6 Commits

Include renames

Author	SHA1	Message	Date
Tool Backend	810ce59af4	change: Implement 5: Save selected images to the caller directory and downscale only behind the resize flag ... Implemented image saving into the caller directory and exact-size downscaling behind --resize. tools/images.py: added stdlib-only helpers `png_chunks`, `decode_png`, `encode_png`, `ico_wrap`, `ico_unwrap`, plus `decode_image`, `resample_rows`, `encode_image` and the `DecodedImage` dataclass; also fixed a real pixel-stride bug in RGBA upsampling (channel step was used as pixel stride, corrupting 8-bit pixels). tools/example.py: removed the inlined codec (now delegated to tools.images); `--download` writes each selected image into the caller-specified directory; each stored file is read back and decoded, so a reported path only names a file that exists and opens as the requested format (`format`/`width`/`height`/`bytes` reported per item, no path for a failed item); `--download` without `--output-dir` is now an `invalid_request` (`request_field`, field=output_dir) instead of a hidden default, and DEFAULT_OUTPUT_DIR is gone; `--resize` stores exactly the requested width/height (one open dimension keeps the source value), re-encodes in the requested format, never upscales and leaves a fitting image's bytes untouched; a per-item save failure is reported in that item and ends the run with `partial_download` (exit 6), including `saved_paths` in the error details. tests/test_download_resize.py: new focused suite over tmp_path dirs and locally built png/ico fixtures (save paths exist and decode, requested-format check, missing --output-dir invalid_request, failing save => partial_download, exact resize dims for png and ico, no upscaling, byte/dimension preservation without --resize, plus codec round-trip tests). tests/test_cli.py: updated the aspect-ratio resize expectation to the required exact 100x100 stored size. Target branch task/1-task is already the current head (d5bb24f), so the tested revision contains the target. Tests: Facade commands run from /workspace with .venv/bin on PATH (prepared via prepare_environment tools restore): `python -m compileall tools tests` (tools.validate) OK; `python -m tools.example examples/input.json` (tools.smoke) exit 0 with status dry_run; `python -m tools.example --help` (acceptance help-command) exit 0; `python -c "import tools, tools.example, tools.images"` (package_import) OK; `pytest` (tools.test, the exact facade entrypoint) => 142 passed. New tests: tests/test_download_resize.py (13 tests). No network-dependent live-source check was run (sandbox has no egress); all fixtures are locally built real png/ico payloads. Progress: Completed issue 7: stdlib PNG/ICO decode/encode/wrap/unwrap plus resample helpers live in tools/images.py; tools/example.py saves selected images into the caller directory, requires --output-dir with --download (invalid_request otherwise), verifies every stored file by decoding it before reporting its path, downscales to exactly the requested width/height in the requested format only when --resize is set (never upscaling, otherwise byte-identical), and reports a per-item save failure with the partial_download status/exit code. New tests in tests/test_download_resize.py cover all of that; tests/test_cli.py resize expectation aligned with the exact-size requirement. Facade validate/smoke/help/import all pass and `pytest` reports 142 passed on the revision that already contains target task/1-task. Published by developer-agent through Tool Backend. Closes #7	2026-11-03 02:25:41 +00:00
Tool Backend	d5bb24fc01	change: Implement 4: Filter candidates by real format and minimum size from downloaded bytes ... Fix for review-fix on issue #6: non-PNG/ICO hits in the bounded candidate window are now classified as format mismatches instead of access problems, so working searches no longer degrade to `partial_download` (exit 6) or `access_error` (exit 5). Changes: - `tools/images.py`: `detect_format` now also names the formats a live image source commonly serves instead (svg, tiff, pdf) alongside png/ico/jpeg/gif/bmp/webp, so "the topic has images, but not in the requested format" is describable; `ImageFormatError` is documented as reserved for genuinely unusable bytes (unknown container, or a png/ico header that is truncated/corrupt). - `tools/example.py`: `_select()` counts any payload whose detected format differs from the request (including payloads that are not a recognised image at all, e.g. an HTML error page) as a format rejection and records `observed_formats`; only an unfetchable candidate is a failure. `Selection` gained `other_formats`. `_execute()` now returns `no_size` when format-matching items were all too small, `no_format` when the topic only offered other formats (or no hits at all), and `access_error` only when not a single candidate could be fetched; a stray unfetchable candidate no longer demotes a successful search to `partial_download`. Status/exit-code vocabulary (`ok`/`no_format` 3/`no_size` 4/`access_error` 5/`partial_download` 6) is unchanged. - `tests/test_images.py`: added svg/tiff/pdf fixtures and detection cases, plus regression tests for the reviewed behaviour — mixed-format window still `ok` (exit 0), ico request with only foreign formats yields `no_format` (exit 3) with `observed_formats`, too-small format matches still yield `no_size` (exit 4), non-image body yields `no_format` not `access_error`, unfetchable candidate does not demote a usable search, corrupt png does not hide a valid match, empty window is `no_format`, fetch failure still `access_error`. - `tests/test_cli.py`: the former "unreadable payload is an access error" test now asserts `no_format` for a non-image body, matching the classification. Filtering still uses only the parsed downloaded bytes (PNG IHDR / ICO directory) and never source-reported metadata, and the candidate cap derived from `--count` is unchanged. Tests: Ran the facade commands from /workspace with the prepared virtualenv on PATH: `python -m compileall tools tests` (validate, ok), `python -m pytest` (test, 129 passed), `python -m tools.example examples/input.json` (smoke, exit 0, dry_run envelope) and `python -m tools.example --help` (acceptance, exit 0). Dependencies restored via prepare_environment (component tools, mode restore). Target branch origin/task/1-task is already an ancestor of the work branch HEAD (verified with git merge-base), so no extra integration commit was needed; the checks ran on the integrated revision. Progress: Addressed changes-requested review on PR #14: the CLI no longer turns common live-source results into partial_download/access_error. Any payload whose real format is not the requested png/ico (jpeg, gif, bmp, webp, svg, tiff, pdf, or an unrecognised body such as an HTML page) is now a format mismatch feeding no_format, with observed_formats/other_format_candidates detail; only unfetchable candidates are failures, and access_error is emitted only when nothing at all could be fetched. images.detect_format gained svg/tiff/pdf recognition and ImageFormatError is scoped to genuinely unusable png/ico payloads. Statuses/exits unchanged (no_format 3, no_size 4, access_error 5). Tests extended in tests/test_images.py (svg/tiff/pdf detection, mixed-format window stays ok, ico-only-foreign yields no_format, no_size preserved alongside other formats, non-image body is no_format, unfetchable candidate does not demote a search, corrupt png does not hide a match, empty window) and tests/test_cli.py aligned. Facade validate/test/smoke/help all green (129 passed) on the integrated branch; no protected facade paths touched. Published by developer-agent through Tool Backend. Closes #6	2026-11-03 02:19:10 +00:00
Tool Backend	b544968cb8	change: Implement 4: Filter candidates by real format and minimum size from downloaded bytes ... Added tools/images.py with stdlib-only detection of the real image format and pixel dimensions from downloaded bytes (PNG signature + IHDR chunk, ICO ICONDIR/ICONDIRENTRY directory including the 0-means-256 encoding) plus stdlib-only builders (png_bytes, ico_bytes) so tests/sources can produce genuine payloads. Wired it into the CLI search path in tools/example.py: a bounded candidate window derived from --count (candidate_cap = count * 4, capped at 50) is fetched from the selected source, each payload is parsed, and only items whose actual format equals the requested png/ico and whose actual width and height are both at least the requested minimum are kept. Kept items report source, url, format and the actual width/height (plus size_bytes and metadata as extras); verified bytes are reused for --download instead of being fetched twice. Statuses stay distinct: ok (0), no_format (3) when hits were only other formats, no_size (4) when format-matching items were all too small, access_error (5) when bytes could not be fetched or are not readable images, partial_download (6) when some candidates could not be read. All metadata-based filtering (FORMAT_MATCHERS, _matches_format, _matches_size) was removed so only parsed bytes decide. Added tests/test_images.py with detector/builder unit tests and CLI tests using an injected fake byte-fetch callable (no network), and updated tests/test_cli.py fixtures to real image bytes where the previous metadata-only assumption no longer holds. Tests: Facade commands run with .venv/bin on PATH after prepare_environment(component=tools, mode=restore): `python -m compileall tools tests` (tools.validate) succeeded; `pytest` (tools.test) => 114 passed, including the new tests/test_images.py (31 tests); `python -m tools.example examples/input.json` (tools.smoke) exited 0 with a single dry_run JSON document; acceptance commands `python -m tools.example --help` exited 0 and `import tools` resolved. Target branch task/1-task is at the same commit as this work branch (git rev-list count 0 ahead / 0 behind), so it is already integrated; facade/protected paths untouched. Progress: Implemented byte-based format and minimum-size filtering for the image search path. New tools/images.py: detect_format (png/ico plus jpeg/gif/bmp/webp so "other format" is distinguishable from "unreadable"), read_png_dimensions (IHDR), read_ico_dimensions (ICONDIR entries, 0 => 256, largest entry), inspect_image/ImageInfo, ImageFormatError (code image_unreadable), and png_bytes/ico_bytes builders. tools/example.py: candidate_cap(count) bounded window (count*4, max 50) requested from the source, _select() fetches and parses every candidate, KeptImage.to_item() reports source/url/format/actual width+height/size_bytes, _download() writes the already verified bytes, and _execute() distinguishes ok/no_format/no_size/access_error/partial_download with the existing exit codes; metadata-based filtering removed. tests/test_images.py added (31 tests) and tests/test_cli.py fixtures aligned to real image bytes. All facade commands pass: compileall, pytest (114 passed), smoke, help, package import. Published by developer-agent through Tool Backend. Closes #6	2026-11-03 02:11:36 +00:00
Tool Backend	6a7538c07f	change: Implement 3: Complete the native-process CLI, JSON request handling, status vocabulary and exit codes ... Review fix for issue #5: aligned the public CLI's size/resize semantics with the active product obligations instead of the previous exact-match behaviour. - `tools/example.py`: the requested width/height are now a target (minimum) size, never an exact-match filter: any image at least that large is selected regardless of `--resize`, and smaller images still produce `no_size`. `--resize` no longer widens selection; it now performs real downscaling of downloaded images to a shrink-to-fit box (aspect ratio preserved, never upscaled), implemented with a dependency-free stdlib PNG codec plus ICO container support (PNG-compressed ICO entries), and each item records `from`/`to`/`applied` dimensions. Undecodable payloads raise the new typed `ImageResizeError` and are reported as failed downloads, so the status vocabulary and exit codes stay unchanged (`ok`/`dry_run` 0, `invalid_request` 2, `no_format` 3, `no_size` 4, `access_error` 5, `partial_download` 6). Help text and the `no_size` error details were updated to describe the minimum-size contract. - `tools/errors.py`: added `ImageResizeError` (code `resize_failed`) to the typed hierarchy and `__all__`. - `tests/test_cli.py`: replaced the old "exact size unless --resize" test with minimum-size tests (larger image accepted without `--resize`, exact size accepted, `--resize` without `--download` keeps the reported source size) and added offline fixtures/tests for real downscaling: PNG stored at 100x50 for a 100x100 box, original bytes preserved without `--resize`, no upscaling (`already_fits`), PNG-in-ICO downscaling, and `resize_failed` surfacing as `access_error`. - `examples/input.json` remains the canonical offline `dry_run` request, so the smoke command stays green and network-free. No protected facade/control paths were touched and no dependency declarations changed. Tests: Ran the facade commands on the final revision through the prepared environment (`.venv` on PATH): `python -m compileall tools tests` → exit 0 (tools.validate); `pytest` → 83 passed (tools.test); `python -m tools.example examples/input.json` → exit 0 with one JSON document, status `dry_run` (tools.smoke); `python -m tools.example --help` → exit 0 (help-command acceptance). `tools/envelope.py` is unchanged and still exports the fixed status/exit-code table checked by tests. Target branch `task/1-task` is already an ancestor of the work branch head, so no integration was needed. Progress: Addressed the changes-requested review on PR #13: requested width/height are now a minimum target size (selection no longer depends on --resize), and --resize now actually downscales downloaded images (stdlib PNG codec + ICO/PNG container support, shrink-to-fit, never upscaling) with per-item resize metadata and typed ImageResizeError for undecodable payloads. Updated tests/test_cli.py with minimum-size and offline resize/download contract tests (83 passed). Facade validate/test/smoke/help all green; examples/input.json stays the canonical dry_run request. Published by developer-agent through Tool Backend. Closes #5	2026-11-03 02:03:37 +00:00
Tool Backend	f20e58e85c	change: Implement 3: Complete the native-process CLI, JSON request handling, status vocabulary and exit codes ... Completed the native-process CLI for `python -m tools.example`. Added `tools/envelope.py` with the fixed status vocabulary (`ok`, `dry_run`, `invalid_request`, `no_format`, `no_size`, `access_error`, `partial_download`) mapped to exit codes 0/0/2/3/4/5/6 and the `Response`/`ResponseBuilder` JSON envelope (keys: status, source, request, items, error). Extended `tools/errors.py` with request errors (`RequestError`, `RequestDocumentError`, `RequestFieldError`, `UnsupportedFormatError`) and `ACCEPTED_FORMATS`. Rewrote `tools/example.py` as a plain executable CLI accepting either a JSON request document or flags (--topic, --width, --height, --format png|ico, --source, --count, --download, --output-dir, --resize, --dry-run), with a custom parser that turns usage errors into `invalid_request` envelopes, unknown sources reported together with the available source ids, unsupported formats rejected with the accepted values, and format/size filtering, download and partial-download handling mapped onto the shared status table. `examples/input.json` is now the canonical offline dry_run request, and `tests/test_cli.py` (new) plus a rewritten `tests/test_example.py` cover the status/exit-code table, single-JSON-document output, help text, unknown source/unsupported format rejection and the download branches. Tests: Facade checks run from /workspace with the prepared .venv: `python -m compileall tools tests` (tools.validate) OK; `pytest` (tools.test) 76 passed; `python -m tools.example examples/input.json` (tools.smoke) exit 0 with a single `dry_run` JSON document; `python -m tools.example --help` exits 0 and documents every parameter plus the JSON document contract. Origin target branch `origin/task/1-task` (1f929c5) is an ancestor of the work branch, so the target is integrated and the checks above ran on the integrated revision. No dependency changes were needed, so no prepare_environment update was required. Progress: Implemented envelope status/exit-code table and response builder, request errors, the full flag/JSON-document CLI in tools/example.py, canonical offline examples/input.json, and CLI tests; facade validate/test/smoke and --help all pass (76 tests). Published by developer-agent through Tool Backend. Closes #5	2026-11-03 01:48:00 +00:00
Tool Backend	d0b7f8f5aa	change: Implement 1: Bootstrap the toolkit tree, project config and initial dependency lock ... Bootstrapped the repository-root toolkit tree required by the facade. - `tools/__init__.py`: minimal importable package exposing `__version__` only. - `tools/example.py`: offline CLI (`python -m tools.example REQUEST`) that loads and validates a JSON request document with a validate-only `dry_run` semantics, prints exactly one structured JSON envelope (sorted keys) and fixes the status/exit vocabulary now (0 ok, 1 invalid request document, 2 usage error). `--help` supported. - `examples/input.json`: canonical offline validate-only request document. - `tests/test_example.py`: focused tests for the smoke command, envelope shape, request-document validation, `--help`, and the module entrypoint (13 tests). - `pyproject.toml`: project metadata, setuptools build backend with `packages = ["tools"]`, `dev` extra with `pytest>=8.0`, pytest options `pythonpath=["."]` / `testpaths=["tests"]` so the repository root is importable. - `uv.lock`: created through the supported dependency update operation for the `tools` component; pinned only from the approved dependency mirror (no public-index fallbacks, no credentials). - `.gitignore` / `.env.example`: keep build artifacts out of the tree and document that no environment variables are required. Note: `requires-python` was raised to `>=3.11` because the approved mirror only offers `exceptiongroup<1.0.0rc8`, which cannot satisfy pytest on Python <3.11; the prepared environment (CPython 3.12.3) and CI bootstrap interpreter satisfy this. No protected facade paths were modified. Tests: Ran the facade commands with the prepared environment (`.venv/bin/python`): - `python -m compileall -q tools tests` -> OK (tools.validate) - `python -m pytest -q` -> 13 passed (tools.test) - `python -m tools.example examples/input.json` -> single JSON envelope, exit 0 (tools.smoke) - `python -m tools.example --help` -> usage text, exit 0 (help-command) Dependency lock/materialization performed via prepare_environment(component='tools', mode='update'): resolved + materialized, `uv.lock` written. Target branch `origin/task/1-task` is already an ancestor of HEAD, so the branch contains the current target. Progress: Created tools package (tools/__init__.py, tools/example.py), examples/input.json, tests/test_example.py, .gitignore, .env.example; updated pyproject.toml (metadata, setuptools packages, dev pytest extra, pytest pythonpath/testpaths, requires-python >=3.11 due to approved-mirror exceptiongroup constraint) and generated uv.lock via prepare_environment update. Facade validate/test/smoke and --help all pass locally. Published by developer-agent through Tool Backend. Closes #3	2026-11-02 23:23:26 +00:00