Add weixin 4.0 OCR support

Author: lixungeng

README.md

@@ -1,23 +1,48 @@
 # wcocr: demonstrate how to use WeChatOCR.exe
 
 Great thanks to IEEE by his [Project IEEE/QQImpl](https://github.com/EEEEhex/qqimpl)] and [article](https://bbs.kanxue.com/thread-278161.htm).
-This project is based on it. 
+This project is based on it and reduced the product size by using `protobuf-lite` instead of `protobuf`.
 
-This project reduced the product size by using `protobuf-lite` instead of `protobuf`, 
-and provided a direct Python interface for calling in sync mode.
+This project provided a direct Python interface for calling in sync mode as well as other languages support including but not limited with c++/java/c#.
 
-To use this project, you need to prepare the `wechatocr.exe` and the wechat folder.
-For example, `wechatocr.exe` might be:
+# Prepare for usage
+
+To work with this project, you need to prepare the wechat OCR binary and the wechat runtime folder.
+
+For wechat 3.x, the wechat OCR binary is `wechatocr.exe`, it might be:
 
 ```
 C:\Users\yourname\AppData\Roaming\Tencent\WeChat\XPlugin\Plugins\WeChatOCR\7061\extracted\WeChatOCR.exe
 ```
-and the wechat folder might be:
+and the wechat runtime folder might be:
 ```
 C:\Program Files (x86)\Tencent\WeChat\[3.9.8.25]
 ```
 
+**Wechat 4.0 is now supported!**
+
+For wechat 4.0, the wechat OCR binary is `wxocr.dll`, it might be:
+
+```
+C:\Users\yourname\AppData\Roaming\Tencent\xwechat\XPlugin\plugins\WeChatOcr\8011\extracted\wxocr.dll
+```
+
+and the wechat runtime folder might be:
+
+```
+C:\Program Files\Tencent\Weixin\4.0.0.26
+```
+
+## Warning
+
+Wechat 4.0 OCR binary is `wxocr.dll`, but this project built a dll named `wcocr.dll`
+
+**Their names are similar, DO NOT confuse them.**
+
+
+
 ## C++ interface
+
 You can use the following code to test it:
 ```cpp
 CWeChatOCR ocr(wechatocr_path, wechat_path);

compile-proto.bat

@@ -0,0 +1,5 @@
+@echo off
+cd /d %~dp0
+cd src
+..\spt\protoc.exe --cpp_out=lite:. -I ../pb ocr_common.proto ocr_wx3.proto ocr_wx4.proto
+pause

java/Test.java

@@ -18,7 +18,7 @@ interface SetResCallback extends Callback {
     public static void main(String [] args) {
         System.out.println("ocr begin...");
         String ocr_exe = "C:\\Users\\xungeng\\AppData\\Roaming\\Tencent\\WeChat\\XPlugin\\Plugins\\WeChatOCR\\7079\\extracted\\WeChatOCR.exe";
-        String wechat_dir = "C:\\Program Files (x86)\\Tencent\\WeChat\\[3.9.10.19]";
+        String wechat_dir = "C:\\Program Files\\Tencent\\Weixin\\4.0.0.26";
         String tstpng = "test.png";
         AtomicReference<String> result = new AtomicReference<>();
         WechatOCR.dll.wechat_ocr(new WString(ocr_exe), new WString(wechat_dir), tstpng, new WechatOCR.SetResCallback() {

pb/ocr_common.proto

@@ -0,0 +1,32 @@
+syntax = "proto3";
+package ocr_common;
+
+message Point{
+	optional float x = 1;
+	optional float y = 2;
+}
+
+message Box{
+	optional Point topleft = 1;
+	optional Point topright = 2;
+	optional Point bottomright = 3;
+	optional Point bottomleft = 4;
+}
+
+message OCRResultChar{
+	optional Box char_box = 1;
+	optional string chars = 2;
+}
+
+message OCRResultLine {
+	Box line_box = 1;
+	string text = 2;				//UTF8格式的字符串
+	float rate = 3;					//单行的识别率
+	repeated OCRResultChar blocks = 4;
+	float left = 5;					//识别矩形的left\top\right\bottom的坐标
+	float top = 6;
+	float right = 7;
+	float bottom = 8;
+	optional bool unknown_0 = 9;	//未知
+	optional Box box10 = 10;		//未知
+}

pb/ocr_protobuf.proto

@@ -0,0 +1,36 @@
+syntax = "proto3";
+package wx3;
+import "ocr_common.proto";
+
+message OcrInputBuffer {
+	// 看起来有两种传递图片的方式。
+	// 第一种是只传文件路径 pic_path = "C:/path/to/xxx.png"
+	// 第二种是用字节流传，猜测需要用到u2,u3,pic_data变量。暂时微信没有用到，可能是为将来保留
+	optional string pic_path = 1;
+	optional uint32 u2 = 2;
+	optional uint32 u3 = 3;
+	optional bytes pic_data = 4;
+}
+
+message OcrOutputBuffer {
+	repeated ocr_common.OCRResultLine lines = 1;	//repeated 每行的结果
+	optional uint32 img_width = 2;
+	optional uint32 img_height = 3;
+	optional string unk4 = 4;
+}
+
+message OcrRespond {
+	optional int32 type = 1;									// type=1像是初始化成功回调。如果是正常OCR请求，回答的type=0
+	optional uint64 task_id = 2;
+	optional int32 err_code = 3;
+	optional OcrOutputBuffer ocr_result = 4;
+}
+
+message OcrRequest {
+	int32 type = 1;	 //为0执行ocr，为1会直接返回init信息. 与OcrRespond.type意义相同
+	// 经过反复核查，在腾讯proto文件中，这个task_id确实是64位的。但在，在执行过程中，高32位会被丢弃，且第32位为1会出错。
+	// 也就是协议上是有64位的uint64，实际上只能有31位。必须是>0的整形数字，范围是[1,2147483647]
+	// 由于 task_id = 1会被用于初始化，所以最好取值为 [2, 0x7fffFFFF]
+	uint64 task_id = 2;
+	OcrInputBuffer input = 3;
+}

pb/ocr_wx3.proto

@@ -0,0 +1,44 @@
+syntax = "proto3";
+package wx4;
+import "ocr_common.proto";
+
+message OCRSupportMessage {
+	optional bool supported = 1;
+}
+
+message ReqType {
+	optional bool t1 = 1;
+	optional bool t2 = 2;
+	optional bool t3 = 3;
+}
+
+message ParseOCRReqMessage {
+	optional uint64 task_id = 1;
+	optional string pic_path = 2;
+	optional uint32 xx3 = 3;
+	optional uint32 xx4 = 4;
+	optional bytes pic_data = 5;
+	optional ReqType rt = 6;
+}
+
+message OCRResultInfo {
+	repeated ocr_common.OCRResultLine lines = 3;
+	optional uint32 img_width = 4;
+	optional uint32 img_height = 5;
+	optional string cpu_report = 6;
+	optional uint64 time_used = 7;
+}
+
+message QRResultInfo {
+}
+message MMFGResultInfo {
+}
+
+message ParseOCRRespMessage {
+	optional uint64 task_id = 1;
+	optional int32 err_code = 2;
+	optional OCRResultInfo res = 3;
+	optional ReqType rt = 4;
+	optional bytes qrcode = 5; // ¶þάÂëʶ±ð
+	optional bytes mmfg = 6; // what is mmfg?
+}

pb/ocr_wx4.proto

@@ -47,7 +47,8 @@ bool wechat_ocr(const wchar_t* ocr_exe, const wchar_t * wechat_dir, const char *
 		return false;
 	}
 	CWeChatOCR::result_t res;
-	wechat_ocr.doOCR(imgfn, &res);
+	if (!wechat_ocr.doOCR(imgfn, &res))
+		return false;
 	string json;
 	json += "{";
 	json += "\"errcode\":" + std::to_string(res.errcode) + ",";
@@ -95,10 +96,12 @@ HRESULT DllRegisterServer(void)
 
 	CWeChatOCR wechat_ocr(_wgetenv(L"WECHATOCR_EXE"), _wgetenv(L"WECHAT_DIR"));
 	if (!wechat_ocr.wait_connection(5000)) {
+		fprintf(stderr, "wechat_ocr.wait_connection failed\n");
 		return E_FAIL;
 	}
 	wechat_ocr.doOCR(getenv("TEST_PNG"), nullptr);
 	wechat_ocr.wait_done(-1);
+	fprintf(stderr, "debug play ocr DONE!\n");
 	return S_OK;
 }
 #endif

src/main.cpp

@@ -2,6 +2,12 @@
 #include "mojocall.h"
 #include "mmmojo.h"
 
+#ifdef _DEBUG
+#define DBG_PRINT(fmt, ...) fprintf(stderr, "[DEBUG]" fmt, __VA_ARGS__)
+#else
+#define DBG_PRINT(fmt, ...)
+#endif
+
 CMojoCall::~CMojoCall()
 {
 	Stop();
@@ -46,30 +52,38 @@ bool CMojoCall::Start(LPCWSTR exepath)
 		//设置回调函数
 		SetMMMojoEnvironmentCallbacks(env, MMMojoEnvironmentCallbackType::kMMUserData, this);
 		void (*ReadOnPush)(uint32_t request_id, const void* request_info, void* user_data) = [](uint32_t request_id, const void* request_info, void* user_data) {
+			DBG_PRINT("ReadOnPush: %u\n", request_id);
 			return ((CMojoCall*)user_data)->ReadOnPush(request_id, request_info);
-			};
+		};
 		void (*ReadOnPull)(uint32_t request_id, const void* request_info, void* user_data) = [](uint32_t request_id, const void* request_info, void* user_data) {
+			DBG_PRINT("ReadOnPull: %u\n", request_id);
 			return ((CMojoCall*)user_data)->ReadOnPull(request_id, request_info);
-			};
+		};
 		void (*ReadOnShared)(uint32_t request_id, const void* request_info, void* user_data) = [](uint32_t request_id, const void* request_info, void* user_data) {
+			DBG_PRINT("ReadOnShared: %u\n", request_id);
 			return ((CMojoCall*)user_data)->ReadOnShared(request_id, request_info);
-			};
+		};
 
 		void (*OnConnect)(bool is_connected, void* user_data) = [](bool is_connected, void* user_data) {
+			DBG_PRINT("OnConnect: %d\n", is_connected);
 			return ((CMojoCall*)user_data)->OnRemoteConnect(is_connected);
-			};
+		};
 		void (*OnDisConnect)(void* user_data) = [](void* user_data) {
+			DBG_PRINT("OnDisConnect\n");
 			return ((CMojoCall*)user_data)->OnRemoteDisConnect();
-			};
+		};
 		void (*OnProcessLaunched)(void* user_data) = [](void* user_data) {
+			DBG_PRINT("OnProcessLaunched\n");
 			return ((CMojoCall*)user_data)->OnRemoteProcessLaunched();
-			};
+		};
 		void (*OnProcessLaunchFailed)(int error_code, void* user_data) = [](int error_code, void* user_data) {
+			DBG_PRINT("OnProcessLaunchFailed: %d\n", error_code);
 			return ((CMojoCall*)user_data)->OnRemoteProcessLaunchFailed(error_code);
-			};
+		};
 		void (*OnError)(const void* errorbuf, int errorsize, void* user_data) = [](const void* errorbuf, int errorsize, void* user_data) {
+			DBG_PRINT("OnError: %.*s\n", errorsize, (const char*)errorbuf);
 			return ((CMojoCall*)user_data)->OnRemoteError(errorbuf, errorsize);
-			};
+		};
 
 		SetMMMojoEnvironmentCallbacks(env, MMMojoEnvironmentCallbackType::kMMReadPush, ReadOnPush);
 		SetMMMojoEnvironmentCallbacks(env, MMMojoEnvironmentCallbackType::kMMReadPull, ReadOnPull);
@@ -123,30 +137,39 @@ bool CMojoCall::SendPbSerializedData(const void* pb_data, size_t data_size, int
 
 void CMojoCall::OnRemoteConnect(bool is_connected)
 {
-	std::lock_guard<std::mutex> lock(m_mutex_conn);
-	m_connected = is_connected;
-	m_cv_conn.notify_all();
+	std::lock_guard<std::mutex> lock(m_mutex_state);
+	m_state = is_connected ? MJC_CONNECTED : MJC_FAILED;
+	m_cv_state.notify_all();
 }
 
 void CMojoCall::OnRemoteDisConnect() {
-	std::lock_guard<std::mutex> lock(m_mutex_conn);
-	m_connected = false;
-	m_cv_conn.notify_all();
+	std::lock_guard<std::mutex> lock(m_mutex_state);
+	m_state = MJC_FAILED;
+	m_cv_state.notify_all();
 }
 
 bool CMojoCall::wait_connection(int timeout)
 {
 	if (timeout < 0) {
-		std::unique_lock<std::mutex> lock(m_mutex_conn);
-		m_cv_conn.wait(lock, [this] {return m_connected; });
+		std::unique_lock<std::mutex> lock(m_mutex_state);
+		m_cv_state.wait(lock, [this] {return m_state != MJC_PENDING; });
 	}
 	else
 	{
-		std::unique_lock<std::mutex> lock(m_mutex_conn);
-		if (!m_cv_conn.wait_for(lock, std::chrono::milliseconds(timeout), [this] {return m_connected; }))
+		std::unique_lock<std::mutex> lock(m_mutex_state);
+		if (!m_cv_state.wait_for(lock, std::chrono::milliseconds(timeout), [this] {return m_state != MJC_PENDING; }))
 		{
 			return false;
 		}
 	}
-	return m_connected;
+	return m_state >= MJC_CONNECTED;
 }
+
+void CMojoCall::OnRemoteProcessLaunchFailed(int error_code)
+{
+	std::lock_guard<std::mutex> lock(m_mutex_state);
+	if (m_state == MJC_PENDING) {
+		m_state = MJC_FAILED;
+		m_cv_state.notify_all();
+	}
+}