aiecs 1.5.1__py3-none-any.whl

aiecs/scripts/dependance_check/setup_nlp_data.sh

@@ -0,0 +1,217 @@
+#!/bin/bash
+# Setup NLP Data for AIECS ClassifierTool
+# This script downloads required NLTK and spaCy data packages
+
+set -e  # Exit on any error
+
+SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
+PROJECT_ROOT="$(cd "$SCRIPT_DIR/../.." && pwd)"
+
+echo "=================================================="
+echo "AIECS NLP Data Setup"
+echo "=================================================="
+echo "This script will download required NLP data for:"
+echo "  - NLTK stopwords corpus"
+echo "  - spaCy English model (en_core_web_sm)"
+echo "  - spaCy Chinese model (zh_core_web_sm)"
+echo "=================================================="
+echo
+
+# Function to check if a command exists
+command_exists() {
+    command -v "$1" >/dev/null 2>&1
+}
+
+# Function to activate virtual environment if it exists
+activate_venv() {
+    if [[ -n "$VIRTUAL_ENV" ]]; then
+        echo "✅ Virtual environment already active: $VIRTUAL_ENV"
+        return 0
+    fi
+    
+    # Check for common virtual environment locations
+    local venv_paths=(
+        "$PROJECT_ROOT/venv"
+        "$PROJECT_ROOT/.venv" 
+        "$PROJECT_ROOT/env"
+        "$PROJECT_ROOT/.env"
+    )
+    
+    for venv_path in "${venv_paths[@]}"; do
+        if [[ -f "$venv_path/bin/activate" ]]; then
+            echo "📦 Activating virtual environment: $venv_path"
+            source "$venv_path/bin/activate"
+            return 0
+        fi
+    done
+    
+    echo "⚠️  No virtual environment found. Using system Python."
+    return 1
+}
+
+# Function to check Python and packages
+check_dependencies() {
+    echo "🔍 Checking dependencies..."
+    
+    if ! command_exists python3; then
+        echo "❌ Python 3 is not installed or not in PATH"
+        exit 1
+    fi
+    
+    local python_version=$(python3 --version 2>&1 | cut -d' ' -f2)
+    echo "✅ Python version: $python_version"
+    
+    # Check if we're in the project directory and can import aiecs
+    cd "$PROJECT_ROOT"
+    if python3 -c "import aiecs" 2>/dev/null; then
+        echo "✅ AIECS package is available"
+    else
+        echo "⚠️  AIECS package not found. You may need to install it first:"
+        echo "    pip install -e ."
+    fi
+}
+
+# Function to run the Python download script
+run_download_script() {
+    echo
+    echo "🚀 Starting NLP data download..."
+    echo
+    
+    cd "$PROJECT_ROOT"
+    
+    # Try multiple ways to run the script
+    if python3 -m aiecs.scripts.download_nlp_data; then
+        echo "✅ NLP data download completed successfully!"
+        return 0
+    elif python3 aiecs/scripts/download_nlp_data.py; then
+        echo "✅ NLP data download completed successfully!"
+        return 0
+    else
+        echo "❌ Failed to run NLP data download script"
+        return 1
+    fi
+}
+
+# Function to verify installation
+verify_installation() {
+    echo
+    echo "🔍 Verifying NLP data installation..."
+    
+    # Test NLTK
+    if python3 -c "
+import nltk
+from nltk.corpus import stopwords
+stopwords.words('english')
+print('✅ NLTK stopwords data available')
+" 2>/dev/null; then
+        echo "✅ NLTK verification passed"
+    else
+        echo "⚠️  NLTK verification failed"
+    fi
+    
+    # Test spaCy English model
+    if python3 -c "
+import spacy
+nlp = spacy.load('en_core_web_sm')
+doc = nlp('This is a test.')
+print('✅ spaCy English model available')
+" 2>/dev/null; then
+        echo "✅ spaCy English model verification passed"
+    else
+        echo "⚠️  spaCy English model verification failed"
+    fi
+    
+    # Test spaCy Chinese model (optional)
+    if python3 -c "
+import spacy
+nlp = spacy.load('zh_core_web_sm')
+doc = nlp('这是测试。')
+print('✅ spaCy Chinese model available')
+" 2>/dev/null; then
+        echo "✅ spaCy Chinese model verification passed"
+    else
+        echo "⚠️  spaCy Chinese model not available (optional)"
+    fi
+}
+
+# Function to display usage information
+show_usage() {
+    echo "Usage: $0 [OPTIONS]"
+    echo
+    echo "Options:"
+    echo "  -h, --help     Show this help message"
+    echo "  -v, --verify   Only verify existing installations"
+    echo "  --no-venv      Skip virtual environment activation"
+    echo
+    echo "This script downloads required NLP data for AIECS ClassifierTool:"
+    echo "  - NLTK stopwords corpus"
+    echo "  - spaCy English model (en_core_web_sm)"
+    echo "  - spaCy Chinese model (zh_core_web_sm, optional)"
+    echo
+}
+
+# Parse command line arguments
+VERIFY_ONLY=false
+SKIP_VENV=false
+
+while [[ $# -gt 0 ]]; do
+    case $1 in
+        -h|--help)
+            show_usage
+            exit 0
+            ;;
+        -v|--verify)
+            VERIFY_ONLY=true
+            shift
+            ;;
+        --no-venv)
+            SKIP_VENV=true
+            shift
+            ;;
+        *)
+            echo "Unknown option: $1"
+            show_usage
+            exit 1
+            ;;
+    esac
+done
+
+# Main execution
+main() {
+    echo "📍 Project root: $PROJECT_ROOT"
+    echo
+    
+    # Activate virtual environment if available and not skipped
+    if [[ "$SKIP_VENV" != true ]]; then
+        activate_venv || true
+    fi
+    
+    # Check dependencies
+    check_dependencies
+    
+    if [[ "$VERIFY_ONLY" == true ]]; then
+        echo
+        echo "🔍 Running verification only..."
+        verify_installation
+    else
+        # Download NLP data
+        if run_download_script; then
+            verify_installation
+            echo
+            echo "🎉 NLP data setup completed successfully!"
+            echo "AIECS ClassifierTool is ready to use."
+        else
+            echo
+            echo "❌ NLP data setup failed. Please check the errors above."
+            exit 1
+        fi
+    fi
+    
+    echo
+    echo "=================================================="
+    echo "Setup complete!"
+    echo "=================================================="
+}
+
+# Run main function
+main "$@"

aiecs/scripts/dependance_patch/__init__.py

@@ -0,0 +1,7 @@
+"""
+依赖补丁工具
+
+提供各种依赖库的补丁修复功能。
+"""
+
+__all__ = []

aiecs/scripts/dependance_patch/fix_weasel/README_WEASEL_PATCH.md

@@ -0,0 +1,126 @@
+# Weasel Library Patch Scripts
+
+这个目录包含用于修复weasel库重复验证器函数错误的脚本，该错误导致测试失败。
+
+## 问题描述
+
+错误发生是因为weasel库中有重复的验证器函数，但没有`allow_reuse=True`参数：
+
+```
+pydantic.v1.errors.ConfigError: duplicate validator function "weasel.schemas.ProjectConfigSchema.check_legacy_keys"; if this is intended, set `allow_reuse=True`
+```
+
+**注意**: 实际的问题是在 `@root_validator` 装饰器中，而不是 `@validator`。
+
+## 可用脚本
+
+### 1. `run_weasel_patch.sh` (推荐使用)
+最简单的解决方案，通过poetry运行补丁：
+- 自动检测poetry虚拟环境
+- 在正确的环境中运行Python补丁脚本
+- 提供清晰的反馈和说明
+
+**使用方法:**
+```bash
+cd python-middleware
+./scripts/run_weasel_patch.sh
+```
+
+### 2. `fix_weasel_validator.py`
+基于Python的方法：
+- 使用正则表达式模式识别和修复验证器装饰器
+- 提供详细的前后对比
+- 创建带时间戳的备份
+
+**使用方法:**
+```bash
+cd python-middleware
+poetry run python3 ./scripts/fix_weasel_validator.py
+```
+
+### 3. `fix_weasel_validator.sh`
+基于bash的方法，使用sed进行文本替换：
+- 简单的sed文本替换
+- 创建备份
+- 显示前后内容
+
+**使用方法:**
+```bash
+cd python-middleware
+./scripts/fix_weasel_validator.sh
+```
+
+### 4. `patch_weasel_library.sh`
+综合解决方案：
+- 结合多种方法
+- 包含测试验证
+- 详细的错误处理
+
+**使用方法:**
+```bash
+cd python-middleware
+./scripts/patch_weasel_library.sh
+```
+
+## 脚本功能
+
+所有脚本都执行以下操作：
+
+1. **定位虚拟环境**: 找到poetry虚拟环境路径
+2. **找到问题文件**: 在site-packages中定位`weasel/schemas.py`
+3. **创建备份**: 制作原始文件的带时间戳备份
+4. **应用补丁**: 向需要的验证器装饰器添加`allow_reuse=True`
+5. **验证修复**: 检查补丁是否正确应用
+
+## 修复内容
+
+脚本将以下代码：
+```python
+@root_validator(pre=True)
+def check_legacy_keys(cls, obj: Dict[str, Any]) -> Dict[str, Any]:
+```
+
+修改为：
+```python
+@root_validator(pre=True, allow_reuse=True)
+def check_legacy_keys(cls, obj: Dict[str, Any]) -> Dict[str, Any]:
+```
+
+脚本同时支持 `@validator` 和 `@root_validator` 装饰器的修复。
+
+## 回滚
+
+如果需要恢复更改，每个脚本都会创建备份文件。您可以使用以下命令恢复：
+```bash
+cp /path/to/backup/file /path/to/original/schemas.py
+```
+
+确切的路径将在脚本输出中显示。
+
+## 故障排除
+
+1. **找不到Poetry**: 确保已安装poetry并且您在项目目录中
+2. **找不到虚拟环境**: 运行`poetry install`创建虚拟环境
+3. **找不到文件**: weasel包可能未安装 - 检查您的依赖项
+4. **权限错误**: 您可能需要使用适当的权限运行
+
+## 运行补丁后
+
+1. 再次尝试运行您的测试：`poetry run pytest`
+2. 如果问题仍然存在，重启您的Python环境/IDE
+3. 修复应该解决重复验证器函数错误
+
+## 注意事项
+
+- 这些脚本可以安全地多次运行
+- 它们在进行更改之前检查补丁是否已应用
+- 所有更改都会自动备份
+- 脚本针对特定问题，不会影响其他功能
+
+## 推荐使用流程
+
+1. 首先尝试使用 `run_weasel_patch.sh`（最简单）
+2. 如果遇到问题，可以尝试直接运行 `poetry run python3 scripts/fix_weasel_validator.py`
+3. 作为最后手段，可以使用 `patch_weasel_library.sh`（最全面）
+
+所有脚本都会在修改前创建备份，因此您可以安全地尝试不同的方法。

aiecs/scripts/dependance_patch/fix_weasel/__init__.py

@@ -0,0 +1,11 @@
+"""
+Weasel 库补丁工具
+
+修复 Weasel 库的验证器问题。
+"""
+
+from .fix_weasel_validator import main as fix_weasel_validator_main
+
+__all__ = [
+    "fix_weasel_validator_main",
+]

aiecs/scripts/dependance_patch/fix_weasel/fix_weasel_validator.py

@@ -0,0 +1,128 @@
+#!/usr/bin/env python3
+"""
+Script to fix weasel library duplicate validator function error.
+This script patches the weasel schemas.py file to add allow_reuse=True to duplicate validators.
+"""
+
+import os
+import sys
+import shutil
+from datetime import datetime
+import re
+
+
+def get_weasel_path():
+    """Get the weasel package path in the current Python environment."""
+    try:
+        import weasel
+        import inspect
+
+        weasel_file = inspect.getfile(weasel)
+        weasel_dir = os.path.dirname(weasel_file)
+        return os.path.join(weasel_dir, "schemas.py")
+    except ImportError:
+        print("❌ Error: weasel package not found")
+        print("Please install aiecs with all dependencies")
+        return None
+    except Exception as e:
+        print(f"❌ Error finding weasel package: {e}")
+        return None
+
+
+def backup_file(file_path):
+    """Create a backup of the file."""
+    timestamp = datetime.now().strftime("%Y%m%d_%H%M%S")
+    backup_path = f"{file_path}.backup.{timestamp}"
+    shutil.copy2(file_path, backup_path)
+    return backup_path
+
+
+def fix_weasel_schemas(schemas_file_path):
+    """Fix the weasel schemas.py file by adding allow_reuse=True to validators."""
+
+    print(f"📁 Processing file: {schemas_file_path}")
+
+    # Read the original file
+    with open(schemas_file_path, "r", encoding="utf-8") as f:
+        content = f.read()
+
+    # Check if already patched
+    if "allow_reuse=True" in content:
+        print("✅ File already patched with allow_reuse=True")
+        return True
+
+    # Create backup
+    backup_path = backup_file(schemas_file_path)
+    print(f"💾 Created backup at: {backup_path}")
+
+    # Show current problematic area
+    lines = content.split("\n")
+    print("\n📖 Current content around line 89:")
+    for i, line in enumerate(lines[84:94], 85):
+        print(f"{i:3d} | {line}")
+
+    # Pattern to match both @validator and @root_validator decorators without
+    # allow_reuse
+    validator_pattern = r"(@(?:root_)?validator\([^)]*)\)(?!\s*,\s*allow_reuse=True)"
+
+    # Replace @validator(...) or @root_validator(...) with allow_reuse=True if
+    # not already present
+    def replace_validator(match):
+        validator_call = match.group(1)
+        # Check if allow_reuse is already in the parameters
+        if "allow_reuse" in validator_call:
+            return match.group(0)  # Return unchanged
+        else:
+            return f"{validator_call}, allow_reuse=True)"
+
+    # Apply the fix
+    fixed_content = re.sub(validator_pattern, replace_validator, content)
+
+    # Write the fixed content back
+    with open(schemas_file_path, "w", encoding="utf-8") as f:
+        f.write(fixed_content)
+
+    # Show the fixed content
+    fixed_lines = fixed_content.split("\n")
+    print("\n📖 Patched content around line 89:")
+    for i, line in enumerate(fixed_lines[84:94], 85):
+        print(f"{i:3d} | {line}")
+
+    # Verify the fix
+    if "allow_reuse=True" in fixed_content:
+        print("✅ Verification successful: allow_reuse=True found in file")
+        return True
+    else:
+        print("⚠️  Warning: allow_reuse=True not found after patching")
+        return False
+
+
+def main():
+    """Main function to execute the patch."""
+    print("🔧 Starting weasel library patch for duplicate validator function...")
+
+    # Get weasel schemas.py path
+    schemas_file = get_weasel_path()
+    if not schemas_file:
+        sys.exit(1)
+
+    print(f"📍 Found weasel schemas.py at: {schemas_file}")
+
+    if not os.path.exists(schemas_file):
+        print(f"❌ Error: weasel schemas.py file not found at {schemas_file}")
+        sys.exit(1)
+
+    # Apply the fix
+    success = fix_weasel_schemas(schemas_file)
+
+    if success:
+        print("\n🎉 Weasel library patch completed successfully!")
+        print("\nYou can now run your tests again.")
+        print("\nIf you need to revert the changes, restore from the backup file.")
+    else:
+        print("\n❌ Patch may not have been applied correctly. Please check manually.")
+        sys.exit(1)
+
+
+if __name__ == "__main__":
+    main()

aiecs/scripts/dependance_patch/fix_weasel/fix_weasel_validator.sh

@@ -0,0 +1,82 @@
+#!/bin/bash
+
+# Script to fix weasel library duplicate validator function error
+# This script patches the weasel schemas.py file to add allow_reuse=True
+
+set -e
+
+echo "🔧 Starting weasel library patch for duplicate validator function..."
+
+# Get the poetry virtual environment path
+VENV_PATH=$(poetry env info --path 2>/dev/null || echo "")
+
+if [ -z "$VENV_PATH" ]; then
+    echo "❌ Error: Could not find poetry virtual environment"
+    echo "Please make sure you're in the project directory and poetry is installed"
+    exit 1
+fi
+
+echo "📍 Found virtual environment at: $VENV_PATH"
+
+# Path to the problematic weasel schemas.py file
+WEASEL_SCHEMAS_FILE="$VENV_PATH/lib/python3.10/site-packages/weasel/schemas.py"
+
+if [ ! -f "$WEASEL_SCHEMAS_FILE" ]; then
+    echo "❌ Error: weasel schemas.py file not found at $WEASEL_SCHEMAS_FILE"
+    exit 1
+fi
+
+echo "📁 Found weasel schemas.py at: $WEASEL_SCHEMAS_FILE"
+
+# Create backup of original file
+BACKUP_FILE="${WEASEL_SCHEMAS_FILE}.backup.$(date +%Y%m%d_%H%M%S)"
+cp "$WEASEL_SCHEMAS_FILE" "$BACKUP_FILE"
+echo "💾 Created backup at: $BACKUP_FILE"
+
+# Check if the file already has allow_reuse=True
+if grep -q "allow_reuse=True" "$WEASEL_SCHEMAS_FILE"; then
+    echo "✅ File already patched with allow_reuse=True"
+    exit 0
+fi
+
+# Apply the patch using sed
+# Look for @validator and @root_validator decorators and add allow_reuse=True if not present
+echo "🔨 Applying patch..."
+
+# First, let's check the current content around the problematic line
+echo "📖 Current content around line 89:"
+sed -n '85,95p' "$WEASEL_SCHEMAS_FILE"
+
+# Apply the patch - add allow_reuse=True to validator decorators that don't have it
+sed -i.tmp '
+/^[[:space:]]*@\(root_\)\?validator(/,/^[[:space:]]*def/ {
+    /^[[:space:]]*@\(root_\)\?validator(/ {
+        # If the line contains @validator or @root_validator but not allow_reuse, add it
+        /allow_reuse/! {
+            s/@\(root_\)\?validator(\([^)]*\))/@\1validator(\2, allow_reuse=True)/
+        }
+    }
+}
+' "$WEASEL_SCHEMAS_FILE"
+
+# Remove the temporary file
+rm -f "${WEASEL_SCHEMAS_FILE}.tmp"
+
+echo "✅ Patch applied successfully!"
+
+# Show the patched content
+echo "📖 Patched content around line 89:"
+sed -n '85,95p' "$WEASEL_SCHEMAS_FILE"
+
+# Verify the patch by checking if allow_reuse=True is now present
+if grep -q "allow_reuse=True" "$WEASEL_SCHEMAS_FILE"; then
+    echo "✅ Verification successful: allow_reuse=True found in file"
+else
+    echo "⚠️  Warning: allow_reuse=True not found after patching"
+fi
+
+echo "🎉 Weasel library patch completed!"
+echo "📝 Backup saved at: $BACKUP_FILE"
+echo ""
+echo "You can now run your tests again. If you need to revert the changes:"
+echo "cp '$BACKUP_FILE' '$WEASEL_SCHEMAS_FILE'"