增加web_api

src/gasflux/blueprints/upload.py

@@ -0,0 +1,134 @@
+"""
+Upload Blueprint
+Handles file upload and processing initiation endpoints.
+"""
+
+import uuid
+import threading
+from pathlib import Path
+from flask import Blueprint, request, current_app
+from werkzeug.utils import secure_filename
+import yaml
+from io import BytesIO
+
+
+from ..app import process_data_async
+from ..shared import _format_response, log_performance, logger, ALLOWED_DATA_EXTENSIONS, ALLOWED_CONFIG_EXTENSIONS, allowed_file,update_task_status, TASK_STATUS_PENDING, TASK_STATUS_FAILED
+
+# Create blueprint
+upload_bp = Blueprint('upload', __name__, url_prefix='/upload')
+
+
+@upload_bp.route('', methods=['POST'])
+@log_performance
+def upload_file():
+    logger.info("Received upload request")
+    logger.info(f"Request content length: {request.content_length} bytes")
+
+    # Check if data file is present
+    if 'file' not in request.files:
+        logger.warning("Upload failed: No data file part in request")
+        return _format_response(400, "未找到数据文件部分")
+
+    data_file = request.files['file']
+    config_file = request.files.get('config')
+
+    # Log file details
+    logger.info(f"Data file: {data_file.filename} (size: {getattr(data_file, 'content_length', 'unknown')} bytes)")
+    if config_file:
+        logger.info(f"Config file: {config_file.filename} (size: {getattr(config_file, 'content_length', 'unknown')} bytes)")
+    else:
+        logger.info("No custom config file provided, will use default")
+
+    if data_file.filename == '':
+        logger.warning("Upload failed: No data file selected (empty filename)")
+        return _format_response(400, "未选择数据文件")
+
+    if not allowed_file(data_file.filename, ALLOWED_DATA_EXTENSIONS):
+        logger.warning(f"Upload failed: Invalid data file type {data_file.filename} - allowed: {ALLOWED_DATA_EXTENSIONS}")
+        return _format_response(400, "无效的数据文件类型。只允许 .xlsx 和 .xls 格式。")
+
+    # Generate unique job ID
+    job_id = str(uuid.uuid4())
+    logger.info(f"Generated job ID: {job_id}")
+
+    # 1) Parse config content (parse in memory without saving first)
+    if config_file and config_file.filename != '':
+        if not allowed_file(config_file.filename, ALLOWED_CONFIG_EXTENSIONS):
+            return _format_response(400, "无效的配置文件类型。只允许 .yaml 和 .yml 格式。")
+        config_file.stream.seek(0)
+        config_text = config_file.read().decode('utf-8', errors='ignore')
+        try:
+            active_config = yaml.safe_load(config_text)
+        except Exception:
+            return _format_response(400, "配置文件解析失败")
+        # Reset stream for saving
+        config_file.stream = BytesIO(config_text.encode('utf-8'))
+    else:
+        default_config_path = Path(__file__).parent.parent / "gasflux_config.yaml"
+        with open(default_config_path, 'r', encoding='utf-8') as f:
+            active_config = yaml.safe_load(f)
+
+    # 2) Create job directories based on config['output_dir']
+    output_base = Path(active_config['output_dir']).expanduser()
+    job_upload_dir = output_base / "uploads" / job_id
+    job_output_dir = output_base / "outputs" / job_id
+    job_upload_dir.mkdir(parents=True, exist_ok=True)
+    job_output_dir.mkdir(parents=True, exist_ok=True)
+    logger.info(f"Job {job_id}: Created directories - Upload: {job_upload_dir}, Output: {job_output_dir}")
+
+    # 3) Save data file to job_upload_dir
+    data_filename = secure_filename(data_file.filename)
+    data_path = job_upload_dir / data_filename
+    try:
+        data_file.seek(0)
+        data_file.save(str(data_path))
+        logger.info(f"Job {job_id}: Data file saved successfully - Path: {data_path}")
+    except Exception as e:
+        logger.error(f"Job {job_id}: Failed to save data file {data_filename}: {str(e)}")
+        return _format_response(500, "保存数据文件失败")
+
+    # 4) Save config file to job_upload_dir
+    if config_file and config_file.filename != '':
+        config_filename = secure_filename(config_file.filename)
+        config_path = job_upload_dir / config_filename
+        try:
+            config_file.seek(0)
+            config_file.save(str(config_path))
+            active_config_path = config_path
+            logger.info(f"Job {job_id}: Custom config saved successfully - Path: {config_path}")
+        except Exception as e:
+            logger.error(f"Job {job_id}: Failed to save config file {config_filename}: {str(e)}")
+            return _format_response(500, "保存配置文件失败")
+    else:
+        # Copy default config for record keeping
+        config_path = job_upload_dir / "config.yaml"
+        with open(config_path, 'w', encoding='utf-8') as f:
+            yaml.safe_dump(active_config, f, allow_unicode=True)
+        active_config_path = config_path
+        logger.info(f"Job {job_id}: Default config saved for record - Path: {config_path}")
+
+    # Initialize task status
+    update_task_status(job_id, TASK_STATUS_PENDING, "Task queued for processing")
+    logger.info(f"Job {job_id}: Task status initialized as PENDING")
+
+    # Start background processing
+    try:
+        thread = threading.Thread(
+            target=process_data_async,
+            args=(job_id, data_path, active_config_path, job_output_dir)
+        )
+        thread.daemon = True
+        thread.start()
+        logger.info(f"Job {job_id}: Background processing thread started successfully")
+    except Exception as e:
+        logger.error(f"Job {job_id}: Failed to start background processing thread: {str(e)}")
+        update_task_status(job_id, TASK_STATUS_FAILED, error=str(e))
+        return _format_response(500, "启动处理失败")
+
+    logger.info(f"Job {job_id}: Upload process completed successfully, returning job ID to client")
+    return _format_response(202, "任务已接受并加入处理队列", {
+        "status": "accepted",
+        "job_id": job_id,
+        "task_status_url": f"/task/{job_id}"
+    })