+	outfile = argv[optind++];
 
-	if (dorebuild)
-		error = pdf_rebuildxref(&xt, filename);
-	else
-		error = pdf_openxref(&xt, filename);
-	if (error) fz_abort(error);
-
-	if (doencrypt && keylen > 40 && xt->version < 1.4)
-		xt->version = 1.4;
-
-	id = fz_dictgets(xt->trailer, "ID");
-	obj = fz_dictgets(xt->trailer, "Encrypt");
-	if (fz_isindirect(obj)) {
-		pdf_loadindirect(&obj, xt, obj, nil);
-	}
-	if (obj && id) {
-		pdf_newdecrypt(&xt->crypt, obj, id);
-		error = pdf_setpassword(xt->crypt, password);
-		if (error) fz_abort(error);
-	}
-
-	id = fz_dictgets(xt->trailer, "ID");
-	if (!id)
-		fz_parseobj(&id, "[ (foobar) (foobar) ]");
-
-	if (doencrypt)
-		pdf_newencrypt(&encrypt, &encryptdict, userpw, ownerpw, perms, keylen, id);
-
-	out = fopen(argv[optind + 1], "w");
-	if (!out) {
-		fz_abort(fz_throw("open(%s): %s", argv[optind + 1], strerror(errno)));
-	}
-
-	fprintf(out, "%%PDF-%.1f\n\n", xt->version);
-
-	ofslist = fz_malloc(sizeof(int) * (xt->size + 1));
-	genlist = fz_malloc(sizeof(int) * (xt->size + 1));
-	uselist = fz_malloc(sizeof(int) * (xt->size + 1));
-
-	lastfree = 0;
-
-	ofslist[0] = 0;
-	genlist[0] = 65535;
-	uselist[0] = 0;
-
-	for (i = 1; i < xt->size; i++) {
-		ofslist[i] = 0;
-		genlist[i] = 0;
-		uselist[i] = 1;
-	}
+	error = pdf_newxref(&xref);
+	if (error)
+		fz_abort(error);
 
-	/* garbage collect from roots in trailer */
-	if (dogc)
-	{
-		for (i = 1; i < xt->size; i++)
-			uselist[i] = 0;
-
-		obj = fz_dictgets(xt->trailer, "Info");
-		if (fz_isindirect(obj))
-			garbagecollect(obj);
-
-		obj = fz_dictgets(xt->trailer, "Root");
-		if (fz_isindirect(obj))
-			garbagecollect(obj);
+	if (dorepair)
+		error = pdf_repairxref(xref, infile);
+	else
+		error = pdf_openxref(xref, infile);
+	if (error)
+		fz_abort(error);
 
-		obj = fz_dictgets(xt->trailer, "ID");
-		if (fz_isindirect(obj))
-			garbagecollect(obj);
-	}
+	error = pdf_decryptxref(xref);
+	if (error)
+		fz_abort(error);
 
-	/* pretty print objects */
-	for (i = 0; i < xt->size; i++)
+	if (xref->crypt)
 	{
-		if (xt->table[i].type == 0)
-			uselist[i] = 0;
-
-		if (xt->table[i].type == 0)
-			genlist[i] = xt->table[i].gen;
-		if (xt->table[i].type == 1)
-			genlist[i] = xt->table[i].gen;
-		if (xt->table[i].type == 2)
-			genlist[i] = 0;
-
-		if (dogc && !uselist[i])
-			continue;
-
-		if (xt->table[i].type == 1 || xt->table[i].type == 2)
-		{
-			ofslist[i] = ftell(out);
-			saveobject(i, genlist[i]);
-		}
-	}
-
-	/* add encryption dictionary if we crypted */
-	if (encryptdict) {
-		xt->size ++;
-		ofslist[xt->size - 1] = ftell(out);
-		genlist[xt->size - 1] = 0;
-		uselist[xt->size - 1] = 1;
-		fprintf(out, "%d %d obj\n", xt->size - 1, 0);
-		printobj(out, encryptdict);
-		fprintf(out, "\nendobj\n\n");
-	}
-
-	/* construct linked list of free object slots */
-	lastfree = 0;
-	for (i = 1; i < xt->size; i++) {
-		if (!uselist[i]) {
-			genlist[i] ++;
-			ofslist[lastfree] = i;
-			lastfree = i;
-		}
+		error = pdf_setpassword(xref->crypt, password);
+		if (error) fz_abort(error);
 	}
 
-	savexref();
+	if (doexpand)
+		expandstreams(xref);
 
-	pdf_closexref(xt);
+printf("saving %s...\n", outfile);
+	error = pdf_savepdf(xref, outfile);
+	if (error)
+		fz_abort(error);
 
-	fclose(out);
+	pdf_closexref(xref);
 
 	return 0;
 }
diff --git a/test/pdfdebug.c b/test/pdfdebug.c
index d63a720c..ddc10dff 100644
--- a/test/pdfdebug.c
+++ b/test/pdfdebug.c
@@ -5,11 +5,10 @@ static char *password = "";
 static int dodecode = 0;
 static int dorepair = 0;
 static int doprintxref = 0;
-static int doprintpages = 0;
 
 void usage()
 {
-	fprintf(stderr, "usage: mupdf [-drxp] [-u password] file.pdf\n");
+	fprintf(stderr, "usage: pdfdebug [-drx] [-u password] file.pdf [oid ...]\n");
 	exit(1);
 }
 
@@ -127,114 +126,11 @@ void printobject(pdf_xref *xref, int oid, int gid)
 	fz_dropobj(obj);
 }
 
-/*
- * Draw page
- */
-
-void runcsi(pdf_xref *xref, pdf_csi *csi, pdf_resources *rdb, fz_obj *stmref)
-{
-	fz_error *error;
-
-	error = pdf_openstream(xref, stmref);
-	if (error) fz_abort(error);
-
-	error = pdf_runcsi(csi, rdb, xref->file);
-	if (error) fz_abort(error);
-
-	pdf_closestream(xref);
-}
-
-void showpage(pdf_xref *xref, fz_obj *page)
-{
-	fz_error *error;
-	pdf_csi *csi;
-	pdf_resources *rdb = nil;
-	fz_obj *resources;
-	fz_obj *contents;
-	int i;
-
-	fz_debugobj(page);
-	printf("\n");
-
-	resources = fz_dictgets(page, "Resources");
-	if (resources)
-	{
-		error = pdf_resolve(&resources, xref);
-		if (error) fz_abort(error);
-
-		error = pdf_loadresources(&rdb, xref, resources);
-		if (error) fz_abort(error);
-
-		// parse resources into native res dict
-		fz_dropobj(resources);
-	}
-	else
-		fz_abort(fz_throw("syntaxerror: missing resource dictionary"));
-
-printf("resources:\n");
-printf("  font:\n");
-fz_debugobj(rdb->font);
-printf("\n  extgstate:\n");
-fz_debugobj(rdb->extgstate);
-printf("\nfitz tree:\n");
-
-	error = pdf_newcsi(&csi);
-	if (error) fz_abort(error);
-
-	contents = fz_dictgets(page, "Contents");
-	if (contents)
-	{
-		if (fz_isarray(contents))
-		{
-			for (i = 0; i < fz_arraylen(contents); i++)
-			{
-				runcsi(xref, csi, rdb, fz_arrayget(contents, i));
-			}
-		}
-		else
-		{
-			// XXX resolve and check if it is an array
-			runcsi(xref, csi, rdb, contents);
-		}
-	}
-
-	fz_debugtree(csi->tree);
-
-	{
-		fz_pixmap *pix;
-		fz_renderer *gc;
-		fz_matrix ctm;
-
-#define W 612
-#define H 792
-
-#define xW 1106
-#define xH 1548
-
-		fz_newrenderer(&gc);
-		fz_newpixmap(&pix, 0, 0, W, H, 1, 0);
-		ctm = fz_concat(fz_translate(0, -H), fz_scale(1,-1));
-
-		memset(pix->samples, 0x00, pix->stride * pix->h * 2);
-
-printf("rendering!\n");
-		fz_rendernode(gc, csi->tree->root, ctm, pix);
-printf("done!\n");
-		fz_debugpixmap(pix);
-
-		fz_freepixmap(pix);
-		fz_freerenderer(gc);
-	}
-
-	pdf_freecsi(csi);
-}
-
 int main(int argc, char **argv)
 {
 	fz_error *error;
 	char *filename;
 	pdf_xref *xref;
-	pdf_pagetree *pages;
 	int c;
 
 	while ((c = getopt(argc, argv, "drxopu:")) != -1)
@@ -250,9 +146,6 @@ int main(int argc, char **argv)
 		case 'x':
 			doprintxref ++;
 			break;
-		case 'p':
-			doprintpages ++;
-			break;
 		case 'u':
 			password = optarg;
 			break;
@@ -290,48 +183,19 @@ int main(int argc, char **argv)
 	if (doprintxref)
 		pdf_debugxref(xref);
 
-	if (doprintpages)
+	if (optind == argc)
 	{
-		error = pdf_loadpagetree(&pages, xref);
-		if (error) fz_abort(error);
-
-		if (optind == argc)
-		{
-			printf("pagetree\n");
-			pdf_debugpagetree(pages);
-			printf("\n");
-		}
-		else
-		{
-			for ( ; optind < argc; optind++)
-			{
-				int page = atoi(argv[optind]);
-				if (page < 1 || page > pages->count)
-					fprintf(stderr, "page out of bounds: %d\n", page);
-				printf("page %d\n", page);
-				showpage(xref, pages->pobj[page - 1]);
-			}
-		}
+		printf("trailer\n");
+		fz_debugobj(xref->trailer);
+		printf("\n");
 	}
 
-	else
+	for ( ; optind < argc; optind++)
 	{
-		if (optind == argc)
-		{
-			printf("trailer\n");
-			fz_debugobj(xref->trailer);
-			printf("\n");
-		}
-
-		for ( ; optind < argc; optind++)
-		{
-			printobject(xref, atoi(argv[optind]), 0);
-			printf("\n");
-		}
+		printobject(xref, atoi(argv[optind]), 0);
+		printf("\n");
 	}
 
-printf("done.\n");
-
 	pdf_closexref(xref);
 
 	return 0;