Handle duplicated image in HN.

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

1 files changed, 24 insertions, 2 deletions

diff --git a/src/cnki_pdf.c b/src/cnki_pdf.c
index 90ba28e..cfcba25 100644
--- a/src/cnki_pdf.c
+++ b/src/cnki_pdf.c
@@ -500,6 +500,8 @@ cnki_pdf_hn(cnki_t **param)
 
 	char buf[64];
 
+	pdf_object_t *tmp;
+
 	int cnt = 0;
 	int *root_kid = malloc((*param)->file_stat->page * sizeof(int));
 
@@ -971,6 +973,28 @@ cnki_pdf_hn(cnki_t **param)
 				margin_y = (3507.874 - dim[1]) / 2;
 			}
 
+			/* Remove duplicated image, ptr->image_length is sometimes squared */
+			for (int i = 1; i < ptr->image_length; i++) {
+				if ((ptr->image_data[i].x > 0 || ptr->image_data[i].y > 0) &&
+					dim[i * 2] < dim[0] && dim[i * 2 + 1] < dim[1])
+					continue;
+
+				for (int j = i; j < ptr->image_length; j++) {
+					pdf_get_obj(&pdf, ids[j], &tmp);
+					pdf_obj_del(&pdf, ids[j]);
+
+					tmp->next = NULL;
+					pdf_obj_destroy(&tmp);
+
+					dim[j * 2] = -1;
+					dim[j * 2 + 1] = -1;
+
+					pdf_obj_append(&pdf, ids[j], NULL, NULL, NULL, 0);
+				}
+
+				break;
+			}
+
 			for (int i = 0; i < ptr->image_length; i++) {
 				if (dim[i * 2] <= 0 || dim[i * 2 + 1] <= 0)
 					continue;
@@ -1150,8 +1174,6 @@ cnki_pdf_hn(cnki_t **param)
 		return 1;
 	}
 
-	pdf_object_t *tmp;
-
 	/* Add /Parent to page object */
 	for (int i = 0; i < (*param)->file_stat->page; i++) {
 		if (pdf_get_obj(&pdf, root_kid[i], &tmp) != 0) {