| 1 | import json |
| 2 | import logging |
| 3 | import os |
| 4 | import shutil |
| 5 | import threading |
| 6 | import time |
| 7 | import uuid |
| 8 | from datetime import datetime |
| 9 | from typing import Any, Dict, Optional |
| 10 | |
| 11 | from config import settings |
| 12 | from .events import publish_task_event |
| 13 | |
| 14 | logger = logging.getLogger(__name__) |
| 15 | _task_store_lock = threading.RLock() |
| 16 | |
| 17 | TASK_DATA_DIR = os.path.join(settings.CODE_DIR, "data", "tasks") |
| 18 | TASK_RESULT_DIR = os.path.join(settings.RESULT_DIR, "task") |
| 19 | |
| 20 | |
| 21 | def ensure_task_dirs() -> None: |
| 22 | os.makedirs(TASK_DATA_DIR, exist_ok=True) |
| 23 | os.makedirs(TASK_RESULT_DIR, exist_ok=True) |
| 24 | |
| 25 | |
| 26 | def new_task_id() -> str: |
| 27 | return f"{datetime.now().strftime('%Y%m%d_%H%M%S')}_{uuid.uuid4().hex[:8]}" |
| 28 | |
| 29 | |
| 30 | def task_metadata_path(task_id: str) -> str: |
| 31 | return os.path.join(TASK_DATA_DIR, f"{task_id}.json") |
| 32 | |
| 33 | |
| 34 | def task_output_dir(task_id: str) -> str: |
| 35 | return os.path.join(TASK_RESULT_DIR, task_id) |
| 36 | |
| 37 | |
| 38 | def now_iso() -> str: |
| 39 | return datetime.now().isoformat() |
| 40 | |
| 41 | |
| 42 | def save_task(metadata: Dict[str, Any]) -> None: |
| 43 | with _task_store_lock: |
| 44 | ensure_task_dirs() |
| 45 | path = task_metadata_path(metadata["task_id"]) |
| 46 | tmp_path = f"{path}.{uuid.uuid4().hex}.tmp" |
| 47 | with open(tmp_path, "w", encoding="utf-8") as f: |
| 48 | json.dump(metadata, f, ensure_ascii=False, indent=2) |
| 49 | os.replace(tmp_path, path) |
| 50 | |
| 51 | |
| 52 | def load_task(task_id: str) -> Optional[Dict[str, Any]]: |
| 53 | with _task_store_lock: |
| 54 | path = task_metadata_path(task_id) |
| 55 | if not os.path.exists(path): |
| 56 | return None |
| 57 | with open(path, "r", encoding="utf-8") as f: |
| 58 | return json.load(f) |
| 59 | |
| 60 | |
| 61 | def delete_task(task_id: str) -> bool: |
| 62 | with _task_store_lock: |
| 63 | metadata = load_task(task_id) |
| 64 | if not metadata: |
| 65 | return False |
| 66 | |
| 67 | metadata_path = task_metadata_path(task_id) |
| 68 | output_dir = metadata.get("output_dir") or task_output_dir(task_id) |
| 69 | if os.path.exists(metadata_path): |
| 70 | os.remove(metadata_path) |
| 71 | if output_dir and os.path.exists(output_dir): |
| 72 | shutil.rmtree(output_dir) |
| 73 | logger.info("Deleted pipeline task: task_id=%s output_dir=%s", task_id, output_dir) |
| 74 | return True |
| 75 | |
| 76 | |
| 77 | def list_tasks(limit: int = 100) -> list[Dict[str, Any]]: |
| 78 | with _task_store_lock: |
| 79 | ensure_task_dirs() |
| 80 | records = [] |
| 81 | for filename in os.listdir(TASK_DATA_DIR): |
| 82 | if not filename.endswith(".json"): |
| 83 | continue |
| 84 | try: |
| 85 | with open(os.path.join(TASK_DATA_DIR, filename), "r", encoding="utf-8") as f: |
| 86 | records.append(json.load(f)) |
| 87 | except Exception: |
| 88 | continue |
| 89 | records.sort(key=lambda item: item.get("created_at", ""), reverse=True) |
| 90 | return records[:limit] |
| 91 | |
| 92 | |
| 93 | def create_task(pipeline: str, input_params: Dict[str, Any]) -> Dict[str, Any]: |
| 94 | with _task_store_lock: |
| 95 | ensure_task_dirs() |
| 96 | task_id = new_task_id() |
| 97 | output_dir = task_output_dir(task_id) |
| 98 | os.makedirs(output_dir, exist_ok=True) |
| 99 | metadata = { |
| 100 | "task_id": task_id, |
| 101 | "pipeline": pipeline, |
| 102 | "status": "pending", |
| 103 | "progress": 0, |
| 104 | "message": "Task created", |
| 105 | "input": input_params, |
| 106 | "output": {}, |
| 107 | "artifacts": [], |
| 108 | "error": None, |
| 109 | "created_at": now_iso(), |
| 110 | "updated_at": now_iso(), |
| 111 | "started_at": None, |
| 112 | "completed_at": None, |
| 113 | "duration_seconds": None, |
| 114 | "output_dir": output_dir, |
| 115 | } |
| 116 | save_task(metadata) |
| 117 | logger.info("Created pipeline task: task_id=%s pipeline=%s output_dir=%s", task_id, pipeline, output_dir) |
| 118 | return metadata |
| 119 | |
| 120 | |
| 121 | def update_task(task_id: str, **updates: Any) -> Dict[str, Any]: |
| 122 | with _task_store_lock: |
| 123 | metadata = load_task(task_id) |
| 124 | if not metadata: |
| 125 | raise FileNotFoundError(f"Task not found: {task_id}") |
| 126 | metadata.update(updates) |
| 127 | metadata["updated_at"] = now_iso() |
| 128 | save_task(metadata) |
| 129 | if "progress" in updates or "status" in updates: |
| 130 | logger.info( |
| 131 | "Task update: task_id=%s status=%s progress=%s message=%s", |
| 132 | task_id, |
| 133 | metadata.get("status"), |
| 134 | metadata.get("progress"), |
| 135 | metadata.get("message"), |
| 136 | ) |
| 137 | publish_task_event(task_id, { |
| 138 | "type": "progress", |
| 139 | "status": metadata.get("status"), |
| 140 | "progress": metadata.get("progress", 0), |
| 141 | }) |
| 142 | return metadata |
| 143 | |
| 144 | |
| 145 | def append_artifact(task_id: str, new_artifact: Dict[str, Any]) -> Dict[str, Any]: |
| 146 | with _task_store_lock: |
| 147 | metadata = load_task(task_id) |
| 148 | if not metadata: |
| 149 | raise FileNotFoundError(f"Task not found: {task_id}") |
| 150 | |
| 151 | if not new_artifact.get("created_at"): |
| 152 | new_artifact = {**new_artifact, "created_at": now_iso()} |
| 153 | |
| 154 | artifacts = list(metadata.get("artifacts") or []) |
| 155 | key = (new_artifact.get("kind"), new_artifact.get("name"), new_artifact.get("path")) |
| 156 | if not any((item.get("kind"), item.get("name"), item.get("path")) == key for item in artifacts): |
| 157 | artifacts.append(new_artifact) |
| 158 | logger.info( |
| 159 | "Task artifact: task_id=%s kind=%s name=%s path=%s", |
| 160 | task_id, |
| 161 | new_artifact.get("kind"), |
| 162 | new_artifact.get("name"), |
| 163 | new_artifact.get("path"), |
| 164 | ) |
| 165 | metadata["artifacts"] = artifacts |
| 166 | metadata["updated_at"] = now_iso() |
| 167 | save_task(metadata) |
| 168 | publish_task_event(task_id, { |
| 169 | "type": "artifact", |
| 170 | "status": metadata.get("status"), |
| 171 | "progress": metadata.get("progress", 0), |
| 172 | "artifact": new_artifact, |
| 173 | }) |
| 174 | return metadata |
| 175 | |
| 176 | |
| 177 | def mark_running(task_id: str) -> Dict[str, Any]: |
| 178 | return update_task(task_id, status="running", progress=1, message="Task running", started_at=now_iso()) |
| 179 | |
| 180 | |
| 181 | def mark_completed(task_id: str, output: Dict[str, Any], artifacts: list[Dict[str, Any]]) -> Dict[str, Any]: |
| 182 | metadata = load_task(task_id) or {"task_id": task_id} |
| 183 | started_at = metadata.get("started_at") |
| 184 | duration = None |
| 185 | if started_at: |
| 186 | try: |
| 187 | duration = time.time() - datetime.fromisoformat(started_at).timestamp() |
| 188 | except Exception: |
| 189 | duration = None |
| 190 | existing_artifacts = list(metadata.get("artifacts") or []) |
| 191 | merged_artifacts = list(existing_artifacts) |
| 192 | seen = { |
| 193 | (item.get("kind"), item.get("name"), item.get("path")) |
| 194 | for item in merged_artifacts |
| 195 | } |
| 196 | for item in artifacts or []: |
| 197 | key = (item.get("kind"), item.get("name"), item.get("path")) |
| 198 | if key in seen: |
| 199 | continue |
| 200 | merged_artifacts.append({**item, "created_at": item.get("created_at") or now_iso()}) |
| 201 | seen.add(key) |
| 202 | |
| 203 | metadata = update_task( |
| 204 | task_id, |
| 205 | status="completed", |
| 206 | progress=100, |
| 207 | message="Task completed", |
| 208 | output=output, |
| 209 | artifacts=merged_artifacts, |
| 210 | error=None, |
| 211 | completed_at=now_iso(), |
| 212 | duration_seconds=duration, |
| 213 | ) |
| 214 | publish_task_event(task_id, { |
| 215 | "type": "completed", |
| 216 | "status": "completed", |
| 217 | "progress": 100, |
| 218 | }) |
| 219 | return metadata |
| 220 | |
| 221 | |
| 222 | def mark_failed(task_id: str, error: str) -> Dict[str, Any]: |
| 223 | metadata = update_task( |
| 224 | task_id, |
| 225 | status="failed", |
| 226 | message="Task failed", |
| 227 | error=error, |
| 228 | completed_at=now_iso(), |
| 229 | ) |
| 230 | publish_task_event(task_id, { |
| 231 | "type": "failed", |
| 232 | "status": "failed", |
| 233 | "progress": metadata.get("progress", 0), |
| 234 | }) |
| 235 | return metadata |
| 236 |