Merge remote-tracking branch 'origin/master' into data_key_collect

command/Log.bat

@@ -34,7 +34,6 @@ echo cnocr: %log_cnocrVer%
 echo gooey: %log_gooeyVer%
 echo colored: %log_coloredVer%
 echo Python Levenshtein: %log_LevenshteinVer%
-echo pyocr: %log_pyocrVer%) >> %deploy_log_file%
 call :Log_common
 goto :eof
 
@@ -57,7 +56,6 @@ echo mxnet: %log_mxnetVer%
 echo cnocr: %log_cnocrVer%
 echo gooey: %log_gooeyVer%
 echo colored: %log_coloredVer%
-echo pyocr: %log_pyocrVer%) >> %deploy_log_file%
 call :Log_common
 goto :eof
 
@@ -74,8 +72,6 @@ for /f "tokens=2 delims=-" %%i in ('dir /b "pip*.dist-info"') do ( set "log_pipV
 set "log_pipVer=%log_pipVer:.dist=%"
 for /f "tokens=2 delims=-" %%i in ('dir /b "adbutils*.dist-info"') do ( set "log_adbutilsVer=%%i" )
 set "log_adbutilsVer=%log_adbutilsVer:.dist=%"
-for /f "tokens=2 delims=-" %%i in ('dir /b "pyocr*.dist-info"') do ( set "log_pyocrVer=%%i" )
-set "log_pyocrVer=%log_pyocrVer:.dist=%"
 for /f "tokens=2 delims=-" %%i in ('dir /b "numpy*.dist-info"') do ( set "log_numpyVer=%%i" )
 set "log_numpyVer=%log_numpyVer:.dist=%"
 for /f "tokens=2 delims=-" %%i in ('dir /b "scipy*.dist-info"') do ( set "log_scipyVer=%%i" )

module/ocr/ocr.py

@@ -9,9 +9,6 @@ from module.exception import ScriptError
 from module.logger import logger
 from module.ocr.al_ocr import AlOcr
 
-if server.server == 'jp':
-    import pyocr
-
 OCR_MODEL = {
     # Folder: ./bin/cnocr_models/azur_lane
     # Size: 3.25MB
@@ -32,6 +29,8 @@ OCR_MODEL = {
     # Charset: Number, English character, Chinese character, symbols, <space>
     # _num_classes: 6426
     'cnocr': AlOcr(model_name='densenet-lite-gru', model_epoch=39, root='./bin/cnocr_models/cnocr', name='cnocr'),
+
+    'jp': AlOcr(model_name='densenet-lite-gru', model_epoch=125, root='./bin/cnocr_models/jp', name='jp'),
 }
 
 
@@ -102,72 +101,6 @@ class Ocr:
         return result_list
 
 
-class OcrJapanese:
-    def __init__(self, buttons, letter=(255, 255, 255), threshold=128, name=None):
-        """
-        Args:
-            buttons (Button, tuple, list[Button], list[tuple]): OCR area.
-            letter (tuple(int)): Letter RGB.
-            threshold (int):
-            name (str):
-        """
-        self.name = str(buttons) if isinstance(buttons, Button) else name
-        self.buttons = buttons if isinstance(buttons, list) else [buttons]
-        self.buttons = [button.area if isinstance(
-            button, Button) else button for button in self.buttons]
-        self.letter = letter
-        self.threshold = threshold
-        self.ocr_tool = self.find_tesseract_tool()
-
-    def find_tesseract_tool(self):
-        """
-        Returns:
-            pyocr tesseract wrapper module
-        """
-        ocr_tool = pyocr.get_available_tools()
-        if len(ocr_tool) == 0:
-            raise ScriptError(
-                'No ocr-tool found, please install tesseract by yourself and make sure to set correct env vars.')
-        ocr_tool = ocr_tool[0]
-        ocr_langs = ocr_tool.get_available_languages()
-        if 'jpn' not in ocr_langs:
-            raise ScriptError(
-                'No jpn found in tesseract langs, please install japanese data files.')
-        return ocr_tool
-
-    def pre_process(self, image):
-        """
-        Args:
-            image (np.ndarray): Shape (height, width, channel)
-
-        Returns:
-            PIL.Image
-        """
-        image = extract_letters(
-            image, letter=self.letter, threshold=self.threshold)
-
-        return Image.fromarray(image)
-
-    def ocr(self, image):
-        start_time = time.time()
-
-        image_list = [self.pre_process(np.array(image.crop(area)))
-                      for area in self.buttons]
-
-        # This will show the images feed to OCR model
-        # self.cnocr.debug(image_list)
-
-        result_list = [self.ocr_tool.image_to_string(
-            img, lang='jpn') for img in image_list]
-
-        if len(self.buttons) == 1:
-            result_list = result_list[0]
-        logger.attr(name='%s %ss' % (self.name, float2str(time.time() - start_time)),
-                    text=str(result_list))
-
-        return result_list
-
-
 class Digit(Ocr):
     def __init__(self, buttons, lang='azur_lane', letter=(255, 255, 255), threshold=128, alphabet='0123456789',
                  name=None):

module/reward/commission.py

@@ -13,7 +13,7 @@ from module.base.utils import color_similar_1d, random_rectangle_point
 from module.exception import GameStuckError
 from module.handler.info_handler import InfoHandler
 from module.logger import logger
-from module.ocr.ocr import Ocr, OcrJapanese
+from module.ocr.ocr import Ocr
 from module.reward.assets import *
 from module.ui.page import page_reward, page_commission, CAMPAIGN_CHECK
 from module.ui.switch import Switch
@@ -53,19 +53,19 @@ dictionary_en = {
 }
 dictionary_jp = {
     'major_comm': ['初級自主訓練', '中級自主訓練', '上級自主訓練', '初級対抗演習', '中級対抗演習', '上級対抗演習', '初級科学研究', '中級科学研究', '上級科学研究', '初級資材整理', '中級資材整理', '上級資材整理', '初級戦術課程', '中級戦術課程', '上級戦術課程', '初級貨物輸送', '中級貨物輸送', '上級貨物輸送'],
-    'daily_comm': ['日常資源開発Ⅰ', '日常資源開発Ⅱ', '日常資源開発Ⅲ', '日常資源開発Ⅳ', '日常資源開発Ⅴ', '日常資源開発Ⅵ', '覚醒実証研究Ⅰ', '覚醒実証研究Ⅱ'],
+    'daily_comm': ['日常資源開発', '覚醒実証研究'],
     'extra_drill': ['短距離練習航海', '中距離練習航海', '外洋練習航海', '近海防衛巡回', '前線基地防衛巡回', '海域浮標保守作業'],
     'extra_part': ['初級木材輸送護衛', '中級木材輸送護衛', '上級木材輸送護衛', '初級鉄鋼輸送護衛', '中級鉄鋼輸送護衛', '上級鉄鋼輸送護衛'],
     'extra_cube': ['船団護衛演習', '艦隊輸送演習', '艦隊実弾演習', '装備慣熟演習', '艦隊慣熟演習', '艦隊運動演習'],
-    'extra_oil': ['小型油田開発Ⅰ', '小型油田開発Ⅱ', '小型油田開発Ⅲ', '中型油田開発Ⅰ', '中型油田開発Ⅱ', '中型油田開発Ⅲ', '大型油田開発Ⅰ', '大型油田開発Ⅱ', '大型油田開発Ⅲ'],
+    'extra_oil': ['小型油田開発', '中型油田開発', '大型油田開発'],
     'extra_book': ['小型船団護衛', '中型船団護衛', '大型船団護衛'],
-    'urgent_drill': ['敵偵察部隊迎撃', '敵主力艦隊撃破', '敵精鋭部隊撃破', '輸送部隊護衛Ⅰ', '輸送部隊護衛Ⅱ', '輸送部隊護衛Ⅲ'],
+    'urgent_drill': ['敵偵察部隊迎撃', '敵主力艦隊撃破', '敵精鋭部隊撃破', '輸送部隊護衛'],
     'urgent_part': ['近海掃海任務', '近海航行展示', '離島火力支援', '離島兵員輸送', '外敵生態調査', '兵站航路確保'],
-    'urgent_book': ['離島物資輸送', '近海パトロール', '離島漸減支援', '外的動静哨戒', '前線部隊支援', '外敵中枢偵察'],
+    'urgent_book': ['離島物資輸送', '近海パトロール', '離島漸減支援', '外敵動静哨戒', '前線部隊支援', '外敵中枢偵察'],
-    'urgent_box': ['BIW装備輸送', 'NYB装備輸送', 'BIW物資交換', 'NYB物資交換', 'BIW装備試験', 'NYB装備試験'],
+    'urgent_box': ['装備輸送', '物資交換', '装備試験'],
-    'urgent_cube': ['船団救出Ⅰ', '船団救出Ⅱ', '船団救出Ⅲ', '敵襲Ⅰ', '敵襲Ⅱ', '敵襲Ⅲ'],
+    'urgent_cube': ['船団救出', '敵襲'],
-    'urgent_gem': ['BIW要人護衛', 'NYB要人護衛', 'BIW休暇護衛', 'NYB休暇護衛'],
+    'urgent_gem': ['要人護衛', '休暇護衛'],
-    'urgent_ship': ['小型観覧式', '連合艦隊観覧式', '多国連合観覧式']
+    'urgent_ship': ['小型観艦式', '連合艦隊観艦式', '多国連合観艦式']
 }
 COMMISSION_SWITCH = Switch('Commission_switch', is_selector=True)
 COMMISSION_SWITCH.add_status('daily', COMMISSION_DAILY)
@@ -132,9 +132,9 @@ class Commission:
         # Name
         area = area_offset((176, 23, 420, 53), self.area[0:2])
         button = Button(area=area, color=(), button=area, name='COMMISSION')
-        ocr_jpn = OcrJapanese(button)
+        ocr = Ocr(button, lang='jp')
         self.button = button
-        self.name = ocr_jpn.ocr(self.image)
+        self.name = ocr.ocr(self.image)
         self.genre = self.commission_name_parse(self.name)
 
         # Duration time
@@ -281,7 +281,7 @@ class Commission:
         """
         min_key = ''
         min_distance = 100
-        string = re.sub(' ', '', string)
+        string = re.sub(r'[\x00-\x7F]', '', string)
         for key, value in dictionary_jp.items():
             for keyword in value:
                 distance = Levenshtein.distance(keyword, string)

requirements.txt

@@ -12,5 +12,4 @@ mxnet==1.6.0
 cnocr==1.2.2
 gooey
 colored
-python-Levenshtein
+python-Levenshtein
-pyocr