def _on_bind(self, _input):
from ezmlex.util import is_string
if _input is None:
self._do_extend = self._do_extend_default
return _input
# If string given as input, we'll walk through it using our own seek
# pointer.
if is_string(_input):
self._tell = 0
self._do_extend = self._do_extend_by_string
return _input
# If it's something that have 'read()' and 'readline()' methods, we'll
# read characters in chunks.
try:
if callable(_input.read) and callable(_input.readline):
self._do_extend = self._do_extend_by_read
return _input
except AttributeError:
pass
# Otherwise we'll try to read input's native portions using iterator
# interface
_input = iter(_input)
self._do_extend = self._do_extend_by_iter
return _input
def recompile(regex, *args):
import re
from ezmlex.util import is_string
try:
regex = regex.pattern
except AttributeError:
if not is_string(regex):
raise TypeError("not a pattern: %r" % regex)
return re.compile(regex, *args)
def _do_assign(self, content, **kw):
from ezmlex.util import is_string, is_list
if content is None:
self._content = []
self._fsm_str = self._make_fsm_str([])
return
if is_string(content):
self._content = content.splitlines()
elif not is_list(content):
self._content = list(content)
else:
self._content = content
if 'fsm_str' in kw:
self._fsm_str = kw['fsm_str']
else:
self._fsm_str = self._make_fsm_str(self._content)
def ccat(*reseq):
"""Concatenate sequence of regular expressions into one.
Example
"""
import re
from ezmlex.util import is_string
result = r''
for ri in reseq:
if is_string(ri):
result += ri
else:
try:
# NOTE: If `ri' is not a string, then we expect it is a
# compiled regular expression, so it should have `pattern'
# attribute and it should be a string (if it's not a string,
# the += operator throws TypeError).
result += ri.pattern
except AttributeError, TypeError:
raise TypeError('not a regular expression: %r' % ri)
