TSG-9869 修复Edit_Element，在命中Start_Indicator时MARK标记问题v4.6.10-20220311

2 files changed, 108 insertions, 14 deletions

diff --git a/plugin/business/pangu-http/src/edit_element.cpp b/plugin/business/pangu-http/src/edit_element.cpp
index ed686e2..988ba6e 100644
--- a/plugin/business/pangu-http/src/edit_element.cpp
+++ b/plugin/business/pangu-http/src/edit_element.cpp
@@ -100,14 +100,14 @@ int construct_cjson_by_treatment(cJSON *a_element, char **node, int *step, int *
 			{
 				if(*node != NULL && strcasecmp(*node, start_indicator) != 0)
 				{
-					return 0;
+					return -2;
 				}
 			}
 			if(a_element->type == cJSON_Array)
 			{
 				if(a_element->string != NULL && strcasecmp(a_element->string, start_indicator))
 				{
-					return 0;
+					return -2;
 				}
 			}
 		}
@@ -135,14 +135,14 @@ int construct_cjson_by_treatment(cJSON *a_element, char **node, int *step, int *
 			{
 				if(*node != NULL && strcasecmp(*node, start_indicator) != 0)
 				{
-					return 0;
+					return -2;
 				}
 			}
 			if(a_element->type == cJSON_Array)
 			{
 				if(a_element->string != NULL && strcasecmp(a_element->string, start_indicator))
 				{
-					return 0;
+					return -2;
 				}
 			}
 		}
@@ -157,7 +157,7 @@ int construct_cjson_by_treatment(cJSON *a_element, char **node, int *step, int *
 		}
 	}
 
-	return 0;
+	return 1;
 }
 
 int construct_html_by_treatment(const struct edit_element_rule * rules, xmlNodePtr node, xmlNodePtr *parent_array, size_t *n_parent, int *match)
@@ -271,15 +271,15 @@ int cjson_dump_array(cJSON *a, int *depth, int *step, int *step_array_level, cha
 		}
 		if(*depth == 0)
 		{
-			construct_cjson_by_treatment(a_element, node, step, step_array_level, rules);
+			xret = construct_cjson_by_treatment(a_element, node, step, step_array_level, rules);
 		}
-		if(xret == 1)
+		if(xret == 1 || xret == -2)
 		{
 			*step = (*step >= 2047) ? 2047 : *step;
 			step_array_level[*step] = array_cnt;
 			*node = a_element->string;
 			*depth = *depth -1;
-			return 1;
+			return xret;
 		}
 		array_cnt++;
         a_element = a_element->next;
@@ -302,13 +302,13 @@ int cjson_dump_object(cJSON *a, int *depth, int *step, int *step_array_level, ch
 		}
 		if(*depth == 0)
 		{
-			construct_cjson_by_treatment(a_element, node, step, step_array_level, rules);
+			xret = construct_cjson_by_treatment(a_element, node, step, step_array_level, rules);
 		}
-		if(xret == 1)
+		if(xret == 1 || xret == -2)
 		{
 			*node = a_element->string;
 			*depth = *depth -1;
-			return 1;
+			return xret;
 		}
     }
 	return xret;
@@ -526,7 +526,7 @@ size_t parse_json_output_unformatted(const char * in, size_t in_sz, const struct
 		}
 	}
 
-	if(match == 0)
+	if(match == 0 || match == -2)
 	{
 		goto finish;
 	}
@@ -587,7 +587,7 @@ size_t format_json_file_type(const char * in, size_t in_sz, const struct edit_el
 		match_num--;
 	}
 
-	if(match == 0)
+	if(match == 0 || match == -2)
 	{
 		goto finish;
 	}
@@ -751,7 +751,12 @@ size_t format_input_html(const char * in, size_t in_sz, const struct edit_elemen
 	{
 		if(doc->children != NULL && doc->children->next != NULL)
 		{
-			xmlNewProp(doc->children->next, (const xmlChar *)"need_check", (const xmlChar *)"true");
+			xmlNodePtr node = doc->children->next;
+			while(node->type != XML_ELEMENT_NODE)
+			{
+				node = node->next;
+			}
+			xmlNewProp(node, (const xmlChar *)"need_check", (const xmlChar *)"true");
 		}
 		else if(doc->children != NULL)
 		{
diff --git a/plugin/business/pangu-http/src/test_edit_element.cpp b/plugin/business/pangu-http/src/test_edit_element.cpp
index 07b2eb8..5893846 100644
--- a/plugin/business/pangu-http/src/test_edit_element.cpp
+++ b/plugin/business/pangu-http/src/test_edit_element.cpp
@@ -134,6 +134,95 @@ TEST(EditElement, Libxml_Inside_Remove_Facebook)
 	free(output);
 }
 
+TEST(EditElement, Libxml_Whole_Remove_Google_Test)
+{
+	char* output=NULL;
+	size_t output_sz=0,input_len=0;
+
+	const char* filename="./test_data/google_search_gtest_cnblog.html";
+
+	FILE* fp=NULL;
+	struct stat file_info;
+	stat(filename, &file_info);
+	size_t input_sz=file_info.st_size;
+
+	fp=fopen(filename,"r");
+	ASSERT_FALSE(fp==NULL);
+	if(fp==NULL)
+	{
+		return;
+	}
+	char* input=(char*)malloc(input_sz);
+	fread(input,1,input_sz,fp);
+	fclose(fp);
+
+	const char *user_region = "{\"rules\":[{\"anchor_element\":{\"search_scope\":\"whole_file\",\"start_indicator\":\"g\",\"contained_keyword\":\"thrift-service-framework/src/test/java/com/cnblogs/yjmyzz at master\"},\
+	\"target_element\":{\"target_distance_from_matching\":4,\"element_treatment\":\"remove\"}}]}";
+
+	simple_edit_element(user_region, input, input_len, &output, &output_sz, 0);
+	EXPECT_TRUE(output_sz>0);
+
+	EXPECT_TRUE(NULL==strstr(output, "thrift-service-framework/src/test/java/com/cnblogs/yjmyzz at master"));
+	free(input);
+	free(output);
+}
+
+TEST(EditElement, Cjson_Inside_Mark_NO_Match)
+{
+	char* output=NULL;
+	size_t output_sz=0;
+
+	const char *input ="{\"A\":\"B\",\"C\":[\"D\", \"E\"],\"F\":{\"G\":{\"H\":\"I\"}},\"J\":{\"K\":[{\"L\":\"M\"}],\"N\":\"Q\"}}\n\n{\"data\":{\"user\":{\"id\":\"1\",\"units\":\
+			{\"edges\":[{\"eligible\":true,\"story\":{\"message\":{\"text\":\"Firefox 3.6 with System\\uff0c\\u7f16\\u8f91\"}}}]}}}}";
+
+	const char *user_region = "{\"rules\":[{\"anchor_element\":{\"search_scope\":\"inside_element\",\"start_indicator\":\"error_message\",\"contained_keyword\":\"Firefox 3.6 with System\"},\
+	\"target_element\":{\"target_distance_from_matching\":1,\"element_treatment\":\"mark\"}}]}";
+
+	simple_edit_element(user_region, input, strlen(input), &output, &output_sz, 0);
+	EXPECT_TRUE(output_sz==0);
+
+	free(output);
+}
+
+TEST(EditElement, Cjson_Inside_Mark_Match)
+{
+	char* output=NULL;
+	size_t output_sz=0;
+
+	const char *input ="{\"A\":\"B\",\"C\":[\"D\", \"E\"],\"F\":{\"G\":{\"H\":\"I\"}},\"J\":{\"K\":[{\"L\":\"M\"}],\"N\":\"Q\"}}\n\n{\"data\":{\"user\":{\"id\":\"1\",\"units\":\
+		{\"edges\":[{\"eligible\":true,\"story\":{\"message\":{\"text\":\"Firefox 3.6 with System\\uff0c\\u7f16\\u8f91\"}}}]}}}}";
+
+	const char *user_region = "{\"rules\":[{\"anchor_element\":{\"search_scope\":\"inside_element\",\"start_indicator\":\"message\",\"contained_keyword\":\"Firefox 3.6 with System\"},\
+	\"target_element\":{\"target_distance_from_matching\":1,\"element_treatment\":\"mark\"}}]}";
+
+	simple_edit_element(user_region, input, strlen(input), &output, &output_sz, 0);
+	EXPECT_TRUE(output_sz>0);
+
+	EXPECT_TRUE(NULL!=strstr(output, "need_filter"));
+	EXPECT_TRUE(NULL!=strstr(output, "need_check"));
+
+	free(output);
+}
+
+TEST(EditElement, LibXML_Inside_Mark_Match)
+{
+	char* output=NULL;
+	size_t output_sz=0;
+
+	const char *input = "<!DOCTYPE html>\n<!-- saved from url=(0038)https://www.facebook.com -->\n<html id=\"facebook\">\n\
+<head>\n<title>Facebook</title>\n<head>\n<body>\n<div class=\"f530mmz5\"><div class=\"c1et5uql\">Firefox 3.6 with System</div></div>\n</body>";
+
+	const char *user_region = "{\"rules\":[{\"anchor_element\":{\"search_scope\":\"inside_element\",\"start_indicator\":\"f530mmz5\",\"contained_keyword\":\"Firefox 3.6 with System\"},\
+	\"target_element\":{\"target_distance_from_matching\":1,\"element_treatment\":\"mark\"}}]}";
+
+	simple_edit_element(user_region, input, strlen(input), &output, &output_sz, 0);
+	EXPECT_TRUE(output_sz>0);
+
+	EXPECT_TRUE(NULL!=strstr(output, "need_filter"));
+	EXPECT_TRUE(NULL!=strstr(output, "need_check"));
+	free(output);
+}
+
 int main(int argc, char ** argv)
 {