1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005
|
# Copyright 2010-2025 Kurt McKee <contactme@kurtmckee.org>
# Copyright 2004-2008 Mark Pilgrim
# All rights reserved.
#
# Redistribution and use in source and binary forms, with or without modification,
# are permitted provided that the following conditions are met:
#
# * Redistributions of source code must retain the above copyright notice,
# this list of conditions and the following disclaimer.
# * Redistributions in binary form must reproduce the above copyright notice,
# this list of conditions and the following disclaimer in the documentation
# and/or other materials provided with the distribution.
#
# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS 'AS IS'
# AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
# IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
# ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE
# LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
# CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
# SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
# INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
# CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
# ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
# POSSIBILITY OF SUCH DAMAGE.
__author__ = "Kurt McKee <contactme@kurtmckee.org>"
__license__ = "BSD 2-clause"
import codecs
import datetime
import glob
import http.server
import io
import os
import posixpath
import pprint
import re
import struct
import sys
import threading
import time
import unittest
import urllib.error
import urllib.request
import warnings
import xml.sax
import zlib
import feedparser
import feedparser.api
import feedparser.datetimes
import feedparser.http
import feedparser.mixin
import feedparser.sanitizer
import feedparser.urls
import feedparser.util
from feedparser.datetimes.asctime import _parse_date_asctime
from feedparser.datetimes.greek import _parse_date_greek
from feedparser.datetimes.hungarian import _parse_date_hungarian
from feedparser.datetimes.iso8601 import _parse_date_iso8601
from feedparser.datetimes.korean import _parse_date_onblog, _parse_date_nate
from feedparser.datetimes.perforce import _parse_date_perforce
from feedparser.datetimes.rfc822 import _parse_date_rfc822
from feedparser.datetimes.w3dtf import _parse_date_w3dtf
# ---------- custom HTTP server (used to serve test feeds) ----------
_PORT = 8097 # not really configurable, must match hardcoded port in tests
_HOST = '127.0.0.1' # also not really configurable
class FeedParserTestRequestHandler(http.server.SimpleHTTPRequestHandler):
headers_re = re.compile(br"^Header:\s+([^:]+):(.+)$", re.MULTILINE)
def send_head(self):
"""Send custom headers defined in test case
Example:
<!--
Header: Content-type: application/atom+xml
Header: X-Foo: bar
-->
"""
# Short-circuit the HTTP status test `test_redirect_to_304()`
if self.path == '/-/return-304.xml':
self.send_response(304)
self.send_header('Content-type', 'text/xml')
self.end_headers()
return io.BytesIO(b'')
path = self.translate_path(self.path)
# the compression tests' filenames determine the header sent
if self.path.startswith('/tests/compression'):
if self.path.endswith('gz'):
headers = {'Content-Encoding': 'gzip'}
else:
headers = {'Content-Encoding': 'deflate'}
headers['Content-type'] = 'application/xml'
else:
with open(path, 'rb') as f:
blob = f.read()
headers = {
k.decode('utf-8'): v.decode('utf-8').strip()
for k, v in self.headers_re.findall(blob)
}
f = open(path, 'rb')
if (
self.headers.get('if-modified-since') == headers.get('Last-Modified', 'nom')
or self.headers.get('if-none-match') == headers.get('ETag', 'nomatch')
):
status = 304
else:
status = 200
headers.setdefault('Status', status)
self.send_response(int(headers['Status']))
headers.setdefault('Content-type', self.guess_type(path))
self.send_header("Content-type", headers['Content-type'])
self.send_header("Content-Length", str(os.stat(f.name)[6]))
for k, v in headers.items():
if k not in ('Status', 'Content-type'):
self.send_header(k, v)
self.end_headers()
return f
def log_request(self, *args):
pass
class FeedParserTestServer(threading.Thread):
"""HTTP Server that runs in a thread and handles a predetermined number of requests"""
def __init__(self, requests):
threading.Thread.__init__(self)
self.requests = requests
self.ready = threading.Event()
self.httpd = None
def run(self):
self.httpd = http.server.HTTPServer((_HOST, _PORT), FeedParserTestRequestHandler)
self.ready.set()
while self.requests:
self.httpd.handle_request()
self.requests -= 1
self.ready.clear()
self.httpd.shutdown()
# ---------- dummy test case class (test methods are added dynamically) ----------
# _bytes is only used in everything_is_unicode().
# In Python 2 it's str, and in Python 3 it's bytes.
_bytes = type(b'')
def everything_is_unicode(d):
"""Takes a dictionary, recursively verifies that every value is unicode"""
for k, v in d.items():
if isinstance(v, dict) and k != 'headers':
if not everything_is_unicode(v):
return False
elif isinstance(v, list):
for i in v:
if isinstance(i, dict) and not everything_is_unicode(i):
return False
elif isinstance(i, _bytes):
return False
elif isinstance(v, _bytes):
return False
return True
def fail_unless_eval(self, xmlfile, eval_string, msg=None):
"""Fail unless eval(eval_string, env)"""
env = feedparser.parse(xmlfile)
if not eval(eval_string, globals(), env):
failure = msg or 'not eval(%s) \nWITH env(%s)' % (eval_string, pprint.pformat(env))
raise self.failureException(failure)
if not everything_is_unicode(env):
raise self.failureException("not everything is unicode \nWITH env(%s)" % (pprint.pformat(env), ))
class BaseTestCase(unittest.TestCase):
fail_unless_eval = fail_unless_eval
class TestCase(BaseTestCase):
pass
class TestTemporaryFallbackBehavior(unittest.TestCase):
"""These tests are temporarily here because of issues 310 and 328"""
def test_issue_328_fallback_behavior(self):
warnings.filterwarnings('error')
d = feedparser.util.FeedParserDict()
d['published'] = 'pub string'
d['published_parsed'] = 'pub tuple'
d['updated'] = 'upd string'
d['updated_parsed'] = 'upd tuple'
# Ensure that `updated` doesn't map to `published` when it exists
self.assertTrue('published' in d)
self.assertTrue('published_parsed' in d)
self.assertTrue('updated' in d)
self.assertTrue('updated_parsed' in d)
self.assertEqual(d['published'], 'pub string')
self.assertEqual(d['published_parsed'], 'pub tuple')
self.assertEqual(d['updated'], 'upd string')
self.assertEqual(d['updated_parsed'], 'upd tuple')
d = feedparser.util.FeedParserDict()
d['published'] = 'pub string'
d['published_parsed'] = 'pub tuple'
# Ensure that `updated` doesn't actually exist
self.assertTrue('updated' not in d)
self.assertTrue('updated_parsed' not in d)
# Ensure that accessing `updated` throws a DeprecationWarning
try:
d['updated']
except DeprecationWarning:
# Expected behavior
pass
else:
# Wrong behavior
self.assertEqual(True, False)
try:
d['updated_parsed']
except DeprecationWarning:
# Expected behavior
pass
else:
# Wrong behavior
self.assertEqual(True, False)
# Ensure that `updated` maps to `published`
warnings.filterwarnings('ignore')
self.assertEqual(d['updated'], 'pub string')
self.assertEqual(d['updated_parsed'], 'pub tuple')
warnings.resetwarnings()
class TestEverythingIsUnicode(unittest.TestCase):
"""Ensure that `everything_is_unicode()` is working appropriately"""
def test_everything_is_unicode(self):
self.assertTrue(everything_is_unicode(
{'a': 'a', 'b': ['b', {'c': 'c'}], 'd': {'e': 'e'}}
))
def test_not_everything_is_unicode(self):
self.assertFalse(everything_is_unicode({'a': b'a'}))
self.assertFalse(everything_is_unicode({'a': [b'a']}))
self.assertFalse(everything_is_unicode({'a': {'b': b'b'}}))
self.assertFalse(everything_is_unicode({'a': [{'b': b'b'}]}))
class TestLooseParser(BaseTestCase):
"""Test the sgmllib-based parser by manipulating feedparser into believing no XML parsers are installed"""
def __init__(self, arg):
unittest.TestCase.__init__(self, arg)
self._xml_available = feedparser.api._XML_AVAILABLE
def setUp(self):
feedparser.api._XML_AVAILABLE = False
def tearDown(self):
feedparser.api._XML_AVAILABLE = self._xml_available
class TestStrictParser(BaseTestCase):
pass
class TestEncodings(BaseTestCase):
def test_doctype_replacement(self):
"""Ensure that non-ASCII-compatible encodings don't hide disallowed ENTITY declarations"""
doc = """<?xml version="1.0" encoding="utf-16be"?>
<!DOCTYPE feed [
<!ENTITY exponential1 "bogus ">
<!ENTITY exponential2 "&exponential1;&exponential1;">
<!ENTITY exponential3 "&exponential2;&exponential2;">
]>
<feed><title type="html">&exponential3;</title></feed>"""
doc = codecs.BOM_UTF16_BE + doc.encode('utf-16be')
result = feedparser.parse(doc)
self.assertEqual(result['feed']['title'], '&exponential3')
def test_gb2312_converted_to_gb18030_in_xml_encoding(self):
# \u55de was chosen because it exists in gb18030 but not gb2312
feed = '''<?xml version="1.0" encoding="gb2312"?>
<feed><title>\u55de</title></feed>'''
result = feedparser.parse(feed.encode('gb18030'), response_headers={
'Content-Type': 'text/xml'
})
self.assertEqual(result.encoding, 'gb18030')
class TestFeedParserDict(unittest.TestCase):
"""Ensure that FeedParserDict returns values as expected and won't crash"""
def setUp(self):
self.d = feedparser.util.FeedParserDict()
def _check_key(self, k):
self.assertTrue(k in self.d)
self.assertTrue(hasattr(self.d, k))
self.assertEqual(self.d[k], 1)
self.assertEqual(getattr(self.d, k), 1)
def _check_no_key(self, k):
self.assertTrue(k not in self.d)
self.assertTrue(not hasattr(self.d, k))
def test_empty(self):
keys = (
'a', 'entries', 'id', 'guid', 'summary', 'subtitle', 'description',
'category', 'enclosures', 'license', 'categories',
)
for k in keys:
self._check_no_key(k)
self.assertTrue('items' not in self.d)
self.assertTrue(hasattr(self.d, 'items')) # dict.items() exists
def test_neutral(self):
self.d['a'] = 1
self._check_key('a')
def test_single_mapping_target_1(self):
self.d['id'] = 1
self._check_key('id')
self._check_key('guid')
def test_single_mapping_target_2(self):
self.d['guid'] = 1
self._check_key('id')
self._check_key('guid')
def test_multiple_mapping_target_1(self):
self.d['summary'] = 1
self._check_key('summary')
self._check_key('description')
def test_multiple_mapping_target_2(self):
self.d['subtitle'] = 1
self._check_key('subtitle')
self._check_key('description')
def test_multiple_mapping_mapped_key(self):
self.d['description'] = 1
self._check_key('summary')
self._check_key('description')
def test_license(self):
self.d['links'] = []
self.assertNotIn('license', self.d)
self.d['links'].append({'rel': 'license'})
self.assertNotIn('license', self.d)
self.d['links'].append({'rel': 'license', 'href': 'http://dom.test/'})
self.assertIn('license', self.d)
self.assertEqual(self.d['license'], 'http://dom.test/')
def test_category(self):
self.d['tags'] = []
self.assertNotIn('category', self.d)
self.d['tags'] = [{}]
self.assertNotIn('category', self.d)
self.d['tags'] = [{'term': 'cat'}]
self.assertIn('category', self.d)
self.assertEqual(self.d['category'], 'cat')
self.d['tags'].append({'term': 'dog'})
self.assertEqual(self.d['category'], 'cat')
class TestOpenResource(unittest.TestCase):
"""Ensure that `_open_resource()` interprets its arguments as URIs, file-like objects, or in-memory feeds as expected"""
def test_fileobj(self):
r = feedparser.api._open_resource(io.BytesIO(b''), '', '', '', '', [], {}, {})
self.assertEqual(r, b'')
def test_feed(self):
f = feedparser.parse('feed://localhost:8097/tests/http/target.xml')
self.assertEqual(f.href, 'http://localhost:8097/tests/http/target.xml')
def test_feed_http(self):
f = feedparser.parse('feed:http://localhost:8097/tests/http/target.xml')
self.assertEqual(f.href, 'http://localhost:8097/tests/http/target.xml')
def test_bytes(self):
s = b'<feed><item><title>text</title></item></feed>'
r = feedparser.api._open_resource(s, '', '', '', '', [], {}, {})
self.assertEqual(s, r)
def test_string(self):
s = b'<feed><item><title>text</title></item></feed>'
r = feedparser.api._open_resource(s, '', '', '', '', [], {}, {})
self.assertEqual(s, r)
def test_unicode_1(self):
s = b'<feed><item><title>text</title></item></feed>'
r = feedparser.api._open_resource(s, '', '', '', '', [], {}, {})
self.assertEqual(s, r)
def test_unicode_2(self):
s = br'<feed><item><title>t\u00e9xt</title></item></feed>'
r = feedparser.api._open_resource(s, '', '', '', '', [], {}, {})
self.assertEqual(s, r)
def test_http_client_ascii_unicode_encode_error(self):
"""Confirm that a Unicode character doesn't cause a UnicodeEncodeError crash."""
url = 'https://0.0.0.0/รด'
with self.assertRaises(urllib.error.URLError):
feedparser.api._open_resource(url, '', '', '', '', [], {}, {})
def test_http_client_basic_auth_type_error(self):
"""Confirm an in-URL username/password doesn't cause a TypeError."""
url = 'https://username@password@0.0.0.0/feed'
with self.assertRaises(urllib.error.URLError):
feedparser.api._open_resource(url, '', '', '', '', [], {}, {})
def test_http_client_urllib_error(self):
"""Confirm urllib.error.URLError is caught correctly.
urllib.request.AbstractHTTPHandler.do_open() may explicitly raise
urllib.error.URLError in e.g. Python 3.9.
"""
original_do_open = urllib.request.AbstractHTTPHandler.do_open
exception = urllib.error.URLError('bogus')
def do_open(*args, **kwargs):
raise exception
urllib.request.AbstractHTTPHandler.do_open = do_open
result = feedparser.parse('https://bogus.example')
assert result['bozo'] is True
assert result['bozo_exception'] is exception
urllib.request.AbstractHTTPHandler.do_open = original_do_open
def make_safe_uri_test(rel, expect, doc):
def fn(self):
value = feedparser.urls.make_safe_absolute_uri(self.base, rel)
self.assertEqual(value, expect)
fn.__doc__ = doc
return fn
class TestMakeSafeAbsoluteURI(unittest.TestCase):
"""Exercise the URI joining and sanitization code"""
base = 'http://d.test/d/f.ext'
# make the test cases; the call signature is:
# (relative_url, expected_return_value, test_doc_string)
test_abs = make_safe_uri_test('https://s.test/', 'https://s.test/', 'absolute uri')
test_rel = make_safe_uri_test('/new', 'http://d.test/new', 'relative uri')
test_bad = make_safe_uri_test('x://bad.test/', '', 'unacceptable uri protocol')
test_mag = make_safe_uri_test('magnet:?xt=a', 'magnet:?xt=a', 'magnet uri')
def test_catch_ValueError(self):
"""catch ValueError in Python 2.7 and up"""
uri = 'http://bad]test/'
value1 = feedparser.urls.make_safe_absolute_uri(uri)
value2 = feedparser.urls.make_safe_absolute_uri(self.base, uri)
swap = feedparser.urls.ACCEPTABLE_URI_SCHEMES
feedparser.urls.ACCEPTABLE_URI_SCHEMES = ()
value3 = feedparser.urls.make_safe_absolute_uri(self.base, uri)
feedparser.urls.ACCEPTABLE_URI_SCHEMES = swap
# Only Python 2.7 and up throw a ValueError, otherwise uri is returned
self.assertTrue(value1 in (uri, ''))
self.assertTrue(value2 in (uri, ''))
self.assertTrue(value3 in (uri, ''))
class TestConvertToIdn(unittest.TestCase):
"""Test IDN support (unavailable in Jython as of Jython 2.5.2)"""
# this is the greek test domain
hostname = '\u03c0\u03b1\u03c1\u03ac\u03b4\u03b5\u03b9\u03b3\u03bc\u03b1'
hostname += '.\u03b4\u03bf\u03ba\u03b9\u03bc\u03ae'
def test_control(self):
r = feedparser.urls.convert_to_idn('http://example.test/')
self.assertEqual(r, 'http://example.test/')
def test_idn(self):
r = feedparser.urls.convert_to_idn('http://%s/' % (self.hostname,))
self.assertEqual(r, 'http://xn--hxajbheg2az3al.xn--jxalpdlp/')
def test_port(self):
r = feedparser.urls.convert_to_idn('http://%s:8080/' % (self.hostname,))
self.assertEqual(r, 'http://xn--hxajbheg2az3al.xn--jxalpdlp:8080/')
class TestCompression(unittest.TestCase):
"""Test the gzip and deflate support in the HTTP code"""
def test_gzip_good(self):
f = feedparser.parse('http://localhost:8097/tests/compression/gzip.gz')
self.assertEqual(f.version, 'atom10')
def test_gzip_not_compressed(self):
f = feedparser.parse('http://localhost:8097/tests/compression/gzip-not-compressed.gz')
self.assertEqual(f.bozo, 1)
self.assertTrue(isinstance(f.bozo_exception, IOError))
self.assertEqual(f['feed']['title'], 'gzip')
def test_gzip_struct_error(self):
f = feedparser.parse('http://localhost:8097/tests/compression/gzip-struct-error.gz')
self.assertEqual(f.bozo, 1)
# Python 3.4 throws an EOFError that gets overwritten as xml.sax.SAXException later.
self.assertTrue(isinstance(f.bozo_exception, (xml.sax.SAXException, struct.error)))
def test_zlib_good(self):
f = feedparser.parse('http://localhost:8097/tests/compression/deflate.z')
self.assertEqual(f.version, 'atom10')
def test_zlib_no_headers(self):
f = feedparser.parse('http://localhost:8097/tests/compression/deflate-no-headers.z')
self.assertEqual(f.version, 'atom10')
def test_zlib_not_compressed(self):
f = feedparser.parse('http://localhost:8097/tests/compression/deflate-not-compressed.z')
self.assertEqual(f.bozo, 1)
self.assertTrue(isinstance(f.bozo_exception, zlib.error))
self.assertEqual(f['feed']['title'], 'deflate')
class TestHTTPStatus(unittest.TestCase):
"""Test HTTP redirection and other status codes"""
def test_301(self):
f = feedparser.parse('http://localhost:8097/tests/http/http_status_301.xml')
self.assertEqual(f.status, 301)
self.assertEqual(f.href, 'http://localhost:8097/tests/http/target.xml')
self.assertEqual(f.entries[0].title, 'target')
def test_302(self):
f = feedparser.parse('http://localhost:8097/tests/http/http_status_302.xml')
self.assertEqual(f.status, 302)
self.assertEqual(f.href, 'http://localhost:8097/tests/http/target.xml')
self.assertEqual(f.entries[0].title, 'target')
def test_302_no_location(self):
"""Confirm no AttributeErrors when an HTTP 3xx response has no Location header."""
f = feedparser.parse('http://localhost:8097/tests/http/http_status_302-no-location.xml')
self.assertEqual(f.status, 302)
def test_303(self):
f = feedparser.parse('http://localhost:8097/tests/http/http_status_303.xml')
self.assertEqual(f.status, 303)
self.assertEqual(f.href, 'http://localhost:8097/tests/http/target.xml')
self.assertEqual(f.entries[0].title, 'target')
def test_307(self):
f = feedparser.parse('http://localhost:8097/tests/http/http_status_307.xml')
self.assertEqual(f.status, 307)
self.assertEqual(f.href, 'http://localhost:8097/tests/http/target.xml')
self.assertEqual(f.entries[0].title, 'target')
def test_304(self):
# first retrieve the url
u = 'http://localhost:8097/tests/http/http_status_304.xml'
f = feedparser.parse(u)
self.assertEqual(f.status, 200)
self.assertEqual(f.entries[0].title, 'title 304')
# extract the etag and last-modified headers
e = [v for k, v in f.headers.items() if k.lower() == 'etag'][0]
ms = f.updated
mt = f.updated_parsed
md = datetime.datetime(*mt[0:7])
self.assertTrue(isinstance(mt, time.struct_time))
self.assertTrue(isinstance(md, datetime.datetime))
# test that sending back the etag results in a 304
f = feedparser.parse(u, etag=e)
self.assertEqual(f.status, 304)
# test that sending back last-modified (string) results in a 304
f = feedparser.parse(u, modified=ms)
self.assertEqual(f.status, 304)
# test that sending back last-modified (9-tuple) results in a 304
f = feedparser.parse(u, modified=mt)
self.assertEqual(f.status, 304)
# test that sending back last-modified (datetime) results in a 304
f = feedparser.parse(u, modified=md)
self.assertEqual(f.status, 304)
def test_404(self):
f = feedparser.parse('http://localhost:8097/tests/http/http_status_404.xml')
self.assertEqual(f.status, 404)
def test_redirect_to_304(self):
# ensure that an http redirect to an http 304 doesn't
# trigger a bozo_exception
u = 'http://localhost:8097/tests/http/http_redirect_to_304.xml'
f = feedparser.parse(u)
self.assertTrue(f.bozo == 0)
self.assertTrue(f.status == 304)
class TestDateParsers(unittest.TestCase):
"""Test the various date parsers; most of the test cases are constructed dynamically based on the contents of the `date_tests` dict, below"""
def test_None(self):
self.assertTrue(feedparser.datetimes._parse_date(None) is None)
def _check_date(self, fn, dt, expected_value):
try:
parsed_value = fn(dt)
except (OverflowError, ValueError):
parsed_value = None
self.assertEqual(parsed_value, expected_value)
def test_year_10000_date(self):
# On some systems this date string will trigger an OverflowError.
# On Jython and x64 systems, however, it's interpreted just fine.
try:
date = _parse_date_rfc822('Sun, 31 Dec 9999 23:59:59 -9999')
except OverflowError:
date = None
self.assertTrue(date in (None, (10000, 1, 5, 4, 38, 59, 2, 5, 0)))
date_tests = {
_parse_date_greek: (
('', None), # empty string
('\u039a\u03c5\u03c1, 11 \u0399\u03bf\u03cd\u03bb 2004 12:00:00 EST', (2004, 7, 11, 17, 0, 0, 6, 193, 0)),
),
_parse_date_hungarian: (
('', None), # empty string
('2004-j\u00falius-13T9:15-05:00', (2004, 7, 13, 14, 15, 0, 1, 195, 0)),
),
_parse_date_iso8601: (
('', None), # empty string
('-0312', (2003, 12, 1, 0, 0, 0, 0, 335, 0)), # 2-digit year/month only variant
('031231', (2003, 12, 31, 0, 0, 0, 2, 365, 0)), # 2-digit year/month/day only, no hyphens
('03-12-31', (2003, 12, 31, 0, 0, 0, 2, 365, 0)), # 2-digit year/month/day only
('-03-12', (2003, 12, 1, 0, 0, 0, 0, 335, 0)), # 2-digit year/month only
('03335', (2003, 12, 1, 0, 0, 0, 0, 335, 0)), # 2-digit year/ordinal, no hyphens
('2003-12-31T10:14:55.1234Z', (2003, 12, 31, 10, 14, 55, 2, 365, 0)), # fractional seconds
# Special case for Google's extra zero in the month
('2003-012-31T10:14:55+00:00', (2003, 12, 31, 10, 14, 55, 2, 365, 0)),
),
_parse_date_nate: (
('', None), # empty string
('2004-05-25 \uc624\ud6c4 11:23:17', (2004, 5, 25, 14, 23, 17, 1, 146, 0)),
),
_parse_date_onblog: (
('', None), # empty string
('2004\ub144 05\uc6d4 28\uc77c 01:31:15', (2004, 5, 27, 16, 31, 15, 3, 148, 0)),
),
_parse_date_perforce: (
('', None), # empty string
('Fri, 2006/09/15 08:19:53 EDT', (2006, 9, 15, 12, 19, 53, 4, 258, 0)),
),
_parse_date_rfc822: (
('', None), # empty string
('Thu, 30 Apr 2015 08:57:00 MET', (2015, 4, 30, 7, 57, 0, 3, 120, 0)),
('Thu, 30 Apr 2015 08:57:00 MEST', (2015, 4, 30, 6, 57, 0, 3, 120, 0)),
('Thu, 01 Jan 0100 00:00:01 +0100', (99, 12, 31, 23, 0, 1, 3, 365, 0)), # ancient date
('Thu, 01 Jan 04 19:48:21 GMT', (2004, 1, 1, 19, 48, 21, 3, 1, 0)), # 2-digit year
('Thu, 01 Jan 2004 19:48:21 GMT', (2004, 1, 1, 19, 48, 21, 3, 1, 0)), # 4-digit year
('Thu, 5 Apr 2012 10:00:00 GMT', (2012, 4, 5, 10, 0, 0, 3, 96, 0)), # 1-digit day
('Wed, 19 Aug 2009 18:28:00 Etc/GMT', (2009, 8, 19, 18, 28, 0, 2, 231, 0)), # etc/gmt timezone
('Wed, 19 Feb 2012 22:40:00 GMT-01:01', (2012, 2, 19, 23, 41, 0, 6, 50, 0)), # gmt+hh:mm timezone
('Wed, 19 Feb 2012 22:40:00 -01:01', (2012, 2, 19, 23, 41, 0, 6, 50, 0)), # +hh:mm timezone
('Mon, 13 Feb, 2012 06:28:00 UTC', (2012, 2, 13, 6, 28, 0, 0, 44, 0)), # extraneous comma
('Thu, 01 Jan 2004 00:00 GMT', (2004, 1, 1, 0, 0, 0, 3, 1, 0)), # no seconds
('Thu, 01 Jan 2004', (2004, 1, 1, 0, 0, 0, 3, 1, 0)), # no time
# Additional tests to handle Disney's long month names and invalid timezones
('Mon, 26 January 2004 16:31:00 AT', (2004, 1, 26, 20, 31, 0, 0, 26, 0)),
('Mon, 26 January 2004 16:31:00 ET', (2004, 1, 26, 21, 31, 0, 0, 26, 0)),
('Mon, 26 January 2004 16:31:00 CT', (2004, 1, 26, 22, 31, 0, 0, 26, 0)),
('Mon, 26 January 2004 16:31:00 MT', (2004, 1, 26, 23, 31, 0, 0, 26, 0)),
('Mon, 26 January 2004 16:31:00 PT', (2004, 1, 27, 0, 31, 0, 1, 27, 0)),
# Swapped month and day
('Thu Aug 30 2012 17:26:16 +0200', (2012, 8, 30, 15, 26, 16, 3, 243, 0)),
('Sun, 16 Dec 2012 1:2:3:4 GMT', None), # invalid time
('Sun, 16 zzz 2012 11:47:32 GMT', None), # invalid month
('Sun, Dec x 2012 11:47:32 GMT', None), # invalid day (swapped day/month)
('Sun, 16 Dec zz 11:47:32 GMT', None), # invalid year
('Sun, 16 Dec 2012 11:47:32 +zz:00', None), # invalid timezone hour
('Sun, 16 Dec 2012 11:47:32 +00:zz', None), # invalid timezone minute
('Sun, 99 Jun 2009 12:00:00 GMT', None), # out-of-range day
),
_parse_date_asctime: (
('Sun Jan 4 16:29:06 2004', (2004, 1, 4, 16, 29, 6, 6, 4, 0)),
('Sun Jul 15 01:16:00 +0000 2012', (2012, 7, 15, 1, 16, 0, 6, 197, 0)),
),
_parse_date_w3dtf: (
('', None), # empty string
('2003-12-31T10:14:55Z', (2003, 12, 31, 10, 14, 55, 2, 365, 0)), # UTC
('2003-12-31T10:14:55-08:00', (2003, 12, 31, 18, 14, 55, 2, 365, 0)), # San Francisco timezone
('2003-12-31T18:14:55+08:00', (2003, 12, 31, 10, 14, 55, 2, 365, 0)), # Tokyo timezone
('2007-04-23T23:25:47.538+10:00', (2007, 4, 23, 13, 25, 47, 0, 113, 0)), # fractional seconds
('2003-12-31', (2003, 12, 31, 0, 0, 0, 2, 365, 0)), # year/month/day only
('2003-12', (2003, 12, 1, 0, 0, 0, 0, 335, 0)), # year/month only
('2003', (2003, 1, 1, 0, 0, 0, 2, 1, 0)), # year only
# Special cases for rollovers in leap years
('2004-02-28T18:14:55-08:00', (2004, 2, 29, 2, 14, 55, 6, 60, 0)), # feb 28 in leap year
('2003-02-28T18:14:55-08:00', (2003, 3, 1, 2, 14, 55, 5, 60, 0)), # feb 28 in non-leap year
('2000-02-28T18:14:55-08:00', (2000, 2, 29, 2, 14, 55, 1, 60, 0)), # feb 28 in leap year on century divisible by 400
# Out-of-range times
('9999-12-31T23:59:59-99:99', None), # Date is out-of-range
('2003-12-31T25:14:55Z', None), # invalid (25 hours)
('2003-12-31T10:61:55Z', None), # invalid (61 minutes)
('2003-12-31T10:14:61Z', None), # invalid (61 seconds)
# Invalid formats
('22013', None), # Year is too long
('013', None), # Year is too short
('2013-01-27-01', None), # Date has to many parts
('2013-01-28T11:30:00-06:00Textra', None), # Too many 't's
# Non-integer values
('2013-xx-27', None), # Date
('2013-01-28T09:xx:00Z', None), # Time
('2013-01-28T09:00:00+00:xx', None), # Timezone
# MSSQL-style dates
('2004-07-08 23:56:58 -00:20', (2004, 7, 9, 0, 16, 58, 4, 191, 0)), # with timezone
('2004-07-08 23:56:58', (2004, 7, 8, 23, 56, 58, 3, 190, 0)), # without timezone
('2004-07-08 23:56:58.0', (2004, 7, 8, 23, 56, 58, 3, 190, 0)), # with fractional second
)
}
def make_date_test(f, s, t):
return lambda self: self._check_date(f, s, t)
for func, items in date_tests.items():
for i, (dtstring, dttuple) in enumerate(items):
uniqfunc = make_date_test(func, dtstring, dttuple)
setattr(TestDateParsers, 'test_%s_%02i' % (func.__name__, i), uniqfunc)
def make_html_guess_test(text, expect, doc):
def fn(self):
value = bool(feedparser.mixin._FeedParserMixin.looks_like_html(text))
self.assertEqual(value, expect)
fn.__doc__ = doc
return fn
class TestHTMLGuessing(unittest.TestCase):
"""Exercise the HTML sniffing code"""
test_text_1 = make_html_guess_test('plain text', False, 'plain text')
test_text_2 = make_html_guess_test('2 < 3', False, 'plain text with angle bracket')
test_html_1 = make_html_guess_test('<a href="">a</a>', True, 'anchor tag')
test_html_2 = make_html_guess_test('<i>i</i>', True, 'italics tag')
test_html_3 = make_html_guess_test('<b>b</b>', True, 'bold tag')
test_html_4 = make_html_guess_test('<code>', False, 'allowed tag, no end tag')
test_html_5 = make_html_guess_test('<rss> .. </rss>', False, 'disallowed tag')
test_entity_1 = make_html_guess_test('AT&T', False, 'corporation name')
test_entity_2 = make_html_guess_test('©', True, 'named entity reference')
test_entity_3 = make_html_guess_test('©', True, 'numeric entity reference')
test_entity_4 = make_html_guess_test('©', True, 'hex numeric entity reference')
# ---------- additional api unit tests, not backed by files
class TestBuildRequest(unittest.TestCase):
"""Test that HTTP request objects are created as expected"""
def test_extra_headers(self):
"""You can pass in extra headers and they go into the request object."""
request = feedparser.http._build_urllib2_request(
'http://example.com/feed',
'agent-name',
None, None, None, None, None,
{'Cache-Control': 'max-age=0'})
# nb, urllib2 folds the case of the headers
self.assertEqual(
request.get_header('Cache-control'), 'max-age=0')
class TestLxmlBug(unittest.TestCase):
def test_lxml_etree_bug(self):
try:
import lxml.etree
except ImportError:
pass
else:
doc = b"<feed>&illformed_charref</feed>"
# Importing lxml.etree currently causes libxml2 to
# throw SAXException instead of SAXParseException.
feedparser.parse(io.BytesIO(doc))
self.assertTrue(True)
class TestParseFlags(unittest.TestCase):
feed_xml = b"""
<rss version="2.0">
<channel>
<item>
<body><script>alert("boo!")</script></body>
</item>
<item>
<body><a href="/boo.html">boo</a></body>
</item>
</channel>
</rss>
"""
def test_sanitize_html_default(self):
d = feedparser.parse(io.BytesIO(self.feed_xml))
self.assertEqual(u'', d.entries[0].content[0].value)
def test_sanitize_html_on(self):
d = feedparser.parse(io.BytesIO(self.feed_xml), sanitize_html=True)
self.assertEqual(u'', d.entries[0].content[0].value)
def test_sanitize_html_off(self):
d = feedparser.parse(io.BytesIO(self.feed_xml), sanitize_html=False)
self.assertEqual(u'<script>alert("boo!")</script>', d.entries[0].content[0].value)
def test_resolve_relative_uris_default(self):
d = feedparser.parse(io.BytesIO(self.feed_xml),
response_headers={'content-location': 'http://example.com/feed'})
self.assertEqual(u'<a href="http://example.com/boo.html">boo</a>', d.entries[1].content[0].value)
def test_resolve_relative_uris_on(self):
d = feedparser.parse(io.BytesIO(self.feed_xml),
response_headers={'content-location': 'http://example.com/feed'},
resolve_relative_uris=True)
self.assertEqual(u'<a href="http://example.com/boo.html">boo</a>', d.entries[1].content[0].value)
def test_resolve_relative_uris_off(self):
d = feedparser.parse(io.BytesIO(self.feed_xml),
response_headers={'content-location': 'http://example.com/feed.xml'},
resolve_relative_uris=False)
self.assertEqual(u'<a href="/boo.html">boo</a>', d.entries[1].content[0].value)
class TestSanitizer(unittest.TestCase):
def test_style_attr_is_enabled(self):
html = """<p style="margin: 15em;">example</p>"""
result = feedparser.sanitizer._sanitize_html(html, None, 'text/html')
self.assertEqual(result, html)
def test_style_attr_can_be_disabled(self):
html = """<p style="margin: 15em;">example</p>"""
expected = """<p>example</p>"""
original_attrs = feedparser.sanitizer._HTMLSanitizer.acceptable_attributes
feedparser.sanitizer._HTMLSanitizer.acceptable_attributes = set()
result = feedparser.sanitizer._sanitize_html(html, None, 'text/html')
feedparser.sanitizer._HTMLSanitizer.acceptable_attributes = original_attrs
self.assertEqual(result, expected)
# ---------- parse test files and create test methods ----------
def convert_to_utf8(data):
"""Identify data's encoding using its byte order mark and convert it to its utf-8 equivalent"""
if data[:4] == b'\x4c\x6f\xa7\x94':
return data.decode('cp037').encode('utf-8')
elif data[:4] == b'\x00\x00\xfe\xff':
return data.decode('utf-32be').encode('utf-8')
elif data[:4] == b'\xff\xfe\x00\x00':
return data.decode('utf-32le').encode('utf-8')
elif data[:4] == b'\x00\x00\x00\x3c':
return data.decode('utf-32be').encode('utf-8')
elif data[:4] == b'\x3c\x00\x00\x00':
return data.decode('utf-32le').encode('utf-8')
elif data[:4] == b'\x00\x3c\x00\x3f':
return data.decode('utf-16be').encode('utf-8')
elif data[:4] == b'\x3c\x00\x3f\x00':
return data.decode('utf-16le').encode('utf-8')
elif (data[:2] == b'\xfe\xff') and (data[2:4] != b'\x00\x00'):
return data[2:].decode('utf-16be').encode('utf-8')
elif (data[:2] == b'\xff\xfe') and (data[2:4] != b'\x00\x00'):
return data[2:].decode('utf-16le').encode('utf-8')
elif data[:3] == b'\xef\xbb\xbf':
return data[3:]
# no byte order mark was found
return data
skip_re = re.compile(br"SkipUnless:\s*(.*?)\n")
desc_re = re.compile(br"Description:\s*(.*?)\s*Expect:\s*(.*)\s*-->")
def get_description(xmlfile, data):
"""Extract test data
Each test case is an XML file which contains not only a test feed
but also the description of the test and the condition that we
would expect the parser to create when it parses the feed. Example:
<!--
Description: feed title
Expect: feed['title'] == 'Example feed'
-->
"""
skip_results = skip_re.search(data)
if skip_results:
skip_unless = skip_results.group(1).strip()
else:
skip_unless = '1'
search_results = desc_re.search(data)
if not search_results:
raise RuntimeError("can't parse %s" % xmlfile)
description, eval_string = [s.strip() for s in list(search_results.groups())]
description = xmlfile + ": " + description.decode('utf8')
return description, eval_string, skip_unless
def build_test_case(xmlfile, description, eval_string):
def fn(self, xmlfile=xmlfile, eval_string=eval_string):
self.fail_unless_eval(xmlfile, eval_string)
fn.__doc__ = description
return fn
def runtests():
"""Read the files in the tests/ directory, dynamically add tests to the TestCases above, spawn the HTTP server, and run the test suite"""
allfiles = glob.glob(os.path.join('.', 'tests', '**', '**', '*.xml'))
wellformedfiles = glob.glob(os.path.join('.', 'tests', 'wellformed', '**', '*.xml'))
illformedfiles = glob.glob(os.path.join('.', 'tests', 'illformed', '*.xml'))
encodingfiles = glob.glob(os.path.join('.', 'tests', 'encoding', '*.xml'))
entitiesfiles = glob.glob(os.path.join('.', 'tests', 'entities', '*.xml'))
# there are several compression test cases that must be accounted for
# as well as a number of http status tests that redirect to a target
# and a few `_open_resource`-related tests
httpcount = 7 + 16 + 2
httpcount += len([f for f in allfiles if 'http' in f])
httpcount += len([f for f in wellformedfiles if 'http' in f])
httpcount += len([f for f in illformedfiles if 'http' in f])
httpcount += len([f for f in encodingfiles if 'http' in f])
for c, xmlfile in enumerate(allfiles + encodingfiles + illformedfiles + entitiesfiles):
add_to = TestCase
if xmlfile in encodingfiles:
add_to = TestEncodings
elif xmlfile in entitiesfiles:
add_to = (TestStrictParser, TestLooseParser)
elif xmlfile in wellformedfiles:
add_to = (TestStrictParser, TestLooseParser)
f = open(xmlfile, 'rb')
data = f.read()
f.close()
if 'encoding' in xmlfile:
data = convert_to_utf8(data)
if data is None:
# convert_to_utf8 found a byte order mark for utf_32
# but it's not supported in this installation of Python
if 'http' in xmlfile:
httpcount -= 1 + (xmlfile in wellformedfiles)
continue
description, eval_string, skip_unless = get_description(xmlfile, data)
test_name = 'test_%06d' % c
is_http = 'http' in xmlfile
try:
if not eval(skip_unless):
raise NotImplementedError
except (ImportError, LookupError, NotImplementedError, AttributeError):
if is_http:
httpcount -= 1 + (xmlfile in wellformedfiles)
continue
if is_http:
xmlfile = 'http://%s:%s/%s' % (_HOST, _PORT, posixpath.normpath(xmlfile.replace('\\', '/')))
test_func = build_test_case(xmlfile, description, eval_string)
if isinstance(add_to, tuple):
setattr(add_to[0], test_name, test_func)
setattr(add_to[1], test_name, test_func)
else:
setattr(add_to, test_name, test_func)
if httpcount:
httpd = FeedParserTestServer(httpcount)
httpd.daemon = True
httpd.start()
httpd.ready.wait()
testsuite = unittest.TestSuite()
testloader = unittest.TestLoader()
testsuite.addTest(testloader.loadTestsFromTestCase(TestCase))
testsuite.addTest(testloader.loadTestsFromTestCase(TestStrictParser))
testsuite.addTest(testloader.loadTestsFromTestCase(TestLooseParser))
testsuite.addTest(testloader.loadTestsFromTestCase(TestEncodings))
testsuite.addTest(testloader.loadTestsFromTestCase(TestDateParsers))
testsuite.addTest(testloader.loadTestsFromTestCase(TestHTMLGuessing))
testsuite.addTest(testloader.loadTestsFromTestCase(TestHTTPStatus))
testsuite.addTest(testloader.loadTestsFromTestCase(TestCompression))
testsuite.addTest(testloader.loadTestsFromTestCase(TestConvertToIdn))
testsuite.addTest(testloader.loadTestsFromTestCase(TestOpenResource))
testsuite.addTest(testloader.loadTestsFromTestCase(TestFeedParserDict))
testsuite.addTest(testloader.loadTestsFromTestCase(TestMakeSafeAbsoluteURI))
testsuite.addTest(testloader.loadTestsFromTestCase(TestEverythingIsUnicode))
testsuite.addTest(testloader.loadTestsFromTestCase(TestTemporaryFallbackBehavior))
testsuite.addTest(testloader.loadTestsFromTestCase(TestLxmlBug))
testsuite.addTest(testloader.loadTestsFromTestCase(TestParseFlags))
testsuite.addTest(testloader.loadTestsFromTestCase(TestBuildRequest))
testsuite.addTest(testloader.loadTestsFromTestCase(TestSanitizer))
testresults = unittest.TextTestRunner(verbosity=1).run(testsuite)
# Return 0 if successful, 1 if there was a failure
sys.exit(not testresults.wasSuccessful())
if __name__ == "__main__":
runtests()
|