Avoid revisiting the same page tree node repeatedly.

Only attempt to progress to new page tree nodes if a new node to
process has actually been found. Previously if e.g. doc.findPage(1)
was called in murun for a single page document the error message was
"cycle in page tree" because the single page tree node was visited
twice. Now the error message is "cannot find page 1 in page tree",
which is more accurate.

1 files changed, 7 insertions, 1 deletions

diff --git a/source/pdf/pdf-page.c b/source/pdf/pdf-page.c
index 9d8a27ec..f5ab501b 100644
--- a/source/pdf/pdf-page.c
+++ b/source/pdf/pdf-page.c
@@ -174,7 +174,13 @@ pdf_lookup_page_loc_imp(fz_context *ctx, pdf_document *doc, pdf_obj *node, int *
 				}
 			}
 		}
-		while (hit == NULL);
+		/* If i < len && hit != NULL the desired page was found in the
+		Kids array, done. If i < len && hit == NULL the found page tree
+		node contains a Kids array that contains the desired page, loop
+		back to top to extract it. When i == len the Kids array has been
+		exhausted without finding the desired page, give up.
+		*/
+		while (hit == NULL && i < len);
 	}
 	fz_always(ctx)
 	{