Handle invalid result from PDF parser.

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

2 files changed, 34 insertions, 22 deletions

diff --git a/src/cnki_pdf.c b/src/cnki_pdf.c
index cfcba25..af38aa6 100644
--- a/src/cnki_pdf.c
+++ b/src/cnki_pdf.c
@@ -160,10 +160,10 @@ _pdf_obj_dedup(cnki_t **param, pdf_object_t **pdf)
 	pdf_object_t *ptr;
 
 	if ((*param)->stat > 1)
-		printf("Deleting duplicated object\n\t%8s\n", "id");
+		printf("Deleting duplicated object\n");
 
 	ptr = *pdf;
-	while (ptr->next != NULL) {
+	while (ptr != NULL && ptr->next != NULL) {
 		if (ptr->id == ptr->next->id) {
 			pdf_get_obj(&ptr, ptr->id, &tmp);
 			pdf_obj_del(&ptr, ptr->id);
@@ -174,7 +174,7 @@ _pdf_obj_dedup(cnki_t **param, pdf_object_t **pdf)
 			ret++;
 
 			if ((*param)->stat > 1)
-				printf("\t%8d\n", ptr->id);
+				printf("Deleted duplicated object %d.\n", ptr->id);
 		}
 
 		ptr = ptr->next;
@@ -247,18 +247,18 @@ cnki_pdf(cnki_t **param)
 	int *parent = NULL;
 	pdf_get_parent_id(&pdf, &parent);
 
-	if (parent[0] == 0)
-		return 1;
-
 	if ((*param)->stat > 0)
 		printf("Discovered %d parent object(s)\n", parent[0]);
 
-	int8_t *parent_missing = malloc(parent[0] * sizeof(int8_t));
+	int8_t *parent_missing;
+	int *kid;
 
-	if (parent_missing == NULL)
-		return 1;
+	if (parent[0] > 0) {
+		parent_missing = malloc(parent[0] * sizeof(int8_t));
 
-	int *kid;
+		if (parent_missing == NULL)
+			return 1;
+	}
 
 	for (int i = 1; i <= parent[0]; i++) {
 		if ((*param)->stat > 1)
@@ -326,7 +326,7 @@ cnki_pdf(cnki_t **param)
 	if ((*param)->stat > 1)
 		printf("Searching for root object\n");
 
-	dictionary_size = 128;
+	dictionary_size = 128 + 12 * parent[0];
 	dictionary = malloc(dictionary_size);
 
 	if (dictionary == NULL) {
@@ -400,8 +400,10 @@ cnki_pdf(cnki_t **param)
 				root);
 	}
 
+	if (parent[0] > 0)
+		free(parent_missing);
+
 	free(parent);
-	free(parent_missing);
 
 	int outline = _pdf_cnki_outline(param, &pdf);
 
@@ -1166,14 +1168,6 @@ cnki_pdf_hn(cnki_t **param)
 
 	free(dictionary);
 
-	dictionary_size = 256;
-	dictionary = malloc(dictionary_size);
-
-	if (dictionary == NULL) {
-		free(root_kid);
-		return 1;
-	}
-
 	/* 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) {
@@ -1182,9 +1176,16 @@ cnki_pdf_hn(cnki_t **param)
 			return 1;
 		}
 
-		memset(dictionary, 0, dictionary_size);
+		dictionary_size = tmp->dictionary_size + 24;
+		dictionary = malloc(dictionary_size);
+
+		if (dictionary == NULL) {
+			free(root_kid);
+			return 1;
+		}
 
 		memcpy(dictionary, tmp->dictionary, tmp->dictionary_size);
+		memset(dictionary + tmp->dictionary_size, 0, 24);
 
 		snprintf(buf, 64, "/Parent %d 0 R\n>>", root);
 		strcat(dictionary, buf);
@@ -1194,10 +1195,20 @@ cnki_pdf_hn(cnki_t **param)
 			free(root_kid);
 			return 1;
 		}
+
+		free(dictionary);
 	}
 
 	free(root_kid);
 
+	dictionary_size = 128;
+	dictionary = malloc(dictionary_size);
+
+	if (dictionary == NULL) {
+		free(root_kid);
+		return 1;
+	}
+
 	memset(dictionary, 0, dictionary_size);
 
 	if ((*param)->stat > 0)
diff --git a/src/pdf_parser.c b/src/pdf_parser.c
index 2585e6f..781bafa 100644
--- a/src/pdf_parser.c
+++ b/src/pdf_parser.c
@@ -119,7 +119,8 @@ _locate(pdf_object_t **pdf, FILE **fp, int size_buf)
 			fseek(*fp, tail + 7, SEEK_SET);
 			head = tail = 0;
 		} else if (head > 0 && tail > 0) {
-			fseek(*fp, head, SEEK_SET);
+			if (cur + size_buf < end)
+				fseek(*fp, head, SEEK_SET);
 			tail = 0;
 		} else {
 			fseek(*fp, -7, SEEK_CUR);