Handle duplicated object in CAJ.

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

1 files changed, 47 insertions, 26 deletions

diff --git a/src/cnki_pdf.c b/src/cnki_pdf.c
index 6cb8c9f..90ba28e 100644
--- a/src/cnki_pdf.c
+++ b/src/cnki_pdf.c
@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2020-2022, yzrh <yzrh@noema.org>
+ * Copyright (c) 2020-2023, yzrh <yzrh@noema.org>
  *
  * SPDX-License-Identifier: Apache-2.0
  */
@@ -145,11 +145,46 @@ _pdf_obj_sort(cnki_t **param, pdf_object_t **pdf)
 
 	ret = pdf_obj_sort(pdf);
 
+	if ((*param)->stat > 0)
+		printf("Sorted object(s)\n");
+
+	return ret;
+}
+
+static int
+_pdf_obj_dedup(cnki_t **param, pdf_object_t **pdf)
+{
+	int ret = 0;
+
+	pdf_object_t *tmp;
+	pdf_object_t *ptr;
+
+	if ((*param)->stat > 1)
+		printf("Deleting duplicated object\n\t%8s\n", "id");
+
+	ptr = *pdf;
+	while (ptr->next != NULL) {
+		if (ptr->id == ptr->next->id) {
+			pdf_get_obj(&ptr, ptr->id, &tmp);
+			pdf_obj_del(&ptr, ptr->id);
+
+			tmp->next = NULL;
+			pdf_obj_destroy(&tmp);
+
+			ret++;
+
+			if ((*param)->stat > 1)
+				printf("\t%8d\n", ptr->id);
+		}
+
+		ptr = ptr->next;
+	}
+
 	if ((*param)->stat > 0) {
 		if (ret == 0)
-			printf("Sorted object(s)\n");
+			printf("No duplicated object\n");
 		else
-			printf("Object(s) not sorted\n");
+			printf("Deleted %d duplicated object(s)\n", ret);
 	}
 
 	return ret;
@@ -338,12 +373,9 @@ cnki_pdf(cnki_t **param)
 			printf("Generating root object\n");
 
 		snprintf(buf, 64,
-			"<<\n/Type /Pages\n/Kids ");
+			"<<\n/Type /Pages\n/Kids [");
 		strcat(dictionary, buf);
 
-		if (parent[0] > 1)
-			strcat(dictionary, "[");
-
 		for (int i = 0, j = 0; i < parent[0]; i++) {
 			if (parent_missing[i] == 1) {
 				snprintf(buf, 64, "%d 0 R", parent[i + 1]);
@@ -354,12 +386,7 @@ cnki_pdf(cnki_t **param)
 			}
 		}
 
-		if (parent[0] > 1)
-			strcat(dictionary, "]");
-
-		strcat(dictionary, "\n");
-
-		snprintf(buf, 64, "/Count %d\n", (*param)->file_stat->page);
+		snprintf(buf, 64, "]\n/Count %d\n", (*param)->file_stat->page);
 		strcat(dictionary, buf);
 
 		strcat(dictionary, ">>");
@@ -442,6 +469,8 @@ cnki_pdf(cnki_t **param)
 
 	_pdf_obj_sort(param, &pdf);
 
+	_pdf_obj_dedup(param, &pdf);
+
 	_pdf_dump(param, &pdf);
 
 	pdf_obj_destroy(&pdf);
@@ -721,12 +750,12 @@ cnki_pdf_hn(cnki_t **param)
 				if ((*param)->stat > 2)
 					printf("Not extracted.\n");
 
-				pdf_obj_append(&pdf, ids[i], "null", NULL, NULL, 0);
+				pdf_obj_append(&pdf, ids[i], NULL, NULL, NULL, 0);
 			} else {
 				if ((*param)->stat > 2)
 					printf("Unsupported format.\n");
 
-				pdf_obj_append(&pdf, ids[i], "null", NULL, NULL, 0);
+				pdf_obj_append(&pdf, ids[i], NULL, NULL, NULL, 0);
 			}
 		}
 
@@ -1028,7 +1057,7 @@ cnki_pdf_hn(cnki_t **param)
 			free(stream);
 		} else {
 			pdf_obj_append(&pdf, ids[ptr->image_length + 1],
-				"null", NULL, NULL, 0);
+				NULL, NULL, NULL, 0);
 		}
 
 		memset(dictionary, 0, dictionary_size);
@@ -1094,12 +1123,9 @@ cnki_pdf_hn(cnki_t **param)
 
 	int root = pdf_get_free_id(&pdf);
 
-	snprintf(buf, 64, "<<\n/Type /Pages\n/Kids ");
+	snprintf(buf, 64, "<<\n/Type /Pages\n/Kids [");
 	strcat(dictionary, buf);
 
-	if ((*param)->file_stat->page > 1)
-		strcat(dictionary, "[");
-
 	for (int i = 0; i < (*param)->file_stat->page; i++) {
 		snprintf(buf, 64, "%d 0 R", root_kid[i]);
 		strcat(dictionary, buf);
@@ -1107,12 +1133,7 @@ cnki_pdf_hn(cnki_t **param)
 			strcat(dictionary, " ");
 	}
 
-	if ((*param)->file_stat->page > 1)
-		strcat(dictionary, "]");
-
-	strcat(dictionary, "\n");
-
-	snprintf(buf, 64, "/Count %d\n", (*param)->file_stat->page);
+	snprintf(buf, 64, "]\n/Count %d\n", (*param)->file_stat->page);
 	strcat(dictionary, buf);
 
 	strcat(dictionary, ">>");