formatting and clean-up 🧹

.gitignore

@@ -5,4 +5,5 @@ venv
 dist
 build
 .pytest_cache
-.vscode
+.vscode
+.aider*

openrecall/app.py

@@ -137,9 +137,7 @@ def timeline():
 def search():
     q = request.args.get("q")
     entries = get_all_entries()
-    embeddings = [
+    embeddings = [np.frombuffer(entry.embedding, dtype=np.float64) for entry in entries]
-        np.frombuffer(entry.embedding, dtype=np.float64) for entry in entries
-    ]
     query_embedding = get_embedding(q)
     similarities = [cosine_similarity(query_embedding, emb) for emb in embeddings]
     indices = np.argsort(similarities)[::-1]

openrecall/config.py

@@ -2,9 +2,7 @@ import os
 import sys
 import argparse
 
-parser = argparse.ArgumentParser(
+parser = argparse.ArgumentParser(description="OpenRecall")
-    description="OpenRecall"
-)
 
 parser.add_argument(
     "--storage-path",
@@ -38,11 +36,12 @@ def get_appdata_folder(app_name="openrecall"):
         os.makedirs(path)
     return path
 
+
 if args.storage_path:
     appdata_folder = args.storage_path
     screenshots_path = os.path.join(appdata_folder, "screenshots")
     db_path = os.path.join(appdata_folder, "recall.db")
-else:   
+else:
     appdata_folder = get_appdata_folder()
     db_path = os.path.join(appdata_folder, "recall.db")
     screenshots_path = os.path.join(appdata_folder, "screenshots")

openrecall/screenshot.py

@@ -9,7 +9,11 @@ from openrecall.config import screenshots_path, args
 from openrecall.database import insert_entry
 from openrecall.nlp import get_embedding
 from openrecall.ocr import extract_text_from_image
-from openrecall.utils import get_active_app_name, get_active_window_title, is_user_active
+from openrecall.utils import (
+    get_active_app_name,
+    get_active_window_title,
+    is_user_active,
+)
 
 
 def mean_structured_similarity_index(img1, img2, L=255):
@@ -45,18 +49,19 @@ def take_screenshots(monitor=1):
 
             if args.primary_monitor_only and monitor != 1:
                 continue
-            
+
             monitor_ = sct.monitors[monitor]
             screenshot = np.array(sct.grab(monitor_))
             screenshot = screenshot[:, :, [2, 1, 0]]
             screenshots.append(screenshot)
-    
+
     return screenshots
 
 
 def record_screenshots_thread():
     # TODO: fix the error from huggingface tokenizers
     import os
+
     os.environ["TOKENIZERS_PARALLELISM"] = "false"
 
     last_screenshots = take_screenshots()
@@ -65,11 +70,11 @@ def record_screenshots_thread():
         if not is_user_active():
             time.sleep(3)
             continue
-    
+
         screenshots = take_screenshots()
-    
+
         for i, screenshot in enumerate(screenshots):
-            
+
             last_screenshot = last_screenshots[i]
 
             if not is_similar(screenshot, last_screenshot):
@@ -88,5 +93,5 @@ def record_screenshots_thread():
                 insert_entry(
                     text, timestamp, embedding, active_app_name, active_window_title
                 )
-                
+
         time.sleep(3)

openrecall/utils.py

@@ -1,5 +1,6 @@
 import sys
 
+
 def human_readable_time(timestamp):
     import datetime
 
@@ -35,6 +36,7 @@ def get_active_app_name_osx():
     except:
         return ""
 
+
 def get_active_window_title_osx():
     from Quartz import (
         CGWindowListCopyWindowInfo,
@@ -80,11 +82,12 @@ def get_active_window_title_windows():
 
 
 def get_active_app_name_linux():
-    return ''
+    return ""
 
 
 def get_active_window_title_linux():
-    return ''
+    return ""
+
 
 def get_active_app_name():
     if sys.platform == "win32":
@@ -107,28 +110,29 @@ def get_active_window_title():
     else:
         raise NotImplementedError("This platform is not supported")
 
+
 def is_user_active_osx():
     import subprocess
 
     try:
         # Run the 'ioreg' command to get idle time information
         output = subprocess.check_output(["ioreg", "-c", "IOHIDSystem"]).decode()
-        
+
         # Find the line containing "HIDIdleTime"
-        for line in output.split('\n'):
+        for line in output.split("\n"):
             if "HIDIdleTime" in line:
                 # Extract the idle time value
-                idle_time = int(line.split('=')[-1].strip())
+                idle_time = int(line.split("=")[-1].strip())
-                
+
                 # Convert idle time from nanoseconds to seconds
                 idle_seconds = idle_time / 1000000000
-                
+
                 # If idle time is less than 5 seconds, consider the user not idle
                 return idle_seconds < 5
-        
+
         # If "HIDIdleTime" is not found, assume the user is not idle
         return True
-    
+
     except subprocess.CalledProcessError:
         # If there's an error running the command, assume the user is not idle
         return True
@@ -137,6 +141,7 @@ def is_user_active_osx():
         # If there's any other error, assume the user is not idle
         return True
 
+
 def is_user_active():
     if sys.platform == "win32":
         return True
@@ -146,4 +151,3 @@ def is_user_active():
         return True
     else:
         raise NotImplementedError("This platform is not supported")
-    

setup.py

@@ -17,12 +17,17 @@ install_requires = [
     "shapely==2.0.4",
     "h5py==3.11.0",
     "rapidfuzz==3.9.3",
-    "Pillow==10.3.0"
+    "Pillow==10.3.0",
 ]
 
 # Define OS-specific dependencies
-extras_require = {"windows": ["pywin32", "psutil"], "macos": ["pyobjc==10.3"], "linux": [],
+extras_require = {
-    'python-doctr': ['python-doctr @ git+https://github.com/koenvaneijk/doctr.git@af711bc04eb8876a7189923fb51ec44481ee18cd']
+    "windows": ["pywin32", "psutil"],
+    "macos": ["pyobjc==10.3"],
+    "linux": [],
+    "python-doctr": [
+        "python-doctr @ git+https://github.com/koenvaneijk/doctr.git@af711bc04eb8876a7189923fb51ec44481ee18cd"
+    ],
 }
 
 # Determine the current OS

tests/test_config.py

@@ -2,29 +2,35 @@ import pytest
 from unittest import mock
 from openrecall.config import get_appdata_folder
 
+
 def test_get_appdata_folder_windows(tmp_path):
-    with mock.patch('sys.platform', 'win32'):
+    with mock.patch("sys.platform", "win32"):
-        with mock.patch.dict('os.environ', {'APPDATA': str(tmp_path)}):
+        with mock.patch.dict("os.environ", {"APPDATA": str(tmp_path)}):
-            expected_path = tmp_path / 'openrecall'
+            expected_path = tmp_path / "openrecall"
             assert get_appdata_folder() == str(expected_path)
             assert expected_path.exists()
 
+
 def test_get_appdata_folder_windows_no_appdata():
-    with mock.patch('sys.platform', 'win32'):
+    with mock.patch("sys.platform", "win32"):
-        with mock.patch.dict('os.environ', {}, clear=True):
+        with mock.patch.dict("os.environ", {}, clear=True):
-            with pytest.raises(EnvironmentError, match="APPDATA environment variable is not set."):
+            with pytest.raises(
+                EnvironmentError, match="APPDATA environment variable is not set."
+            ):
                 get_appdata_folder()
 
+
 def test_get_appdata_folder_darwin(tmp_path):
-    with mock.patch('sys.platform', 'darwin'):
+    with mock.patch("sys.platform", "darwin"):
-        with mock.patch('os.path.expanduser', return_value=str(tmp_path)):
+        with mock.patch("os.path.expanduser", return_value=str(tmp_path)):
-            expected_path = tmp_path / 'Library' / 'Application Support' / 'openrecall'
+            expected_path = tmp_path / "Library" / "Application Support" / "openrecall"
             assert get_appdata_folder() == str(expected_path)
             assert expected_path.exists()
 
+
 def test_get_appdata_folder_linux(tmp_path):
-    with mock.patch('sys.platform', 'linux'):
+    with mock.patch("sys.platform", "linux"):
-        with mock.patch('os.path.expanduser', return_value=str(tmp_path)):
+        with mock.patch("os.path.expanduser", return_value=str(tmp_path)):
-            expected_path = tmp_path / '.local' / 'share' / 'openrecall'
+            expected_path = tmp_path / ".local" / "share" / "openrecall"
             assert get_appdata_folder() == str(expected_path)
             assert expected_path.exists()

tests/test_nlp.py

@@ -2,34 +2,42 @@ import pytest
 import numpy as np
 from openrecall.nlp import cosine_similarity
 
+
 def test_cosine_similarity_identical_vectors():
     a = np.array([1, 0, 0])
     b = np.array([1, 0, 0])
     assert cosine_similarity(a, b) == 1.0
 
+
 def test_cosine_similarity_orthogonal_vectors():
     a = np.array([1, 0, 0])
     b = np.array([0, 1, 0])
     assert cosine_similarity(a, b) == 0.0
 
+
 def test_cosine_similarity_opposite_vectors():
     a = np.array([1, 0, 0])
     b = np.array([-1, 0, 0])
     assert cosine_similarity(a, b) == -1.0
 
+
 def test_cosine_similarity_non_unit_vectors():
     a = np.array([3, 0, 0])
     b = np.array([1, 0, 0])
     assert cosine_similarity(a, b) == 1.0
 
+
 def test_cosine_similarity_arbitrary_vectors():
     a = np.array([1, 2, 3])
     b = np.array([4, 5, 6])
     expected_similarity = np.dot(a, b) / (np.linalg.norm(a) * np.linalg.norm(b))
     assert cosine_similarity(a, b) == pytest.approx(expected_similarity)
 
+
 def test_cosine_similarity_zero_vector():
     a = np.array([0, 0, 0])
     b = np.array([1, 0, 0])
     result = cosine_similarity(a, b)
-    assert np.isnan(result), "Expected result to be NaN when one of the vectors is a zero vector"
+    assert np.isnan(
+        result
+    ), "Expected result to be NaN when one of the vectors is a zero vector"