def __init__(self, filename, **kwargs):
self.tags = kwargs.pop('tags', None)
self.tagspec = '.*/seg/fs'
self.xml_tool = XML_Tool(filename, 'ann_morphosyntax.xml')
XMLCorpusView.__init__(
self, self.xml_tool.build_preprocessed_file(), self.tagspec
)
def __init__(self, fileid, sent, tag, strip_space, stem):
"""
:param fileid: The name of the underlying file.
:param sent: If true, include sentence bracketing.
:param tag: The name of the tagset to use, or None for no tags.
:param strip_space: If true, strip spaces from word tokens.
:param stem: If true, then substitute stems for words.
"""
if sent:
tagspec = ".*/s"
else:
tagspec = ".*/s/(.*/)?(c|w)"
self._sent = sent
self._tag = tag
self._strip_space = strip_space
self._stem = stem
self.title = None #: Title of the document.
self.author = None #: Author of the document.
self.editor = None #: Editor
self.resps = None #: Statement of responsibility
XMLCorpusView.__init__(self, fileid, tagspec)
# Read in a tasty header.
self._open()
self.read_block(self._stream, ".*/teiHeader$", self.handle_header)
self.close()
# Reset tag context.
self._tag_context = {0: ()}
def __init__(self, fileid, sent, tag, strip_space, stem):
"""
:param fileid: The name of the underlying file.
:param sent: If true, include sentence bracketing.
:param tag: The name of the tagset to use, or None for no tags.
:param strip_space: If true, strip spaces from word tokens.
:param stem: If true, then substitute stems for words.
"""
if sent:
tagspec = '.*/s'
else:
tagspec = '.*/s/(.*/)?(c|w)'
self._sent = sent
self._tag = tag
self._strip_space = strip_space
self._stem = stem
self.title = None #: Title of the document.
self.author = None #: Author of the document.
self.editor = None #: Editor
self.resps = None #: Statement of responsibility
XMLCorpusView.__init__(self, fileid, tagspec)
# Read in a tasty header.
self._open()
self.read_block(self._stream, '.*/teiHeader$', self.handle_header)
self.close()
# Reset tag context.
self._tag_context = {0: ()}
def __init__(self, filename, **kwargs):
self.tags = kwargs.pop('tags', None)
self.tagspec = '.*/seg/fs'
self.xml_tool = XML_Tool(filename, 'ann_morphosyntax.xml')
XMLCorpusView.__init__(
self, self.xml_tool.build_preprocessed_file(), self.tagspec
)
def __init__(self, filename, **kwargs):
"""
HEADER_MODE
A stream backed corpus view specialized for use with
header.xml files in NKJP corpus.
"""
self.tagspec = ".*/sourceDesc$"
XMLCorpusView.__init__(self, filename + 'header.xml', self.tagspec)
def __init__(self, filename, **kwargs):
self.mode = kwargs.pop('mode', 0)
self.tagspec = '.*/div/ab'
self.segm_dict = dict()
#xml preprocessing
self.xml_tool = XML_Tool(filename, 'text.xml')
#base class init
XMLCorpusView.__init__(self, self.xml_tool.build_preprocessed_file(), self.tagspec)
def __init__(self, filename, **kwargs):
"""
HEADER_MODE
A stream backed corpus view specialized for use with
header.xml files in NKJP corpus.
"""
self.tagspec = ".*/sourceDesc$"
XMLCorpusView.__init__(self, filename + 'header.xml', self.tagspec)
def __init__(self, filename, **kwargs):
self.mode = kwargs.pop('mode', 0)
self.tagspec = '.*/div/ab'
self.segm_dict = dict()
#xml preprocessing
self.xml_tool = XML_Tool(filename, 'text.xml')
#base class init
XMLCorpusView.__init__(self, self.xml_tool.build_preprocessed_file(), self.tagspec)
def __init__(self, filename, **kwargs):
self.tagspec = '.*p/.*s'
#intersperse NKJPCorpus_Text_View
self.text_view = NKJPCorpus_Text_View(filename, mode=NKJPCorpus_Text_View.SENTS_MODE)
self.text_view.handle_query()
#xml preprocessing
self.xml_tool = XML_Tool(filename, 'ann_segmentation.xml')
#base class init
XMLCorpusView.__init__(self, self.xml_tool.build_preprocessed_file(), self.tagspec)
def __init__(self, filename, **kwargs):
self.tagspec = '.*p/.*s'
#intersperse NKJPCorpus_Text_View
self.text_view = NKJPCorpus_Text_View(filename, mode=NKJPCorpus_Text_View.SENTS_MODE)
self.text_view.handle_query()
#xml preprocessing
self.xml_tool = XML_Tool(filename, 'ann_segmentation.xml')
#base class init
XMLCorpusView.__init__(self, self.xml_tool.build_preprocessed_file(), self.tagspec)
def __init__(self, fileid, unit, bracket_sent, pos_tag, sem_tag):
"""
:param fileid: The name of the underlying file.
:param unit: One of `'token'`, `'word'`, or `'chunk'`.
:param bracket_sent: If true, include sentence bracketing.
:param pos_tag: Whether to include part-of-speech tags.
:param sem_tag: Whether to include semantic tags, namely WordNet lemma
and OOV named entity status.
"""
if bracket_sent: tagspec = '.*/s'
else: tagspec = '.*/s/(punc|wf)'
self._unit = unit
self._sent = bracket_sent
self._pos_tag = pos_tag
self._sem_tag = sem_tag
XMLCorpusView.__init__(self, fileid, tagspec)
def __init__(self, fileid, unit, bracket_sent, pos_tag, sem_tag, wordnet):
"""
:param fileid: The name of the underlying file.
:param unit: One of `'token'`, `'word'`, or `'chunk'`.
:param bracket_sent: If true, include sentence bracketing.
:param pos_tag: Whether to include part-of-speech tags.
:param sem_tag: Whether to include semantic tags, namely WordNet lemma
and OOV named entity status.
"""
if bracket_sent: tagspec = '.*/s'
else: tagspec = '.*/s/(punc|wf)'
self._unit = unit
self._sent = bracket_sent
self._pos_tag = pos_tag
self._sem_tag = sem_tag
self._wordnet = wordnet
XMLCorpusView.__init__(self, fileid, tagspec)
def __init__(self, fileid, tagspec, elt_handler=None):
XMLCorpusView.__init__(self, fileid, tagspec, elt_handler)
def __init__(self, fileid, tagspec, elt_handler=None):
XMLCorpusView.__init__(self, fileid, tagspec, elt_handler)
def __init__(self, filename, **kwargs):
self.tags = kwargs.pop("tags", None)
self.tagspec = ".*/seg/fs"
self.xml_tool = XML_Tool(filename, "ann_morphosyntax.xml")
XMLCorpusView.__init__(self, self.xml_tool.build_preprocessed_file(),
self.tagspec)
def __init__(self, filename, **kwargs):
self.tagspec = '.*/seg/fs'
self.xml_tool = XML_Tool(filename, 'ann_named.xml')
XMLCorpusView.__init__(self, self.xml_tool.build_preprocessed_file(),
self.tagspec)
