paddleocr-skills 1.0.0 → 1.1.0

package/templates/ppocrv5/scripts/ppocrv5/configure.py

@@ -1,346 +1,346 @@
-#!/usr/bin/env python3
-"""
-Configuration Wizard for PP-OCRv5 API Skill
-
-Supports two modes:
-1. Interactive mode (default): python configure.py
-2. CLI mode: python configure.py --api-url URL --token TOKEN
-
-Supports pasting Python code format (e.g., API_URL = "...")
-"""
-
-import argparse
-import re
-import sys
-from pathlib import Path
-
-# Add parent directory to path for imports
-sys.path.insert(0, str(Path(__file__).parent.parent))
-
-
-def parse_input(user_input: str) -> str:
-    """
-    Intelligently parse user input, supporting multiple formats:
-    - API_URL = "https://..."
-    - "https://..."
-    - https://...
-    - TOKEN = "abc123..."
-
-    Returns the extracted value
-    """
-    user_input = user_input.strip()
-
-    # Format 1: KEY = "value" or KEY = 'value'
-    match = re.match(r'^\w+\s*=\s*["\'](.+?)["\']$', user_input)
-    if match:
-        return match.group(1)
-
-    # Format 2: "value" or 'value'
-    match = re.match(r'^["\'](.+?)["\']$', user_input)
-    if match:
-        return match.group(1)
-
-    # Format 3: value (direct input)
-    return user_input
-
-
-def normalize_api_url(url: str) -> str:
-    """
-    Normalize API URL
-    Supports:
-    - https://xxx.aistudio-app.com/ocr
-    - https://xxx.aistudio-app.com
-    - xxx.aistudio-app.com
-
-    Returns: https://xxx.aistudio-app.com/ocr
-    """
-    url = url.strip()
-
-    # Remove http:// or https://
-    url = re.sub(r'^https?://', '', url)
-
-    # Remove trailing /ocr or other paths
-    url = re.sub(r'/.*$', '', url)
-
-    # Return complete API URL
-    return f"https://{url}/ocr"
-
-
-def mask_token(token: str) -> str:
-    """Mask token, only show first and last parts"""
-    if len(token) <= 8:
-        return "****"
-    return f"{token[:4]}...{token[-4:]}"
-
-
-def test_connection(api_url: str, token: str) -> bool:
-    """Test API connection (optional)"""
-    try:
-        import httpx
-
-        print("\nTesting connection...")
-
-        # Simple test request (using a small base64 image)
-        test_payload = {
-            "file": "iVBORw0KGgoAAAANSUhEUgAAAAEAAAABCAYAAAAfFcSJAAAADUlEQVR42mNk+M9QDwADhgGAWjR9awAAAABJRU5ErkJggg==",
-            "fileType": 1,
-            "visualize": False
-        }
-
-        headers = {
-            "Authorization": f"token {token}",
-            "Content-Type": "application/json"
-        }
-
-        client = httpx.Client(timeout=10.0)
-        try:
-            resp = client.post(api_url, json=test_payload, headers=headers)
-            resp_json = resp.json()
-
-            if resp.status_code == 200 and resp_json.get("errorCode") == 0:
-                print("✓ API connection successful!")
-                print("✓ OCR function is working!")
-                return True
-            elif resp.status_code == 403:
-                print("✗ Token verification failed, please check if the Token is correct")
-                return False
-            elif resp.status_code == 429:
-                print("⚠ API quota exhausted, but connection is working")
-                return True
-            else:
-                print(f"⚠ API returned error: {resp_json.get('errorMsg', 'Unknown error')}")
-                return False
-        finally:
-            client.close()
-
-    except ImportError:
-        print("⚠ httpx not installed, skipping connection test")
-        print("   Install with: pip install httpx")
-        return True
-    except Exception as e:
-        print(f"✗ Connection test failed: {e}")
-        return False
-
-
-def save_config(api_url: str, token: str, project_root: Path, quiet: bool = False) -> bool:
-    """
-    Save configuration to .env file
-
-    Args:
-        api_url: Normalized API URL
-        token: Access token
-        project_root: Project root directory
-        quiet: If True, suppress output messages
-
-    Returns:
-        True if successful, False otherwise
-    """
-    env_file = project_root / ".env"
-
-    # Read existing configuration (if exists)
-    existing_config = {}
-    if env_file.exists():
-        if not quiet:
-            print(f"\nDetected existing configuration file: {env_file}")
-            overwrite = input("Overwrite? [Y/n]: ").strip().lower()
-            if overwrite == 'n':
-                print("Configuration cancelled")
-                return False
-
-        # Preserve other configuration items
-        with open(env_file, 'r', encoding='utf-8') as f:
-            for line in f:
-                line = line.strip()
-                if line and not line.startswith('#'):
-                    if '=' in line and not line.startswith('API_URL=') and not line.startswith('PADDLE_OCR_TOKEN='):
-                        key, value = line.split('=', 1)
-                        existing_config[key.strip()] = value.strip()
-
-    # Write new configuration
-    try:
-        with open(env_file, 'w', encoding='utf-8') as f:
-            f.write("# PP-OCRv5 API Configuration\n")
-            f.write("# This file was automatically generated by configure.py\n\n")
-
-            f.write(f"API_URL={api_url}\n")
-            f.write(f"PADDLE_OCR_TOKEN={token}\n")
-
-            # Write other preserved configurations
-            if existing_config:
-                f.write("\n# Other configurations\n")
-                for key, value in existing_config.items():
-                    f.write(f"{key}={value}\n")
-
-        if not quiet:
-            print(f"\n✓ Configuration saved to: {env_file}")
-        return True
-
-    except Exception as e:
-        print(f"\n✗ Failed to save configuration: {e}")
-        return False
-
-
-def main():
-    # Parse command-line arguments
-    parser = argparse.ArgumentParser(
-        description='PP-OCRv5 API Configuration Tool',
-        formatter_class=argparse.RawDescriptionHelpFormatter,
-        epilog="""
-Examples:
-  # Interactive mode
-  python configure.py
-
-  # CLI mode (non-interactive)
-  python configure.py --api-url "https://xxx.aistudio-app.com/ocr" --token "your_token"
-        """
-    )
-    parser.add_argument('--api-url', help='API URL (non-interactive mode)')
-    parser.add_argument('--token', help='Access token (non-interactive mode)')
-    parser.add_argument('--quiet', action='store_true', help='Suppress output messages')
-
-    args = parser.parse_args()
-
-    # Get project root directory (parent of scripts directory)
-    project_root = Path(__file__).parent.parent
-
-    # ========================================
-    # CLI Mode (non-interactive)
-    # ========================================
-    if args.api_url and args.token:
-        try:
-            # Normalize API URL
-            api_url = normalize_api_url(parse_input(args.api_url))
-            token = parse_input(args.token)
-
-            # Validate
-            if len(token) < 16:
-                print("Error: Token seems too short. Please check and try again.")
-                sys.exit(1)
-
-            # Save configuration
-            if save_config(api_url, token, project_root, quiet=args.quiet):
-                if not args.quiet:
-                    print("\n✓ Configuration complete!")
-                    print(f"  API_URL: {api_url}")
-                    print(f"  TOKEN: {mask_token(token)}")
-                sys.exit(0)
-            else:
-                sys.exit(1)
-
-        except Exception as e:
-            print(f"Error: {e}")
-            sys.exit(1)
-
-    elif args.api_url or args.token:
-        print("Error: Both --api-url and --token are required for CLI mode")
-        print("Run without arguments for interactive mode")
-        sys.exit(1)
-
-    # ========================================
-    # Interactive Mode
-    # ========================================
-    print("\n" + "=" * 60)
-    print("PP-OCRv5 API Skill - Configuration Wizard")
-    print("=" * 60)
-    print("\nCopy your configuration from Paddle AI Studio\n")
-
-    # ========================================
-    # Step 1: Get API URL
-    # ========================================
-    print("[Step 1/2] Please enter your API URL")
-    print('Tip: You can paste directly, for example:')
-    print('  API_URL = "https://33a9f5e7p5r9bek9.aistudio-app.com/ocr"')
-    print('  or: https://33a9f5e7p5r9bek9.aistudio-app.com/ocr')
-    print()
-
-    while True:
-        api_url_input = input("> ").strip()
-
-        if not api_url_input:
-            print("Error: API URL cannot be empty, please enter again")
-            continue
-
-        # Parse input
-        api_url_raw = parse_input(api_url_input)
-
-        # Normalize
-        try:
-            api_url = normalize_api_url(api_url_raw)
-            print(f"✓ Recognized: {api_url}\n")
-            break
-        except Exception as e:
-            print(f"Error: Cannot parse API URL: {e}")
-            print("Please enter again\n")
-
-    # ========================================
-    # Step 2: Get Token
-    # ========================================
-    print("[Step 2/2] Please enter your Access Token")
-    print('Tip: You can paste directly, for example:')
-    print('  TOKEN = "1505a1bd17e9b74004f9c7e54e0d707ddddca7dc"')
-    print('  or: 1505a1bd17e9b74004f9c7e54e0d707ddddca7dc')
-    print()
-
-    while True:
-        token_input = input("> ").strip()
-
-        if not token_input:
-            print("Error: Token cannot be empty, please enter again")
-            continue
-
-        # Parse input
-        token = parse_input(token_input)
-
-        if len(token) < 16:
-            print("⚠ Token length seems too short, please confirm if correct")
-            confirm = input("Continue? [y/N]: ").strip().lower()
-            if confirm != 'y':
-                continue
-
-        print(f"✓ Recognized: {mask_token(token)}\n")
-        break
-
-    # ========================================
-    # Save configuration
-    # ========================================
-    print("=" * 60)
-    print("Saving configuration...")
-    print("=" * 60)
-
-    if not save_config(api_url, token, project_root):
-        sys.exit(1)
-
-    # ========================================
-    # Test connection (optional)
-    # ========================================
-    print("\n" + "=" * 60)
-    test_choice = input("Test connection? [Y/n]: ").strip().lower()
-
-    if test_choice != 'n':
-        success = test_connection(api_url, token)
-        if not success:
-            print("\n⚠ Connection test failed, but configuration has been saved")
-            print("  Please check if API URL and Token are correct")
-
-    # ========================================
-    # Complete
-    # ========================================
-    print("\n" + "=" * 60)
-    print("Configuration complete!")
-    print("=" * 60)
-    print("\nYou can now use the OCR function:")
-    print(f"  cd {project_root}")
-    print('  python scripts/ocr_caller.py --file-url "https://example.com/image.jpg"')
-    print("\nTo reconfigure, run this script again.")
-    print("=" * 60 + "\n")
-
-
-if __name__ == "__main__":
-    try:
-        main()
-    except KeyboardInterrupt:
-        print("\n\nConfiguration cancelled")
-        sys.exit(0)
-    except Exception as e:
-        print(f"\n\nError: {e}")
-        sys.exit(1)
+#!/usr/bin/env python3
+"""
+Configuration Wizard for PP-OCRv5 API Skill
+
+Supports two modes:
+1. Interactive mode (default): python configure.py
+2. CLI mode: python configure.py --api-url URL --token TOKEN
+
+Supports pasting Python code format (e.g., API_URL = "...")
+"""
+
+import argparse
+import re
+import sys
+from pathlib import Path
+
+# Add parent directory to path for imports
+sys.path.insert(0, str(Path(__file__).parent.parent))
+
+
+def parse_input(user_input: str) -> str:
+    """
+    Intelligently parse user input, supporting multiple formats:
+    - API_URL = "https://..."
+    - "https://..."
+    - https://...
+    - TOKEN = "abc123..."
+
+    Returns the extracted value
+    """
+    user_input = user_input.strip()
+
+    # Format 1: KEY = "value" or KEY = 'value'
+    match = re.match(r'^\w+\s*=\s*["\'](.+?)["\']$', user_input)
+    if match:
+        return match.group(1)
+
+    # Format 2: "value" or 'value'
+    match = re.match(r'^["\'](.+?)["\']$', user_input)
+    if match:
+        return match.group(1)
+
+    # Format 3: value (direct input)
+    return user_input
+
+
+def normalize_api_url(url: str) -> str:
+    """
+    Normalize API URL
+    Supports:
+    - https://xxx.aistudio-app.com/ocr
+    - https://xxx.aistudio-app.com
+    - xxx.aistudio-app.com
+
+    Returns: https://xxx.aistudio-app.com/ocr
+    """
+    url = url.strip()
+
+    # Remove http:// or https://
+    url = re.sub(r'^https?://', '', url)
+
+    # Remove trailing /ocr or other paths
+    url = re.sub(r'/.*$', '', url)
+
+    # Return complete API URL
+    return f"https://{url}/ocr"
+
+
+def mask_token(token: str) -> str:
+    """Mask token, only show first and last parts"""
+    if len(token) <= 8:
+        return "****"
+    return f"{token[:4]}...{token[-4:]}"
+
+
+def test_connection(api_url: str, token: str) -> bool:
+    """Test API connection (optional)"""
+    try:
+        import httpx
+
+        print("\nTesting connection...")
+
+        # Simple test request (using a small base64 image)
+        test_payload = {
+            "file": "iVBORw0KGgoAAAANSUhEUgAAAAEAAAABCAYAAAAfFcSJAAAADUlEQVR42mNk+M9QDwADhgGAWjR9awAAAABJRU5ErkJggg==",
+            "fileType": 1,
+            "visualize": False
+        }
+
+        headers = {
+            "Authorization": f"token {token}",
+            "Content-Type": "application/json"
+        }
+
+        client = httpx.Client(timeout=10.0)
+        try:
+            resp = client.post(api_url, json=test_payload, headers=headers)
+            resp_json = resp.json()
+
+            if resp.status_code == 200 and resp_json.get("errorCode") == 0:
+                print("✓ API connection successful!")
+                print("✓ OCR function is working!")
+                return True
+            elif resp.status_code == 403:
+                print("✗ Token verification failed, please check if the Token is correct")
+                return False
+            elif resp.status_code == 429:
+                print("⚠ API quota exhausted, but connection is working")
+                return True
+            else:
+                print(f"⚠ API returned error: {resp_json.get('errorMsg', 'Unknown error')}")
+                return False
+        finally:
+            client.close()
+
+    except ImportError:
+        print("⚠ httpx not installed, skipping connection test")
+        print("   Install with: pip install httpx")
+        return True
+    except Exception as e:
+        print(f"✗ Connection test failed: {e}")
+        return False
+
+
+def save_config(api_url: str, token: str, project_root: Path, quiet: bool = False) -> bool:
+    """
+    Save configuration to .env file
+
+    Args:
+        api_url: Normalized API URL
+        token: Access token
+        project_root: Project root directory
+        quiet: If True, suppress output messages
+
+    Returns:
+        True if successful, False otherwise
+    """
+    env_file = project_root / ".env"
+
+    # Read existing configuration (if exists)
+    existing_config = {}
+    if env_file.exists():
+        if not quiet:
+            print(f"\nDetected existing configuration file: {env_file}")
+            overwrite = input("Overwrite? [Y/n]: ").strip().lower()
+            if overwrite == 'n':
+                print("Configuration cancelled")
+                return False
+
+        # Preserve other configuration items
+        with open(env_file, 'r', encoding='utf-8') as f:
+            for line in f:
+                line = line.strip()
+                if line and not line.startswith('#'):
+                    if '=' in line and not line.startswith('API_URL=') and not line.startswith('PADDLE_OCR_TOKEN='):
+                        key, value = line.split('=', 1)
+                        existing_config[key.strip()] = value.strip()
+
+    # Write new configuration
+    try:
+        with open(env_file, 'w', encoding='utf-8') as f:
+            f.write("# PP-OCRv5 API Configuration\n")
+            f.write("# This file was automatically generated by configure.py\n\n")
+
+            f.write(f"API_URL={api_url}\n")
+            f.write(f"PADDLE_OCR_TOKEN={token}\n")
+
+            # Write other preserved configurations
+            if existing_config:
+                f.write("\n# Other configurations\n")
+                for key, value in existing_config.items():
+                    f.write(f"{key}={value}\n")
+
+        if not quiet:
+            print(f"\n✓ Configuration saved to: {env_file}")
+        return True
+
+    except Exception as e:
+        print(f"\n✗ Failed to save configuration: {e}")
+        return False
+
+
+def main():
+    # Parse command-line arguments
+    parser = argparse.ArgumentParser(
+        description='PP-OCRv5 API Configuration Tool',
+        formatter_class=argparse.RawDescriptionHelpFormatter,
+        epilog="""
+Examples:
+  # Interactive mode
+  python configure.py
+
+  # CLI mode (non-interactive)
+  python configure.py --api-url "https://xxx.aistudio-app.com/ocr" --token "your_token"
+        """
+    )
+    parser.add_argument('--api-url', help='API URL (non-interactive mode)')
+    parser.add_argument('--token', help='Access token (non-interactive mode)')
+    parser.add_argument('--quiet', action='store_true', help='Suppress output messages')
+
+    args = parser.parse_args()
+
+    # Get project root directory (parent of scripts directory)
+    project_root = Path(__file__).parent.parent
+
+    # ========================================
+    # CLI Mode (non-interactive)
+    # ========================================
+    if args.api_url and args.token:
+        try:
+            # Normalize API URL
+            api_url = normalize_api_url(parse_input(args.api_url))
+            token = parse_input(args.token)
+
+            # Validate
+            if len(token) < 16:
+                print("Error: Token seems too short. Please check and try again.")
+                sys.exit(1)
+
+            # Save configuration
+            if save_config(api_url, token, project_root, quiet=args.quiet):
+                if not args.quiet:
+                    print("\n✓ Configuration complete!")
+                    print(f"  API_URL: {api_url}")
+                    print(f"  TOKEN: {mask_token(token)}")
+                sys.exit(0)
+            else:
+                sys.exit(1)
+
+        except Exception as e:
+            print(f"Error: {e}")
+            sys.exit(1)
+
+    elif args.api_url or args.token:
+        print("Error: Both --api-url and --token are required for CLI mode")
+        print("Run without arguments for interactive mode")
+        sys.exit(1)
+
+    # ========================================
+    # Interactive Mode
+    # ========================================
+    print("\n" + "=" * 60)
+    print("PP-OCRv5 API Skill - Configuration Wizard")
+    print("=" * 60)
+    print("\nCopy your configuration from Paddle AI Studio\n")
+
+    # ========================================
+    # Step 1: Get API URL
+    # ========================================
+    print("[Step 1/2] Please enter your API URL")
+    print('Tip: You can paste directly, for example:')
+    print('  API_URL = "https://33a9f5e7p5r9bek9.aistudio-app.com/ocr"')
+    print('  or: https://33a9f5e7p5r9bek9.aistudio-app.com/ocr')
+    print()
+
+    while True:
+        api_url_input = input("> ").strip()
+
+        if not api_url_input:
+            print("Error: API URL cannot be empty, please enter again")
+            continue
+
+        # Parse input
+        api_url_raw = parse_input(api_url_input)
+
+        # Normalize
+        try:
+            api_url = normalize_api_url(api_url_raw)
+            print(f"✓ Recognized: {api_url}\n")
+            break
+        except Exception as e:
+            print(f"Error: Cannot parse API URL: {e}")
+            print("Please enter again\n")
+
+    # ========================================
+    # Step 2: Get Token
+    # ========================================
+    print("[Step 2/2] Please enter your Access Token")
+    print('Tip: You can paste directly, for example:')
+    print('  TOKEN = "1505a1bd17e9b74004f9c7e54e0d707ddddca7dc"')
+    print('  or: 1505a1bd17e9b74004f9c7e54e0d707ddddca7dc')
+    print()
+
+    while True:
+        token_input = input("> ").strip()
+
+        if not token_input:
+            print("Error: Token cannot be empty, please enter again")
+            continue
+
+        # Parse input
+        token = parse_input(token_input)
+
+        if len(token) < 16:
+            print("⚠ Token length seems too short, please confirm if correct")
+            confirm = input("Continue? [y/N]: ").strip().lower()
+            if confirm != 'y':
+                continue
+
+        print(f"✓ Recognized: {mask_token(token)}\n")
+        break
+
+    # ========================================
+    # Save configuration
+    # ========================================
+    print("=" * 60)
+    print("Saving configuration...")
+    print("=" * 60)
+
+    if not save_config(api_url, token, project_root):
+        sys.exit(1)
+
+    # ========================================
+    # Test connection (optional)
+    # ========================================
+    print("\n" + "=" * 60)
+    test_choice = input("Test connection? [Y/n]: ").strip().lower()
+
+    if test_choice != 'n':
+        success = test_connection(api_url, token)
+        if not success:
+            print("\n⚠ Connection test failed, but configuration has been saved")
+            print("  Please check if API URL and Token are correct")
+
+    # ========================================
+    # Complete
+    # ========================================
+    print("\n" + "=" * 60)
+    print("Configuration complete!")
+    print("=" * 60)
+    print("\nYou can now use the OCR function:")
+    print(f"  cd {project_root}")
+    print('  python scripts/ocr_caller.py --file-url "https://example.com/image.jpg"')
+    print("\nTo reconfigure, run this script again.")
+    print("=" * 60 + "\n")
+
+
+if __name__ == "__main__":
+    try:
+        main()
+    except KeyboardInterrupt:
+        print("\n\nConfiguration cancelled")
+        sys.exit(0)
+    except Exception as e:
+        print(f"\n\nError: {e}")
+        sys.exit(1)