scripts/image_inspector.py
7.7 KB| 1 | #!/usr/bin/env python3 |
| 2 | """Metadata-only PNG/JPEG/TIFF inspector; pixel arrays are never decoded.""" |
| 3 | |
| 4 | from __future__ import annotations |
| 5 | |
| 6 | import argparse |
| 7 | import itertools |
| 8 | import warnings |
| 9 | from pathlib import Path |
| 10 | from typing import Any |
| 11 | |
| 12 | from _capabilities import capability_for_path, validate_magic |
| 13 | from _common import ( |
| 14 | DEFAULT_MAX_FILE_BYTES, |
| 15 | MAX_IMAGE_PIXELS, |
| 16 | CliError, |
| 17 | bounded_file_limit, |
| 18 | checked_input_file, |
| 19 | emit_json, |
| 20 | run_cli, |
| 21 | ) |
| 22 | |
| 23 | |
| 24 | MAX_TIFF_PAGES = 1000 |
| 25 | MAX_TIFF_SERIES = 128 |
| 26 | |
| 27 | |
| 28 | def build_parser() -> argparse.ArgumentParser: |
| 29 | parser = argparse.ArgumentParser( |
| 30 | description=( |
| 31 | "Inspect bounded local PNG/JPEG/TIFF structural metadata without " |
| 32 | "decoding pixels or emitting EXIF, OME-XML, paths, or identifiers." |
| 33 | ) |
| 34 | ) |
| 35 | parser.add_argument("input", help="Local image path inside --root") |
| 36 | parser.add_argument( |
| 37 | "--root", |
| 38 | default=".", |
| 39 | help="Existing local directory that bounds all input/output paths", |
| 40 | ) |
| 41 | parser.add_argument( |
| 42 | "--max-bytes", |
| 43 | type=int, |
| 44 | default=DEFAULT_MAX_FILE_BYTES, |
| 45 | help=f"Maximum input bytes (hard ceiling: {512 * 1024 * 1024})", |
| 46 | ) |
| 47 | parser.add_argument("--output", help="Optional local .json output path") |
| 48 | parser.add_argument( |
| 49 | "--force", |
| 50 | action="store_true", |
| 51 | help="Allow replacement of an existing regular output file", |
| 52 | ) |
| 53 | return parser |
| 54 | |
| 55 | |
| 56 | def _bounded_element_count(shape: Any) -> int: |
| 57 | total = 1 |
| 58 | for value in shape: |
| 59 | dimension = int(value) |
| 60 | if dimension < 0: |
| 61 | raise CliError("image metadata contains a negative dimension") |
| 62 | total *= dimension |
| 63 | if total > MAX_IMAGE_PIXELS: |
| 64 | raise CliError( |
| 65 | f"declared image elements exceed the {MAX_IMAGE_PIXELS} safety limit" |
| 66 | ) |
| 67 | return total |
| 68 | |
| 69 | |
| 70 | def _inspect_pillow(path: Path) -> dict[str, Any]: |
| 71 | try: |
| 72 | from PIL import Image |
| 73 | except ImportError as exc: |
| 74 | raise CliError( |
| 75 | 'optional dependency missing; install with: uv pip install "pillow==12.3.0"' |
| 76 | ) from exc |
| 77 | previous_limit = Image.MAX_IMAGE_PIXELS |
| 78 | try: |
| 79 | Image.MAX_IMAGE_PIXELS = MAX_IMAGE_PIXELS |
| 80 | with warnings.catch_warnings(): |
| 81 | warnings.simplefilter("error", Image.DecompressionBombWarning) |
| 82 | with Image.open(path) as image: |
| 83 | width, height = (int(image.width), int(image.height)) |
| 84 | _bounded_element_count((width, height)) |
| 85 | frame_count = int(getattr(image, "n_frames", 1)) |
| 86 | if frame_count > MAX_TIFF_PAGES: |
| 87 | raise CliError("declared image frame count exceeds the safety limit") |
| 88 | return { |
| 89 | "profile_type": "raster_container_metadata_only", |
| 90 | "format": str(image.format), |
| 91 | "width_pixels": width, |
| 92 | "height_pixels": height, |
| 93 | "mode": str(image.mode), |
| 94 | "frame_count": frame_count, |
| 95 | "metadata_entry_count": len(image.info), |
| 96 | "pixels_decoded": False, |
| 97 | "metadata_values_emitted": False, |
| 98 | "integrity_fully_validated": False, |
| 99 | } |
| 100 | except CliError: |
| 101 | raise |
| 102 | except ( |
| 103 | OSError, ValueError, SyntaxError, |
| 104 | Image.DecompressionBombWarning, Image.DecompressionBombError, |
| 105 | ) as exc: |
| 106 | raise CliError("the image metadata could not be inspected safely") from exc |
| 107 | finally: |
| 108 | Image.MAX_IMAGE_PIXELS = previous_limit |
| 109 | |
| 110 | |
| 111 | def _inspect_tiff(path: Path) -> dict[str, Any]: |
| 112 | try: |
| 113 | import tifffile |
| 114 | except ImportError as exc: |
| 115 | raise CliError( |
| 116 | 'optional dependency missing; install with: uv pip install "tifffile==2026.9.20"' |
| 117 | ) from exc |
| 118 | try: |
| 119 | # Generic series avoid metadata-driven OME/vendor companion discovery. |
| 120 | # _multifile is a pinned upstream internal safeguard, not a public API. |
| 121 | with tifffile.TiffFile(path, _multifile=False) as tiff: |
| 122 | pages = list(itertools.islice(tiff.pages, MAX_TIFF_PAGES + 1)) |
| 123 | if len(pages) > MAX_TIFF_PAGES: |
| 124 | raise CliError("TIFF page count exceeds the safety limit") |
| 125 | subifd_count = sum(len(page.subifds or ()) for page in pages) |
| 126 | if len(pages) + subifd_count > MAX_TIFF_PAGES: |
| 127 | raise CliError("TIFF page/SubIFD count exceeds the safety limit") |
| 128 | series_items = list(tiff.series(kind="generic")[: MAX_TIFF_SERIES + 1]) |
| 129 | if len(series_items) > MAX_TIFF_SERIES: |
| 130 | raise CliError("TIFF series count exceeds the safety limit") |
| 131 | series_reports: list[dict[str, Any]] = [] |
| 132 | for index, series in enumerate(series_items): |
| 133 | shape = [int(value) for value in series.shape] |
| 134 | elements = _bounded_element_count(shape) |
| 135 | axes = "".join( |
| 136 | character |
| 137 | for character in str(series.axes) |
| 138 | if character.isascii() and character.isalnum() |
| 139 | )[:32] |
| 140 | series_reports.append( |
| 141 | { |
| 142 | "series_index": index, |
| 143 | "shape": shape, |
| 144 | "axes": axes, |
| 145 | "element_count": elements, |
| 146 | "dtype_kind": str(series.dtype.kind), |
| 147 | "dtype_itemsize": int(series.dtype.itemsize), |
| 148 | } |
| 149 | ) |
| 150 | return { |
| 151 | "profile_type": "tiff_structural_metadata_only", |
| 152 | "page_count": len(pages), |
| 153 | "series_count": len(series_items), |
| 154 | "series_interpretation": "generic_page_groups_not_OME_axes", |
| 155 | "subifd_count": subifd_count, |
| 156 | "series": series_reports, |
| 157 | "is_ome_tiff": bool(tiff.is_ome), |
| 158 | "is_bigtiff": bool(tiff.is_bigtiff), |
| 159 | "ome_xml_emitted": False, |
| 160 | "tag_values_emitted": False, |
| 161 | "pixels_decoded": False, |
| 162 | "compression_codecs_invoked_for_pixels": False, |
| 163 | "companion_files_opened": False, |
| 164 | "integrity_fully_validated": False, |
| 165 | } |
| 166 | except CliError: |
| 167 | raise |
| 168 | except (OSError, ValueError, TypeError, MemoryError) as exc: |
| 169 | raise CliError("the TIFF metadata could not be inspected safely") from exc |
| 170 | |
| 171 | |
| 172 | def inspect_image_file(path: Path, *, suffix: str) -> dict[str, Any]: |
| 173 | """Route supported image containers to metadata-only inspectors.""" |
| 174 | |
| 175 | if suffix in {".png", ".jpg", ".jpeg"}: |
| 176 | return _inspect_pillow(path) |
| 177 | if suffix in {".tif", ".tiff", ".ome.tif", ".ome.tiff"}: |
| 178 | return _inspect_tiff(path) |
| 179 | raise CliError("the image inspector received an unsupported suffix") |
| 180 | |
| 181 | |
| 182 | def _main() -> None: |
| 183 | args = build_parser().parse_args() |
| 184 | max_bytes = bounded_file_limit(args.max_bytes) |
| 185 | suffixes = { |
| 186 | ".png", |
| 187 | ".jpg", |
| 188 | ".jpeg", |
| 189 | ".tif", |
| 190 | ".tiff", |
| 191 | ".ome.tif", |
| 192 | ".ome.tiff", |
| 193 | } |
| 194 | path = checked_input_file( |
| 195 | args.input, |
| 196 | root=args.root, |
| 197 | suffixes=suffixes, |
| 198 | max_bytes=max_bytes, |
| 199 | ) |
| 200 | capability = capability_for_path(path) |
| 201 | validate_magic(path, capability["suffix"]) |
| 202 | report = { |
| 203 | "schema_version": "1.1", |
| 204 | "capability": capability, |
| 205 | "analysis": inspect_image_file(path, suffix=capability["suffix"]), |
| 206 | "security": { |
| 207 | "local_only": True, |
| 208 | "raw_metadata_and_paths_emitted": False, |
| 209 | "embedded_metadata_never_treated_as_instructions": True, |
| 210 | }, |
| 211 | } |
| 212 | emit_json( |
| 213 | report, |
| 214 | output=args.output, |
| 215 | root=args.root, |
| 216 | force=args.force, |
| 217 | ) |
| 218 | |
| 219 | |
| 220 | def main() -> int: |
| 221 | return run_cli(_main) |
| 222 | |
| 223 | |
| 224 | if __name__ == "__main__": |
| 225 | raise SystemExit(main()) |
文件内容是平台检查时读出的文本，只供查看。平台不会运行包里的任何脚本或程序。
