Allow Path-like objects for file arguments (GH-337)

Use "PyOS_FSPath()" if available (Py3.6+). Otherwise, manually check for "__fspath__", in case an object defines it.

5 files changed, 84 insertions, 3 deletions

diff --git a/src/lxml/tests/common_imports.py b/src/lxml/tests/common_imports.py
index 57097e3c..68db7c2b 100644
--- a/src/lxml/tests/common_imports.py
+++ b/src/lxml/tests/common_imports.py
@@ -251,6 +251,12 @@ class LargeFileLikeUnicode(LargeFileLike):
             yield self.chars
         yield _str('</root>')
 
+class SimpleFSPath(object):
+    def __init__(self, path):
+        self.path = path
+    def __fspath__(self):
+        return self.path
+
 def fileInTestDir(name):
     _testdir = os.path.dirname(__file__)
     return os.path.join(_testdir, name)
diff --git a/src/lxml/tests/test_dtd.py b/src/lxml/tests/test_dtd.py
index 779f9e84..5c9b1c02 100644
--- a/src/lxml/tests/test_dtd.py
+++ b/src/lxml/tests/test_dtd.py
@@ -9,7 +9,7 @@ import unittest, sys
 from .common_imports import (
     etree, html, BytesIO, _bytes, _str,
     HelperTestCase, make_doctest, skipIf,
-    fileInTestDir, fileUrlInTestDir
+    fileInTestDir, fileUrlInTestDir, SimpleFSPath
 )
 
 
@@ -24,6 +24,14 @@ class ETreeDtdTestCase(HelperTestCase):
 
         dtd = etree.DTD(fileInTestDir("test.dtd"))
         self.assertTrue(dtd.validate(root))
+    
+    def test_dtd_file_pathlike(self):
+        parse = etree.parse
+        tree = parse(fileInTestDir("test.xml"))
+        root = tree.getroot()
+
+        dtd = etree.DTD(SimpleFSPath(fileInTestDir("test.dtd")))
+        self.assertTrue(dtd.validate(root))
 
     def test_dtd_stringio(self):
         root = etree.XML(_bytes("<b/>"))
diff --git a/src/lxml/tests/test_etree.py b/src/lxml/tests/test_etree.py
index ef5c54b7..e5f08469 100644
--- a/src/lxml/tests/test_etree.py
+++ b/src/lxml/tests/test_etree.py
@@ -25,6 +25,7 @@ from .common_imports import etree, StringIO, BytesIO, HelperTestCase
 from .common_imports import fileInTestDir, fileUrlInTestDir, read_file, path2url, tmpfile
 from .common_imports import SillyFileLike, LargeFileLikeUnicode, doctest, make_doctest
 from .common_imports import canonicalize, _str, _bytes
+from .common_imports import SimpleFSPath
 
 print("""
 TESTED VERSION: %s""" % etree.__version__ + """
@@ -4599,6 +4600,20 @@ class ETreeOnlyTestCase(HelperTestCase):
         self.assertEqual('child1', c2.getprevious().tag)
         self.assertEqual('abc', c2.getprevious().tail)
 
+    def test_parse_source_pathlike(self):
+        etree = self.etree
+        tounicode = self.etree.tounicode
+
+        tree = etree.parse(SimpleFSPath(fileInTestDir('test.xml')))
+        self.assertEqual(_bytes('<a><b></b></a>'),
+                         canonicalize(tounicode(tree)))
+    
+    def test_iterparse_source_pathlike(self):
+        iterparse = self.etree.iterparse
+
+        events = list(iterparse(SimpleFSPath(fileInTestDir('test.xml'))))
+        self.assertEqual(2, len(events))
+
     # helper methods
 
     def _writeElement(self, element, encoding='us-ascii', compression=0):
@@ -4883,6 +4898,14 @@ class ETreeC14NTestCase(HelperTestCase):
             data = read_file(filename, 'rb')
         self.assertEqual(_bytes('<a><b></b></a>'),
                           data)
+    
+    def test_c14n_file_pathlike(self):
+        tree = self.parse(_bytes('<a><b/></a>'))
+        with tmpfile() as filename:
+            tree.write_c14n(SimpleFSPath(filename))
+            data = read_file(filename, 'rb')
+        self.assertEqual(_bytes('<a><b></b></a>'),
+                        data)
 
     def test_c14n_file_gzip(self):
         tree = self.parse(_bytes('<a>'+'<b/>'*200+'</a>'))
@@ -4892,6 +4915,15 @@ class ETreeC14NTestCase(HelperTestCase):
                 data = f.read()
         self.assertEqual(_bytes('<a>'+'<b></b>'*200+'</a>'),
                           data)
+    
+    def test_c14n_file_gzip_pathlike(self):
+        tree = self.parse(_bytes('<a>'+'<b/>'*200+'</a>'))
+        with tmpfile() as filename:
+            tree.write_c14n(SimpleFSPath(filename), compression=9)
+            with gzip.open(filename, 'rb') as f:
+                data = f.read()
+        self.assertEqual(_bytes('<a>'+'<b></b>'*200+'</a>'),
+                        data)
 
     def test_c14n2_file_gzip(self):
         tree = self.parse(_bytes('<a>'+'<b/>'*200+'</a>'))
@@ -5182,6 +5214,14 @@ class ETreeWriteTestCase(HelperTestCase):
             data = read_file(filename, 'rb')
         self.assertEqual(_bytes('<a><b/></a>'),
                           data)
+    
+    def test_write_file_pathlike(self):
+        tree = self.parse(_bytes('<a><b/></a>'))
+        with tmpfile() as filename:
+            tree.write(SimpleFSPath(filename))
+            data = read_file(filename, 'rb')
+        self.assertEqual(_bytes('<a><b/></a>'),
+                        data)
 
     def test_write_file_gzip(self):
         tree = self.parse(_bytes('<a>'+'<b/>'*200+'</a>'))
@@ -5192,6 +5232,15 @@ class ETreeWriteTestCase(HelperTestCase):
         self.assertEqual(_bytes('<a>'+'<b/>'*200+'</a>'),
                           data)
 
+    def test_write_file_gzip_pathlike(self):
+        tree = self.parse(_bytes('<a>'+'<b/>'*200+'</a>'))
+        with tmpfile() as filename:
+            tree.write(SimpleFSPath(filename), compression=9)
+            with gzip.open(filename, 'rb') as f:
+                data = f.read()
+        self.assertEqual(_bytes('<a>'+'<b/>'*200+'</a>'),
+                        data)
+
     def test_write_file_gzip_parse(self):
         tree = self.parse(_bytes('<a>'+'<b/>'*200+'</a>'))
         with tmpfile() as filename:
diff --git a/src/lxml/tests/test_xmlschema.py b/src/lxml/tests/test_xmlschema.py
index c5653c1e..dbfc251a 100644
--- a/src/lxml/tests/test_xmlschema.py
+++ b/src/lxml/tests/test_xmlschema.py
@@ -8,7 +8,7 @@ from __future__ import absolute_import
 
 import unittest
 
-from .common_imports import etree, BytesIO, HelperTestCase, fileInTestDir, make_doctest
+from .common_imports import etree, BytesIO, HelperTestCase, fileInTestDir, make_doctest, SimpleFSPath
 
 
 class ETreeXMLSchemaTestCase(HelperTestCase):
@@ -387,6 +387,11 @@ class ETreeXMLSchemaTestCase(HelperTestCase):
         etree.XMLSchema(schema_element)
         etree.XMLSchema(schema_element)
 
+    def test_xmlschema_pathlike(self):
+        schema = etree.XMLSchema(file=SimpleFSPath(fileInTestDir('test.xsd')))
+        tree_valid = self.parse('<a><b></b></a>')
+        self.assertTrue(schema.validate(tree_valid))
+
 
 class ETreeXMLSchemaResolversTestCase(HelperTestCase):
     resolver_schema_int = BytesIO("""\
diff --git a/src/lxml/tests/test_xslt.py b/src/lxml/tests/test_xslt.py
index cde23357..0ef07669 100644
--- a/src/lxml/tests/test_xslt.py
+++ b/src/lxml/tests/test_xslt.py
@@ -29,7 +29,7 @@ except NameError: # Python 3
     basestring = str
 
 from .common_imports import (
-    etree, BytesIO, HelperTestCase, fileInTestDir, _bytes, make_doctest, skipif
+    etree, BytesIO, HelperTestCase, fileInTestDir, _bytes, make_doctest, skipif, SimpleFSPath
 )
 
 
@@ -195,6 +195,19 @@ class ETreeXSLTTestCase(HelperTestCase):
                     res[0] = f.read().decode("UTF-16")
             finally:
                 os.unlink(f.name)
+    
+    def test_xslt_write_output_file_pathlike(self):
+        with self._xslt_setup() as res:
+            f = NamedTemporaryFile(delete=False)
+            try:
+                try:
+                    res[0].write_output(SimpleFSPath(f.name), compression=9)
+                finally:
+                    f.close()
+                with gzip.GzipFile(f.name) as f:
+                    res[0] = f.read().decode("UTF-16")
+            finally:
+                os.unlink(f.name)
 
     def test_xslt_write_output_file_path_urlescaped(self):
         # libxml2 should not unescape file paths.