cpython: 846c165cf643 (original) (raw)

--- a/Lib/test/test_sax.py +++ b/Lib/test/test_sax.py @@ -16,10 +16,11 @@ from xml.sax.handler import feature_name from xml.sax.xmlreader import InputSource, AttributesImpl, AttributesNSImpl from io import BytesIO, StringIO import codecs +import gc import os.path import shutil from test import support -from test.support import findfile, run_unittest +from test.support import findfile, run_unittest, TESTFN TEST_XMLFILE = findfile("test.xml", subdir="xmltestdata") TEST_XMLFILE_OUT = findfile("test.xml.out", subdir="xmltestdata") @@ -95,6 +96,126 @@ class XmlTestBase(unittest.TestCase): self.assertEqual(attrs["attr"], "val") self.assertEqual(attrs.getQNameByName("attr"), "attr") + +def xml_str(doc, encoding=None):

if encoding is None:
```
   return doc[](#l1.23)
```
return '\n%s' % (encoding, doc)

+ +def xml_bytes(doc, encoding, decl_encoding=...):

if decl_encoding is ...:
```
   decl_encoding = encoding[](#l1.28)
```
return xml_str(doc, decl_encoding).encode(encoding, 'xmlcharrefreplace')

+ +def make_xml_file(doc, encoding, decl_encoding=...):

if decl_encoding is ...:
```
   decl_encoding = encoding[](#l1.33)
```
with open(TESTFN, 'w', encoding=encoding, errors='xmlcharrefreplace') as f:

   f.write(xml_str(doc, decl_encoding))[](#l1.35)

+ + +class ParseTest(unittest.TestCase):

data = '$\xa3\u20ac\U0001017b'

def tearDown(self):
```
   support.unlink(TESTFN)[](#l1.42)
```

def check_parse(self, f):
```
   from xml.sax import parse[](#l1.45)
```
```
   result = StringIO()[](#l1.46)
```

   parse(f, XMLGenerator(result, 'utf-8'))[](#l1.47)

   self.assertEqual(result.getvalue(), xml_str(self.data, 'utf-8'))[](#l1.48)

def test_parse_text(self):

   encodings = ('us-ascii', 'iso-8859-1', 'utf-8',[](#l1.51)

                'utf-16', 'utf-16le', 'utf-16be')[](#l1.52)

```
   for encoding in encodings:[](#l1.53)
```

       self.check_parse(StringIO(xml_str(self.data, encoding)))[](#l1.54)

       make_xml_file(self.data, encoding)[](#l1.55)

       with open(TESTFN, 'r', encoding=encoding) as f:[](#l1.56)

           self.check_parse(f)[](#l1.57)

       self.check_parse(StringIO(self.data))[](#l1.58)

       make_xml_file(self.data, encoding, None)[](#l1.59)

       with open(TESTFN, 'r', encoding=encoding) as f:[](#l1.60)

           self.check_parse(f)[](#l1.61)

def test_parse_bytes(self):

   # UTF-8 is default encoding, US-ASCII is compatible with UTF-8,[](#l1.64)

```
   # UTF-16 is autodetected[](#l1.65)
```

   encodings = ('us-ascii', 'utf-8', 'utf-16', 'utf-16le', 'utf-16be')[](#l1.66)

```
   for encoding in encodings:[](#l1.67)
```

       self.check_parse(BytesIO(xml_bytes(self.data, encoding)))[](#l1.68)

       make_xml_file(self.data, encoding)[](#l1.69)

       self.check_parse(TESTFN)[](#l1.70)

       with open(TESTFN, 'rb') as f:[](#l1.71)

           self.check_parse(f)[](#l1.72)

       self.check_parse(BytesIO(xml_bytes(self.data, encoding, None)))[](#l1.73)

       make_xml_file(self.data, encoding, None)[](#l1.74)

       self.check_parse(TESTFN)[](#l1.75)

       with open(TESTFN, 'rb') as f:[](#l1.76)

           self.check_parse(f)[](#l1.77)

```
   # accept UTF-8 with BOM[](#l1.78)
```

   self.check_parse(BytesIO(xml_bytes(self.data, 'utf-8-sig', 'utf-8')))[](#l1.79)

   make_xml_file(self.data, 'utf-8-sig', 'utf-8')[](#l1.80)

```
   self.check_parse(TESTFN)[](#l1.81)
```

   with open(TESTFN, 'rb') as f:[](#l1.82)

```
       self.check_parse(f)[](#l1.83)
```

   self.check_parse(BytesIO(xml_bytes(self.data, 'utf-8-sig', None)))[](#l1.84)

   make_xml_file(self.data, 'utf-8-sig', None)[](#l1.85)

```
   self.check_parse(TESTFN)[](#l1.86)
```

   with open(TESTFN, 'rb') as f:[](#l1.87)

```
       self.check_parse(f)[](#l1.88)
```

   # accept data with declared encoding[](#l1.89)

   self.check_parse(BytesIO(xml_bytes(self.data, 'iso-8859-1')))[](#l1.90)

   make_xml_file(self.data, 'iso-8859-1')[](#l1.91)

```
   self.check_parse(TESTFN)[](#l1.92)
```

   with open(TESTFN, 'rb') as f:[](#l1.93)

```
       self.check_parse(f)[](#l1.94)
```

   # fail on non-UTF-8 incompatible data without declared encoding[](#l1.95)

   with self.assertRaises(SAXException):[](#l1.96)

       self.check_parse(BytesIO(xml_bytes(self.data, 'iso-8859-1', None)))[](#l1.97)

   make_xml_file(self.data, 'iso-8859-1', None)[](#l1.98)

   with support.check_warnings(('unclosed file', ResourceWarning)):[](#l1.99)

       # XXX Failed parser leaks an opened file.[](#l1.100)

       with self.assertRaises(SAXException):[](#l1.101)

           self.check_parse(TESTFN)[](#l1.102)

       # Collect leaked file.[](#l1.103)

```
       gc.collect()[](#l1.104)
```

   with open(TESTFN, 'rb') as f:[](#l1.105)

       with self.assertRaises(SAXException):[](#l1.106)

           self.check_parse(f)[](#l1.107)

def test_parse_InputSource(self):

   # accept data without declared but with explicitly specified encoding[](#l1.110)

   make_xml_file(self.data, 'iso-8859-1', None)[](#l1.111)

   with open(TESTFN, 'rb') as f:[](#l1.112)

```
       input = InputSource()[](#l1.113)
```

       input.setByteStream(f)[](#l1.114)

       input.setEncoding('iso-8859-1')[](#l1.115)

       self.check_parse(input)[](#l1.116)

def check_parseString(self, s):

   from xml.sax import parseString[](#l1.119)

```
   result = StringIO()[](#l1.120)
```

   parseString(s, XMLGenerator(result, 'utf-8'))[](#l1.121)

   self.assertEqual(result.getvalue(), xml_str(self.data, 'utf-8'))[](#l1.122)

def test_parseString_bytes(self):

   # UTF-8 is default encoding, US-ASCII is compatible with UTF-8,[](#l1.125)

```
   # UTF-16 is autodetected[](#l1.126)
```

   encodings = ('us-ascii', 'utf-8', 'utf-16', 'utf-16le', 'utf-16be')[](#l1.127)

   for encoding in encodings:[](#l1.128)

       self.check_parseString(xml_bytes(self.data, encoding))[](#l1.129)

       self.check_parseString(xml_bytes(self.data, encoding, None))[](#l1.130)

```
   # accept UTF-8 with BOM[](#l1.131)
```

   self.check_parseString(xml_bytes(self.data, 'utf-8-sig', 'utf-8'))[](#l1.132)

   self.check_parseString(xml_bytes(self.data, 'utf-8-sig', None))[](#l1.133)

   # accept data with declared encoding[](#l1.134)

   self.check_parseString(xml_bytes(self.data, 'iso-8859-1'))[](#l1.135)

   # fail on non-UTF-8 incompatible data without declared encoding[](#l1.136)

   with self.assertRaises(SAXException):[](#l1.137)

       self.check_parseString(xml_bytes(self.data, 'iso-8859-1', None))[](#l1.138)

+ class MakeParserTest(unittest.TestCase): def test_make_parser2(self): # Creating parsers several times in a row should succeed. @@ -1115,6 +1236,7 @@ class XmlReaderTest(XmlTestBase): def test_main(): run_unittest(MakeParserTest,

            ParseTest,[](#l1.147)
            SaxutilsTest,[](#l1.148)
            PrepareInputSourceTest,[](#l1.149)
            StringXmlgenTest,[](#l1.150)