def _get_doc_by_raw_offset(self, doc_id):
"""
Load document from xml using bytes offset information.
XXX: this is not tested under Windows.
"""
bounds = self._get_meta()[str(doc_id)].bounds
return xml_utils.load_chunk(self.filename, bounds)
def _get_doc_by_line_offset(self, doc_id):
"""
Load document from xml using line offset information.
This is much slower than _get_doc_by_raw_offset but should
work everywhere.
"""
bounds = self._get_meta()[str(doc_id)].bounds
return xml_utils.load_chunk(self.filename, bounds, slow=True)
def _compute_document_meta(self):
"""
Returns documents meta information that can
be used for fast document lookups. Meta information
consists of documents titles, categories and positions
in file.
"""
meta = compat.OrderedDict()
bounds_iter = xml_utils.bounds(self.filename,
r'<text id="(\d+)"[^>]*name="([^"]*)"',
r'</text>',
)
for match, bounds in bounds_iter:
doc_id, title = str(match.group(1)), match.group(2)
title = xml_utils.unescape_attribute(title)
# cache categories
xml = xml_utils.load_chunk(self.filename, bounds)
doc = Document(compat.ElementTree.XML(xml.encode('utf8')))
meta[doc_id] = _DocumentMeta(title, bounds, doc.categories())
return meta
