1 files changed, 112 insertions, 5 deletions

diff --git a/src/xref.c b/src/xref.c
index 8239f45..452aa19 100644
--- a/src/xref.c
+++ b/src/xref.c
@@ -499,12 +499,123 @@ nspdferror decode_trailers(struct pdf_doc *doc)
     return decode_xref_trailer(doc, startxref);
 }
 
+/**
+ * recursively decodes a page tree
+ */
+nspdferror
+decode_page_tree(struct pdf_doc *doc,
+                 struct cos_object *page_tree_node,
+                 unsigned int *page_index)
+{
+    nspdferror res;
+    const char *type;
+
+    // Type = Pages
+    res = cos_get_dictionary_name(doc, page_tree_node, "Type", &type);
+    if (res != NSPDFERROR_OK) {
+        return res;
+    }
+
+    if (strcmp(type, "Pages") == 0) {
+        struct cos_object *kids;
+        unsigned int kids_size;
+        unsigned int kids_index;
+
+        if (doc->page_table == NULL) {
+            /* allocate top level page table */
+            int64_t count;
+
+            res = cos_get_dictionary_int(doc, page_tree_node, "Count", &count);
+            if (res != NSPDFERROR_OK) {
+                return res;
+            }
+
+            doc->page_table = calloc(count, sizeof(struct page_table_entry));
+            if (doc->page_table == NULL) {
+                return NSPDFERROR_NOMEM;
+            }
+            doc->page_table_size = count;
+        }
+
+        res = cos_get_dictionary_array(doc, page_tree_node, "Kids", &kids);
+        if (res != NSPDFERROR_OK) {
+            return res;
+        }
+
+        res = cos_get_array_size(doc, kids, &kids_size);
+        if (res != NSPDFERROR_OK) {
+            return res;
+        }
+
+        for (kids_index = 0; kids_index < kids_size; kids_index++) {
+            struct cos_object *kid;
+
+            res = cos_get_array_dictionary(doc, kids, kids_index, &kid);
+            if (res != NSPDFERROR_OK) {
+                return res;
+            }
+
+            res = decode_page_tree(doc, kid, page_index);
+            if (res != NSPDFERROR_OK) {
+                return res;
+            }
+        }
+
+    } else if (strcmp(type, "Page") == 0) {
+        struct page_table_entry *page;
+
+        page = doc->page_table + (*page_index);
+
+        /* required heritable resources */
+        res = cos_heritable_dictionary_dictionary(doc,
+                                                  page_tree_node,
+                                                  "Resources",
+                                                  &(page->resources));
+        if (res != NSPDFERROR_OK) {
+            return res;
+        }
+
+        /* required heritable mediabox */
+        res = cos_heritable_dictionary_array(doc,
+                                             page_tree_node,
+                                             "MediaBox",
+                                             &(page->mediabox));
+        if (res != NSPDFERROR_OK) {
+            return res;
+        }
+
+        /* optional page contents */
+        res = cos_get_dictionary_value(doc,
+                                       page_tree_node,
+                                       "Contents",
+                                       &(page->contents));
+        if ((res != NSPDFERROR_OK) &&
+            (res != NSPDFERROR_NOTFOUND)) {
+            return res;
+        }
+
+        printf("page index:%d page:%p resources:%p mediabox:%p contents:%p\n",
+               *page_index,
+               page,
+               page->resources,
+               page->mediabox,
+               page->contents);
+
+        (*page_index)++;
+        res = NSPDFERROR_OK;
+    } else {
+        res = NSPDFERROR_FORMAT;
+    }
+    return res;
+}
+
 nspdferror decode_catalog(struct pdf_doc *doc)
 {
     nspdferror res;
     struct cos_object *catalog;
     const char *type;
     struct cos_object *pages;
+    unsigned int page_index = 0;
 
     res = cos_get_dictionary(doc, doc->root, &catalog);
     if (res != NSPDFERROR_OK) {
@@ -526,14 +637,10 @@ nspdferror decode_catalog(struct pdf_doc *doc)
         return res;
     }
 
-    // Type = Pages
-    res = cos_get_dictionary_name(doc, pages, "Type", &type);
+    res = decode_page_tree(doc, pages, &page_index);
     if (res != NSPDFERROR_OK) {
         return res;
     }
-    if (strcmp(type, "Pages") != 0) {
-        return NSPDFERROR_FORMAT;
-    }
 
     return res;
 }