Fix HN text parsing.

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

src/cnki_pdf.c

@@ -850,45 +850,7 @@ cnki_pdf_hn(cnki_t **param)
 			for (int i = 0, j = 0; i < ptr->text_size - 1;) {
 				switch (((unsigned char) ptr->text[i + 1] << 8) + (unsigned char) ptr->text[i]) {
 					case 0x8001:
-						if (ptr->address_next > ptr->address)
-							strcat(dictionary, "T*\n");
-					case 0x8070:
-						if (ptr->address_next > ptr->address) {
-							i += 4;
-
-							for (;;) {
-								if (i + 3 >= ptr->text_size ||
-									(unsigned char) ptr->text[i + 1] == 0x80)
-									break;
-
-								conv_src[0] = ptr->text[i + 3];
-								conv_src[1] = ptr->text[i + 2];
-
-								//snprintf(buf, 64, "%f %f Td\n");
-								//strcat(dictionary, buf);
-
-								conv_size = 6;
-
-								if (strconv(&conv_dst, "UTF-16BE",
-									conv_src, "GB18030", &conv_size) == 0) {
-									if (conv_size - 2 > 0) {
-										strcat(dictionary, "<feff");
-										for (int k = 0; k < conv_size - 2; k++) {
-											snprintf(conv_hex, 3,
-												"%02x", (unsigned char) conv_dst[k]);
-											strcat(dictionary, conv_hex);
-										}
-										strcat(dictionary, "> Tj\n");
-									}
-									free(conv_dst);
-								}
-
-								i += 4;
-							}
-
-							break;
-						}
-
+						if (ptr->address_next <= ptr->address) {
 							if (i + 7 >= ptr->text_size) {
 								i += 2;
 								break;
@@ -897,7 +859,7 @@ cnki_pdf_hn(cnki_t **param)
 							conv_src[0] = ptr->text[i + 7];
 							conv_src[1] = ptr->text[i + 6];
 
-						//snprintf(buf, 64, "%f %f Td\n");
+							//snprintf(buf, 64, "1 0 0 1 %d %d Tm\n")
 							//strcat(dictionary, buf);
 
 							conv_size = 6;
@@ -905,7 +867,7 @@ cnki_pdf_hn(cnki_t **param)
 							if (strconv(&conv_dst, "UTF-16BE",
 								conv_src, "GB18030", &conv_size) == 0) {
 								if (conv_size - 2 > 0) {
-								strcat(dictionary, "<feff");
+									strcat(dictionary, "<");
 									for (int k = 0; k < conv_size - 2; k++) {
 										snprintf(conv_hex, 3,
 											"%02x", (unsigned char) conv_dst[k]);
@@ -917,6 +879,46 @@ cnki_pdf_hn(cnki_t **param)
 							}
 
 							i += 8;
+							break;
+						}
+
+						strcat(dictionary, "T*\n");
+					case 0x8070:
+						i += 4;
+
+						if (ptr->address_next <= ptr->address)
+							break;
+
+						for (;;) {
+							if (i + 3 >= ptr->text_size ||
+								(unsigned char) ptr->text[i + 1] == 0x80)
+								break;
+
+							conv_src[0] = ptr->text[i + 3];
+							conv_src[1] = ptr->text[i + 2];
+
+							//snprintf(buf, 64, "1 0 0 1 %d %d Tm\n")
+							//strcat(dictionary, buf);
+
+							conv_size = 6;
+
+							if (strconv(&conv_dst, "UTF-16BE",
+								conv_src, "GB18030", &conv_size) == 0) {
+								if (conv_size - 2 > 0) {
+									strcat(dictionary, "<");
+									for (int k = 0; k < conv_size - 2; k++) {
+										snprintf(conv_hex, 3,
+											"%02x", (unsigned char) conv_dst[k]);
+										strcat(dictionary, conv_hex);
+									}
+									strcat(dictionary, "> Tj\n");
+								}
+								free(conv_dst);
+							}
+
+							i += 4;
+						}
+
 						break;
 					case 0x800a:
 						if (i + 27 >= ptr->text_size || j >= ptr->image_length) {

src/jbig2.c

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2022, yzrh <yzrh@noema.org>
+ * Copyright (c) 2022-2023, yzrh <yzrh@noema.org>
  *
  * SPDX-License-Identifier: Apache-2.0
  */
@@ -31,5 +31,6 @@ strdec_jbig2(char **bitmap,
 	}
 
 	jbig2_release_page(ctx, image);
+	jbig2_ctx_free(ctx);
 	return 0;
 }

src/pdf_writer.c

@@ -1,19 +1,39 @@
 /*
- * Copyright (c) 2020-2022, yzrh <yzrh@noema.org>
+ * Copyright (c) 2020-2023, yzrh <yzrh@noema.org>
  *
  * SPDX-License-Identifier: Apache-2.0
  */
 
 #include <stdlib.h>
+#include <string.h>
 #include <time.h>
 
+#include "version.h"
 #include "md5.h"
 #include "pdf.h"
 
+static int
+_info_obj(pdf_object_t **pdf)
+{
+	char dictionary[128] = "<<\n"
+		"/Producer (Melon " VERSION "." RELEASE "." PATCH EXTRA ")\n"
+		"/CreationDate (D:";
+
+	char buf[64];
+
+	time_t timestamp = time(NULL);
+
+	strftime(buf, 64, "%Y%m%d%H%M%S", gmtime(&timestamp));
+	strcat(dictionary, buf);
+	strcat(dictionary, "+00'00')\n>>");
+
+	return pdf_obj_append(pdf, 0, NULL, dictionary, NULL, 0);
+}
+
 int
 pdf_dump_obj(pdf_object_t **pdf, FILE **fp)
 {
-	if (*pdf == NULL || *fp == NULL)
+	if (*pdf == NULL || *fp == NULL || _info_obj(pdf) != 0)
 		return 1;
 
 	long cur;
@@ -152,18 +172,11 @@ pdf_dump_trailer(pdf_object_t **pdf, FILE **fp, int xref)
 	while (ptr->next != NULL)
 		ptr = ptr->next;
 
-	/*
-	 * TODO: Document information dictionary
-	 * `"/Producer (Melon)"'
-	 * `"/CreationDate (D:YYYYMMDDHHmmSS+00'00')"'
-	 *
-	 * Trailer dictionary
-	 * `"/Info %d 0 R"'
-	 */
 	fprintf(*fp,
-		"/Size %d\n/Root %d 0 R\n",
+		"/Size %d\n/Root %d 0 R\n/Info %d 0 R\n",
 		ptr->id + 1,
-		pdf_get_catalog_id(pdf));
+		pdf_get_catalog_id(pdf),
+		ptr->id);
 
 	fputs("/ID [", *fp);
 

src/version.h

@@ -5,6 +5,6 @@
  */
 
 #define VERSION	"0"
-#define RELEASE	"2"
-#define PATCH	"5"
+#define RELEASE	"3"
+#define PATCH	"0"
 #define EXTRA	""