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

[formatter] implement 'H' conversion (#4164)

Browse Source 
to remove HTML tags and unescape HTML entities
This commit is contained in:
Mike Fährmann
2023-06-15 13:07:51 +02:00
parent deff3b434d
commit ce93c460a6
3 changed files with 20 additions and 2 deletions
Download Patch File Download Diff File Expand all files Collapse all files

12
docs/formatting.md
View File

@@ -94,6 +94,18 @@ Conversion specifiers allow to *convert* the value to a different form or type.
<td><code>{created!d}</code></td> <td><code>{created!d}</code></td>
<td><code>2010-01-01 00:00:00</code></td> <td><code>2010-01-01 00:00:00</code></td>
</tr> </tr>
<tr>
<td align="center"><code>U</code></td>
<td>Convert HTML entities</td>
<td><code>{html!U}</code></td>
<td><code>&lt;p&gt;foo &amp; bar&lt;/p&gt;</code></td>
</tr>
<tr>
<td align="center"><code>H</code></td>
<td>Convert HTML entities &amp; remove HTML tags</td>
<td><code>{html!H}</code></td>
<td><code>foo &amp; bar</code></td>
</tr>
<tr> <tr>
<td align="center"><code>s</code></td> <td align="center"><code>s</code></td>
<td>Convert value to <a href="https://docs.python.org/3/library/stdtypes.html#text-sequence-type-str" rel="nofollow"><code>str</code></a></td> <td>Convert value to <a href="https://docs.python.org/3/library/stdtypes.html#text-sequence-type-str" rel="nofollow"><code>str</code></a></td>

1
gallery_dl/formatter.py
View File

@@ -437,6 +437,7 @@ _CONVERSIONS = {
"T": util.datetime_to_timestamp_string, "T": util.datetime_to_timestamp_string,
"d": text.parse_timestamp, "d": text.parse_timestamp,
"U": text.unescape, "U": text.unescape,
"H": lambda s: text.unescape(text.remove_html(s)),
"g": text.slugify, "g": text.slugify,
"S": util.to_string, "S": util.to_string,
"s": str, "s": str,

9
test/test_formatter.py
View File

@@ -28,6 +28,7 @@ class TestFormatter(unittest.TestCase):
"l": ["a", "b", "c"], "l": ["a", "b", "c"],
"n": None, "n": None,
"s": " \n\r\tSPACE ", "s": " \n\r\tSPACE ",
"h": "<p>foo </p> &amp; bar <p> </p>",
"u": "&#x27;&lt; / &gt;&#x27;", "u": "&#x27;&lt; / &gt;&#x27;",
"t": 1262304000, "t": 1262304000,
"dt": datetime.datetime(2010, 1, 1), "dt": datetime.datetime(2010, 1, 1),
@@ -47,6 +48,10 @@ class TestFormatter(unittest.TestCase):
self._run_test("{s!t}", "SPACE") self._run_test("{s!t}", "SPACE")
self._run_test("{a!U}", self.kwdict["a"]) self._run_test("{a!U}", self.kwdict["a"])
self._run_test("{u!U}", "'< / >'") self._run_test("{u!U}", "'< / >'")
self._run_test("{a!H}", self.kwdict["a"])
self._run_test("{h!H}", "foo & bar")
self._run_test("{u!H}", "'< / >'")
self._run_test("{n!H}", "")
self._run_test("{a!s}", self.kwdict["a"]) self._run_test("{a!s}", self.kwdict["a"])
self._run_test("{a!r}", "'" + self.kwdict["a"] + "'") self._run_test("{a!r}", "'" + self.kwdict["a"] + "'")
self._run_test("{a!a}", "'" + self.kwdict["a"] + "'") self._run_test("{a!a}", "'" + self.kwdict["a"] + "'")
@@ -434,10 +439,10 @@ def noarg():
fmt4 = formatter.parse("\fM " + path + ":lengths") fmt4 = formatter.parse("\fM " + path + ":lengths")
self.assertEqual(fmt1.format_map(self.kwdict), "'Title' by Name") self.assertEqual(fmt1.format_map(self.kwdict), "'Title' by Name")
self.assertEqual(fmt2.format_map(self.kwdict), "96") self.assertEqual(fmt2.format_map(self.kwdict), "126")
self.assertEqual(fmt3.format_map(self.kwdict), "'Title' by Name") self.assertEqual(fmt3.format_map(self.kwdict), "'Title' by Name")
self.assertEqual(fmt4.format_map(self.kwdict), "96") self.assertEqual(fmt4.format_map(self.kwdict), "126")
with self.assertRaises(TypeError): with self.assertRaises(TypeError):
self.assertEqual(fmt0.format_map(self.kwdict), "") self.assertEqual(fmt0.format_map(self.kwdict), "")