binary

LunaTranslator/LunaTranslator/gui/showword.py

@@ -55,19 +55,21 @@ class AnkiWindow(QWidget):
             )
 
     @threader
-    def asyncocr(self, fname):
-        self.__ocrsettext.emit(ocr_run(fname))
+    def asyncocr(self, img):
+        self.__ocrsettext.emit(ocr_run(img))
 
     def crop(self):
         def ocroncefunction(rect):
-            img = imageCut(0, rect[0][0], rect[0][1], rect[1][0], rect[1][1])
+            img = imageCut(
+                0, rect[0][0], rect[0][1], rect[1][0], rect[1][1], False, True
+            )
             fname = "./cache/ocr/cropforanki.png"
             os.makedirs("./cache/ocr", exist_ok=True)
             img.save(fname)
             self.editpath.setText("")
             self.editpath.setText(os.path.abspath(fname))
             if globalconfig["ankiconnect"]["ocrcroped"]:
-                self.asyncocr(fname)
+                self.asyncocr(img)
 
         rangeselct_function(self, ocroncefunction, False, False)
 
@@ -378,6 +380,8 @@ class AnkiWindow(QWidget):
         return wid
 
     def wrappedpixmap(self, src):
+        if not src:
+            return
         pix = QPixmap.fromImage(QImage(src))
         rate = self.devicePixelRatioF()
         pix.setDevicePixelRatio(rate)

LunaTranslator/LunaTranslator/gui/translatorUI.py

@@ -356,10 +356,7 @@ class QUnFrameWindow(resizableframeless):
         @threader
         def ocroncefunction(rect):
             img = imageCut(0, rect[0][0], rect[0][1], rect[1][0], rect[1][1])
-            fname = "./cache/ocr/once.png"
-            os.makedirs("./cache/ocr", exist_ok=True)
-            img.save(fname)
-            text = ocr_run(fname)
+            text = ocr_run(img)
             gobject.baseobject.textgetmethod(text, False)
 
         rangeselct_function(self, ocroncefunction, False, False)

LunaTranslator/LunaTranslator/myutils/ocrutil.py

@@ -11,24 +11,31 @@ from traceback import print_exc
 import gobject, winsharedutils
 
 
+def qimage2binary(qimage: QImage):
+    byte_array = QByteArray()
+    buffer = QBuffer(byte_array)
+    buffer.open(QBuffer.WriteOnly)
+    qimage.save(buffer, "BMP")
+    buffer.close()
+    image_data = byte_array.data()
+    return image_data
+
+
+def binary2qimage(binary):
+    image = QImage()
+    image.loadFromData(binary)
+    return image
+
+
 def togray(image):
     gray_image = image.convertToFormat(QImage.Format_Grayscale8)
     return gray_image
 
 
 def otsu_threshold_fast(image: QImage, thresh):
-
-    byte_array = QByteArray()
-    buffer = QBuffer(byte_array)
-    buffer.open(QBuffer.WriteOnly)
-    image.save(buffer, "BMP")
-    buffer.close()
-    image_data = byte_array.data()
-
+    image_data = qimage2binary(image)
     solved = winsharedutils.otsu_binary(image_data, thresh)
-    image = QImage()
-    image.loadFromData(solved)
-    return image
+    return binary2qimage(solved)
 
 
 def imagesolve(image):
@@ -43,7 +50,7 @@ def imagesolve(image):
     return image2
 
 
-def imageCut(hwnd, x1, y1, x2, y2, viscompare=True):
+def imageCut(hwnd, x1, y1, x2, y2, viscompare=True, rawimage=False) -> QImage:
     screen = QApplication.primaryScreen()
     for _ in range(2):
 
@@ -81,6 +88,8 @@ def imageCut(hwnd, x1, y1, x2, y2, viscompare=True):
             )
 
     image = pix.toImage()
+    if rawimage:
+        return image
     image2 = imagesolve(image)
     if viscompare:
         gobject.baseobject.showocrimage.setimage.emit([image, image2])
@@ -101,7 +110,7 @@ def ocr_end():
     _ocrengine = None
 
 
-def ocr_run(img):
+def ocr_run(qimage: QImage):
     global _nowuseocr, _ocrengine
 
     use = None
@@ -123,7 +132,7 @@ def ocr_run(img):
             aclass = importlib.import_module("ocrengines." + use).OCR
             _ocrengine = aclass(use)
             _nowuseocr = use
-        text = _ocrengine._private_ocr(img)
+        text = _ocrengine._private_ocr(qimage2binary(qimage))
     except Exception as e:
         if isinstance(e, ArgsEmptyExc):
             msg = str(e)

LunaTranslator/LunaTranslator/ocrengines/baiduocr_X.py

@@ -36,7 +36,7 @@ class OCR(baseocr):
                 + self.secretKey
             ).json()["access_token"]
 
-    def ocr(self, imgfile):
+    def ocr(self, imagebinary):
         self.checkchange()
         if self.accstoken == "":
             return ""
@@ -57,9 +57,8 @@ class OCR(baseocr):
         }
 
         params = {"access_token": self.accstoken}  # '',
-        with open(imgfile, "rb") as ff:
-            f = ff.read()
-        b64 = base64.b64encode(f)
+        
+        b64 = base64.b64encode(imagebinary)
 
         data = {
             "image": b64,

LunaTranslator/LunaTranslator/ocrengines/baseocrclass.py

@@ -9,7 +9,7 @@ class baseocr(commonbase):
     def initocr(self):
         pass
 
-    def ocr(self, imgpath):
+    def ocr(self, imagebinary):
         raise Exception
 
     def end(self):
@@ -101,11 +101,11 @@ class baseocr(commonbase):
             raise e
         self.needinit = False
 
-    def _private_ocr(self, imgpath):
+    def _private_ocr(self, imagebinary):
         if self.needinit:
             self.level2init()
         try:
-            text = self.ocr(imgpath)
+            text = self.ocr(imagebinary)
         except Exception as e:
             self.needinit = True
             raise e

LunaTranslator/LunaTranslator/ocrengines/docsumo.py

@@ -3,7 +3,7 @@ from ocrengines.baseocrclass import baseocr
 
 class OCR(baseocr):
 
-    def ocr(self, imgfile):
+    def ocr(self, imagebinary):
         self.checkempty(["token"])
 
         headers = {
@@ -29,7 +29,7 @@ class OCR(baseocr):
             '------WebKitFormBoundaryUjYOv45hug6CFh3t\r\nContent-Disposition: form-data; name="file"; filename="screenshot.png"\r\nContent-Type: application/octet-stream\r\n\r\n'.encode(
                 "latin-1"
             )
-            + open(imgfile, "rb").read()
+            + imagebinary
             + "\r\n------WebKitFormBoundaryUjYOv45hug6CFh3t--\r\n".encode("latin-1")
         )
 

LunaTranslator/LunaTranslator/ocrengines/feishu.py

@@ -24,11 +24,9 @@ class OCR(baseocr):
             self.tokens[(app_id, app_secret)] = token
         return self.tokens[(app_id, app_secret)]
 
-    def ocr(self, imgfile):
+    def ocr(self, imagebinary):
         token = self.check()
-        with open(imgfile, "rb") as ff:
-            f = ff.read()
-        b64 = base64.b64encode(f)
+        b64 = base64.b64encode(imagebinary)
         res = self.session.post(
             "https://open.feishu.cn/open-apis/optical_char_recognition/v1/image/basic_recognize",
             headers={

LunaTranslator/LunaTranslator/ocrengines/googlecloudvision.py

@@ -4,7 +4,7 @@ import base64
 
 class OCR(baseocr):
 
-    def ocr(self, imgfile):
+    def ocr(self, imagebinary):
         # https://github.com/dmotz/thing-translator/blob/d1fec3f38d24e973af49766669f9ee00bd9e98a8/src/effects/snap.js
         # https://cloud.google.com/vision/docs/ocr?hl=zh-cn
         # https://cloud.google.com/vision/docs/reference/rest/v1/AnnotateImageResponse#EntityAnnotation
@@ -12,9 +12,7 @@ class OCR(baseocr):
         ocr_url = (
             "https://vision.googleapis.com/v1/images:annotate?key=" + self.config["key"]
         )
-        with open(imgfile, "rb") as f:
-            data = f.read()
-            encodestr = str(base64.b64encode(data), "utf-8")
+        encodestr = str(base64.b64encode(imagebinary), "utf-8")
         data = {
             "requests": [
                 {

LunaTranslator/LunaTranslator/ocrengines/local.py

@@ -1,9 +1,35 @@
 import os
 from myutils.config import globalconfig, _TR, static_data
 from ocrengines.baseocrclass import baseocr
-from ctypes import CDLL, c_char_p, create_string_buffer, c_uint32, POINTER, c_int32
+from ctypes import (
+    CDLL,
+    c_char_p,
+    create_string_buffer,
+    c_size_t,
+    c_void_p,
+    c_int32,
+    POINTER,
+    Structure,
+    pointer,
+    cast,
+    c_char_p,
+)
 import os
 import gobject
+from traceback import print_exc
+
+
+class ocrpoints(Structure):
+    _fields_ = [
+        ("x1", c_int32),
+        ("y1", c_int32),
+        ("x2", c_int32),
+        ("y2", c_int32),
+        ("x3", c_int32),
+        ("y3", c_int32),
+        ("x4", c_int32),
+        ("y4", c_int32),
+    ]
 
 
 class ocrwrapper:
@@ -13,7 +39,7 @@ class ocrwrapper:
     def _OcrInit(self, szDetModel, szRecModel, szKeyPath, szClsModel="", nThreads=4):
 
         _OcrInit = self.dll.OcrInit
-        _OcrInit.restype = POINTER(c_uint32)
+        _OcrInit.restype = c_void_p
         self.pOcrObj = _OcrInit(
             c_char_p(szDetModel.encode("utf8")),
             c_char_p(szClsModel.encode("utf8")),
@@ -22,22 +48,53 @@ class ocrwrapper:
             nThreads,
         )
 
-    def _OcrDetect(self, imgPath, imgName, angle):
+    def _OcrDetect(self, data: bytes, angle):
         _OcrDetect = self.dll.OcrDetect
-        return _OcrDetect(
-            self.pOcrObj,
-            c_char_p(imgPath.encode("utf8")),
-            c_char_p(imgName.encode("utf8")),
-            c_int32(angle),
+        _OcrDetect.argtypes = (
+            c_void_p,
+            c_void_p,
+            c_size_t,
+            c_int32,
+            POINTER(c_int32),
+            POINTER(POINTER(ocrpoints)),
+            POINTER(POINTER(c_char_p)),
         )
 
-    def _OcrGet(self):
-        _OcrGetLen = self.dll.OcrGetLen
-        _OcrGetResult = self.dll.OcrGetResult
-        length = _OcrGetLen(self.pOcrObj)
-        buff = create_string_buffer(length)
-        _OcrGetResult(self.pOcrObj, buff, length)
-        return buff.value
+        _OcrFreeptr = self.dll.OcrFreeptr
+        _OcrFreeptr.argtypes = c_int32, c_void_p, c_void_p
+
+        num = c_int32()
+        ps = POINTER(ocrpoints)()
+        chars = POINTER(c_char_p)()
+        res = _OcrDetect(
+            self.pOcrObj,
+            data,
+            len(data),
+            c_int32(angle),
+            pointer(num),
+            pointer(ps),
+            pointer(chars),
+        )
+        if not res:
+            return [], []
+        texts = []
+        pss = []
+        for i in range((num.value)):
+            texts.append(chars[i].decode("utf8"))
+            pss.append(
+                (
+                    ps[i].x1,
+                    ps[i].y1,
+                    ps[i].x2,
+                    ps[i].y2,
+                    ps[i].x3,
+                    ps[i].y3,
+                    ps[i].x4,
+                    ps[i].y4,
+                )
+            )
+        _OcrFreeptr(num, ps, chars)
+        return pss, texts
 
     def _OcrDestroy(self):
         _OcrDestroy = self.dll.OcrDestroy
@@ -46,15 +103,12 @@ class ocrwrapper:
     def init(self, det, rec, key):
         self._OcrInit(det, rec, key)
 
-    def ocr(self, path, name, angle=0):
+    def ocr(self, data, angle=0):
         try:
-            if self._OcrDetect(path, name, angle):
-                return self._OcrGet().decode("utf8")
-            else:
-                return ""
+            return self._OcrDetect(data, angle)
         except:
-
-            return ""
+            print_exc()
+            return [], []
 
     def trydestroy(self):
         try:
@@ -93,20 +147,12 @@ class OCR(baseocr):
         self._ocr.init(path + "/det.onnx", path + "/rec.onnx", path + "/dict.txt")
         self._savelang = self.srclang
 
-    def ocr(self, imgfile):
+    def ocr(self, imagebinary):
         self.checkchange()
 
-        s = self._ocr.ocr(
-            os.path.dirname(imgfile) + "/",
-            os.path.basename(imgfile),
+        pss, texts = self._ocr.ocr(
+            imagebinary,
             globalconfig["verticalocr"],
         )
 
-        ls = s.split("\n")
-        box = []
-        text = []
-        for i in range(len(ls) // 2):
-            box.append([int(_) for _ in ls[i * 2].split(",")])
-            text.append(ls[i * 2 + 1])
-
-        return self.common_solve_text_orientation(box, text)
+        return self.common_solve_text_orientation(pss, texts)

LunaTranslator/LunaTranslator/ocrengines/mangaocr.py

@@ -1,20 +1,24 @@
 import requests
 from ocrengines.baseocrclass import baseocr
-import os
+import os, uuid
+from myutils.ocrutil import binary2qimage
 
 
 class OCR(baseocr):
 
-    def ocr(self, img_path):
-
+    def ocr(self, imagebinary):
+        qimage = binary2qimage(imagebinary)
+        os.makedirs("./cache/ocr", exist_ok=True)
+        fname = "./cache/ocr/" + str(uuid.uuid4()) + ".png"
+        qimage.save(fname)
         self.checkempty(["Port"])
         self.port = self.config["Port"]
 
-        absolute_img_path = os.path.abspath(img_path)
+        absolute_img_path = os.path.abspath(fname)
         params = {"image_path": absolute_img_path}
 
         response = requests.get(f"http://127.0.0.1:{self.port}/image", params=params)
-
+        os.remove(absolute_img_path)
         try:
             return response.json()["text"]
         except Exception as e:

LunaTranslator/LunaTranslator/ocrengines/ocrspace.py

@@ -15,7 +15,7 @@ class OCR(baseocr):
             "cht": "cht",
         }
 
-    def ocr(self, imgfile):
+    def ocr(self, imagebinary):
         self.checkempty(["apikey"])
         apikey = self.config["apikey"]
         if self.config["interface"] == 1:
@@ -39,9 +39,7 @@ class OCR(baseocr):
             "user-agent": "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/105.0.0.0 Safari/537.36 Edg/105.0.1343.53",
         }
 
-        with open(imgfile, "rb") as ff:
-            f = ff.read()
-        b64 = base64.b64encode(f)
+        b64 = base64.b64encode(imagebinary)
         data = {
             "language": self.srclang,
             "base64Image": "data:image/jpeg;base64," + str(b64, encoding="utf8"),

LunaTranslator/LunaTranslator/ocrengines/tesseract5.py

@@ -1,6 +1,6 @@
-import os
+import os, uuid
 from myutils.config import _TR, ocrsetting
-
+from myutils.ocrutil import binary2qimage
 from ocrengines.baseocrclass import baseocr
 from myutils.subproc import subproc_w
 
@@ -19,11 +19,16 @@ class OCR(baseocr):
     def initocr(self):
         self.langs = list_langs()
 
-    def ocr(self, imgfile):
+    def ocr(self, imagebinary):
         self.checkempty(["路径"])
         path = self.config["路径"]
         if os.path.exists(path) == False:
             raise Exception(_TR("路径不存在"))
+        qimage = binary2qimage(imagebinary)
+        os.makedirs("./cache/ocr", exist_ok=True)
+        fname = "./cache/ocr/" + str(uuid.uuid4()) + ".png"
+        qimage.save(fname)
+        imgfile = os.path.abspath(fname)
         _ = subproc_w(
             '"{}" "{}" - -l {} {}'.format(
                 path, imgfile, self.langs[self.config["语言"]], self.config["附加参数"]
@@ -32,6 +37,7 @@ class OCR(baseocr):
             encoding="utf8",
             run=True,
         )
+        os.remove(imgfile)
         res = _.stdout
         err = _.stderr
         if len(err):

LunaTranslator/LunaTranslator/ocrengines/txocr.py

@@ -19,12 +19,10 @@ class OCR(baseocr):
             "th": "tha",
         }
 
-    def ocr(self, imgfile):
+    def ocr(self, imagebinary):
         self.checkempty(["SecretId", "SecretKey"])
 
-        with open(imgfile, "rb") as f:
-            data = f.read()
-            encodestr = str(base64.b64encode(data), "utf-8")
+        encodestr = str(base64.b64encode(imagebinary), "utf-8")
         req_para = {
             "LanguageType": self.srclang,
             "Action": "GeneralBasicOCR",

LunaTranslator/LunaTranslator/ocrengines/txocrtrans.py

@@ -8,12 +8,10 @@ class OCR(baseocr):
         # https://cloud.tencent.com/document/product/551/17232
         return {"cht": "zh-TW"}
 
-    def ocr(self, imgfile):
+    def ocr(self, imagebinary):
         self.checkempty(["SecretId", "SecretKey"])
 
-        with open(imgfile, "rb") as f:
-            data = f.read()
-            encodestr = str(base64.b64encode(data), "utf-8")
+        encodestr = str(base64.b64encode(imagebinary), "utf-8")
         req_para = {
             "Source": self.srclang,
             "Target": self.tgtlang,

LunaTranslator/LunaTranslator/ocrengines/volcengine.py

@@ -1740,7 +1740,7 @@ from ocrengines.baseocrclass import baseocr
 
 class OCR(baseocr):
 
-    def ocr(self, imgfile):
+    def ocr(self, imagebinary):
         visual_service = VisualService()
         self.checkempty(["Access Key ID", "Secret Access Key"])
         # call below method if you dont set ak and sk in $HOME/.volc/config
@@ -1753,11 +1753,8 @@ class OCR(baseocr):
         # if you cannot find the needed one, please check other example files in the same dir
         # or contact us for further help
         form = dict()
-        import base64
 
-        with open(imgfile, "rb") as ff:
-            f = ff.read()
-        b64 = base64.b64encode(f)
+        b64 = base64.b64encode(imagebinary)
         form["image_base64"] = b64
         resp = visual_service.ocr_api("MultiLanguageOCR", form, self.proxy)
         try:

LunaTranslator/LunaTranslator/ocrengines/windowsocr.py

@@ -25,7 +25,7 @@ class OCR(baseocr):
             v = self.supportmap.pop("zh-Hant")
             self.supportmap["cht"] = v
 
-    def ocr(self, imgfile):
+    def ocr(self, imagebinary):
         if self.srclang not in self.supportmap:
             idx = static_data["language_list_translator_inner"].index(self.srclang)
             raise Exception(
@@ -39,9 +39,7 @@ class OCR(baseocr):
         else:
             space = " "
 
-        ret = winrtutils.OCR_f(
-            os.path.abspath(imgfile), self.supportmap[self.srclang], space
-        )
+        ret = winrtutils.OCR_f(imagebinary, self.supportmap[self.srclang], space)
         boxs = [_[1:] for _ in ret]
         texts = [_[0] for _ in ret]
 

LunaTranslator/LunaTranslator/ocrengines/xunfei.py

@@ -11,7 +11,7 @@ import json
 
 
 class OCR(baseocr):
-    def ocr(self, imgfile):
+    def ocr(self, imagebinary):
         self.checkempty(["APPId", "APISecret", "APIKey"])
 
         APPId = self.config["APPId"]
@@ -50,9 +50,8 @@ class OCR(baseocr):
                 u = Url(host, path, schema)
                 return u
 
-            def get_body(self, file_path):
-                file = open(file_path, "rb")
-                buf = file.read()
+            def get_body(self, imagebinary):
+                buf = imagebinary
                 body = {
                     "header": {"app_id": self.appid, "status": 3},
                     "parameter": {
@@ -120,7 +119,7 @@ class OCR(baseocr):
         }
         # print("request_url:", request_url)
 
-        body = printed_word_recognition.get_body(file_path=imgfile)
+        body = printed_word_recognition.get_body(file_path=imagebinary)
         response = self.session.post(
             request_url, data=json.dumps(body), headers=headers
         )

LunaTranslator/LunaTranslator/ocrengines/youdaocr.py

@@ -10,7 +10,7 @@ class OCR(baseocr):
     def langmap(self):
         return {"zh": "zh-CHS", "cht": "zh-CHT"}
 
-    def freetest(self, imgfile):
+    def freetest(self, imagebinary):
         headers = {
             "authority": "aidemo.youdao.com",
             "accept": "*/*",
@@ -26,9 +26,7 @@ class OCR(baseocr):
             "sec-fetch-site": "same-site",
             "user-agent": "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/106.0.0.0 Safari/537.36",
         }
-        with open(imgfile, "rb") as ff:
-            f = ff.read()
-        b64 = base64.b64encode(f)
+        b64 = base64.b64encode(imagebinary)
         data = {
             "imgBase": "data:image/jpeg;base64," + str(b64, encoding="utf8"),
             "lang": "",
@@ -50,7 +48,7 @@ class OCR(baseocr):
         except:
             raise Exception(response.text)
 
-    def ocrapi(self, imgfile):
+    def ocrapi(self, imagebinary):
         def truncate(q):
             if q is None:
                 return None
@@ -65,9 +63,7 @@ class OCR(baseocr):
         self.checkempty(["APP_KEY", "APP_SECRET"])
         APP_KEY, APP_SECRET = self.config["APP_KEY"], self.config["APP_SECRET"]
         YOUDAO_URL = "https://openapi.youdao.com/ocrapi"
-        file = open(imgfile, "rb")
-        content = base64.b64encode(file.read()).decode("utf-8")
-        file.close()
+        content = base64.b64encode(imagebinary).decode("utf-8")
 
         data = {}
         data["img"] = content
@@ -99,10 +95,10 @@ class OCR(baseocr):
         except:
             raise Exception(response.text)
 
-    def ocr(self, imgfile):
+    def ocr(self, imagebinary):
         interfacetype = self.config["接口"]
         if interfacetype == 0:
-            return self.freetest(imgfile)
+            return self.freetest(imagebinary)
         elif interfacetype == 1:
-            return self.ocrapi(imgfile)
+            return self.ocrapi(imagebinary)
         raise Exception("unknown")

LunaTranslator/LunaTranslator/ocrengines/youdaocrtrans.py

@@ -9,7 +9,7 @@ class OCR(baseocr):
     def langmap(self):
         return {"zh": "zh-CHS", "cht": "zh-CHT"}
 
-    def freetest(self, imgfile):
+    def freetest(self, imagebinary):
 
         headers = {
             "authority": "aidemo.youdao.com",
@@ -26,9 +26,7 @@ class OCR(baseocr):
             "sec-fetch-site": "same-site",
             "user-agent": "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/106.0.0.0 Safari/537.36",
         }
-        with open(imgfile, "rb") as ff:
-            f = ff.read()
-        b64 = base64.b64encode(f)
+        b64 = base64.b64encode(imagebinary)
         data = {
             "imgBase": "data:image/jpeg;base64," + str(b64, encoding="utf8"),
             "lang": "",
@@ -46,13 +44,11 @@ class OCR(baseocr):
         except:
             raise Exception(response.text)
 
-    def ocrapi(self, imgfile):
+    def ocrapi(self, imagebinary):
 
         self.checkempty(["APP_KEY", "APP_SECRET"])
         APP_KEY, APP_SECRET = self.config["APP_KEY"], self.config["APP_SECRET"]
-
-        # 待翻译图片路径, 例windows路径：PATH = "C:\\youdao\\media.jpg"
-        PATH = imgfile
+ 
 
         """
         添加鉴权相关参数 -
@@ -121,7 +117,7 @@ class OCR(baseocr):
             type = "1"
 
             # 数据的base64编码
-            q = readFileAsBase64(PATH)
+            q = readFileAsBase64(imagebinary)
             data = {
                 "q": q,
                 "from": lang_from,
@@ -142,10 +138,8 @@ class OCR(baseocr):
             elif "post" == method:
                 return self.session.post(url, params, header)
 
-        def readFileAsBase64(path):
-            f = open(path, "rb")
-            data = f.read()
-            return str(base64.b64encode(data), "utf-8")
+        def readFileAsBase64(imagebinary):
+            return str(base64.b64encode(imagebinary), "utf-8")
 
         self.countnum()
 
@@ -161,10 +155,10 @@ class OCR(baseocr):
         except:
             raise Exception(response.text)
 
-    def ocr(self, imgfile):
+    def ocr(self, imagebinary):
         interfacetype = self.config["接口"]
         if interfacetype == 0:
-            return self.freetest(imgfile)
+            return self.freetest(imagebinary)
         elif interfacetype == 1:
-            return self.ocrapi(imgfile)
+            return self.ocrapi(imagebinary)
         raise Exception("unknown")

LunaTranslator/LunaTranslator/textsource/ocrtext.py

@@ -2,7 +2,7 @@ import time
 from myutils.config import globalconfig
 import winsharedutils
 from gui.rangeselect import rangeadjust
-from myutils.ocrutil import imageCut, ocr_run, ocr_end
+from myutils.ocrutil import imageCut, ocr_run, ocr_end,qimage2binary
 import time, gobject, os
 from PyQt5.QtWidgets import QApplication
 from PyQt5.QtGui import QImage
@@ -131,7 +131,7 @@ class ocrtext(basetext):
                     ok = False
             if ok == False:
                 continue
-            text = self.ocrtest(imgr)
+            text = ocr_run(imgr)
             self.lastocrtime[i] = time.time()
 
             if self.savelasttext[i] is not None:
@@ -154,7 +154,7 @@ class ocrtext(basetext):
                 return
             img = imageCut(self.hwnd, rect[0][0], rect[0][1], rect[1][0], rect[1][1])
 
-            text = self.ocrtest(img)
+            text = ocr_run(img)
             imgr1 = qimge2np(img)
             self.savelastimg[i] = imgr1
             self.savelastrecimg[i] = imgr1
@@ -163,14 +163,6 @@ class ocrtext(basetext):
             __text.append(text)
         return "\n".join(__text)
 
-    def ocrtest(self, img):
-        os.makedirs("./cache/ocr", exist_ok=True)
-        fname = "./cache/ocr/{}.png".format(self.timestamp)
-        img.save(fname)
-        # print(fname)
-        text = ocr_run(fname)
-        # print(text)
-        return text
 
     def end(self):
         globalconfig["ocrregions"] = [_.getrect() for _ in self.range_ui]

LunaTranslator/LunaTranslator/winrtutils.py

@@ -5,6 +5,7 @@ from ctypes import (
     c_wchar_p,
     pointer,
     CDLL,
+    c_size_t,
     Structure,
     c_void_p,
 )
@@ -30,7 +31,7 @@ if winrtutilsdll:
         ]
 
     _OCR_f = winrtutilsdll.OCR
-    _OCR_f.argtypes = c_wchar_p, c_wchar_p, c_wchar_p, POINTER(c_uint)
+    _OCR_f.argtypes = c_void_p, c_size_t, c_wchar_p, c_wchar_p, POINTER(c_uint)
     _OCR_f.restype = ocrres
     _freeocrres = winrtutilsdll.freeocrres
     _freeocrres.argtypes = ocrres, c_uint
@@ -54,9 +55,9 @@ if winrtutilsdll:
         _freewstringlist(ret, num.value)
         return _allsupport
 
-    def OCR_f(imgpath, lang, space):
+    def OCR_f(data, lang, space):
         num = c_uint()
-        ret = _OCR_f(imgpath, lang, space, pointer(num))
+        ret = _OCR_f(data, len(data), lang, space, pointer(num))
         res = []
         for i in range(num.value):
             res.append((ret.lines[i], ret.xs[i], ret.ys[i], ret.xs2[i], ret.ys2[i]))

plugins/CMakeLists.txt

@@ -29,7 +29,7 @@ include(generate_product_version)
 
 set(VERSION_MAJOR 2)
 set(VERSION_MINOR 51)
-set(VERSION_PATCH 1)
+set(VERSION_PATCH 2)
 
 add_library(pch pch.cpp)
 target_precompile_headers(pch PUBLIC pch.h)

plugins/winrtutils/define.h

@@ -13,7 +13,7 @@ extern "C"
 
     __declspec(dllexport) bool check_language_valid(wchar_t *);
     __declspec(dllexport) wchar_t **getlanguagelist(int *);
-    __declspec(dllexport) ocrres OCR(wchar_t *fname, wchar_t *lang, wchar_t *, int *);
+    __declspec(dllexport) ocrres OCR(void* ptr, size_t size, wchar_t *lang, wchar_t *, int *);
 
     __declspec(dllexport) void freewstringlist(wchar_t **, int);
     __declspec(dllexport) void freeocrres(ocrres, int);

plugins/winrtutils/winrtocr.cpp

@@ -57,13 +57,13 @@ wchar_t **getlanguagelist(int *num)
     *num = languages.Size();
     return ret;
 }
-ocrres OCR(wchar_t *fname, wchar_t *lang, wchar_t *space, int *num)
+ocrres OCR(void *ptr, size_t size, wchar_t *lang, wchar_t *space, int *num)
 {
-    std::wstring imagePath = fname;
-
-    StorageFile imageFile = StorageFile::GetFileFromPathAsync(imagePath).get();
-    IRandomAccessStream imageStream = imageFile.OpenAsync(FileAccessMode::Read).get();
-    BitmapDecoder decoder = BitmapDecoder::CreateAsync(imageStream).get();
+    IBuffer buffer = CryptographicBuffer::CreateFromByteArray(
+        winrt::array_view<uint8_t>(static_cast<uint8_t *>(ptr), size));
+    InMemoryRandomAccessStream memoryStream;
+    memoryStream.WriteAsync(buffer).get();
+    BitmapDecoder decoder = BitmapDecoder::CreateAsync(memoryStream).get();
 
     SoftwareBitmap softwareBitmap = decoder.GetSoftwareBitmapAsync().get();
     std::wstring l = lang;