Merge pull request #425 from anthrotype/xmlreader-fileobj

make XMLReader accept file objects
This commit is contained in:
Cosimo Lupo 2015-12-08 13:08:19 +00:00
commit 9049d5f866
3 changed files with 113 additions and 27 deletions

View File

@ -13,9 +13,17 @@ BUFSIZE = 0x4000
class XMLReader(object):
def __init__(self, fileName, ttFont, progress=None, quiet=False):
def __init__(self, fileOrPath, ttFont, progress=None, quiet=False):
if fileOrPath == '-':
fileOrPath = sys.stdin
if not hasattr(fileOrPath, "read"):
self.file = open(fileOrPath, "rb")
self._closeStream = True
else:
# assume readable file object
self.file = fileOrPath
self._closeStream = False
self.ttFont = ttFont
self.fileName = fileName
self.progress = progress
self.quiet = quiet
self.root = None
@ -24,11 +32,16 @@ class XMLReader(object):
def read(self):
if self.progress:
import stat
self.progress.set(0, os.stat(self.fileName)[stat.ST_SIZE] // 100 or 1)
file = open(self.fileName, 'rb')
self._parseFile(file)
file.close()
self.file.seek(0, 2)
fileSize = self.file.tell()
self.progress.set(0, fileSize // 100 or 1)
self.file.seek(0)
self._parseFile(self.file)
if self._closeStream:
self.close()
def close(self):
self.file.close()
def _parseFile(self, file):
from xml.parsers.expat import ParserCreate
@ -63,7 +76,13 @@ class XMLReader(object):
elif stackSize == 1:
subFile = attrs.get("src")
if subFile is not None:
subFile = os.path.join(os.path.dirname(self.fileName), subFile)
if hasattr(self.file, 'name'):
# if file has a name, get its parent directory
dirname = os.path.dirname(self.file.name)
else:
# else fall back to using the current working directory
dirname = os.getcwd()
subFile = os.path.join(dirname, subFile)
subReader = XMLReader(subFile, self.ttFont, self.progress, self.quiet)
subReader.read()
self.contentStack.append([])
@ -71,7 +90,7 @@ class XMLReader(object):
tag = ttLib.xmlToTag(name)
msg = "Parsing '%s' table..." % tag
if self.progress:
self.progress.setlabel(msg)
self.progress.setLabel(msg)
elif self.ttFont.verbose:
ttLib.debugmsg(msg)
else:

View File

@ -5,7 +5,7 @@ from fontTools.misc.py23 import *
import os
import unittest
from fontTools.ttLib import TTFont
from .xmlReader import XMLReader
from .xmlReader import XMLReader, ProgressPrinter, BUFSIZE
import tempfile
@ -15,9 +15,9 @@ class TestXMLReader(unittest.TestCase):
class DebugXMLReader(XMLReader):
def __init__(self, fileName, ttFont, progress=None, quiet=False):
def __init__(self, fileOrPath, ttFont, progress=None, quiet=False):
super(DebugXMLReader, self).__init__(
fileName, ttFont, progress, quiet)
fileOrPath, ttFont, progress, quiet)
self.contents = []
def _endElementHandler(self, name):
@ -38,21 +38,19 @@ class TestXMLReader(unittest.TestCase):
</ttFont>
''' % expected
with tempfile.NamedTemporaryFile(delete=False) as tmp:
tmp.write(data.encode('utf-8'))
reader = DebugXMLReader(tmp.name, TTFont(), quiet=True)
reader.read()
os.remove(tmp.name)
content = strjoin(reader.contents[0]).strip()
with BytesIO(data.encode('utf-8')) as tmp:
reader = DebugXMLReader(tmp, TTFont(), quiet=True)
reader.read()
content = strjoin(reader.contents[0]).strip()
self.assertEqual(expected, content)
def test_normalise_newlines(self):
class DebugXMLReader(XMLReader):
def __init__(self, fileName, ttFont, progress=None, quiet=False):
def __init__(self, fileOrPath, ttFont, progress=None, quiet=False):
super(DebugXMLReader, self).__init__(
fileName, ttFont, progress, quiet)
fileOrPath, ttFont, progress, quiet)
self.newlines = []
def _characterDataHandler(self, data):
@ -68,14 +66,83 @@ class TestXMLReader(unittest.TestCase):
' escaped CR and windows newline &#13;\r\n' # &#13;\r\n -> \r\n
' </test>\n' # \n
'</ttFont>')
with tempfile.NamedTemporaryFile(delete=False) as tmp:
tmp.write(data.encode('utf-8'))
reader = DebugXMLReader(tmp.name, TTFont(), quiet=True)
reader.read()
os.remove(tmp.name)
with BytesIO(data.encode('utf-8')) as tmp:
reader = DebugXMLReader(tmp, TTFont(), quiet=True)
reader.read()
expected = ['\n'] * 3 + ['\r', '\n'] * 3 + ['\n']
self.assertEqual(expected, reader.newlines)
def test_progress(self):
class DummyProgressPrinter(ProgressPrinter):
def __init__(self, title, maxval=100):
self.label = title
self.maxval = maxval
self.pos = 0
def set(self, val, maxval=None):
if maxval is not None:
self.maxval = maxval
self.pos = val
def increment(self, val=1):
self.pos += val
def setLabel(self, text):
self.label = text
data = (
'<ttFont>\n'
' <test>\n'
' %s\n'
' </test>\n'
'</ttFont>\n'
% ("z" * 2 * BUFSIZE)
).encode('utf-8')
dataSize = len(data)
progressBar = DummyProgressPrinter('test')
with BytesIO(data) as tmp:
reader = XMLReader(tmp, TTFont(), progress=progressBar)
self.assertEqual(progressBar.pos, 0)
reader.read()
self.assertEqual(progressBar.pos, dataSize // 100)
self.assertEqual(progressBar.maxval, dataSize // 100)
self.assertTrue('test' in progressBar.label)
with BytesIO(b"<ttFont></ttFont>") as tmp:
reader = XMLReader(tmp, TTFont(), progress=progressBar)
reader.read()
# when data size is less than 100 bytes, 'maxval' is 1
self.assertEqual(progressBar.maxval, 1)
def test_close_file_path(self):
with tempfile.NamedTemporaryFile(delete=False) as tmp:
tmp.write(b'<ttFont></ttFont>')
reader = XMLReader(tmp.name, TTFont())
reader.read()
# when reading from path, the file is closed automatically at the end
self.assertTrue(reader.file.closed)
# this does nothing
reader.close()
self.assertTrue(reader.file.closed)
os.remove(tmp.name)
def test_close_file_obj(self):
with tempfile.NamedTemporaryFile(delete=False) as tmp:
tmp.write(b'<ttFont>"hello"</ttFont>')
with open(tmp.name, "rb") as f:
reader = XMLReader(f, TTFont())
reader.read()
# when reading from a file or file-like object, the latter is kept open
self.assertFalse(reader.file.closed)
# ... until the user explicitly closes it
reader.close()
self.assertTrue(reader.file.closed)
os.remove(tmp.name)
if __name__ == '__main__':
unittest.main()

View File

@ -327,7 +327,7 @@ class TTFont(object):
writer.newline()
writer.newline()
def importXML(self, file, progress=None, quiet=False):
def importXML(self, fileOrPath, progress=None, quiet=False):
"""Import a TTX file (an XML-based text format), so as to recreate
a font object.
"""
@ -340,7 +340,7 @@ class TTFont(object):
from fontTools.misc import xmlReader
reader = xmlReader.XMLReader(file, self, progress, quiet)
reader = xmlReader.XMLReader(fileOrPath, self, progress, quiet)
reader.read()
def isLoaded(self, tag):