GameSentenceMiner 2.9.23__py3-none-any.whl → 2.9.25__py3-none-any.whl

GameSentenceMiner/ocr/gsm_ocr_config.py

@@ -5,6 +5,8 @@ from math import floor, ceil
 from dataclasses_json import dataclass_json
 from typing import List, Optional, Union
 
+from GameSentenceMiner.util.configuration import logger
+
 
 @dataclass_json
 @dataclass
@@ -51,14 +53,14 @@ class OCRConfig:
             import pygetwindow as gw
             try:
                 set_dpi_awareness()
-                window = gw.getWindowsWithTitle(self.window)[0]
+                window = get_window(self.window)
                 self.window_geometry = WindowGeometry(
                     left=window.left,
                     top=window.top,
                     width=window.width,
                     height=window.height,
                 )
-                print(f"Window '{self.window}' found with geometry: {self.window_geometry}")
+                logger.info(f"Window '{self.window}' found with geometry: {self.window_geometry}")
             except IndexError:
                 raise ValueError(f"Window with title '{self.window}' not found.")
             for rectangle in self.rectangles:
@@ -69,6 +71,27 @@ class OCRConfig:
                     ceil(rectangle.coordinates[3] * self.window_geometry.height),
                 ]
 
+def get_window(title):
+    import pygetwindow as gw
+    windows = gw.getWindowsWithTitle(title)
+    if not windows:
+        raise ValueError(f"No windows found with title '{title}'.")
+    ret = None
+    if windows:
+        for window in windows:
+            if "cmd.exe" in window.title.lower():
+                logger.info(f"Skipping cmd.exe window with title: {window.title}")
+                continue
+            if len(windows) > 1:
+                logger.info(
+                    f"Warning: More than 1 window with title, Window Title: {window.title}, Geometry: {window.left}, {window.top}, {window.width}, {window.height}")
+            if window.title == title:
+                if window.isMinimized or not window.visible:
+                    logger.info(f"Warning: Window '{title}' is minimized. Attempting to restore it.")
+                    window.restore()
+                return window
+    return ret
+
 # try w10+, fall back to w8.1+
 def set_dpi_awareness():
     per_monitor_awareness = 2

GameSentenceMiner/ocr/owocr_area_selector.py

@@ -8,7 +8,7 @@ import mss
 from PIL import Image, ImageTk, ImageDraw
 
 from GameSentenceMiner import obs  # Import your actual obs module
-from GameSentenceMiner.ocr.gsm_ocr_config import set_dpi_awareness
+from GameSentenceMiner.ocr.gsm_ocr_config import set_dpi_awareness, get_window
 from GameSentenceMiner.util.gsm_utils import sanitize_filename
 
 try:
@@ -88,13 +88,7 @@ class ScreenSelector:
         if not self.window_name:
             return None
         try:
-            windows = gw.getWindowsWithTitle(self.window_name)
-            if windows:
-                if len(windows) > 1:
-                    print(f"Warning: Multiple windows found with title '{self.window_name}'. Using the first one.")
-                return windows[0]
-            else:
-                return None
+            return get_window(self.window_name)
         except Exception as e:
             print(f"Error finding window '{self.window_name}': {e}")
             return None
@@ -269,7 +263,7 @@ class ScreenSelector:
         """Saves rectangles to the config file, using percentages if window is targeted."""
         # Use the window geometry found during __init__ for consistency during save
         window_geom_to_save = self.target_window_geometry
-        save_coord_system = COORD_SYSTEM_ABSOLUTE  # Default if no window
+        save_coord_system = COORD_SYSTEM_PERCENTAGE  # Default if no window
 
         config_path = self.get_scene_ocr_config()
         print(f"Saving rectangles to: {config_path}")
@@ -334,7 +328,7 @@ class ScreenSelector:
                 serializable_rects.append(rect_data)
 
             # Prepare final data structure for JSON
-            if len(rectangles) == 0:
+            if not self.rectangles or len(self.rectangles) == 0:
                 save_coord_system = COORD_SYSTEM_PERCENTAGE
             save_data = {
                 "scene": obs.get_current_scene() or "default_scene",

GameSentenceMiner/ocr/owocr_helper.py

@@ -19,7 +19,7 @@ from GameSentenceMiner.ocr.ss_picker import ScreenCropper
 from GameSentenceMiner.owocr.owocr.run import TextFiltering
 from GameSentenceMiner.util.configuration import get_config, get_app_directory, get_temporary_directory
 from GameSentenceMiner.util.electron_config import get_ocr_scan_rate, get_requires_open_window
-from GameSentenceMiner.ocr.gsm_ocr_config import OCRConfig, set_dpi_awareness
+from GameSentenceMiner.ocr.gsm_ocr_config import OCRConfig, set_dpi_awareness, get_window
 from GameSentenceMiner.owocr.owocr import screen_coordinate_picker, run
 from GameSentenceMiner.util.gsm_utils import sanitize_filename, do_text_replacements, OCR_REPLACEMENTS_FILE
 
@@ -238,7 +238,7 @@ def save_result_image(img):
 async def send_result(text, time):
     if text:
         text = do_text_replacements(text, OCR_REPLACEMENTS_FILE)
-        if get_config().advanced.ocr_sends_to_clipboard or ssonly:
+        if get_config().advanced.ocr_sends_to_clipboard or manual:
             import pyperclip
             pyperclip.copy(text)
         try:
@@ -334,16 +334,13 @@ def run_oneocr(ocr_config: OCRConfig, rectangles):
     global done
     print("Running OneOCR")
     screen_area = None
-    screen_areas = []
-    exclusions = []
-    if not ssonly:
-        screen_areas = [",".join(str(c) for c in rect_config.coordinates) for rect_config in rectangles if not rect_config.is_excluded]
-        exclusions = list(rect.coordinates for rect in list(filter(lambda x: x.is_excluded, rectangles)))
+    screen_areas = [",".join(str(c) for c in rect_config.coordinates) for rect_config in rectangles if not rect_config.is_excluded]
+    exclusions = list(rect.coordinates for rect in list(filter(lambda x: x.is_excluded, rectangles)))
 
     run.init_config(False)
     try:
-        run.run(read_from="screencapture" if not ssonly else "clipboard" if ss_clipboard else "",
-                read_from_secondary="clipboard" if ss_clipboard and not ssonly else None,
+        run.run(read_from="screencapture",
+                read_from_secondary="clipboard" if ss_clipboard else None,
                 write_to="callback",
                 screen_capture_area=screen_area,
                 # screen_capture_monitor=monitor_config['index'],
@@ -358,7 +355,8 @@ def run_oneocr(ocr_config: OCRConfig, rectangles):
                 ocr2=ocr2,
                 gsm_ocr_config=ocr_config,
                 screen_capture_areas=screen_areas,
-                furigana_filter_sensitivity=furigana_filter_sensitivity)
+                furigana_filter_sensitivity=furigana_filter_sensitivity,
+                screen_capture_combo=manual_ocr_hotkey if manual_ocr_hotkey and manual else None)
     except Exception as e:
         logger.exception(f"Error running OneOCR: {e}")
     done = True
@@ -368,13 +366,12 @@ def run_oneocr(ocr_config: OCRConfig, rectangles):
 def add_ss_hotkey(ss_hotkey="ctrl+shift+g"):
     import keyboard
     cropper = ScreenCropper()
-    filtering = TextFiltering()
+    filtering = TextFiltering(lang=language)
     def capture():
         print("Taking screenshot...")
         img = cropper.run()
         do_second_ocr("", datetime.now(), img, filtering)
     try:
-        raise Exception("Forcing keyboard hotkey setup, this is a test.")
         keyboard.add_hotkey(ss_hotkey, capture)
         print(f"Press {ss_hotkey} to take a screenshot.")
     except Exception as e:
@@ -391,21 +388,6 @@ def add_ss_hotkey(ss_hotkey="ctrl+shift+g"):
         except Exception as e:
             logger.error(f"Error setting up screenshot hotkey with pynput, Screenshot Hotkey Will not work: {e}")
 
-
-def get_window(window_name):
-    import pygetwindow as gw
-    try:
-        windows = gw.getWindowsWithTitle(window_name)
-        if windows:
-            if len(windows) > 1:
-                print(f"Warning: Multiple windows found with title '{window_name}'. Using the first one.")
-            return windows[0]
-        else:
-            return None
-    except Exception as e:
-        print(f"Error finding window '{window_name}': {e}")
-        return None
-
 def set_force_stable_hotkey():
     import keyboard
     global force_stable
@@ -420,7 +402,7 @@ def set_force_stable_hotkey():
     print("Press Ctrl+Shift+F to toggle force stable mode.")
 
 if __name__ == "__main__":
-    global ocr1, ocr2, twopassocr, language, ss_clipboard, ss, ocr_config, furigana_filter_sensitivity
+    global ocr1, ocr2, twopassocr, language, ss_clipboard, ss, ocr_config, furigana_filter_sensitivity, area_select_ocr_hotkey
     import sys
 
     import argparse
@@ -430,11 +412,12 @@ if __name__ == "__main__":
     parser.add_argument("--ocr1", type=str, default="oneocr", help="Primary OCR engine (default: oneocr)")
     parser.add_argument("--ocr2", type=str, default="glens", help="Secondary OCR engine (default: glens)")
     parser.add_argument("--twopassocr", type=int, choices=[0, 1], default=1, help="Enable two-pass OCR (default: 1)")
-    parser.add_argument("--ssonly", action="store_true", help="Use screenshot-only mode")
+    parser.add_argument("--manual", action="store_true", help="Use screenshot-only mode")
     parser.add_argument("--clipboard", action="store_true", help="Use clipboard for input")
     parser.add_argument("--window", type=str, help="Specify the window name for OCR")
     parser.add_argument("--furigana_filter_sensitivity", type=float, default=0, help="Furigana Filter Sensitivity for OCR (default: 0)")
-    parser.add_argument("--manual_ocr_hotkey", type=str, default="ctrl+shift+g", help="Hotkey for manual OCR (default: ctrl+shift+g)")
+    parser.add_argument("--manual_ocr_hotkey", type=str, default=None, help="Hotkey for manual OCR (default: None)")
+    parser.add_argument("--area_select_ocr_hotkey", type=str, default="ctrl+shift+o", help="Hotkey for area selection OCR (default: ctrl+shift+o)")
 
     args = parser.parse_args()
 
@@ -442,36 +425,35 @@ if __name__ == "__main__":
     ocr1 = args.ocr1
     ocr2 = args.ocr2 if args.ocr2 else None
     twopassocr = bool(args.twopassocr)
-    ssonly = args.ssonly
+    manual = args.manual
     ss_clipboard = args.clipboard
     window_name = args.window
     furigana_filter_sensitivity = args.furigana_filter_sensitivity
-    ss_hotkey = args.manual_ocr_hotkey.lower()
+    ss_hotkey = args.area_select_ocr_hotkey.lower()
+    manual_ocr_hotkey = args.manual_ocr_hotkey.lower().replace("ctrl", "<ctrl>").replace("shift", "<shift>").replace("alt", "<alt>") if args.manual_ocr_hotkey else None
 
     logger.info(f"Received arguments: {vars(args)}")
     # set_force_stable_hotkey()
-    ocr_config = None
-    if not ssonly:
-        ocr_config: OCRConfig = get_ocr_config(window=window_name)
-        if ocr_config:
-            if ocr_config.window:
-                start_time = time.time()
-                while time.time() - start_time < 30:
-                    if get_window(ocr_config.window):
-                        break
-                    logger.info(f"Window: {ocr_config.window} Could not be found, retrying in 1 second...")
-                    time.sleep(1)
-                else:
-                    logger.error(f"Window '{ocr_config.window}' not found within 30 seconds.")
-                    sys.exit(1)
-            logger.info(f"Starting OCR with configuration: Window: {ocr_config.window}, Rectangles: {ocr_config.rectangles}, Engine 1: {ocr1}, Engine 2: {ocr2}, Two-pass OCR: {twopassocr}")
+    ocr_config: OCRConfig = get_ocr_config(window=window_name)
+    if ocr_config:
+        if ocr_config.window:
+            start_time = time.time()
+            while time.time() - start_time < 30:
+                if get_window(ocr_config.window):
+                    break
+                logger.info(f"Window: {ocr_config.window} Could not be found, retrying in 1 second...")
+                time.sleep(1)
+            else:
+                logger.error(f"Window '{ocr_config.window}' not found within 30 seconds.")
+                sys.exit(1)
+        logger.info(f"Starting OCR with configuration: Window: {ocr_config.window}, Rectangles: {ocr_config.rectangles}, Engine 1: {ocr1}, Engine 2: {ocr2}, Two-pass OCR: {twopassocr}")
     set_dpi_awareness()
-    if ssonly or ocr_config:
+    if manual or ocr_config:
         rectangles = ocr_config.rectangles if ocr_config and ocr_config.rectangles else []
         oneocr_threads = []
         ocr_thread = threading.Thread(target=run_oneocr, args=(ocr_config,rectangles ), daemon=True)
         ocr_thread.start()
-        if not ssonly:
+        if not manual:
             worker_thread = threading.Thread(target=process_task_queue, daemon=True)
             worker_thread.start()
         websocket_server_thread = WebsocketServerThread(read=True)

GameSentenceMiner/owocr/owocr/ocr.py

@@ -307,6 +307,7 @@ class GoogleLens:
         #     json.dump(response_dict, f, indent=4, ensure_ascii=False)
         res = ''
         text = response_dict['objects_response']['text']
+        skipped = []
         if furigana_filter_sensitivity > 0:
             if 'text_layout' in text:
                 for paragraph in text['text_layout']['paragraphs']:
@@ -315,9 +316,11 @@ class GoogleLens:
                             for word in line['words']:
                                 res += word['plain_text'] + word['text_separator']
                         else:
+                            skipped.append(word['plain_text'] for word in line['words'])
                             continue
                         res += '\n'
-
+            # logger.info(
+            #     f"Skipped {len(skipped)} chars due to furigana filter sensitivity: {furigana_filter_sensitivity}")
             # widths = []
             # heights = []
             # if 'text_layout' in text:
@@ -789,7 +792,6 @@ class OneOCR:
     available = False
 
     def __init__(self, config={}):
-        self.kana_kanji_regex = re.compile(r'[\u3041-\u3096\u30A1-\u30FA\u4E00-\u9FFF]')
         if sys.platform == 'win32':
             if int(platform.release()) < 10:
                 logger.warning('OneOCR is not supported on Windows older than 10!')
@@ -836,10 +838,9 @@ class OneOCR:
                 #           encoding='utf-8') as f:
                 #     json.dump(ocr_resp, f, indent=4, ensure_ascii=False)
                 res = ''
+                skipped = []
                 if furigana_filter_sensitivity > 0:
                     for line in ocr_resp['lines']:
-                        if self.kana_kanji_regex.search(line['text']) is None:
-                            continue
                         x1, x2, x3, x4 = line['bounding_rect']['x1'], line['bounding_rect']['x2'], \
                             line['bounding_rect']['x3'], line['bounding_rect']['x4']
                         y1, y2, y3, y4 = line['bounding_rect']['y1'], line['bounding_rect']['y2'], \
@@ -849,8 +850,11 @@ class OneOCR:
                         if width > furigana_filter_sensitivity and height > furigana_filter_sensitivity:
                             res += line['text']
                         else:
+                            skipped.extend(char for char in line['text'])
                             continue
                         res += '\n'
+                    # logger.info(
+                    #     f"Skipped {len(skipped)} chars due to furigana filter sensitivity: {furigana_filter_sensitivity}")
                     # widths, heights = [], []
                     # for line in ocr_resp['lines']:
                     #     for word in line['words']:

GameSentenceMiner/owocr/owocr/run.py

@@ -309,7 +309,7 @@ class RequestHandler(socketserver.BaseRequestHandler):
 class TextFiltering:
     accurate_filtering = False
 
-    def __init__(self):
+    def __init__(self, lang="ja"):
         from pysbd import Segmenter
         self.segmenter = Segmenter(language=lang, clean=True)
         self.kana_kanji_regex = re.compile(r'[\u3041-\u3096\u30A1-\u30FA\u4E00-\u9FFF]')
@@ -1159,7 +1159,6 @@ def run(read_from=None,
 
     if 'screencapture' in (read_from, read_from_secondary):
         global take_screenshot
-        screen_capture_combo = config.get_general('screen_capture_combo')
         last_screenshot_time = 0
         last_result = ([], engine_index)
         if screen_capture_combo != '':
@@ -1216,6 +1215,8 @@ def run(read_from=None,
     user_input_thread = threading.Thread(target=user_input_thread_run, daemon=True)
     user_input_thread.start()
     logger.opt(ansi=True).info(f"Reading from {' and '.join(read_from_readable)}, writing to {write_to_readable} using <{engine_color}>{engine_instances[engine_index].readable_name}</{engine_color}>{' (paused)' if paused else ''}")
+    if screen_capture_combo:
+        logger.opt(ansi=True).info(f'Manual OCR Running... Press <{engine_color}>{screen_capture_combo.replace("<", "").replace(">", "")}</{engine_color}> to run OCR')
 
     while not terminated:
         ocr_start_time = datetime.now()

{gamesentenceminer-2.9.23.dist-info → gamesentenceminer-2.9.25.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: GameSentenceMiner
-Version: 2.9.23
+Version: 2.9.25
 Summary: A tool for mining sentences from games.
 Author-email: Beangate <bpwhelan95@gmail.com>
 License: MIT License

{gamesentenceminer-2.9.23.dist-info → gamesentenceminer-2.9.25.dist-info}/RECORD

@@ -16,17 +16,17 @@ GameSentenceMiner/assets/icon512.png,sha256=HxUj2GHjyQsk8NV433256UxU9phPhtjCY-YB
 GameSentenceMiner/assets/icon64.png,sha256=N8xgdZXvhqVQP9QUK3wX5iqxX9LxHljD7c-Bmgim6tM,9301
 GameSentenceMiner/assets/pickaxe.png,sha256=VfIGyXyIZdzEnVcc4PmG3wszPMO1W4KCT7Q_nFK6eSE,1403829
 GameSentenceMiner/ocr/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-GameSentenceMiner/ocr/gsm_ocr_config.py,sha256=fEQ2o2NXksGRHpueO8c4TfAp75GEdAtAr1ngTFOsdpg,2257
+GameSentenceMiner/ocr/gsm_ocr_config.py,sha256=0_4aug96x8JFsOVdYr7wjs5U-l4PPW7Y_V_09oDML-8,3220
 GameSentenceMiner/ocr/ocrconfig.py,sha256=_tY8mjnzHMJrLS8E5pHqYXZjMuLoGKYgJwdhYgN-ny4,6466
-GameSentenceMiner/ocr/owocr_area_selector.py,sha256=HvzGDW5UaJ3AmrKV_IqOfCvZ7D7X6S3th_XnXk_LIU0,47259
-GameSentenceMiner/ocr/owocr_helper.py,sha256=idjVB8Ln9PwNUFZMOtRkq7wAhbteU7-haRM7Cf_wmY4,20979
+GameSentenceMiner/ocr/owocr_area_selector.py,sha256=qLsnM_UEP14V7t8vpERji_-_TwpwEI3Gd4BFck4vYaY,47029
+GameSentenceMiner/ocr/owocr_helper.py,sha256=QeQ0PrUcfFASVlsdbTXG5sxdke3x0d-gPnfRQw6L6RY,20666
 GameSentenceMiner/ocr/ss_picker.py,sha256=Ck2K9JliXhY_7a7nS0BVsoAmzwPFFtsW_TxHCcfxT00,4274
 GameSentenceMiner/owocr/owocr/__init__.py,sha256=87hfN5u_PbL_onLfMACbc0F5j4KyIK9lKnRCj6oZgR0,49
 GameSentenceMiner/owocr/owocr/__main__.py,sha256=XQaqZY99EKoCpU-gWQjNbTs7Kg17HvBVE7JY8LqIE0o,157
 GameSentenceMiner/owocr/owocr/config.py,sha256=qM7kISHdUhuygGXOxmgU6Ef2nwBShrZtdqu4InDCViE,8103
 GameSentenceMiner/owocr/owocr/lens_betterproto.py,sha256=oNoISsPilVVRBBPVDtb4-roJtAhp8ZAuFTci3TGXtMc,39141
-GameSentenceMiner/owocr/owocr/ocr.py,sha256=uEv0Pjvq7n6XJOkV4i_EiR0wxXqzi2EVpWsK2-i3SXs,51932
-GameSentenceMiner/owocr/owocr/run.py,sha256=bNh9nF0tzFa7BuW-ixcipyf6vTrNyaNwtgG6W4a-uIo,55063
+GameSentenceMiner/owocr/owocr/ocr.py,sha256=fWrbKomSrdkSdlEiMGTKb6-F7wCgfaZZNBUo2gCqmuA,52247
+GameSentenceMiner/owocr/owocr/run.py,sha256=wOileOoP1djCpOLgg7d-nWrYS78NctClNTscLZ3kwDc,55198
 GameSentenceMiner/owocr/owocr/screen_coordinate_picker.py,sha256=Na6XStbQBtpQUSdbN3QhEswtKuU1JjReFk_K8t5ezQE,3395
 GameSentenceMiner/util/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 GameSentenceMiner/util/configuration.py,sha256=TAdd-aLMCMvpu7OS1Y5brKQC4svcK29J94nvrEb4Ue4,27535
@@ -61,9 +61,9 @@ GameSentenceMiner/web/templates/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm
 GameSentenceMiner/web/templates/index.html,sha256=HZKiIjiGJV8PGQ9T2aLDUNSfJn71qOwbYCjbRuSIjpY,213583
 GameSentenceMiner/web/templates/text_replacements.html,sha256=tV5c8mCaWSt_vKuUpbdbLAzXZ3ATZeDvQ9PnnAfqY0M,8598
 GameSentenceMiner/web/templates/utility.html,sha256=3flZinKNqUJ7pvrZk6xu__v67z44rXnaK7UTZ303R-8,16946
-gamesentenceminer-2.9.23.dist-info/licenses/LICENSE,sha256=OXLcl0T2SZ8Pmy2_dmlvKuetivmyPd5m1q-Gyd-zaYY,35149
-gamesentenceminer-2.9.23.dist-info/METADATA,sha256=4NIEep_aRefXZXkRen67RttLkQRsm-Zgcb7pi9g801g,7276
-gamesentenceminer-2.9.23.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
-gamesentenceminer-2.9.23.dist-info/entry_points.txt,sha256=2APEP25DbfjSxGeHtwBstMH8mulVhLkqF_b9bqzU6vQ,65
-gamesentenceminer-2.9.23.dist-info/top_level.txt,sha256=V1hUY6xVSyUEohb0uDoN4UIE6rUZ_JYx8yMyPGX4PgQ,18
-gamesentenceminer-2.9.23.dist-info/RECORD,,
+gamesentenceminer-2.9.25.dist-info/licenses/LICENSE,sha256=OXLcl0T2SZ8Pmy2_dmlvKuetivmyPd5m1q-Gyd-zaYY,35149
+gamesentenceminer-2.9.25.dist-info/METADATA,sha256=fH0PSv9qJAA6wBq72A3c1xEhB0Y2jE6fkWyen-i6M1Y,7276
+gamesentenceminer-2.9.25.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
+gamesentenceminer-2.9.25.dist-info/entry_points.txt,sha256=2APEP25DbfjSxGeHtwBstMH8mulVhLkqF_b9bqzU6vQ,65
+gamesentenceminer-2.9.25.dist-info/top_level.txt,sha256=V1hUY6xVSyUEohb0uDoN4UIE6rUZ_JYx8yMyPGX4PgQ,18
+gamesentenceminer-2.9.25.dist-info/RECORD,,