Prioritise incomplete object during deduplication.

Signed-off-by: yzrh <yzrh@noema.org>

1 files changed, 22 insertions, 3 deletions

diff --git a/src/pdf_parser.c b/src/pdf_parser.c
index 781bafa..ed7bfba 100644
--- a/src/pdf_parser.c
+++ b/src/pdf_parser.c
@@ -183,7 +183,7 @@ pdf_load(pdf_object_t **pdf, FILE **fp, int size_buf)
 				ptr->size - (tail - buf) - 3,
 				">>", 2)) != NULL &&
 				memmem(tail + 3,
-				ptr->size - (tail - buf) - 3,
+				(tmp - tail) - 3,
 				"stream\r\n", 8) == NULL)
 				tail = tmp;
 
@@ -226,8 +226,27 @@ pdf_load(pdf_object_t **pdf, FILE **fp, int size_buf)
 
 			free(buf);
 		} else {
-			ptr->object_size = ptr->size;
-			ptr->object = buf;
+			/* Handle incomplete object */
+			head = buf;
+			while ((tmp = _memmem_whitespace(head,
+				ptr->size - (head - buf),
+				" 0 obj", 6)) != NULL)
+				head = tmp + 7;
+
+			if (head - buf > 0) {
+				ptr->object_size = ptr->size - (head - buf);
+				ptr->object = malloc(ptr->object_size);
+
+				if (ptr->object == NULL)
+					return 1;
+
+				memcpy(ptr->object, head, ptr->object_size);
+
+				free(buf);
+			} else {
+				ptr->object_size = ptr->size;
+				ptr->object = buf;
+			}
 		}
 
 		ptr = ptr->next;