Decode JBIG and JPEG during HN conversion.

1 files changed, 6 insertions, 222 deletions

diff --git a/src/cnki_hn.c b/src/cnki_hn.c
index f6a4c24..0e16a76 100644
--- a/src/cnki_hn.c
+++ b/src/cnki_hn.c
@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2020, yzrh <yzrh@noema.org>
+ * Copyright (c) 2020-2021, yzrh <yzrh@noema.org>
  *
  * SPDX-License-Identifier: Apache-2.0
  */
@@ -9,6 +9,8 @@
 
 #include "cnki.h"
 #include "iconv.h"
+#include "zlib.h"
+#include "jpeg.h"
 #include "pdf.h"
 #include "pdf_cnki.h"
 
@@ -131,231 +133,13 @@ cnki_hn(cnki_t **param)
 		ptr = ptr->next;
 	}
 
-	if ((*param)->stat > 1)
-		printf("Loaded %d page(s)\n", (*param)->file_stat->page);
-
-	if ((*param)->stat > 1)
-		printf("Generating PDF object(s)\n");
-
-	pdf_object_t *pdf = NULL;
-
-	if (pdf_obj_create(&pdf) != 0)
-		return 1;
-
-	int buf_size;
-	char *buf;
-
-	int str_size;
-	char *str;
-
-	int conv_size;
-	char *conv_dst;
-	char conv_src[2];
-	char conv_hex[3];
-
-	ptr = (*param)->object_hn;
-	while (ptr != NULL) {
-		if (strncmp(ptr->text + 8, "COMPRESSTEXT", 12) == 0) {
-			cnki_zlib(&buf, &buf_size, ptr->text, ptr->text_size);
-
-			str_size = buf_size / 8 + 7;
-			str = malloc(str_size);
-
-			if (str == NULL)
-				return 1;
-
-			memset(str, 0, str_size);
-
-			strcat(str, "<feff");
-
-			for (int i = 0; i < buf_size; i += 16) {
-				conv_src[0] = buf[i + 7];
-				conv_src[1] = buf[i + 6];
-
-				conv_size = 6;
-
-				if (strconv(&conv_dst, "UTF-16BE",
-					conv_src, "GB18030", &conv_size) == 0) {
-					for (int j = 0; j < conv_size - 2; j++) {
-						snprintf(conv_hex, 3,
-							"%02x", (unsigned char) conv_dst[j]);
-						strcat(str, conv_hex);
-					}
-					free(conv_dst);
-				}
-			}
-			free(buf);
-
-			strcat(str, ">");
-		} else {
-			str_size = ptr->text_size;
-			str = malloc(str_size);
-
-			if (str == NULL)
-				return 1;
-
-			memset(str, 0, str_size);
-
-			strcat(str, "<feff");
-
-			for (int i = 0; i < ptr->text_size; i += 4) {
-				conv_src[0] = ptr->text[i + 3];
-				conv_src[1] = ptr->text[i + 2];
-
-				conv_size = 6;
-
-				if (strconv(&conv_dst, "UTF-16BE",
-					conv_src, "GB18030", &conv_size) == 0) {
-					for (int j = 0; j < conv_size - 2; j++) {
-						snprintf(conv_hex, 3,
-							"%02x", (unsigned char) conv_dst[j]);
-						strcat(str, conv_hex);
-					}
-					free(conv_dst);
-				}
-			}
-
-			strcat(str, ">");
-		}
-
-		pdf_obj_append(&pdf, 0, str, NULL, NULL);
-
-		free(str);
-
-		ptr = ptr->next;
-	}
-
-	if ((*param)->stat > 1) {
-		printf("\t%8s\t%12s\t%12s\t%12s\n",
-			"id",
-			"object",
-			"dictionary",
-			"stream");
-
-		pdf_object_t *ptr = pdf->next;
-		while (ptr != NULL) {
-			printf("\t%8d\t%12d\t%12d\t%12d\n",
-				ptr->id,
-				ptr->object_size,
-				ptr->dictionary_size,
-				ptr->stream_size);
-			ptr = ptr->next;
-		}
-	}
-
 	if ((*param)->stat > 0)
-		printf("Generated %d object(s)\n",
-			pdf_get_count(&pdf));
-
-	int *ids = NULL;
-
-	if ((*param)->file_stat->outline > 0) {
-		if ((*param)->stat > 1)
-			printf("Generating outline object(s)\n\t%8s\n", "id");
-
-		pdf_get_free_ids(&pdf, &ids, (*param)->file_stat->outline + 1);
-		int outline = pdf_cnki_outline(&pdf, &(*param)->object_outline, &ids);
-
-		if ((*param)->stat > 1)
-			for (int i = 0; i < (*param)->file_stat->outline + 1; i++)
-				printf("\t%8d\n", ids[i]);
-
-		if ((*param)->stat > 0) {
-			if (outline != 0)
-				printf("No outline information\n");
-			else
-				printf("Generated %d outline object(s)\n",
-					(*param)->file_stat->outline + 1);
-		}
-	}
-
-	if ((*param)->stat > 1)
-		printf("Writing header\n");
-
-	long cur = 0;
-
-	if ((*param)->stat > 0)
-		cur = ftell((*param)->fp_o);
-
-	if (pdf_dump_header(&pdf, &(*param)->fp_o) != 0) {
-		fprintf(stderr, "Header not written\n");
-		return 1;
-	} else {
-		if ((*param)->stat > 0)
-			printf("Header %ld byte(s) written\n",
-				ftell((*param)->fp_o) - cur);
-	}
-
-	if ((*param)->stat > 1)
-		printf("Writing object(s)\n");
-
-	pdf_dump_obj(&pdf, &(*param)->fp_o);
-
-	if ((*param)->stat > 1) {
-		printf("\t%8s\t%8s\t%8s\t%12s\t%12s\t%12s\n",
-			"address",
-			"size",
-			"id",
-			"object",
-			"dictionary",
-			"stream");
-
-		pdf_object_t *ptr = pdf->next;
-		while (ptr != NULL) {
-			printf("\t%08x\t%8d\t%8d\t%12d\t%12d\t%12d\n",
-				ptr->address,
-				ptr->size,
-				ptr->id,
-				ptr->object_size,
-				ptr->dictionary_size,
-				ptr->stream_size);
-			ptr = ptr->next;
-		}
-	}
-
-	if ((*param)->stat > 0)
-		printf("%d object(s) %ld byte(s) written\n",
-			pdf_get_count(&pdf),
-			ftell((*param)->fp_o));
-
-	long xref = ftell((*param)->fp_o);
-
-	if ((*param)->stat > 1)
-		printf("Writing cross-reference table\n");
-
-	if (pdf_dump_xref(&pdf, &(*param)->fp_o) != 0) {
-		if ((*param)->stat > 0)
-			printf("Cross-reference table not written\n");
-	} else {
-		if ((*param)->stat > 0)
-			printf("Cross-reference table %ld byte(s) written\n",
-				ftell((*param)->fp_o) - xref);
-	}
-
-	if ((*param)->stat > 1)
-		printf("Writing trailer\n");
-
-	if ((*param)->stat > 0)
-		cur = ftell((*param)->fp_o);
-
-	if (pdf_dump_trailer(&pdf, &(*param)->fp_o, xref) != 0) {
-		if ((*param)->stat > 0)
-			printf("Trailer not written\n");
-	} else {
-		if ((*param)->stat > 0)
-			printf("Trailer %ld byte(s) written\n",
-				ftell((*param)->fp_o) - cur);
-	}
-
-	if ((*param)->stat > 0)
-		printf("Total %ld byte(s) written\n",
-			ftell((*param)->fp_o));
+		printf("Loaded %d page(s)\n", (*param)->file_stat->page);
 
-	pdf_obj_destroy(&pdf);
+	cnki_pdf_hn(param);
 
 	if ((*param)->stat > 0)
-		printf("Conversion ended (partial)\n");
+		printf("Conversion ended\n");
 
-	/* TODO: Finish me please :) */
 	return 0;
 }