From c2ad6549fb337ce707e04aa441c9b492171a3b9d Mon Sep 17 00:00:00 2001
From: yzrh <yzrh@noema.org>
Date: Sun, 25 Dec 2022 18:03:01 +0000
Subject: Handle headless HN and page with no image.

Signed-off-by: yzrh <yzrh@noema.org>
---
 src/cnki_pdf.c | 265 ++++++++++++++++++++++++++++++---------------------------
 1 file changed, 139 insertions(+), 126 deletions(-)

(limited to 'src/cnki_pdf.c')

diff --git a/src/cnki_pdf.c b/src/cnki_pdf.c
index c56a45f..603ce01 100644
--- a/src/cnki_pdf.c
+++ b/src/cnki_pdf.c
@@ -481,7 +481,10 @@ cnki_pdf_hn(cnki_t **param)
 		 * page object
 		 */
 		int *ids = NULL;
-		pdf_get_free_ids(&pdf, &ids, ptr->image_length + 3);
+		if (ptr->image_length > 0)
+			pdf_get_free_ids(&pdf, &ids, ptr->image_length + 3);
+		else
+			pdf_get_free_ids(&pdf, &ids, 2);
 
 		int bitmap_size;
 		char *bitmap;
@@ -489,27 +492,31 @@ cnki_pdf_hn(cnki_t **param)
 		int stream_size;
 		char *stream;
 
-		int *dim = malloc(2 * ptr->image_length * sizeof(int));
+		int *dim;
 
-		int ret;
-		int info[3];
+		if (ptr->image_length > 0) {
+			dim = malloc(2 * ptr->image_length * sizeof(int));
 
-		if (dim == NULL) {
-			free(root_kid);
-			free(ids);
-			return 1;
-		}
+			if (dim == NULL) {
+				free(root_kid);
+				free(ids);
+				return 1;
+			}
 
-		dictionary_size = 256;
-		dictionary = malloc(dictionary_size);
+			dictionary_size = 256;
+			dictionary = malloc(dictionary_size);
 
-		if (dictionary == NULL) {
-			free(root_kid);
-			free(ids);
-			free(dim);
-			return 1;
+			if (dictionary == NULL) {
+				free(root_kid);
+				free(ids);
+				free(dim);
+				return 1;
+			}
 		}
 
+		int ret;
+		int info[3];
+
 		for (int i = 0; i < ptr->image_length; i++) {
 			memset(dictionary, 0, dictionary_size);
 
@@ -684,66 +691,42 @@ cnki_pdf_hn(cnki_t **param)
 			}
 		}
 
-		memset(dictionary, 0, dictionary_size);
+		if (ptr->image_length > 0) {
+			memset(dictionary, 0, dictionary_size);
 
-		strcat(dictionary, "<<\n/XObject <<");
+			strcat(dictionary, "<<\n/XObject <<");
 
-		for (int i = 0; i < ptr->image_length; i++) {
-			snprintf(buf, 64, "/Im%d %d 0 R", i, ids[i]);
-			strcat(dictionary, buf);
+			for (int i = 0; i < ptr->image_length; i++) {
+				snprintf(buf, 64, "/Im%d %d 0 R", i, ids[i]);
+				strcat(dictionary, buf);
 
-			if (i + 1 < ptr->image_length)
-				strcat(dictionary, " ");
-		}
+				if (i + 1 < ptr->image_length)
+					strcat(dictionary, " ");
+			}
 
-		strcat(dictionary, ">>\n>>");
+			strcat(dictionary, ">>\n>>");
 
-		pdf_obj_append(&pdf, ids[ptr->image_length], NULL, dictionary, NULL, 0);
+			pdf_obj_append(&pdf, ids[ptr->image_length], NULL, dictionary, NULL, 0);
 
-		free(dictionary);
+			free(dictionary);
+		}
 
 		int conv_size;
 		char *conv_dst;
 		char conv_src[2];
 		char conv_hex[3];
 
-		if (strncmp(ptr->text + 8, "COMPRESSTEXT", 12) == 0) {
-			cnki_zlib(&stream, &stream_size, ptr->text, ptr->text_size);
-
-			dictionary_size = 64 + 2 * stream_size;
-			dictionary = malloc(dictionary_size);
+		if (ptr->text_size > 0) {
+			if (strncmp(ptr->text + 8, "COMPRESSTEXT", 12) == 0 ||
+				strncmp(ptr->text, "COMPRESSTEXT", 12) == 0) {
+				cnki_zlib(&stream, &stream_size, ptr->text, ptr->text_size);
 
-			if (dictionary == NULL) {
-				free(root_kid);
-				free(ids);
-				free(dim);
-				return 1;
-			}
+				free(ptr->text);
 
-			memset(dictionary, 0, dictionary_size);
-
-			strcat(dictionary, "<feff");
-
-			for (int i = 0; i < stream_size; i += 16) {
-				conv_src[0] = stream[i + 7];
-				conv_src[1] = stream[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(dictionary, conv_hex);
-					}
-					free(conv_dst);
-				}
+				ptr->text_size = stream_size;
+				ptr->text = stream;
 			}
-			free(stream);
 
-			strcat(dictionary, ">");
-		} else {
 			dictionary_size = 64 + 2 * ptr->text_size;
 			dictionary = malloc(dictionary_size);
 
@@ -758,9 +741,26 @@ cnki_pdf_hn(cnki_t **param)
 
 			strcat(dictionary, "<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];
+			for (int i = 0; i < ptr->text_size; i += 6) {
+				if (i + 5 >= ptr->text_size)
+					break;
+
+				conv_src[0] = ptr->text[i + 5];
+				conv_src[1] = ptr->text[i + 4];
+
+				if ((conv_src[0] << 8 | conv_src[1]) == 0xa389) {
+					strcat(dictionary, "a389");
+					continue;
+				} else if ((conv_src[0] << 8 | conv_src[1]) == 0xa38a) {
+					strcat(dictionary, "a38a");
+					continue;
+				} else if ((conv_src[0] << 8 | conv_src[1]) == 0xa38d) {
+					strcat(dictionary, "a38d");
+					continue;
+				} else if ((conv_src[0] << 8 | conv_src[1]) == 0xa3a0) {
+					strcat(dictionary, "a3a0");
+					continue;
+				}
 
 				conv_size = 6;
 
@@ -776,12 +776,12 @@ cnki_pdf_hn(cnki_t **param)
 			}
 
 			strcat(dictionary, ">");
-		}
 
-		/* FIXME: Use the text somehow? */
-		free(dictionary);
+			/* FIXME: Use the text somehow? */
+			free(dictionary);
+		}
 
-		dictionary_size = 64 + 64 * ptr->image_length;
+		dictionary_size = 64 + 128 * ptr->image_length;
 		dictionary = malloc(dictionary_size);
 
 		if (dictionary == NULL) {
@@ -791,96 +791,109 @@ cnki_pdf_hn(cnki_t **param)
 			return 1;
 		}
 
-		memset(dictionary, 0, dictionary_size);
-
-		strcat(dictionary, "q\n");
+		if (ptr->image_length > 0) {
+			memset(dictionary, 0, dictionary_size);
 
-		strcat(dictionary, "0.25 0 0 0.25 0 0 cm\n");
+			strcat(dictionary, "q\n");
 
-		double resize_x;
-		double resize_y;
+			strcat(dictionary, "0.25 0 0 0.25 0 0 cm\n");
 
-		for (int i = 0; i < ptr->image_length; i++) {
-			if (dim[i * 2] <= 0 || dim[i * 2 + 1] <= 0)
-				continue;
+			double resize_x;
+			double resize_y;
 
-			/* Scale within bound of A4 paper */
-			resize_x = 595.276 * 4 / dim[i * 2];
-			resize_y = 841.89 * 4 / dim[i * 2 + 1];
+			for (int i = 0; i < ptr->image_length; i++) {
+				if (dim[i * 2] <= 0 || dim[i * 2 + 1] <= 0)
+					continue;
 
-			if (resize_y < resize_x)
-				snprintf(buf, 64, "%f 0 0 %f 0 0 cm\n",
-					resize_y, resize_y);
-			else
-				snprintf(buf, 64, "%f 0 0 %f 0 0 cm\n",
-					resize_x, resize_x);
-			strcat(dictionary, buf);
+				/* Scale within bound of A4 paper */
+				resize_x = 595.276 * 4 / dim[i * 2];
+				resize_y = 841.89 * 4 / dim[i * 2 + 1];
 
-			/* Apply transformation matrix */
-			if (ptr->image_data[i].format == JBIG || ptr->image_data[i].format == DCT_1) {
-				snprintf(buf, 64, "1 0 0 1 0 %d cm\n",
-					dim[i * 2 + 1]);
+				if (resize_y < resize_x)
+					snprintf(buf, 64, "%f 0 0 %f 0 0 cm\n",
+						resize_y, resize_y);
+				else
+					snprintf(buf, 64, "%f 0 0 %f 0 0 cm\n",
+						resize_x, resize_x);
 				strcat(dictionary, buf);
 
-				strcat(dictionary, "1 0 0 -1 0 0 cm\n");
-			}
+				/* Apply transformation matrix */
+				if (ptr->image_data[i].format == JBIG || ptr->image_data[i].format == DCT_1) {
+					snprintf(buf, 64, "1 0 0 1 0 %d cm\n",
+						dim[i * 2 + 1]);
+					strcat(dictionary, buf);
 
-			snprintf(buf, 64, "%d 0 0 %d 0 0 cm\n",
-				dim[i * 2], dim[i * 2 + 1]);
-			strcat(dictionary, buf);
+					strcat(dictionary, "1 0 0 -1 0 0 cm\n");
+				}
 
-			snprintf(buf, 64, "/Im%d Do\n", i);
-			strcat(dictionary, buf);
-		}
+				snprintf(buf, 64, "%d 0 0 %d 0 0 cm\n",
+					dim[i * 2], dim[i * 2 + 1]);
+				strcat(dictionary, buf);
 
-		strcat(dictionary, "Q");
+				snprintf(buf, 64, "/Im%d Do\n", i);
+				strcat(dictionary, buf);
+			}
 
-		if (strdeflate(&stream, &stream_size, dictionary, strlen(dictionary)) != 0) {
-			free(root_kid);
-			free(ids);
-			free(dim);
-			free(dictionary);
-			return 1;
-		}
+			strcat(dictionary, "Q");
 
-		memset(dictionary, 0, dictionary_size);
+			if (strdeflate(&stream, &stream_size, dictionary, strlen(dictionary)) != 0) {
+				free(root_kid);
+				free(ids);
+				free(dim);
+				free(dictionary);
+				return 1;
+			}
 
-		strcat(dictionary, "<<\n");
+			memset(dictionary, 0, dictionary_size);
 
-		snprintf(buf, 64, "/Length %d\n", stream_size);
-		strcat(dictionary, buf);
+			strcat(dictionary, "<<\n");
 
-		strcat(dictionary, "/Filter /FlateDecode\n");
+			snprintf(buf, 64, "/Length %d\n", stream_size);
+			strcat(dictionary, buf);
 
-		strcat(dictionary, ">>");
+			strcat(dictionary, "/Filter /FlateDecode\n");
+
+			strcat(dictionary, ">>");
 
-		pdf_obj_append(&pdf, ids[ptr->image_length + 1],
-			NULL, dictionary, stream, stream_size);
+			pdf_obj_append(&pdf, ids[ptr->image_length + 1],
+				NULL, dictionary, stream, stream_size);
 
-		free(stream);
+			free(stream);
+		}
 
 		memset(dictionary, 0, dictionary_size);
 
 		strcat(dictionary, "<<\n/Type /Page\n");
 
-		snprintf(buf, 64, "/Resources %d 0 R\n", ids[ptr->image_length]);
-		strcat(dictionary, buf);
-
-		snprintf(buf, 64, "/Contents %d 0 R\n", ids[ptr->image_length + 1]);
-		strcat(dictionary, buf);
-
 		/* A4 paper */
 		strcat(dictionary, "/MediaBox [0 0 595.276 841.89]\n");
 
-		/* Add /Parent when we know root */
-		pdf_obj_append(&pdf, ids[ptr->image_length + 2], NULL, dictionary, NULL, 0);
+		if (ptr->image_length > 0) {
+			free(dim);
 
-		free(dictionary);
+			snprintf(buf, 64, "/Resources %d 0 R\n", ids[ptr->image_length]);
+			strcat(dictionary, buf);
+
+			snprintf(buf, 64, "/Contents %d 0 R\n", ids[ptr->image_length + 1]);
+			strcat(dictionary, buf);
+
+			/* Add /Parent when we know root */
+			pdf_obj_append(&pdf, ids[ptr->image_length + 2], NULL, dictionary, NULL, 0);
 
-		root_kid[cnt++] = ids[ptr->image_length + 2];
+			root_kid[cnt++] = ids[ptr->image_length + 2];
+		} else {
+			snprintf(buf, 64, "/Contents %d 0 R\n", ids[ptr->image_length]);
+			strcat(dictionary, buf);
+
+			/* Add /Parent when we know root */
+			pdf_obj_append(&pdf, ids[ptr->image_length + 1], NULL, dictionary, NULL, 0);
+
+			root_kid[cnt++] = ids[ptr->image_length + 1];
+		}
+
+		free(dictionary);
 
 		free(ids);
-		free(dim);
 
 		ptr = ptr->next;
 	}
-- 
cgit v1.2.3