read_html(): rewinding [wip] (pandas-dev#18017)

Author: LiamIm

doc/source/whatsnew/v0.22.0.txt

@@ -107,7 +107,7 @@ Indexing
 I/O
 ^^^
 
--
+- :func:`read_html` now rewinds seekable IO objects after parse failure, before attempting to parse with a new parser. If a parser errors and the object is non-seekable, an informative error is raised suggesting the use of a different parser (:issue:`17975`)
 -
 -
 

pandas/io/html.py

@@ -742,6 +742,18 @@ def _parse(flavor, io, match, attrs, encoding, **kwargs):
         try:
             tables = p.parse_tables()
         except Exception as caught:
+            # if `io` is an io-like object, check if it's seekable
+            # and try to rewind it before trying the next parser
+            if hasattr(io, 'seekable') and io.seekable():
+                io.seek(0)
+            elif hasattr(io, 'seekable') and not io.seekable():
+                # if we couldn't rewind it, let the user know
+                raise ValueError('The flavor {} failed to parse your input. '
+                                 'Since you passed a non-rewindable file '
+                                 'object, we can\'t rewind it to try '
+                                 'another parser. Try read_html() with a '
+                                 'different flavor.'.format(flav))
+
             retained = caught
         else:
             break

pandas/tests/io/test_html.py

@@ -968,3 +968,53 @@ def test_importcheck_thread_safety():
     while helper_thread1.is_alive() or helper_thread2.is_alive():
         pass
     assert None is helper_thread1.err is helper_thread2.err
+
+
+def test_parse_failure_unseekable():
+    # Issue #17975
+    _skip_if_no('lxml')
+
+    class UnseekableStringIO(StringIO):
+        def seekable(self):
+            return False
+
+    good = UnseekableStringIO('''
+        <table><tr><td>spam<br />eggs</td></tr></table>''')
+    bad = UnseekableStringIO('''
+        <table><tr><td>spam<foobr />eggs</td></tr></table>''')
+
+    assert read_html(good)
+    assert read_html(bad, flavor='bs4')
+
+    bad.seek(0)
+
+    with pytest.raises(ValueError,
+                       match='passed a non-rewindable file object'):
+        read_html(bad)
+
+
+def test_parse_failure_rewinds():
+    # Issue #17975
+    _skip_if_no('lxml')
+
+    class MockFile(object):
+        def __init__(self, data):
+            self.data = data
+            self.at_end = False
+
+        def read(self, size=None):
+            data = '' if self.at_end else self.data
+            self.at_end = True
+            return data
+
+        def seek(self, offset):
+            self.at_end = False
+
+        def seekable(self):
+            return True
+
+    good = MockFile('<table><tr><td>spam<br />eggs</td></tr></table>')
+    bad = MockFile('<table><tr><td>spam<foobr />eggs</td></tr></table>')
+
+    assert read_html(good)
+    assert read_html(bad)