Handle HN page with figure only.

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

1 files changed, 20 insertions, 19 deletions

diff --git a/src/cnki_pdf.c b/src/cnki_pdf.c
index 372c5da..45f5324 100644
--- a/src/cnki_pdf.c
+++ b/src/cnki_pdf.c
@@ -218,7 +218,7 @@ cnki_pdf(cnki_t **param)
 	if ((*param)->stat > 0)
 		printf("Discovered %d parent object(s)\n", parent[0]);
 
-	int *parent_missing = malloc(parent[0] * sizeof(int));
+	int8_t *parent_missing = malloc(parent[0] * sizeof(int8_t));
 
 	if (parent_missing == NULL)
 		return 1;
@@ -343,7 +343,7 @@ cnki_pdf(cnki_t **param)
 			strcat(dictionary, "[");
 
 		for (int i = 0, j = 0; i < parent[0]; i++) {
-			if (parent_missing[i]) {
+			if (parent_missing[i] == 1) {
 				snprintf(buf, 64, "%d 0 R", parent[i + 1]);
 				strcat(dictionary, buf);
 
@@ -793,15 +793,11 @@ cnki_pdf_hn(cnki_t **param)
 			strcat(dictionary, "/F0 10 Tf\n");
 
 			for (int i = 0, j = 0; i < ptr->text_size - 1;) {
-				switch (ptr->text[i]) {
-					case 0x01:
-						if (ptr->address_next <= ptr->address) {
-							i += 2;
-							break;
-						}
-
-						strcat(dictionary, "T*\n");
-					case 0x70:
+				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;
 
@@ -867,9 +863,12 @@ cnki_pdf_hn(cnki_t **param)
 
 						i += 8;
 						break;
-					case 0x0a:
+					case 0x800a:
 						if (i + 27 >= ptr->text_size || j >= ptr->image_length) {
 							i += 2;
+
+							if (j >= ptr->image_length)
+								i += 26;
 							break;
 						}
 
@@ -896,7 +895,9 @@ cnki_pdf_hn(cnki_t **param)
 						}
 
 						i += 28;
-						j++;
+
+						if (j == 0 || ptr->image_data[j].x > 0 || ptr->image_data[j].y > 0)
+							j++;
 						break;
 					default:
 						i += 4;
@@ -915,10 +916,10 @@ cnki_pdf_hn(cnki_t **param)
 			double resize_x = 1;
 			double resize_y = 1;
 
-			if (dim[0] > 0 && dim[1] > 0) {
+			if (ptr->image_data[0].x == 0 && ptr->image_data[0].y == 0 && dim[0] > 0 && dim[1] > 0) {
 				/* Scale within bound of A4 paper */
-				resize_x = 4 * 595.2756 / dim[0];
-				resize_y = 4 * 841.8898 / dim[1];
+				resize_x = 2381.102 / dim[0];
+				resize_y = 3367.559 / dim[1];
 
 				if (resize_y < resize_x)
 					snprintf(resize_str, 64, "%f 0 0 %f 0 0 cm\n",
@@ -948,9 +949,9 @@ cnki_pdf_hn(cnki_t **param)
 				}
 
 				/* Translate figure */
-				if (i > 0) {
-					double origin_x = 0.4043339 * ptr->image_data[i].x;
-					double origin_y = 0.4043273 * ptr->image_data[i].y;
+				if (ptr->image_data[i].x > 0 || ptr->image_data[i].y > 0) {
+					double origin_x = ptr->image_data[i].x * dim[i * 2] / ptr->image_data[i].w;
+					double origin_y = ptr->image_data[i].y * dim[i * 2 + 1] / ptr->image_data[i].h;
 
 					if (ptr->image_data[i].format == JBIG || ptr->image_data[i].format == DCT_1)
 						origin_y = -3367.559 + origin_y + dim[i * 2 + 1];