Clean up xps and pdf page access functions.

1 files changed, 180 insertions, 50 deletions

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))