v3.0.1

hot fix!

Author: BBC-Esq

src/initialize.py

@@ -2,8 +2,6 @@
 import yaml
 import platform
 import ctranslate2
-import os
-import shutil
 from pathlib import Path
 
 def get_compute_device_info():
@@ -62,22 +60,10 @@ def update_config_file(**system_info):
     with open(full_config_path, 'w') as stream:
         yaml.safe_dump(config_data, stream)
 
-def move_custom_pdf_loader():
-    current_dir = Path.cwd()
-    user_manual_pdf_path = current_dir / "User_Manual" / "PDF.py"
-    lib_pdf_path = current_dir / "Lib" / "site-packages" / "langchain" / "document_loaders" / "parsers" / "PDF.py"
-
-    user_manual_pdf_size = user_manual_pdf_path.stat().st_size
-    lib_pdf_size = lib_pdf_path.stat().st_size
-
-    if user_manual_pdf_size != lib_pdf_size:
-        shutil.copy(user_manual_pdf_path, lib_pdf_path)
-
 def main():
     compute_device_info = get_compute_device_info()
     platform_info = get_platform_info()
     update_config_file(Compute_Device=compute_device_info, Platform_Info=platform_info)
-    move_custom_pdf_loader()
 
 if __name__ == "__main__":
     main()

src/replace_pdf.py

@@ -1,5 +1,6 @@
 import shutil
 from pathlib import Path
+import hashlib
 
 def find_all_target_directories_with_file(base_path, target_folder, target_file):
     found_directories = []
@@ -20,6 +21,14 @@ def find_closest_directory(directories, base_directory):
     depths = [(dir, get_directory_depth(dir, base_directory)) for dir in directories]
     return min(depths, key=lambda x: x[1])[0]
 
+def hash_file(filepath):
+    """Compute the SHA-256 hash of a file."""
+    hasher = hashlib.sha256()
+    with open(filepath, 'rb') as f:
+        buf = f.read()
+        hasher.update(buf)
+    return hasher.hexdigest()
+
 def replace_pdf_in_parsers():
     script_dir = Path(__file__).parent
     user_manual_pdf_path = script_dir / "User_Manual" / "PDF.py"
@@ -28,7 +37,7 @@ def replace_pdf_in_parsers():
         print("No 'pdf.py' file found in 'User_Manual' directory.")
         return
 
-    base_dir = script_dir.parent.parent  # Move up two levels from the script's location
+    base_dir = script_dir.parent  # Move up one level from the script's location
     target_folder = "parsers"
     target_file = "pdf.py"
     found_paths = find_all_target_directories_with_file(base_dir, target_folder, target_file)
@@ -44,16 +53,16 @@ def replace_pdf_in_parsers():
         print(f"Chosen 'parsers' directory based on path depth: {closest_parsers_path}")
         chosen_pdf_path = closest_parsers_path / target_file
 
-    # File size comparison and replacement
-    user_manual_pdf_size = user_manual_pdf_path.stat().st_size
-    chosen_pdf_size = chosen_pdf_path.stat().st_size
+    # Hash comparison and replacement
+    user_manual_pdf_hash = hash_file(user_manual_pdf_path)
+    chosen_pdf_hash = hash_file(chosen_pdf_path)
 
-    if user_manual_pdf_size != chosen_pdf_size:
+    if user_manual_pdf_hash != chosen_pdf_hash:
         print("Replacing the existing pdf.py with the new one...")
         shutil.copy(user_manual_pdf_path, chosen_pdf_path)
         print(f"PDF.py replaced at: {chosen_pdf_path}")
     else:
-        print("No replacement needed. The files are of the same size.")
+        print("No replacement needed. The files are identical.")
 
 if __name__ == "__main__":
     replace_pdf_in_parsers()