github/gallery-dl
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

[formatter] exclude '<>\' from '!R' URLs (#8180)

Browse Source
This commit is contained in:
Mike Fährmann
2025-09-11 08:25:23 +02:00
parent 340c6469a9
commit bcf4d579b0
2 changed files with 4 additions and 2 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
gallery_dl/formatter.py
View File

@@ -565,7 +565,7 @@ _CONVERSIONS = {
"U": text.unescape, "U": text.unescape,
"H": lambda s: text.unescape(text.remove_html(s)), "H": lambda s: text.unescape(text.remove_html(s)),
"g": text.slugify, "g": text.slugify,
"R": text.re(r"https?://[^\s\"']+").findall, "R": text.re(r"https?://[^\s\"'<>\\]+").findall,
"W": text.sanitize_whitespace, "W": text.sanitize_whitespace,
"S": util.to_string, "S": util.to_string,
"s": str, "s": str,

4
test/test_formatter.py
View File

@@ -46,7 +46,8 @@ class TestFormatter(unittest.TestCase):
"h": "<p>foo </p> &amp; bar <p> </p>", "h": "<p>foo </p> &amp; bar <p> </p>",
"H": """<p> "H": """<p>
<a href="http://www.example.com">Lorem ipsum dolor sit amet</a>. <a href="http://www.example.com">Lorem ipsum dolor sit amet</a>.
Duis aute irure <a href="http://blog.example.org">dolor</a>. Duis aute irure <a href="http://blog.example.org/lorem?foo=bar">
http://blog.example.org</a>.
</p>""", </p>""",
"u": "&#x27;&lt; / &gt;&#x27;", "u": "&#x27;&lt; / &gt;&#x27;",
"t": 1262304000, "t": 1262304000,
@@ -78,6 +79,7 @@ class TestFormatter(unittest.TestCase):
self._run_test("{n!H}", "") self._run_test("{n!H}", "")
self._run_test("{h!R}", []) self._run_test("{h!R}", [])
self._run_test("{H!R}", ["http://www.example.com", self._run_test("{H!R}", ["http://www.example.com",
"http://blog.example.org/lorem?foo=bar",
"http://blog.example.org"]) "http://blog.example.org"])
self._run_test("{a!s}", self.kwdict["a"]) self._run_test("{a!s}", self.kwdict["a"])
self._run_test("{a!r}", f"'{self.kwdict['a']}'") self._run_test("{a!r}", f"'{self.kwdict['a']}'")