Produce PDF directly from KDH.

README.md

@@ -6,15 +6,11 @@ Melon: Converter that produces PDF from CNKI proprietary formats
 Development
 -----------
 
-Currently, PDF, CAJ, and KDH can be converted. Please report 
+Currently, CAJ and KDH can be converted. Please report
 any failures with a sample that can reproduce the behaviour.
 
 HN support is being worked on.
 
-KDH is essentially an invalid PDF file xor'ed with a predetermined key.
-You may want to convert the decrypted KDH to valid PDF, although some
-PDF readers can display the invalid PDF.
-
 Usage
 =====
 

src/cnki_kdh.c

@@ -27,6 +27,11 @@ cnki_kdh(cnki_t **param)
 
 	char buf[(*param)->size_buf];
 
+	FILE *tmp = tmpfile();
+
+	if (tmp == NULL)
+		return 1;
+
 	for (;;) {
 		fread(buf, (*param)->size_buf, 1, (*param)->fp_i);
 
@@ -35,15 +40,27 @@ cnki_kdh(cnki_t **param)
 			key_cur++;
 		}
 
-		fwrite(buf, (*param)->size_buf, 1, (*param)->fp_o);
+		fwrite(buf, (*param)->size_buf, 1, tmp);
 
 		if (ftell((*param)->fp_i) == size)
 			break;
 	}
 
 	if ((*param)->stat > 0)
-		printf("Decryption ended total %ld byte(s) written\n",
+		printf("Decrypted %ld byte(s)\n", ftell(tmp));
-			ftell((*param)->fp_o));
+
+	fseek(tmp, 0, SEEK_SET);
+
+	FILE *orig = (*param)->fp_i;
+	(*param)->fp_i = tmp;
+
+	cnki_pdf(param);
+
+	(*param)->fp_i = orig;
+	fclose(tmp);
+
+	if ((*param)->stat > 0)
+		printf("Conversion ended\n");
 
 	return 0;
 }

src/cnki_pdf.c

@@ -232,7 +232,19 @@ cnki_pdf(cnki_t **param)
 	}
 
 	if ((*param)->stat > 1)
-		printf("Generating '/Catalog' dictionary\n");
+		printf("Searching for catalog object\n");
+
+	int catalog = pdf_get_catalog_id(&pdf);
+
+	if (catalog != 0) {
+		if ((*param)->stat > 0)
+			printf("catalog object is %d.\n", catalog);
+	} else {
+		if ((*param)->stat > 0)
+			printf("catalog object is missing\n");
+
+		if ((*param)->stat > 1)
+			printf("Generating catalog object\n");
 
 		snprintf(buf, 64,
 			"<<\n/Type /Catalog\n/Pages %d 0 R\n",
@@ -250,10 +262,32 @@ cnki_pdf(cnki_t **param)
 
 		pdf_obj_append(&pdf, 0, NULL, dictionary, NULL);
 
-	free(dictionary);
+		if ((*param)->stat > 0)
+			printf("Generated catalog object\n");
+	}
+
+	if ((*param)->stat > 1)
+		printf("Searching for xref object\n");
+
+	int xref = pdf_get_xref_id(&pdf);
+
+	if (xref != 0) {
+		if ((*param)->stat > 0)
+			printf("xref object is %d.\n", xref);
+
+		if ((*param)->stat > 1)
+			printf("Deleting xref object\n");
+
+		pdf_obj_del(&pdf, xref);
 
 		if ((*param)->stat > 0)
-		printf("Generated '/Catalog' dictionary\n");
+			printf("Deleted xref object\n");
+	} else {
+		if ((*param)->stat > 0)
+			printf("xref object is missing\n");
+	}
+
+	free(dictionary);
 
 	if ((*param)->stat > 1)
 		printf("Sorting object(s)\n");
@@ -312,7 +346,7 @@ cnki_pdf(cnki_t **param)
 			pdf_get_count(&pdf),
 			ftell((*param)->fp_o));
 
-	long xref = ftell((*param)->fp_o);
+	long cur_xref = ftell((*param)->fp_o);
 
 	if ((*param)->stat > 1)
 		printf("Writing cross-reference table\n");
@@ -323,7 +357,7 @@ cnki_pdf(cnki_t **param)
 	} else {
 		if ((*param)->stat > 0)
 			printf("Cross-reference table %ld byte(s) written\n",
-				ftell((*param)->fp_o) - xref);
+				ftell((*param)->fp_o) - cur_xref);
 	}
 
 	if ((*param)->stat > 1)
@@ -332,7 +366,7 @@ cnki_pdf(cnki_t **param)
 	if ((*param)->stat > 0)
 		cur = ftell((*param)->fp_o);
 
-	if (pdf_dump_trailer(&pdf, &(*param)->fp_o, xref) != 0) {
+	if (pdf_dump_trailer(&pdf, &(*param)->fp_o, cur_xref) != 0) {
 		if ((*param)->stat > 0)
 			printf("Trailer not written\n");
 	} else {

src/pdf.h

@@ -54,6 +54,7 @@ int pdf_get_size(pdf_object_t **pdf);
 int pdf_get_free_id(pdf_object_t **pdf);
 int pdf_get_free_ids(pdf_object_t **pdf, int **ids, int count);
 int pdf_get_catalog_id(pdf_object_t **pdf);
+int pdf_get_xref_id(pdf_object_t **pdf);
 int pdf_get_parent_id(pdf_object_t **pdf, int **id);
 int pdf_get_kid_id(pdf_object_t **pdf, int id, int **kid);
 int pdf_get_kid_count(pdf_object_t **pdf, int id);

src/pdf_get.c

@@ -161,6 +161,27 @@ pdf_get_catalog_id(pdf_object_t **pdf)
 	return catalog_id;
 }
 
+int
+pdf_get_xref_id(pdf_object_t **pdf)
+{
+	if (*pdf == NULL)
+		return 1;
+
+	int xref_id = 0;
+
+	pdf_object_t *ptr = (*pdf)->next;
+
+	while (ptr != NULL) {
+		if (ptr->dictionary != NULL &&
+			strstr(ptr->dictionary, "/XRef") != NULL)
+			xref_id = ptr->id;
+
+		ptr = ptr->next;
+	}
+
+	return xref_id;
+}
+
 int
 pdf_get_parent_id(pdf_object_t **pdf, int **id)
 {

src/pdf_parser.c

@@ -154,10 +154,16 @@ pdf_load(pdf_object_t **pdf, FILE **fp, int size_buf)
 
 		if ((head = memmem(buf, ptr->size, "<<", 2)) != NULL &&
 			(tail = _memmem_whitespace(buf, ptr->size, ">>", 2)) != NULL) {
-			/* A dictionary object may have nested dictionary */
+			/*
+			 * A dictionary object may have nested dictionary,
+			 * but it should not be in a stream
+			 */
 			while ((tmp = _memmem_whitespace(tail + 2,
 				ptr->size - (tail - buf) - 2,
-				">>", 2)) != NULL)
+				">>", 2)) != NULL &&
+				memmem(tail + 2,
+				ptr->size - (tail - buf) - 2,
+				"stream\r\n", 8) == NULL)
 				tail = tmp;
 
 			ptr->dictionary_size = tail - head + 2;

src/pdf_writer.c

@@ -27,10 +27,10 @@ pdf_dump_obj(pdf_object_t **pdf, FILE **fp)
 		fprintf(*fp, "%d 0 obj\n", ptr->id);
 
 		if (ptr->dictionary != NULL) {
-			fputs(ptr->dictionary, *fp);
+			fwrite(ptr->dictionary, ptr->dictionary_size, 1, *fp);
 			fputs("\n", *fp);
 		} else if (ptr->object != NULL) {
-			fputs(ptr->object, *fp);
+			fwrite(ptr->object, ptr->object_size, 1, *fp);
 			fputs("\n", *fp);
 		} else if (ptr->stream == NULL) {
 			fputs("null\n", *fp);