cpython: 3ac1b21fbb42 (original) (raw)

--- a/Lib/test/test_sax.py +++ b/Lib/test/test_sax.py @@ -15,10 +15,11 @@ from xml.sax.handler import feature_name from xml.sax.xmlreader import InputSource, AttributesImpl, AttributesNSImpl from cStringIO import StringIO import io +import gc import os.path import shutil import test.test_support as support -from test.test_support import findfile, run_unittest +from test.test_support import findfile, run_unittest, TESTFN import unittest TEST_XMLFILE = findfile("test.xml", subdir="xmltestdata") @@ -90,6 +91,111 @@ class XmlTestBase(unittest.TestCase): self.assertEqual(attrs["attr"], "val") self.assertEqual(attrs.getQNameByName("attr"), "attr") + +def xml_unicode(doc, encoding=None):

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

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

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

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

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

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

+ + +class ParseTest(unittest.TestCase):

data = support.u(r''

                r'$\xa3\u20ac\U0001017b</money>')[](#l1.40)

def tearDown(self):
```
   support.unlink(TESTFN)[](#l1.43)
```

def check_parse(self, f):
```
   from xml.sax import parse[](#l1.46)
```
```
   result = StringIO()[](#l1.47)
```

   parse(f, XMLGenerator(result, 'utf-8'))[](#l1.48)

   self.assertEqual(result.getvalue(), xml_bytes(self.data, 'utf-8'))[](#l1.49)

def test_parse_bytes(self):

   # UTF-8 is default encoding, US-ASCII is compatible with UTF-8,[](#l1.52)

```
   # UTF-16 is autodetected[](#l1.53)
```

   encodings = ('us-ascii', 'utf-8', 'utf-16', 'utf-16le', 'utf-16be')[](#l1.54)

```
   for encoding in encodings:[](#l1.55)
```

       self.check_parse(io.BytesIO(xml_bytes(self.data, encoding)))[](#l1.56)

       make_xml_file(self.data, encoding)[](#l1.57)

       self.check_parse(TESTFN)[](#l1.58)

       with io.open(TESTFN, 'rb') as f:[](#l1.59)

           self.check_parse(f)[](#l1.60)

       self.check_parse(io.BytesIO(xml_bytes(self.data, encoding, None)))[](#l1.61)

       make_xml_file(self.data, encoding, None)[](#l1.62)

       self.check_parse(TESTFN)[](#l1.63)

       with io.open(TESTFN, 'rb') as f:[](#l1.64)

           self.check_parse(f)[](#l1.65)

```
   # accept UTF-8 with BOM[](#l1.66)
```

   self.check_parse(io.BytesIO(xml_bytes(self.data, 'utf-8-sig', 'utf-8')))[](#l1.67)

   make_xml_file(self.data, 'utf-8-sig', 'utf-8')[](#l1.68)

```
   self.check_parse(TESTFN)[](#l1.69)
```

   with io.open(TESTFN, 'rb') as f:[](#l1.70)

```
       self.check_parse(f)[](#l1.71)
```

   self.check_parse(io.BytesIO(xml_bytes(self.data, 'utf-8-sig', None)))[](#l1.72)

   make_xml_file(self.data, 'utf-8-sig', None)[](#l1.73)

```
   self.check_parse(TESTFN)[](#l1.74)
```

   with io.open(TESTFN, 'rb') as f:[](#l1.75)

```
       self.check_parse(f)[](#l1.76)
```

   # accept data with declared encoding[](#l1.77)

   self.check_parse(io.BytesIO(xml_bytes(self.data, 'iso-8859-1')))[](#l1.78)

   make_xml_file(self.data, 'iso-8859-1')[](#l1.79)

```
   self.check_parse(TESTFN)[](#l1.80)
```

   with io.open(TESTFN, 'rb') as f:[](#l1.81)

```
       self.check_parse(f)[](#l1.82)
```

   # fail on non-UTF-8 incompatible data without declared encoding[](#l1.83)

   with self.assertRaises(SAXException):[](#l1.84)

       self.check_parse(io.BytesIO(xml_bytes(self.data, 'iso-8859-1', None)))[](#l1.85)

   make_xml_file(self.data, 'iso-8859-1', None)[](#l1.86)

   with self.assertRaises(SAXException):[](#l1.87)

       self.check_parse(TESTFN)[](#l1.88)

   with io.open(TESTFN, 'rb') as f:[](#l1.89)

       with self.assertRaises(SAXException):[](#l1.90)

           self.check_parse(f)[](#l1.91)

def test_parse_InputSource(self):

   # accept data without declared but with explicitly specified encoding[](#l1.94)

   make_xml_file(self.data, 'iso-8859-1', None)[](#l1.95)

   with io.open(TESTFN, 'rb') as f:[](#l1.96)

```
       input = InputSource()[](#l1.97)
```
```
       input.setByteStream(f)[](#l1.98)
```

       input.setEncoding('iso-8859-1')[](#l1.99)

       self.check_parse(input)[](#l1.100)

def check_parseString(self, s):

   from xml.sax import parseString[](#l1.103)

```
   result = StringIO()[](#l1.104)
```

   parseString(s, XMLGenerator(result, 'utf-8'))[](#l1.105)

   self.assertEqual(result.getvalue(), xml_bytes(self.data, 'utf-8'))[](#l1.106)

def test_parseString_bytes(self):

   # UTF-8 is default encoding, US-ASCII is compatible with UTF-8,[](#l1.109)

```
   # UTF-16 is autodetected[](#l1.110)
```

   encodings = ('us-ascii', 'utf-8', 'utf-16', 'utf-16le', 'utf-16be')[](#l1.111)

   for encoding in encodings:[](#l1.112)

       self.check_parseString(xml_bytes(self.data, encoding))[](#l1.113)

       self.check_parseString(xml_bytes(self.data, encoding, None))[](#l1.114)

```
   # accept UTF-8 with BOM[](#l1.115)
```

   self.check_parseString(xml_bytes(self.data, 'utf-8-sig', 'utf-8'))[](#l1.116)

   self.check_parseString(xml_bytes(self.data, 'utf-8-sig', None))[](#l1.117)

   # accept data with declared encoding[](#l1.118)

   self.check_parseString(xml_bytes(self.data, 'iso-8859-1'))[](#l1.119)

   # fail on non-UTF-8 incompatible data without declared encoding[](#l1.120)

   with self.assertRaises(SAXException):[](#l1.121)

       self.check_parseString(xml_bytes(self.data, 'iso-8859-1', None))[](#l1.122)

+ + class MakeParserTest(unittest.TestCase): def test_make_parser2(self): # Creating parsers several times in a row should succeed. @@ -949,6 +1055,7 @@ class XmlReaderTest(XmlTestBase): def test_main(): run_unittest(MakeParserTest,

            ParseTest,[](#l1.132)
            SaxutilsTest,[](#l1.133)
            PrepareInputSourceTest,[](#l1.134)
            StringXmlgenTest,[](#l1.135)