def testSaveIntoOutputBuffer():
"""
Similar to the previous two tests, except this time we invoke the save
methods on the output buffer object and pass in an XML node object.
"""
input = '<foo>Hello</foo>'
expected = '''\
<?xml version="1.0" encoding="UTF-8"?>
<foo>Hello</foo>
'''
f = str_io()
doc = libxml2.parseDoc(input)
buf = libxml2.createOutputBuffer(f, 'UTF-8')
buf.saveFileTo(doc, 'UTF-8')
if f.getvalue() != expected:
print('outputBuffer.saveFileTo() call failed.')
print(' got: %s' % repr(f.getvalue()))
print('expected: %s' % repr(expected))
sys.exit(1)
f = str_io()
buf = libxml2.createOutputBuffer(f, 'UTF-8')
buf.saveFormatFileTo(doc, 'UTF-8', 1)
if f.getvalue() != expected:
print('outputBuffer.saveFormatFileTo() call failed.')
print(' got: %s' % repr(f.getvalue()))
print('expected: %s' % repr(expected))
sys.exit(1)
doc.freeDoc()
def testSaveIntoOutputBuffer():
"""
Similar to the previous two tests, except this time we invoke the save
methods on the output buffer object and pass in an XML node object.
"""
input = '<foo>Hello</foo>'
expected = '''\
<?xml version="1.0" encoding="UTF-8"?>
<foo>Hello</foo>
'''
f = StringIO.StringIO()
doc = libxml2.parseDoc(input)
buf = libxml2.createOutputBuffer(f, 'UTF-8')
buf.saveFileTo(doc, 'UTF-8')
if f.getvalue() != expected:
print 'outputBuffer.saveFileTo() call failed.'
print ' got: %s' % repr(f.getvalue())
print 'expected: %s' % repr(expected)
sys.exit(1)
f = StringIO.StringIO()
buf = libxml2.createOutputBuffer(f, 'UTF-8')
buf.saveFormatFileTo(doc, 'UTF-8', 1)
if f.getvalue() != expected:
print 'outputBuffer.saveFormatFileTo() call failed.'
print ' got: %s' % repr(f.getvalue())
print 'expected: %s' % repr(expected)
sys.exit(1)
doc.freeDoc()
def flatten(string):
doc = libxml2.parseDoc(string)
context = doc.xpathNewContext()
result = context.xpathEval("//use")
for orig in result:
xlink = orig.prop('href')
id = orig.prop('id')
id_length = len(id)
nodes = context.xpathEval('//g[@id="'+xlink[1:]+'"]')
if len(nodes) == 0:
nodes = context.xpathEval('//image[@id="'+xlink[1:]+'"]')
node = nodes[0]
copy = node.copyNode(extended=True)
copy.removeNsDef(None)
copy_context = doc.xpathNewContext()
copy_context.setContextNode(copy)
for copy_id in copy_context.xpathEval('.//@id'):
copy_id.setContent(id + copy_id.content[id_length:])
tx, ty = re.match('translate\((-?\d+\.?\d*.*),(-?\d+\.?\d*.*)\)', orig.prop('transform')).groups()
transform = { 'x': float(tx), 'y': float(ty) }
for c in [ 'x', 'y' ]:
for coord in copy_context.xpathEval('.//@' + c):
coord.setContent(str(int(round(float(coord.content) + transform[c]))))
orig.replaceNode(copy)
f = StringIO()
buf = libxml2.createOutputBuffer(f, None)
doc.saveFileTo(buf, None)
return f.getvalue()
def get_wikidot_content_body(URL):
xmlfile = tempfile.mktemp()
htmlfile = tempfile.mktemp()
os.system("wget -O " + htmlfile + " " + URL)
the_html = file2string(htmlfile)
os.system("rm " + htmlfile)
the_xml = html2xml(the_html)
string2file(the_xml, xmlfile)
doc = libxml2.parseFile(xmlfile)
ctxt = doc.xpathNewContext()
ctxt.xpathRegisterNs('xhtml', 'http://www.w3.org/1999/xhtml')
rows = ctxt.xpathEval('//xhtml:div[@id="page-content"]')
ctxt.xpathFreeContext()
f = StringIO.StringIO()
buf = libxml2.createOutputBuffer(f, 'UTF-8')
rows[0].docSetRootElement(doc)
doc.saveFileTo(buf, 'UTF-8')
os.system("rm " + xmlfile)
result = f.getvalue()
return result
def flatten(string):
doc = libxml2.parseDoc(string)
context = doc.xpathNewContext()
result = context.xpathEval("//use")
for orig in result:
xlink = orig.prop('href')
id = orig.prop('id')
id_length = len(id)
nodes = context.xpathEval('//g[@id="' + xlink[1:] + '"]')
if len(nodes) == 0:
nodes = context.xpathEval('//image[@id="' + xlink[1:] + '"]')
node = nodes[0]
copy = node.copyNode(extended=True)
copy.removeNsDef(None)
copy_context = doc.xpathNewContext()
copy_context.setContextNode(copy)
for copy_id in copy_context.xpathEval('.//@id'):
copy_id.setContent(id + copy_id.content[id_length:])
tx, ty = re.match('translate\((-?\d+\.?\d*.*),(-?\d+\.?\d*.*)\)',
orig.prop('transform')).groups()
transform = {'x': float(tx), 'y': float(ty)}
for c in ['x', 'y']:
for coord in copy_context.xpathEval('.//@' + c):
coord.setContent(
str(int(round(float(coord.content) + transform[c]))))
orig.replaceNode(copy)
f = StringIO()
buf = libxml2.createOutputBuffer(f, None)
doc.saveFileTo(buf, None)
return f.getvalue()
def testSaveFormattedDocToBuffer():
input = "<outer><inner>Some text</inner><inner/></outer>"
# The formatted and non-formatted versions of the output.
expected = (
"""\
<?xml version="1.0" encoding="UTF-8"?>
<outer><inner>Some text</inner><inner/></outer>
""",
"""\
<?xml version="1.0" encoding="UTF-8"?>
<outer>
<inner>Some text</inner>
<inner/>
</outer>
""",
)
doc = libxml2.parseDoc(input)
for i in (0, 1):
f = str_io()
buf = libxml2.createOutputBuffer(f, "UTF-8")
doc.saveFormatFileTo(buf, "UTF-8", i)
if f.getvalue() != expected[i]:
print("xmlDoc.saveFormatFileTo() call failed.")
print(" got: %s" % repr(f.getvalue()))
print("expected: %s" % repr(expected[i]))
sys.exit(1)
doc.freeDoc()
def testSimpleBufferWrites():
f = StringIO.StringIO()
buf = libxml2.createOutputBuffer(f, "ISO-8859-1")
buf.write(3, "foo")
buf.writeString("bar")
buf.close()
if f.getvalue() != "foobar":
print "Failed to save to StringIO"
sys.exit(1)
def testSimpleBufferWrites():
f = str_io()
buf = libxml2.createOutputBuffer(f, "ISO-8859-1")
buf.write(3, "foo")
buf.writeString("bar")
buf.close()
if f.getvalue() != "foobar":
print("Failed to save to StringIO")
sys.exit(1)
def write(self, fd, formatted=True):
"""
Write the Metadata information to the given file descriptor
"""
s = StringIO()
buf = libxml2.createOutputBuffer(s, "UTF-8")
buf.saveFormatFileTo(self.doc, "UTF-8", formatted)
header = self.Header()
header.size = s.len
header.write(fd)
fd.write(s.getvalue())
def SendReport(self, xmldoc):
if xmldoc.type != 'document_xml':
raise Exception, "Input is not XML document"
fbuf = StringIO.StringIO()
xmlbuf = libxml2.createOutputBuffer(fbuf, 'UTF-8')
doclen = xmldoc.saveFileTo(xmlbuf, 'UTF-8')
compr = bz2.BZ2Compressor(9)
cmpr = compr.compress(fbuf.getvalue())
data = base64.b64encode(cmpr + compr.flush())
ret = self.srv.SendReport(self.hostname, data)
print "rtevalclient::SendReport() - Sent %i bytes (XML document length: %i bytes, compression ratio: %.02f%%)" % (len(data), doclen, (1-(float(len(data)) / float(doclen)))*100 )
return ret
def SendReport(self, xmldoc):
if xmldoc.type != 'document_xml':
raise Exception("Input is not XML document")
fbuf = io.StringIO()
xmlbuf = libxml2.createOutputBuffer(fbuf, 'UTF-8')
doclen = xmldoc.saveFileTo(xmlbuf, 'UTF-8')
compr = bz2.BZ2Compressor(9)
cmpr = compr.compress(fbuf.getvalue())
data = base64.b64encode(cmpr + compr.flush())
ret = self.srv.SendReport(self.hostname, data)
print("rtevalclient::SendReport() - Sent %i bytes (XML document length: %i bytes, compression ratio: %.02f%%)" % (len(data), doclen, (1-(float(len(data)) / float(doclen)))*100 ))
return ret
def SendReport(self, xmldoc):
if xmldoc.type != 'document_xml':
raise Exception, "Input is not XML document"
f = open('/home/kchalas/Workspace/rteval/log', 'r')
fbuf = StringIO.StringIO()
xmlbuf = libxml2.createOutputBuffer(fbuf, 'UTF-8')
doclen = xmldoc.saveFileTo(xmlbuf, 'UTF-8')
compr = bz2.BZ2Compressor(9)
cmpr = compr.compress(fbuf.getvalue() + '\n<!--' + f.read() + '\n-->')
data = base64.b64encode(cmpr + compr.flush())
ret = self.srv.SendReport(self.hostname, data)
print "rtevalclient::SendReport() - Sent %i bytes (XML document length: %i bytes, compression ratio: %.02f%%)" % (
len(data), doclen, (1 - (float(len(data)) / float(doclen))) * 100)
return ret
def formatXML(self, fn):
"""
Format XML file.
@param fn: XML filename.
"""
ctxt = libxml2.createFileParserCtxt(fn)
ctxt.parseDocument()
doc = ctxt.doc()
# save formatted file
f = open(fn, 'w')
buf = libxml2.createOutputBuffer(f, 'UTF-8')
doc.saveFormatFileTo(buf, 'UTF-8', 1)
f.close()
doc.freeDoc()
del ctxt
def testSaveDocToBuffer():
"""
Regression test for bug #154294.
"""
input = '<foo>Hello</foo>'
expected = '''\
<?xml version="1.0" encoding="UTF-8"?>
<foo>Hello</foo>
'''
f = StringIO.StringIO()
buf = libxml2.createOutputBuffer(f, 'UTF-8')
doc = libxml2.parseDoc(input)
doc.saveFileTo(buf, 'UTF-8')
doc.freeDoc()
if f.getvalue() != expected:
print 'xmlDoc.saveFileTo() call failed.'
print ' got: %s' % repr(f.getvalue())
print 'expected: %s' % repr(expected)
sys.exit(1)
def testSaveDocToBuffer():
"""
Regression test for bug #154294.
"""
input = '<foo>Hello</foo>'
expected = '''\
<?xml version="1.0" encoding="UTF-8"?>
<foo>Hello</foo>
'''
f = str_io()
buf = libxml2.createOutputBuffer(f, 'UTF-8')
doc = libxml2.parseDoc(input)
doc.saveFileTo(buf, 'UTF-8')
doc.freeDoc()
if f.getvalue() != expected:
print('xmlDoc.saveFileTo() call failed.')
print(' got: %s' % repr(f.getvalue()))
print('expected: %s' % repr(expected))
sys.exit(1)
def testSaveFormattedDocToBuffer():
input = '<outer><inner>Some text</inner><inner/></outer>'
# The formatted and non-formatted versions of the output.
expected = ('''\
<?xml version="1.0" encoding="UTF-8"?>
<outer><inner>Some text</inner><inner/></outer>
''', '''\
<?xml version="1.0" encoding="UTF-8"?>
<outer>
<inner>Some text</inner>
<inner/>
</outer>
''')
doc = libxml2.parseDoc(input)
for i in (0, 1):
f = StringIO.StringIO()
buf = libxml2.createOutputBuffer(f, 'UTF-8')
doc.saveFormatFileTo(buf, 'UTF-8', i)
if f.getvalue() != expected[i]:
print 'xmlDoc.saveFormatFileTo() call failed.'
print ' got: %s' % repr(f.getvalue())
print 'expected: %s' % repr(expected[i])
sys.exit(1)
doc.freeDoc()
def testSaveFormattedDocToBuffer():
input = '<outer><inner>Some text</inner><inner/></outer>'
# The formatted and non-formatted versions of the output.
expected = ('''\
<?xml version="1.0" encoding="UTF-8"?>
<outer><inner>Some text</inner><inner/></outer>
''', '''\
<?xml version="1.0" encoding="UTF-8"?>
<outer>
<inner>Some text</inner>
<inner/>
</outer>
''')
doc = libxml2.parseDoc(input)
for i in (0, 1):
f = str_io()
buf = libxml2.createOutputBuffer(f, 'UTF-8')
doc.saveFormatFileTo(buf, 'UTF-8', i)
if f.getvalue() != expected[i]:
print('xmlDoc.saveFormatFileTo() call failed.')
print(' got: %s' % repr(f.getvalue()))
print('expected: %s' % repr(expected[i]))
sys.exit(1)
doc.freeDoc()
#!/usr/bin/python -u
import sys
import libxml2
import StringIO
print "Skipped"
sys.exit(1)
# Memory debug specific
libxml2.debugMemory(1)
#f = open('res', 'w')
f = StringIO.StringIO()
buf = libxml2.createOutputBuffer(f, "ISO-8859-1")
buf.write(3, "foo")
buf.writeString("bar")
buf.close()
del buf
if f.getvalue() != "foobar":
print "Failed to save to StringIO"
sys.exit(1)
del f
# Memory debug specific
libxml2.cleanupParser()
if libxml2.debugMemory(1) == 0:
print "OK"
else:
print "Memory leak %d bytes" % (libxml2.debugMemory(1))
else:
status = "Validation status: This document (%s) is valid MEDIN metadata" % unique_id
log(status)
else:
status = (
"Validation status: This document (%s) has NOT been validated as conforming to the MEDIN Metadata Standard"
% unique_id
)
log(status)
# save the validation status in the document itself
comment = doc.newDocComment(status)
doc.getRootElement().addPrevSibling(comment)
# output the document
f = StringIO()
buf = libxml2.createOutputBuffer(f, "utf-8")
doc.saveFormatFileTo(buf, "utf-8", True)
doc.freeDoc() # clean up
return (f.getvalue().strip(), warning)
def _createFlush(self, xml, doc, unique_id):
"""
Return a function that outputs a metadata entry
"""
raise NotImplementedError("This method must be overridden")
def __call__(self, iter_metadata, filter_=None):
if filter_ is None:
filter_ = FilterMetadata()
#!/usr/bin/python -u
import sys
import libxml2
import StringIO
print "Skipped"
sys.exit(1)
# Memory debug specific
libxml2.debugMemory(1)
# f = open('res', 'w')
f = StringIO.StringIO()
buf = libxml2.createOutputBuffer(f, "ISO-8859-1")
buf.write(3, "foo")
buf.writeString("bar")
buf.close()
del buf
if f.getvalue() != "foobar":
print "Failed to save to StringIO"
sys.exit(1)
del f
# Memory debug specific
libxml2.cleanupParser()
if libxml2.debugMemory(1) == 0:
print "OK"
else:
print "Memory leak %d bytes" % (libxml2.debugMemory(1))
