|
| 1 | +"""Service managing validation prompt library files.""" |
| 2 | + |
| 3 | +from __future__ import annotations |
| 4 | + |
| 5 | +import json |
| 6 | +import logging |
| 7 | +import re |
| 8 | +from dataclasses import dataclass |
| 9 | +from datetime import datetime, timezone |
| 10 | +from pathlib import Path |
| 11 | +from typing import Any, Dict, List, Optional |
| 12 | + |
| 13 | +from fastapi import status |
| 14 | + |
| 15 | +from simpletuner.simpletuner_sdk.server.services.webui_state import WebUIStateStore |
| 16 | +from simpletuner.simpletuner_sdk.server.utils.paths import get_config_directory |
| 17 | + |
| 18 | +logger = logging.getLogger(__name__) |
| 19 | + |
| 20 | + |
| 21 | +class PromptLibraryError(Exception): |
| 22 | + """Domain error for prompt library operations.""" |
| 23 | + |
| 24 | + def __init__(self, message: str, status_code: int = status.HTTP_400_BAD_REQUEST) -> None: |
| 25 | + super().__init__(message) |
| 26 | + self.message = message |
| 27 | + self.status_code = status_code |
| 28 | + |
| 29 | + |
| 30 | +@dataclass |
| 31 | +class PromptLibraryRecord: |
| 32 | + filename: str |
| 33 | + relative_path: str |
| 34 | + absolute_path: str |
| 35 | + display_name: str |
| 36 | + library_name: str |
| 37 | + prompt_count: int |
| 38 | + updated_at: str |
| 39 | + |
| 40 | + |
| 41 | +class PromptLibraryService: |
| 42 | + """Manages prompt library files stored under a configs directory.""" |
| 43 | + |
| 44 | + _PREFIX = "user_prompt_library" |
| 45 | + _LIBRARIES_SUBDIR = "validation_prompt_libraries" |
| 46 | + _FILENAME_PATTERN = re.compile(rf"^{_PREFIX}(?:-([A-Za-z0-9._-]+))?\.json$") |
| 47 | + |
| 48 | + def __init__(self, config_dir: Optional[Path] = None, libraries_dir: Optional[Path] = None) -> None: |
| 49 | + self._config_dir = Path(config_dir) if config_dir else self._resolve_config_dir() |
| 50 | + self._config_dir.mkdir(parents=True, exist_ok=True) |
| 51 | + if libraries_dir: |
| 52 | + self._libraries_dir = Path(libraries_dir) |
| 53 | + else: |
| 54 | + self._libraries_dir = self._config_dir / self._LIBRARIES_SUBDIR |
| 55 | + self._libraries_dir.mkdir(parents=True, exist_ok=True) |
| 56 | + |
| 57 | + def _resolve_config_dir(self) -> Path: |
| 58 | + try: |
| 59 | + defaults = WebUIStateStore().load_defaults() |
| 60 | + if defaults.configs_dir: |
| 61 | + candidate = Path(defaults.configs_dir).expanduser() |
| 62 | + candidate.mkdir(parents=True, exist_ok=True) |
| 63 | + return candidate |
| 64 | + except Exception as exc: # pragma: no cover - defensive guard |
| 65 | + logger.debug("Unable to resolve configs_dir from WebUI defaults", exc_info=exc) |
| 66 | + default = get_config_directory() |
| 67 | + default.mkdir(parents=True, exist_ok=True) |
| 68 | + return default |
| 69 | + |
| 70 | + def _validate_filename(self, filename: str) -> str: |
| 71 | + if not filename: |
| 72 | + raise PromptLibraryError("Prompt library filename is required.") |
| 73 | + candidate = Path(filename).name |
| 74 | + match = self._FILENAME_PATTERN.fullmatch(candidate) |
| 75 | + if not match: |
| 76 | + raise PromptLibraryError( |
| 77 | + "Prompt library filenames must be user_prompt_library[-name].json and only contain letters, numbers, '.', '_', or '-'." |
| 78 | + ) |
| 79 | + return candidate |
| 80 | + |
| 81 | + def _normalize_entries(self, payload: Any) -> Dict[str, str]: |
| 82 | + if not isinstance(payload, dict): |
| 83 | + raise PromptLibraryError("Prompt library entries must be an object with ID -> prompt mappings.") |
| 84 | + normalized: Dict[str, str] = {} |
| 85 | + for key, value in payload.items(): |
| 86 | + shortname = str(key).strip() |
| 87 | + if not shortname: |
| 88 | + continue |
| 89 | + prompt = "" if value is None else str(value) |
| 90 | + normalized[shortname] = prompt |
| 91 | + return normalized |
| 92 | + |
| 93 | + def _load_entries(self, path: Path) -> Dict[str, str]: |
| 94 | + if not path.exists(): |
| 95 | + raise PromptLibraryError(f"Prompt library '{path.name}' not found", status.HTTP_404_NOT_FOUND) |
| 96 | + try: |
| 97 | + with path.open("r", encoding="utf-8") as handle: |
| 98 | + payload = json.load(handle) |
| 99 | + except json.JSONDecodeError as exc: |
| 100 | + raise PromptLibraryError(f"Invalid JSON in '{path.name}': {exc}", status.HTTP_422_UNPROCESSABLE_CONTENT) from exc |
| 101 | + except OSError as exc: |
| 102 | + raise PromptLibraryError(f"Failed to read '{path.name}': {exc}", status.HTTP_500_INTERNAL_SERVER_ERROR) from exc |
| 103 | + return self._normalize_entries(payload) |
| 104 | + |
| 105 | + def _build_metadata(self, path: Path, entries: Dict[str, str]) -> PromptLibraryRecord: |
| 106 | + match = self._FILENAME_PATTERN.fullmatch(path.name) |
| 107 | + library_name = match.group(1) if match else "" |
| 108 | + relative_path = self._relative_path(path) |
| 109 | + display_name = library_name if library_name else "default" |
| 110 | + absolute_path = str(path.resolve()) |
| 111 | + updated_at = datetime.fromtimestamp(path.stat().st_mtime, tz=timezone.utc).isoformat() |
| 112 | + return PromptLibraryRecord( |
| 113 | + filename=path.name, |
| 114 | + relative_path=relative_path, |
| 115 | + absolute_path=absolute_path, |
| 116 | + display_name=display_name, |
| 117 | + library_name=library_name or "", |
| 118 | + prompt_count=len(entries), |
| 119 | + updated_at=updated_at, |
| 120 | + ) |
| 121 | + |
| 122 | + def _relative_path(self, path: Path) -> str: |
| 123 | + try: |
| 124 | + return path.relative_to(self._config_dir).as_posix() |
| 125 | + except Exception: |
| 126 | + return path.name |
| 127 | + |
| 128 | + def list_libraries(self) -> List[PromptLibraryRecord]: |
| 129 | + records: List[PromptLibraryRecord] = [] |
| 130 | + if not self._libraries_dir.exists(): |
| 131 | + return records |
| 132 | + for path in sorted(self._libraries_dir.iterdir()): |
| 133 | + if not path.is_file(): |
| 134 | + continue |
| 135 | + if not self._FILENAME_PATTERN.fullmatch(path.name): |
| 136 | + continue |
| 137 | + try: |
| 138 | + entries = self._load_entries(path) |
| 139 | + except PromptLibraryError as exc: |
| 140 | + logger.warning("Skipping prompt library '%s': %s", path.name, exc.message) |
| 141 | + continue |
| 142 | + records.append(self._build_metadata(path, entries)) |
| 143 | + records.sort(key=lambda record: record.display_name.lower()) |
| 144 | + return records |
| 145 | + |
| 146 | + def read_library(self, filename: str) -> Dict[str, Any]: |
| 147 | + sanitized = self._validate_filename(filename) |
| 148 | + path = self._libraries_dir / sanitized |
| 149 | + entries = self._load_entries(path) |
| 150 | + metadata = self._build_metadata(path, entries) |
| 151 | + return {"entries": entries, "library": metadata} |
| 152 | + |
| 153 | + def save_library( |
| 154 | + self, |
| 155 | + filename: str, |
| 156 | + entries: Dict[str, str], |
| 157 | + previous_filename: Optional[str] = None, |
| 158 | + ) -> Dict[str, Any]: |
| 159 | + normalized = self._normalize_entries(entries) |
| 160 | + sanitized = self._validate_filename(filename) |
| 161 | + target = self._libraries_dir / sanitized |
| 162 | + target.parent.mkdir(parents=True, exist_ok=True) |
| 163 | + try: |
| 164 | + with target.open("w", encoding="utf-8") as handle: |
| 165 | + json.dump(normalized, handle, indent=2, ensure_ascii=False) |
| 166 | + handle.write("\n") |
| 167 | + except OSError as exc: |
| 168 | + raise PromptLibraryError( |
| 169 | + f"Failed to write '{target.name}': {exc}", status.HTTP_500_INTERNAL_SERVER_ERROR |
| 170 | + ) from exc |
| 171 | + |
| 172 | + if previous_filename: |
| 173 | + previous = self._validate_filename(previous_filename) |
| 174 | + if previous != sanitized: |
| 175 | + previous_path = self._libraries_dir / previous |
| 176 | + try: |
| 177 | + if previous_path.exists(): |
| 178 | + previous_path.unlink() |
| 179 | + except OSError as exc: |
| 180 | + logger.warning("Could not remove old prompt library '%s': %s", previous, exc) |
| 181 | + |
| 182 | + metadata = self._build_metadata(target, normalized) |
| 183 | + return {"entries": normalized, "library": metadata} |
0 commit comments