Clean up xps and pdf page access functions.

13 files changed, 251 insertions, 280 deletions

diff --git a/Makefile b/Makefile
index 30902f1a..72c9a285 100644
--- a/Makefile
+++ b/Makefile
@@ -154,7 +154,6 @@ MUPDF_SRC := \
 	pdf/pdf_nametree.c \
 	pdf/pdf_outline.c \
 	pdf/pdf_page.c \
-	pdf/pdf_pagetree.c \
 	pdf/pdf_parse.c \
 	pdf/pdf_pattern.c \
 	pdf/pdf_repair.c \
diff --git a/apps/pdfapp.c b/apps/pdfapp.c
index 277bdbf0..46459497 100644
--- a/apps/pdfapp.c
+++ b/apps/pdfapp.c
@@ -159,7 +159,7 @@ static void pdfapp_open_pdf(pdfapp_t *app, char *filename, int fd)
 	if (error)
 		pdfapp_error(app, fz_rethrow(error, "cannot load page tree"));
 
-	app->pagecount = pdf_get_page_count(app->xref);
+	app->pagecount = pdf_count_pages(app->xref);
 }
 
 static void pdfapp_open_xps(pdfapp_t *app, char *filename, int fd)
@@ -168,8 +168,7 @@ static void pdfapp_open_xps(pdfapp_t *app, char *filename, int fd)
 
 	close(fd); // TODO: fix this for windows
 
-	app->xps = xps_new_context();
-	error = xps_open_file(app->xps, filename);
+	error = xps_open_file(&app->xps, filename);
 	if (error)
 		pdfapp_error(app, fz_rethrow(error, "cannot open document: '%s'", filename));
 
@@ -278,10 +277,8 @@ static void pdfapp_loadpage_pdf(pdfapp_t *app)
 	pdf_page *page;
 	fz_error error;
 	fz_device *mdev;
-	fz_obj *obj;
 
-	obj = pdf_get_page_object(app->xref, app->pageno);
-	error = pdf_load_page(&page, app->xref, obj);
+	error = pdf_load_page(&page, app->xref, app->pageno - 1);
 	if (error)
 		pdfapp_error(app, error);
 
@@ -310,10 +307,11 @@ static void pdfapp_loadpage_xps(pdfapp_t *app)
 {
 	xps_page *page;
 	fz_device *mdev;
+	fz_error error;
 
-	page = xps_load_page(app->xps, app->pageno - 1);
-	if (!page)
-		pdfapp_error(app, fz_rethrow(-1, "cannot load page %d in file '%s'", app->pageno, app->doctitle));
+	error = xps_load_page(&page, app->xps, app->pageno - 1);
+	if (error)
+		pdfapp_error(app, fz_rethrow(error, "cannot load page %d in file '%s'", app->pageno, app->doctitle));
 
 	app->page_bbox.x0 = 0;
 	app->page_bbox.y0 = 0;
@@ -436,9 +434,11 @@ static void pdfapp_gotouri(pdfapp_t *app, fz_obj *uri)
 
 static void pdfapp_gotopage(pdfapp_t *app, fz_obj *obj)
 {
-	int page;
+	int number;
 
-	page = pdf_find_page_object(app->xref, obj);
+	number = pdf_find_page_number(app->xref, obj);
+	if (number < 0)
+		return;
 
 	if (app->histlen + 1 == 256)
 	{
@@ -446,7 +446,7 @@ static void pdfapp_gotopage(pdfapp_t *app, fz_obj *obj)
 		app->histlen --;
 	}
 	app->hist[app->histlen++] = app->pageno;
-	app->pageno = page;
+	app->pageno = number + 1;
 	pdfapp_showpage(app, 1, 1, 1);
 }
 
diff --git a/apps/pdfclean.c b/apps/pdfclean.c
index 67c7809c..dd85b1a4 100644
--- a/apps/pdfclean.c
+++ b/apps/pdfclean.c
@@ -314,7 +314,7 @@ static void retainpages(int argc, char **argv)
 			dash = strchr(spec, '-');
 
 			if (dash == spec)
-				spage = epage = pdf_get_page_count(xref);
+				spage = epage = pdf_count_pages(xref);
 			else
 				spage = epage = atoi(spec);
 
@@ -323,7 +323,7 @@ static void retainpages(int argc, char **argv)
 				if (strlen(dash) > 1)
 					epage = atoi(dash + 1);
 				else
-					epage = pdf_get_page_count(xref);
+					epage = pdf_count_pages(xref);
 			}
 
 			if (spage > epage)
@@ -331,13 +331,13 @@ static void retainpages(int argc, char **argv)
 
 			if (spage < 1)
 				spage = 1;
-			if (epage > pdf_get_page_count(xref))
-				epage = pdf_get_page_count(xref);
+			if (epage > pdf_count_pages(xref))
+				epage = pdf_count_pages(xref);
 
 			for (page = spage; page <= epage; page++)
 			{
-				fz_obj *pageobj = pdf_get_page_object(xref, page);
-				fz_obj *pageref = pdf_get_page_ref(xref, page);
+				fz_obj *pageobj = xref->page_objs[page-1];
+				fz_obj *pageref = xref->page_refs[page-1];
 
 				fz_dict_puts(pageobj, "Parent", parent);
 
diff --git a/apps/pdfdraw.c b/apps/pdfdraw.c
index f15fdfb4..baaac92c 100644
--- a/apps/pdfdraw.c
+++ b/apps/pdfdraw.c
@@ -87,7 +87,6 @@ static int isrange(char *s)
 static void drawpage(pdf_xref *xref, int pagenum)
 {
 	fz_error error;
-	fz_obj *pageobj;
 	pdf_page *page;
 	fz_display_list *list;
 	fz_device *dev;
@@ -98,8 +97,7 @@ static void drawpage(pdf_xref *xref, int pagenum)
 		start = gettime();
 	}
 
-	pageobj = pdf_get_page_object(xref, pagenum);
-	error = pdf_load_page(&page, xref, pageobj);
+	error = pdf_load_page(&page, xref, pagenum - 1);
 	if (error)
 		die(fz_rethrow(error, "cannot load page %d in file '%s'", pagenum, filename));
 
@@ -253,7 +251,7 @@ static void drawrange(pdf_xref *xref, char *range)
 		dash = strchr(spec, '-');
 
 		if (dash == spec)
-			spage = epage = pdf_get_page_count(xref);
+			spage = epage = pdf_count_pages(xref);
 		else
 			spage = epage = atoi(spec);
 
@@ -262,11 +260,11 @@ static void drawrange(pdf_xref *xref, char *range)
 			if (strlen(dash) > 1)
 				epage = atoi(dash + 1);
 			else
-				epage = pdf_get_page_count(xref);
+				epage = pdf_count_pages(xref);
 		}
 
-		spage = CLAMP(spage, 1, pdf_get_page_count(xref));
-		epage = CLAMP(epage, 1, pdf_get_page_count(xref));
+		spage = CLAMP(spage, 1, pdf_count_pages(xref));
+		epage = CLAMP(epage, 1, pdf_count_pages(xref));
 
 		if (spage < epage)
 			for (page = spage; page <= epage; page++)
diff --git a/apps/pdfinfo.c b/apps/pdfinfo.c
index fcc2cb38..d8948264 100644
--- a/apps/pdfinfo.c
+++ b/apps/pdfinfo.c
@@ -569,8 +569,8 @@ gatherresourceinfo(int page, fz_obj *rsrc)
 	fz_obj *subrsrc;
 	int i;
 
-	pageobj = pdf_get_page_object(xref, page);
-	pageref = pdf_get_page_ref(xref, page);
+	pageobj = xref->page_objs[page-1];
+	pageref = xref->page_refs[page-1];
 
 	if (!pageobj)
 		die(fz_throw("cannot retrieve info from page %d", page));
@@ -632,8 +632,8 @@ gatherpageinfo(int page)
 	fz_obj *pageref;
 	fz_obj *rsrc;
 
-	pageobj = pdf_get_page_object(xref, page);
-	pageref = pdf_get_page_ref(xref, page);
+	pageobj = xref->page_objs[page-1];
+	pageref = xref->page_refs[page-1];
 
 	if (!pageobj)
 		die(fz_throw("cannot retrieve info from page %d", page));
@@ -988,7 +988,7 @@ int main(int argc, char **argv)
 			error = pdf_load_page_tree(xref);
 			if (error)
 				die(fz_rethrow(error, "cannot load page tree: %s", filename));
-			pagecount = pdf_get_page_count(xref);
+			pagecount = pdf_count_pages(xref);
 
 			showglobalinfo();
 			state = NO_INFO_GATHERED;
diff --git a/apps/pdfshow.c b/apps/pdfshow.c
index 4c834035..5e74042b 100644
--- a/apps/pdfshow.c
+++ b/apps/pdfshow.c
@@ -61,10 +61,10 @@ static void showpagetree(void)
 			die(fz_rethrow(error, "cannot load page tree"));
 	}
 
-	count = pdf_get_page_count(xref);
+	count = pdf_count_pages(xref);
 	for (i = 0; i < count; i++)
 	{
-		ref = pdf_get_page_ref(xref, i + 1);
+		ref = xref->page_refs[i];
 		printf("page %d = %d %d R\n", i + 1, fz_to_num(ref), fz_to_gen(ref));
 	}
 	printf("\n");
diff --git a/apps/xpsdraw.c b/apps/xpsdraw.c
index 41d01c14..29738e6b 100644
--- a/apps/xpsdraw.c
+++ b/apps/xpsdraw.c
@@ -90,15 +90,16 @@ static void drawpage(xps_context *ctx, int pagenum)
 	fz_display_list *list;
 	fz_device *dev;
 	int start;
+	int code;
 
 	if (showtime)
 	{
 		start = gettime();
 	}
 
-	page = xps_load_page(ctx, pagenum - 1);
-	if (!page)
-		die(fz_throw("cannot load page %d in file '%s'", pagenum, filename));
+	code = xps_load_page(&page, ctx, pagenum - 1);
+	if (code)
+		die(fz_rethrow(code, "cannot load page %d in file '%s'", pagenum, filename));
 
 	list = NULL;
 
@@ -333,8 +334,7 @@ int main(int argc, char **argv)
 	{
 		filename = argv[fz_optind++];
 
-		ctx = xps_new_context();
-		code = xps_open_file(ctx, filename);
+		code = xps_open_file(&ctx, filename);
 		if (code)
 			die(fz_rethrow(code, "cannot open document: %s", filename));
 
diff --git a/pdf/mupdf.h b/pdf/mupdf.h
index 96ac2da3..daee0396 100644
--- a/pdf/mupdf.h
+++ b/pdf/mupdf.h
@@ -522,13 +522,11 @@ struct pdf_page_s
 
 /* pagetree.c */
 fz_error pdf_load_page_tree(pdf_xref *xref);
-int pdf_get_page_count(pdf_xref *xref);
-fz_obj *pdf_get_page_object(pdf_xref *xref, int p);
-fz_obj *pdf_get_page_ref(pdf_xref *xref, int p);
-int pdf_find_page_object(pdf_xref *xref, fz_obj *pageobj);
+int pdf_find_page_number(pdf_xref *xref, fz_obj *pageobj);
+int pdf_count_pages(pdf_xref *xref);
 
 /* page.c */
-fz_error pdf_load_page(pdf_page **pagep, pdf_xref *xref, fz_obj *ref);
+fz_error pdf_load_page(pdf_page **pagep, pdf_xref *xref, int number);
 void pdf_free_page(pdf_page *page);
 
 /*
diff --git a/pdf/pdf_page.c b/pdf/pdf_page.c
index 685ec3ae..39cae41d 100644
--- a/pdf/pdf_page.c
+++ b/pdf/pdf_page.c
@@ -1,67 +1,123 @@
 #include "fitz.h"
 #include "mupdf.h"
 
-/* we need to combine all sub-streams into one for the content stream interpreter */
+struct info
+{
+	fz_obj *resources;
+	fz_obj *mediabox;
+	fz_obj *cropbox;
+	fz_obj *rotate;
+};
+
+int
+pdf_count_pages(pdf_xref *xref)
+{
+	return xref->page_len;
+}
 
-static fz_error
-pdf_load_page_contents_array(fz_buffer **bigbufp, pdf_xref *xref, fz_obj *list)
+int
+pdf_find_page_number(pdf_xref *xref, fz_obj *page)
 {
-	fz_error error;
-	fz_buffer *big;
-	fz_buffer *one;
-	int i;
+	int i, num = fz_to_num(page);
+	for (i = 0; i < xref->page_len; i++)
+		if (num == fz_to_num(xref->page_refs[i]))
+			return i;
+	return -1;
+}
 
-	pdf_log_page("multiple content streams: %d\n", fz_array_len(list));
+static void
+pdf_load_page_tree_node(pdf_xref *xref, fz_obj *node, struct info info)
+{
+	fz_obj *dict, *kids, *count;
+	fz_obj *obj, *tmp;
+	int i, n;
 
-	/* TODO: openstream, read, close into big buffer at once */
+	/* prevent infinite recursion */
+	if (fz_dict_gets(node, ".seen"))
+		return;
 
-	big = fz_new_buffer(32 * 1024);
+	kids = fz_dict_gets(node, "Kids");
+	count = fz_dict_gets(node, "Count");
 
-	for (i = 0; i < fz_array_len(list); i++)
+	if (fz_is_array(kids) && fz_is_int(count))
 	{
-		fz_obj *stm = fz_array_get(list, i);
-		error = pdf_load_stream(&one, xref, fz_to_num(stm), fz_to_gen(stm));
-		if (error)
+		obj = fz_dict_gets(node, "Resources");
+		if (obj)
+			info.resources = obj;
+		obj = fz_dict_gets(node, "MediaBox");
+		if (obj)
+			info.mediabox = obj;
+		obj = fz_dict_gets(node, "CropBox");
+		if (obj)
+			info.cropbox = obj;
+		obj = fz_dict_gets(node, "Rotate");
+		if (obj)
+			info.rotate = obj;
+
+		tmp = fz_new_null();
+		fz_dict_puts(node, ".seen", tmp);
+		fz_drop_obj(tmp);
+
+		n = fz_array_len(kids);
+		for (i = 0; i < n; i++)
 		{
-			fz_drop_buffer(big);
-			return fz_rethrow(error, "cannot load content stream part %d/%d (%d %d R)", i + 1, fz_array_len(list), fz_to_num(stm), fz_to_gen(stm));
+			obj = fz_array_get(kids, i);
+			pdf_load_page_tree_node(xref, obj, info);
 		}
 
-		if (big->len + one->len + 1 > big->cap)
-			fz_resize_buffer(big, big->len + one->len + 1);
-		memcpy(big->data + big->len, one->data, one->len);
-		big->data[big->len + one->len] = ' ';
-		big->len += one->len + 1;
-
-		fz_drop_buffer(one);
+		fz_dict_dels(node, ".seen");
 	}
+	else
+	{
+		dict = fz_resolve_indirect(node);
+
+		if (info.resources && !fz_dict_gets(dict, "Resources"))
+			fz_dict_puts(dict, "Resources", info.resources);
+		if (info.mediabox && !fz_dict_gets(dict, "MediaBox"))
+			fz_dict_puts(dict, "MediaBox", info.mediabox);
+		if (info.cropbox && !fz_dict_gets(dict, "CropBox"))
+			fz_dict_puts(dict, "CropBox", info.cropbox);
+		if (info.rotate && !fz_dict_gets(dict, "Rotate"))
+			fz_dict_puts(dict, "Rotate", info.rotate);
+
+		if (xref->page_len == xref->page_cap)
+		{
+			fz_warn("found more pages than expected");
+			xref->page_cap ++;
+			xref->page_refs = fz_realloc(xref->page_refs, xref->page_cap, sizeof(fz_obj*));
+			xref->page_objs = fz_realloc(xref->page_objs, xref->page_cap, sizeof(fz_obj*));
+		}
 
-	*bigbufp = big;
-	return fz_okay;
+		xref->page_refs[xref->page_len] = fz_keep_obj(node);
+		xref->page_objs[xref->page_len] = fz_keep_obj(dict);
+		xref->page_len ++;
+	}
 }
 
-static fz_error
-pdf_load_page_contents(fz_buffer **bufp, pdf_xref *xref, fz_obj *obj)
+fz_error
+pdf_load_page_tree(pdf_xref *xref)
 {
-	fz_error error;
+	struct info info;
+	fz_obj *catalog = fz_dict_gets(xref->trailer, "Root");
+	fz_obj *pages = fz_dict_gets(catalog, "Pages");
+	fz_obj *count = fz_dict_gets(pages, "Count");
 
-	if (fz_is_array(obj))
-	{
-		error = pdf_load_page_contents_array(bufp, xref, obj);
-		if (error)
-			return fz_rethrow(error, "cannot load content stream array (%d 0 R)", fz_to_num(obj));
-	}
-	else if (pdf_is_stream(xref, fz_to_num(obj), fz_to_gen(obj)))
-	{
-		error = pdf_load_stream(bufp, xref, fz_to_num(obj), fz_to_gen(obj));
-		if (error)
-			return fz_rethrow(error, "cannot load content stream (%d 0 R)", fz_to_num(obj));
-	}
-	else
-	{
-		fz_warn("page contents missing, leaving page blank");
-		*bufp = fz_new_buffer(0);
-	}
+	if (!fz_is_dict(pages))
+		return fz_throw("missing page tree");
+	if (!fz_is_int(count))
+		return fz_throw("missing page count");
+
+	xref->page_cap = fz_to_int(count);
+	xref->page_len = 0;
+	xref->page_refs = fz_calloc(xref->page_cap, sizeof(fz_obj*));
+	xref->page_objs = fz_calloc(xref->page_cap, sizeof(fz_obj*));
+
+	info.resources = NULL;
+	info.mediabox = NULL;
+	info.cropbox = NULL;
+	info.rotate = NULL;
+
+	pdf_load_page_tree_node(xref, pages, info);
 
 	return fz_okay;
 }
@@ -149,21 +205,92 @@ found:
 	return 1;
 }
 
+/* we need to combine all sub-streams into one for the content stream interpreter */
+
+static fz_error
+pdf_load_page_contents_array(fz_buffer **bigbufp, pdf_xref *xref, fz_obj *list)
+{
+	fz_error error;
+	fz_buffer *big;
+	fz_buffer *one;
+	int i;
+
+	pdf_log_page("multiple content streams: %d\n", fz_array_len(list));
+
+	/* TODO: openstream, read, close into big buffer at once */
+
+	big = fz_new_buffer(32 * 1024);
+
+	for (i = 0; i < fz_array_len(list); i++)
+	{
+		fz_obj *stm = fz_array_get(list, i);
+		error = pdf_load_stream(&one, xref, fz_to_num(stm), fz_to_gen(stm));
+		if (error)
+		{
+			fz_drop_buffer(big);
+			return fz_rethrow(error, "cannot load content stream part %d/%d (%d 0 R)",
+				i + 1, fz_array_len(list), fz_to_num(stm));
+		}
+
+		if (big->len + one->len + 1 > big->cap)
+			fz_resize_buffer(big, big->len + one->len + 1);
+		memcpy(big->data + big->len, one->data, one->len);
+		big->data[big->len + one->len] = ' ';
+		big->len += one->len + 1;
+
+		fz_drop_buffer(one);
+	}
+
+	*bigbufp = big;
+	return fz_okay;
+}
+
+static fz_error
+pdf_load_page_contents(fz_buffer **bufp, pdf_xref *xref, fz_obj *obj)
+{
+	fz_error error;
+
+	if (fz_is_array(obj))
+	{
+		error = pdf_load_page_contents_array(bufp, xref, obj);
+		if (error)
+			return fz_rethrow(error, "cannot load content stream array (%d 0 R)", fz_to_num(obj));
+	}
+	else if (pdf_is_stream(xref, fz_to_num(obj), fz_to_gen(obj)))
+	{
+		error = pdf_load_stream(bufp, xref, fz_to_num(obj), fz_to_gen(obj));
+		if (error)
+			return fz_rethrow(error, "cannot load content stream (%d 0 R)", fz_to_num(obj));
+	}
+	else
+	{
+		fz_warn("page contents missing, leaving page blank");
+		*bufp = fz_new_buffer(0);
+	}
+
+	return fz_okay;
+}
+
 fz_error
-pdf_load_page(pdf_page **pagep, pdf_xref *xref, fz_obj *dict)
+pdf_load_page(pdf_page **pagep, pdf_xref *xref, int number)
 {
 	fz_error error;
 	pdf_page *page;
+	fz_obj *dict;
 	fz_obj *obj;
 	fz_bbox bbox;
 
+	if (number < 0 || number >= xref->page_len)
+		return fz_throw("cannot find page %d", number + 1);
+
 	pdf_log_page("load page {\n");
 
-	// TODO: move this to a more appropriate place
 	/* Ensure that we have a store for resource objects */
 	if (!xref->store)
 		xref->store = pdf_new_store();
 
+	dict = xref->page_objs[number];
+
 	page = fz_malloc(sizeof(pdf_page));
 	page->resources = NULL;
 	page->contents = NULL;
@@ -175,7 +302,7 @@ pdf_load_page(pdf_page **pagep, pdf_xref *xref, fz_obj *dict)
 	bbox = fz_round_rect(pdf_to_rect(obj));
 	if (fz_is_empty_rect(pdf_to_rect(obj)))
 	{
-		fz_warn("cannot find page bounds, guessing page bounds.");
+		fz_warn("cannot find page size for page %d", number + 1);
 		bbox.x0 = 0;
 		bbox.y0 = 0;
 		bbox.x1 = 612;
@@ -195,7 +322,10 @@ pdf_load_page(pdf_page **pagep, pdf_xref *xref, fz_obj *dict)
 	page->mediabox.y1 = MAX(bbox.y0, bbox.y1);
 
 	if (page->mediabox.x1 - page->mediabox.x0 < 1 || page->mediabox.y1 - page->mediabox.y0 < 1)
-		return fz_throw("invalid page size");
+	{
+		fz_warn("invalid page size in page %d", number + 1);
+		page->mediabox = fz_unit_rect;
+	}
 
 	page->rotate = fz_to_int(fz_dict_gets(dict, "Rotate"));
 
@@ -218,7 +348,7 @@ pdf_load_page(pdf_page **pagep, pdf_xref *xref, fz_obj *dict)
 	if (error)
 	{
 		pdf_free_page(page);
-		return fz_rethrow(error, "cannot load page contents (%d %d R)", fz_to_num(obj), fz_to_gen(obj));
+		return fz_rethrow(error, "cannot load page contents for page %d (%d 0 R)", number + 1, fz_to_num(obj));
 	}
 
 	if (page->resources && pdf_resources_use_blending(page->resources))
diff --git a/pdf/pdf_pagetree.c b/pdf/pdf_pagetree.c
deleted file mode 100644
index 9320bb9c..00000000
--- a/pdf/pdf_pagetree.c
+++ /dev/null
@@ -1,141 +0,0 @@
-#include "fitz.h"
-#include "mupdf.h"
-
-struct info
-{
-	fz_obj *resources;
-	fz_obj *mediabox;
-	fz_obj *cropbox;
-	fz_obj *rotate;
-};
-
-int
-pdf_get_page_count(pdf_xref *xref)
-{
-	return xref->page_len;
-}
-
-fz_obj *
-pdf_get_page_object(pdf_xref *xref, int number)
-{
-	if (number > 0 && number <= xref->page_len)
-		return xref->page_objs[number - 1];
-	return NULL;
-}
-
-fz_obj *
-pdf_get_page_ref(pdf_xref *xref, int number)
-{
-	if (number > 0 && number <= xref->page_len)
-		return xref->page_refs[number - 1];
-	return NULL;
-}
-
-int
-pdf_find_page_object(pdf_xref *xref, fz_obj *page)
-{
-	int num = fz_to_num(page);
-	int gen = fz_to_gen(page);
-	int i;
-	for (i = 0; i < xref->page_len; i++)
-		if (num == fz_to_num(xref->page_refs[i]) && gen == fz_to_gen(xref->page_refs[i]))
-			return i + 1;
-	return 0;
-}
-
-static void
-pdf_load_page_tree_node(pdf_xref *xref, fz_obj *node, struct info info)
-{
-	fz_obj *dict, *kids, *count;
-	fz_obj *obj, *tmp;
-	int i, n;
-
-	/* prevent infinite recursion */
-	if (fz_dict_gets(node, ".seen"))
-		return;
-
-	kids = fz_dict_gets(node, "Kids");
-	count = fz_dict_gets(node, "Count");
-
-	if (fz_is_array(kids) && fz_is_int(count))
-	{
-		obj = fz_dict_gets(node, "Resources");
-		if (obj)
-			info.resources = obj;
-		obj = fz_dict_gets(node, "MediaBox");
-		if (obj)
-			info.mediabox = obj;
-		obj = fz_dict_gets(node, "CropBox");
-		if (obj)
-			info.cropbox = obj;
-		obj = fz_dict_gets(node, "Rotate");
-		if (obj)
-			info.rotate = obj;
-
-		tmp = fz_new_null();
-		fz_dict_puts(node, ".seen", tmp);
-		fz_drop_obj(tmp);
-
-		n = fz_array_len(kids);
-		for (i = 0; i < n; i++)
-		{
-			obj = fz_array_get(kids, i);
-			pdf_load_page_tree_node(xref, obj, info);
-		}
-
-		fz_dict_dels(node, ".seen");
-	}
-	else
-	{
-		dict = fz_resolve_indirect(node);
-
-		if (info.resources && !fz_dict_gets(dict, "Resources"))
-			fz_dict_puts(dict, "Resources", info.resources);
-		if (info.mediabox && !fz_dict_gets(dict, "MediaBox"))
-			fz_dict_puts(dict, "MediaBox", info.mediabox);
-		if (info.cropbox && !fz_dict_gets(dict, "CropBox"))
-			fz_dict_puts(dict, "CropBox", info.cropbox);
-		if (info.rotate && !fz_dict_gets(dict, "Rotate"))
-			fz_dict_puts(dict, "Rotate", info.rotate);
-
-		if (xref->page_len == xref->page_cap)
-		{
-			fz_warn("found more pages than expected");
-			xref->page_cap ++;
-			xref->page_refs = fz_realloc(xref->page_refs, xref->page_cap, sizeof(fz_obj*));
-			xref->page_objs = fz_realloc(xref->page_objs, xref->page_cap, sizeof(fz_obj*));
-		}
-
-		xref->page_refs[xref->page_len] = fz_keep_obj(node);
-		xref->page_objs[xref->page_len] = fz_keep_obj(dict);
-		xref->page_len ++;
-	}
-}
-
-fz_error
-pdf_load_page_tree(pdf_xref *xref)
-{
-	struct info info;
-	fz_obj *catalog = fz_dict_gets(xref->trailer, "Root");
-	fz_obj *pages = fz_dict_gets(catalog, "Pages");
-	fz_obj *count = fz_dict_gets(pages, "Count");
-
-	if (!fz_is_dict(pages))
-		return fz_throw("missing page tree");
-	if (!fz_is_int(count))
-		return fz_throw("missing page count");
-
-	xref->page_cap = fz_to_int(count);
-	xref->page_len = 0;
-	xref->page_refs = fz_calloc(xref->page_cap, sizeof(fz_obj*));
-	xref->page_objs = fz_calloc(xref->page_cap, sizeof(fz_obj*));
-
-	info.resources = NULL;
-	info.mediabox = NULL;
-	info.cropbox = NULL;
-	info.rotate = NULL;
-
-	pdf_load_page_tree_node(xref, pages, info);
-
-	return fz_okay;
-}
diff --git a/xps/muxps.h b/xps/muxps.h
index 8a29d674..a6faec67 100644
--- a/xps/muxps.h
+++ b/xps/muxps.h
@@ -105,21 +105,17 @@ void xps_debug_page_list(xps_context *ctx);
 void xps_free_page_list(xps_context *ctx);
 
 int xps_count_pages(xps_context *ctx);
-xps_page *xps_load_page(xps_context *ctx, int number);
+int xps_load_page(xps_page **page, xps_context *ctx, int number);
 void xps_free_page(xps_context *ctx, xps_page *page);
 
 /*
- * Images.
+ * Images, fonts, and colorspaces.
  */
 
 int xps_decode_jpeg(fz_pixmap **imagep, byte *rbuf, int rlen);
 int xps_decode_png(fz_pixmap **imagep, byte *rbuf, int rlen);
 int xps_decode_tiff(fz_pixmap **imagep, byte *rbuf, int rlen);
 
-/*
- * Fonts.
- */
-
 typedef struct xps_glyph_metrics_s xps_glyph_metrics;
 
 struct xps_glyph_metrics_s
@@ -136,10 +132,6 @@ void xps_measure_font_glyph(xps_context *ctx, fz_font *font, int gid, xps_glyph_
 
 void xps_debug_path(xps_context *ctx);
 
-/*
- * Colorspaces and colors.
- */
-
 void xps_parse_color(xps_context *ctx, char *base_uri, char *hexstring, fz_colorspace **csp, float *samples);
 void xps_set_color(xps_context *ctx, fz_colorspace *colorspace, float *samples);
 
@@ -240,8 +232,7 @@ struct xps_context_s
 	fz_device *dev;
 };
 
-xps_context *xps_new_context(void);
-int xps_open_file(xps_context *ctx, char *filename);
-int xps_free_context(xps_context *ctx);
+int xps_open_file(xps_context **ctxp, char *filename);
+void xps_free_context(xps_context *ctx);
 
 #endif
diff --git a/xps/xps_doc.c b/xps/xps_doc.c
index c3255261..5719e7f8 100644
--- a/xps/xps_doc.c
+++ b/xps/xps_doc.c
@@ -305,8 +305,8 @@ xps_load_fixed_page(xps_context *ctx, xps_page *page)
 	return 0;
 }
 
-xps_page *
-xps_load_page(xps_context *ctx, int number)
+int
+xps_load_page(xps_page **pagep, xps_context *ctx, int number)
 {
 	xps_page *page;
 	int code;
@@ -319,16 +319,16 @@ xps_load_page(xps_context *ctx, int number)
 			if (!page->root)
 			{
 				code = xps_load_fixed_page(ctx, page);
-				if (code) {
-					fz_rethrow(code, "cannot load page %d", number + 1);
-					return NULL;
-				}
+				if (code)
+					return fz_rethrow(code, "cannot load page %d", number + 1);
 			}
-			return page;
+			*pagep = page;
+			return fz_okay;
 		}
 		n ++;
 	}
-	return NULL;
+
+	return fz_throw("cannot find page %d", number + 1);
 }
 
 void
diff --git a/xps/xps_zip.c b/xps/xps_zip.c
index e1019ae0..ccf82027 100644
--- a/xps/xps_zip.c
+++ b/xps/xps_zip.c
@@ -391,15 +391,25 @@ xps_read_part(xps_context *ctx, char *partname)
 }
 
 int
-xps_open_file(xps_context *ctx, char *filename)
+xps_open_file(xps_context **ctxp, char *filename)
 {
+	xps_context *ctx;
 	char buf[2048];
 	int code;
 	char *p;
 
+	ctx = fz_malloc(sizeof(xps_context));
+	memset(ctx, 0, sizeof(xps_context));
+	ctx->font_table = xps_hash_new();
+	ctx->colorspace_table = xps_hash_new();
+	ctx->start_part = NULL;
+
 	ctx->file = fopen(filename, "rb");
 	if (!ctx->file)
+	{
+		xps_free_context(ctx);
 		return fz_throw("cannot open file: '%s'", filename);
+	}
 
 	if (strstr(filename, "/_rels/.rels") || strstr(filename, "\\_rels\\.rels"))
 	{
@@ -414,33 +424,23 @@ xps_open_file(xps_context *ctx, char *filename)
 	{
 		code = xps_find_and_read_zip_dir(ctx);
 		if (code < 0)
+		{
+			xps_free_context(ctx);
 			return fz_rethrow(code, "cannot read zip central directory");
+		}
 	}
 
 	code = xps_read_page_list(ctx);
 	if (code)
+	{
+		xps_free_context(ctx);
 		return fz_rethrow(code, "cannot read page list");
+	}
 
+	*ctxp = ctx;
 	return fz_okay;
 }
 
-xps_context *
-xps_new_context(void)
-{
-	xps_context *ctx;
-
-	ctx = fz_malloc(sizeof(xps_context));
-
-	memset(ctx, 0, sizeof(xps_context));
-
-	ctx->font_table = xps_hash_new();
-	ctx->colorspace_table = xps_hash_new();
-
-	ctx->start_part = NULL;
-
-	return ctx;
-}
-
 static void xps_free_key_func(void *ptr)
 {
 	fz_free(ptr);
@@ -456,7 +456,7 @@ static void xps_free_colorspace_func(void *ptr)
 	fz_drop_colorspace(ptr);
 }
 
-int
+void
 xps_free_context(xps_context *ctx)
 {
 	int i;
@@ -464,19 +464,15 @@ xps_free_context(xps_context *ctx)
 	if (ctx->file)
 		fclose(ctx->file);
 
-	if (ctx->start_part)
-		fz_free(ctx->start_part);
-
 	for (i = 0; i < ctx->zip_count; i++)
 		fz_free(ctx->zip_table[i].name);
 	fz_free(ctx->zip_table);
 
 	xps_hash_free(ctx->font_table, xps_free_key_func, xps_free_font_func);
 	xps_hash_free(ctx->colorspace_table, xps_free_key_func, xps_free_colorspace_func);
-
 	xps_free_page_list(ctx);
 
+	fz_free(ctx->start_part);
+	fz_free(ctx->directory);
 	fz_free(ctx);
-
-	return 0;
 }