GameSentenceMiner 2.14.14__tar.gz → 2.14.15__tar.gz

{gamesentenceminer-2.14.14 → gamesentenceminer-2.14.15}/GameSentenceMiner/owocr/owocr/ocr.py

@@ -15,6 +15,7 @@ import numpy as np
 import rapidfuzz.fuzz
 from PIL import Image
 from loguru import logger
+import regex
 import requests
 
 
@@ -377,7 +378,8 @@ class GoogleLens:
                 for line in paragraph['lines']:
                     if furigana_filter_sensitivity:
                         for word in line['words']:
-                            if not self.punctuation_regex.findall(word):
+                            if self.punctuation_regex.findall(word['plain_text']):
+                                res += word['plain_text'] + word['text_separator']
                                 continue
                             if 'geometry' not in word:
                                 res += word['plain_text'] + word['text_separator']
@@ -869,8 +871,10 @@ class OneOCR:
     available = False
 
     def __init__(self, config={}, lang='ja'):
+        import regex
         self.initial_lang = lang
         self.regex = get_regex(lang)
+        self.punctuation_regex = regex.compile(r'[\p{P}\p{S}]')
         if sys.platform == 'win32':
             if int(platform.release()) < 10:
                 logger.warning('OneOCR is not supported on Windows older than 10!')
@@ -939,10 +943,10 @@ class OneOCR:
         if sys.platform == 'win32':
             try:
                 ocr_resp = self.model.recognize_pil(img)
-                # if os.path.exists(os.path.expanduser("~/GSM/temp")):
-                #     with open(os.path.join(os.path.expanduser("~/GSM/temp"), 'oneocr_response.json'), 'w',
-                #                 encoding='utf-8') as f:
-                #         json.dump(ocr_resp, f, indent=4, ensure_ascii=False)
+                if os.path.exists(os.path.expanduser("~/GSM/temp")):
+                    with open(os.path.join(os.path.expanduser("~/GSM/temp"), 'oneocr_response.json'), 'w',
+                                encoding='utf-8') as f:
+                        json.dump(ocr_resp, f, indent=4, ensure_ascii=False)
                 # print(json.dumps(ocr_resp))
                 filtered_lines = [line for line in ocr_resp['lines'] if self.regex.search(line['text'])]
                 x_coords = [line['bounding_rect'][f'x{i}'] for line in filtered_lines for i in range(1, 5)]
@@ -955,17 +959,21 @@ class OneOCR:
                 boxes = []
                 if furigana_filter_sensitivity > 0:
                     for line in filtered_lines:
-                        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'], \
-                            line['bounding_rect']['y3'], line['bounding_rect']['y4']
-                        width = max(x2 - x1, x3 - x4)
-                        height = max(y3 - y1, y4 - y2)
-                        if width > furigana_filter_sensitivity and height > furigana_filter_sensitivity:
-                            res += line['text']
-                        else:
-                            skipped.extend(char for char in line['text'])
-                            continue
+                        for char in line['words']:
+                            if self.punctuation_regex.findall(char['text']):
+                                res += char['text']
+                                continue
+                            x1, x2, x3, x4 = char['bounding_rect']['x1'], char['bounding_rect']['x2'], \
+                            char['bounding_rect']['x3'], char['bounding_rect']['x4']
+                            y1, y2, y3, y4 = char['bounding_rect']['y1'], char['bounding_rect']['y2'], \
+                                char['bounding_rect']['y3'], char['bounding_rect']['y4']
+                            width = max(x2 - x1, x3 - x4)
+                            height = max(y3 - y1, y4 - y2)
+                            if width > furigana_filter_sensitivity and height > furigana_filter_sensitivity:
+                                res += char['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}")

{gamesentenceminer-2.14.14 → gamesentenceminer-2.14.15}/GameSentenceMiner/owocr/owocr/run.py

@@ -59,6 +59,7 @@ from GameSentenceMiner.util.configuration import get_temporary_directory, get_co
 
 config = None
 last_image = None
+last_image_np = None
 
 
 class ClipboardThread(threading.Thread):
@@ -818,9 +819,10 @@ def apply_adaptive_threshold_filter(img):
 
 
 def set_last_image(image):
-    global last_image
+    global last_image, last_image_np
     if image is None:
         last_image = None
+        last_image_np = None
     try:
         if image == last_image:
             return
@@ -833,23 +835,34 @@ def set_last_image(image):
     except Exception:
         pass
     last_image = image
+    last_image_np = np.array(last_image)
     # last_image = apply_adaptive_threshold_filter(image)
 
 
-def are_images_identical(img1, img2):
-    if None in (img1, img2):
+def are_images_identical(img1, img2, img2_np=None):
+    """
+    Compares two images for pixel-wise identity.
+    Optionally, pass a cached np.array for img2 as img2_np to avoid repeated conversion.
+
+    Args:
+        img1: PIL.Image or np.ndarray
+        img2: PIL.Image or np.ndarray
+        img2_np: Optional cached np.ndarray for img2
+
+    Returns:
+        bool: True if images are identical, False otherwise.
+    """
+    if any(v is None for v in (img1, img2, img2_np)):
         return False
 
     try:
-        img1 = np.array(img1)
-        img2 = np.array(img2)
+        img1_np = np.array(img1)
+        img2_np = img2_np if img2_np is not None else np.array(img2)
     except Exception:
-        logger.warning(
-            "Failed to convert images to numpy arrays for comparison.")
-        # If conversion to numpy array fails, consider them not identical
+        logger.warning("Failed to convert images to numpy arrays for comparison.")
         return False
 
-    return (img1.shape == img2.shape) and np.array_equal(img1, img2)
+    return (img1_np.shape == img2_np.shape) and np.array_equal(img1_np, img2_np)
 
 
 import cv2
@@ -997,7 +1010,7 @@ class OBSScreenshotThread(threading.Thread):
 
     def connect_obs(self):
         import GameSentenceMiner.obs as obs
-        obs.connect_to_obs_sync()
+        obs.connect_to_obs_sync(check_output=False)
 
     def scale_down_width_height(self, width, height):
         if width == 0 or height == 0:
@@ -1765,7 +1778,7 @@ def run(read_from=None,
                 #             sleep_time_to_add += .005
                 #         continue
                 # else:
-                if are_images_identical(img, last_image):
+                if are_images_identical(img, last_image, last_image_np):
                     logger.info("Captured screenshot is identical to the last one, sleeping.")
                     if time.time() - last_result_time > 10:
                         sleep_time_to_add += .005

{gamesentenceminer-2.14.14 → gamesentenceminer-2.14.15}/GameSentenceMiner.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: GameSentenceMiner
-Version: 2.14.14
+Version: 2.14.15
 Summary: A tool for mining sentences from games. Update: Overlay?
 Author-email: Beangate <bpwhelan95@gmail.com>
 License: MIT License

{gamesentenceminer-2.14.14 → gamesentenceminer-2.14.15}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: GameSentenceMiner
-Version: 2.14.14
+Version: 2.14.15
 Summary: A tool for mining sentences from games. Update: Overlay?
 Author-email: Beangate <bpwhelan95@gmail.com>
 License: MIT License

{gamesentenceminer-2.14.14 → gamesentenceminer-2.14.15}/pyproject.toml

@@ -7,7 +7,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "GameSentenceMiner"
-version = "2.14.14"
+version = "2.14.15"
 description = "A tool for mining sentences from games. Update: Overlay?"
 readme = "README.md"
 requires-python = ">=3.10"