The_Great_Archive/ArchiveBox
1
0
Fork 0
mirror of https://github.com/ArchiveBox/ArchiveBox.git synced 2026-09-13 18:46:17 +05:00
Code Issues Actions Packages Projects Releases Wiki Activity

Constrain transformed preview images
Some checks are pending
CI / Linters (push) Waiting to run
Details
CI / Install and CLI platform compatibility (push) Waiting to run
Details
CI / Discovered test matrix (push) Waiting to run
Details
CI / Documentation and root tests (push) Waiting to run
Details
CI / CodeQL (push) Waiting to run
Details
CI / Tested Python artifacts (push) Waiting to run
Details
CI / Tested Docker images (push) Waiting to run
Details
CI / All required CI lanes (push) Blocked by required conditions
Details
Deploy Publicsite to GitHub Pages / deploy (push) Waiting to run
Details

Browse Source
This commit is contained in:
Nick Sweeting 2026-07-25 23:28:37 -07:00
parent f3ac229026
commit 317da5ea7c
No known key found for this signature in database
2 changed files with 33 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

24
archivebox/misc/serve_static.py
View File

@ -32,6 +32,19 @@ from archivebox.misc.logging_util import printable_filesize
_HASHES_CACHE: dict[Path, tuple[float, dict[str, str]]] = {}
IMG_SRC_ATTR_RE = re.compile(r'(<img\b[^>]*?\s(?:src|data-src)=["\'])([^"\']+)(["\'])', re.IGNORECASE)
TRANSFORMED_HTML_PREVIEW_STYLE = """<style id="archivebox-static-html-preview-style">
img:not([width]):not([height]) {
max-width: min(100%, 12rem);
max-height: 12rem;
width: auto;
height: auto;
object-fit: contain;
}
a > img:not([width]):not([height]) {
max-width: min(100%, 2.5rem);
max-height: 2.5rem;
}
</style>"""
def _load_hash_map(snapshot_dir: Path) -> dict[str, str] | None:
@ -531,6 +544,14 @@ def _rewrite_html_image_sources_for_request(
)
def _apply_transformed_html_preview_style(html_text: str) -> str:
if "archivebox-static-html-preview-style" in html_text:
return html_text
if re.search(r"</head\s*>", html_text, flags=re.IGNORECASE):
return re.sub(r"</head\s*>", f"{TRANSFORMED_HTML_PREVIEW_STYLE}\\g<0>", html_text, count=1, flags=re.IGNORECASE)
return f"{TRANSFORMED_HTML_PREVIEW_STYLE}\n{html_text}"
def _set_transformed_response_headers(response, fullpath: Path, statobj: os.stat_result, encoding: str | None, config) -> None:
response.headers["Last-Modified"] = http_date(statobj.st_mtime)
response.headers["Cache-Control"] = f"{_cache_policy(config=config)}, max-age=60, stale-while-revalidate=300"
@ -963,6 +984,7 @@ def serve_static_with_byterange_support(request, path, document_root=None, show_
if _looks_like_markdown(markdown_candidate):
wrapped = _render_markdown_document(markdown_candidate)
wrapped, _rewrite_count = _rewrite_html_image_sources_for_request(request, wrapped, document_root, rel_path)
wrapped = _apply_transformed_html_preview_style(wrapped)
response = HttpResponse(wrapped, content_type="text/html; charset=utf-8")
_set_transformed_response_headers(response, fullpath, statobj, encoding, config)
return _apply_archive_replay_headers(
@ -973,6 +995,7 @@ def serve_static_with_byterange_support(request, path, document_root=None, show_
config=config,
)
if rewritten_count:
rewritten_html = _apply_transformed_html_preview_style(rewritten_html)
response = HttpResponse(rewritten_html, content_type=content_type)
_set_transformed_response_headers(response, fullpath, statobj, encoding, config)
return _apply_archive_replay_headers(
@ -983,6 +1006,7 @@ def serve_static_with_byterange_support(request, path, document_root=None, show_
config=config,
)
if escaped_count and escaped_count > tag_count * 2:
decoded = _apply_transformed_html_preview_style(decoded)
response = HttpResponse(decoded, content_type=content_type)
_set_transformed_response_headers(response, fullpath, statobj, encoding, config)
return _apply_archive_replay_headers(

10
archivebox/tests/test_urls.py
View File

@ -57,7 +57,7 @@ def test_static_html_and_markdown_preview_images_rewrite_to_captured_responses(t
html_path = tmp_path / "extractor" / "content.html"
html_path.parent.mkdir()
html_path.write_text('<img src="images/twitter.png">', encoding="utf-8")
html_path.write_text('<img src="images/twitter.png"><img width="48" src="images/twitter.png">', encoding="utf-8")
request = RequestFactory().get("/web/20260722/sweeting.me/snapshot/extractor/content.html")
request.archivebox_snapshot_url = "https://sweeting.me/"
@ -67,7 +67,14 @@ def test_static_html_and_markdown_preview_images_rewrite_to_captured_responses(t
assert response.status_code == 200
assert "ETag" not in response.headers
assert "max-age=60" in response.headers["Cache-Control"]
assert b"archivebox-static-html-preview-style" in response.content
assert b"img:not([width]):not([height])" in response.content
assert b"a > img:not([width]):not([height])" in response.content
assert b'src="../responses/all/20260722T061544__GET__https_3A_2F_2Fsweeting.me_2Fimages_2Ftwitter.png"' in response.content
assert (
b'<img width="48" src="../responses/all/20260722T061544__GET__https_3A_2F_2Fsweeting.me_2Fimages_2Ftwitter.png">'
in response.content
)
text_path = tmp_path / "article" / "content.txt"
text_path.parent.mkdir()
@ -89,6 +96,7 @@ def test_static_html_and_markdown_preview_images_rewrite_to_captured_responses(t
assert response.status_code == 200
assert "ETag" not in response.headers
assert "max-age=60" in response.headers["Cache-Control"]
assert b"archivebox-static-html-preview-style" in response.content
assert b'src="../responses/all/20260722T061544__GET__https_3A_2F_2Fsweeting.me_2Fimages_2Ftwitter.png"' in response.content