Page MenuHomec4science
Files F65080779

BibIndexFulltextTokenizer.py
No OneTemporary
Actions

File Metadata

Created
Fri, May 31, 15:49

BibIndexFulltextTokenizer.py
View Options

# -*- coding: utf-8 -*-
#
# This file is part of Invenio.
# Copyright (C) 2010, 2011, 2012, 2014, 2015 CERN.
#
# Invenio is free software; you can redistribute it and/or
# modify it under the terms of the GNU General Public License as
# published by the Free Software Foundation; either version 2 of the
# License, or (at your option) any later version.
#
# Invenio is distributed in the hope that it will be useful, but
# WITHOUT ANY WARRANTY; without even the implied warranty of
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
# General Public License for more details.
#
# You should have received a copy of the GNU General Public License
# along with Invenio; if not, write to the Free Software Foundation, Inc.,
# 59 Temple Place, Suite 330, Boston, MA 02111-1307, USA.
"""BibIndexFulltextTokenizer: extracts words form a given document.
Document is given by its URL.
"""
import os
import sys
import logging
import urllib2
import re
from six import iteritems
from invenio.config import \
CFG_SOLR_URL, \
CFG_XAPIAN_ENABLED, \
CFG_BIBINDEX_FULLTEXT_INDEX_LOCAL_FILES_ONLY, \
CFG_BIBINDEX_SPLASH_PAGES
from invenio.utils.html import get_links_in_html_page
from invenio.legacy.websubmit.file_converter import convert_file, get_file_converter_logger
from invenio.legacy.miscutil.solrutils_bibindex_indexer import solr_add_fulltext
from invenio.legacy.miscutil.xapianutils_bibindex_indexer import xapian_add
from invenio.legacy.bibdocfile.api import bibdocfile_url_p, \
bibdocfile_url_to_bibdoc, download_url, \
BibRecDocs, InvenioBibDocFileError
from invenio.legacy.bibindex.engine_utils import get_idx_indexer
from invenio.legacy.bibsched.bibtask import write_message
from invenio.ext.logging import register_exception
from intbitset import intbitset
from invenio.modules.indexer.tokenizers.BibIndexDefaultTokenizer import BibIndexDefaultTokenizer
fulltext_added = intbitset() # stores ids of records whose fulltexts have been added
class BibIndexFulltextTokenizer(BibIndexDefaultTokenizer):
"""
Exctracts all the words contained in document specified by url.
"""
def __init__(self, stemming_language = None, remove_stopwords = False, remove_html_markup = False, remove_latex_markup = False):
self.verbose = 3
BibIndexDefaultTokenizer.__init__(self, stemming_language,
remove_stopwords,
remove_html_markup,
remove_latex_markup)
def set_verbose(self, verbose):
"""Allows to change verbosity level during indexing"""
self.verbose = verbose
def tokenize_for_words_default(self, phrase):
"""Default tokenize_for_words inherited from default tokenizer"""
return super(BibIndexFulltextTokenizer, self).tokenize_for_words(phrase)
def get_words_from_fulltext(self, url_direct_or_indirect):
"""Returns all the words contained in the document specified by
URL_DIRECT_OR_INDIRECT with the words being split by various
SRE_SEPARATORS regexp set earlier. If FORCE_FILE_EXTENSION is
set (e.g. to "pdf", then treat URL_DIRECT_OR_INDIRECT as a PDF
file. (This is interesting to index Indico for example.) Note
also that URL_DIRECT_OR_INDIRECT may be either a direct URL to
the fulltext file or an URL to a setlink-like page body that
presents the links to be indexed. In the latter case the
URL_DIRECT_OR_INDIRECT is parsed to extract actual direct URLs
to fulltext documents, for all knows file extensions as
specified by global CONV_PROGRAMS config variable.
"""
write_message("... reading fulltext files from %s started" % url_direct_or_indirect, verbose=2)
try:
if bibdocfile_url_p(url_direct_or_indirect):
write_message("... %s is an internal document" % url_direct_or_indirect, verbose=2)
try:
bibdoc = bibdocfile_url_to_bibdoc(url_direct_or_indirect)
except InvenioBibDocFileError:
# Outdated 8564 tag
return []
indexer = get_idx_indexer('fulltext')
if indexer != 'native':
# A document might belong to multiple records
for rec_link in bibdoc.bibrec_links:
recid = rec_link["recid"]
# Adds fulltexts of all files once per records
if not recid in fulltext_added:
bibrecdocs = BibRecDocs(recid)
try:
text = bibrecdocs.get_text()
except InvenioBibDocFileError:
# Invalid PDF
continue
if indexer == 'SOLR' and CFG_SOLR_URL:
solr_add_fulltext(recid, text)
elif indexer == 'XAPIAN' and CFG_XAPIAN_ENABLED:
xapian_add(recid, 'fulltext', text)
fulltext_added.add(recid)
# we are relying on an external information retrieval system
# to provide full-text indexing, so dispatch text to it and
# return nothing here:
return []
else:
text = ""
if hasattr(bibdoc, "get_text"):
text = bibdoc.get_text()
return self.tokenize_for_words_default(text)
else:
if CFG_BIBINDEX_FULLTEXT_INDEX_LOCAL_FILES_ONLY:
write_message("... %s is external URL but indexing only local files" % url_direct_or_indirect, verbose=2)
return []
write_message("... %s is an external URL" % url_direct_or_indirect, verbose=2)
urls_to_index = set()
for splash_re, url_re in iteritems(CFG_BIBINDEX_SPLASH_PAGES):
if re.match(splash_re, url_direct_or_indirect):
write_message("... %s is a splash page (%s)" % (url_direct_or_indirect, splash_re), verbose=2)
html = urllib2.urlopen(url_direct_or_indirect).read()
urls = get_links_in_html_page(html)
write_message("... found these URLs in %s splash page: %s" % (url_direct_or_indirect, ", ".join(urls)), verbose=3)
for url in urls:
if re.match(url_re, url):
write_message("... will index %s (matched by %s)" % (url, url_re), verbose=2)
urls_to_index.add(url)
if not urls_to_index:
urls_to_index.add(url_direct_or_indirect)
write_message("... will extract words from %s" % ', '.join(urls_to_index), verbose=2)
words = {}
for url in urls_to_index:
tmpdoc = download_url(url)
file_converter_logger = get_file_converter_logger()
old_logging_level = file_converter_logger.getEffectiveLevel()
if self.verbose > 3:
file_converter_logger.setLevel(logging.DEBUG)
try:
try:
tmptext = convert_file(tmpdoc, output_format='.txt')
text = open(tmptext).read()
os.remove(tmptext)
indexer = get_idx_indexer('fulltext')
if indexer != 'native':
if indexer == 'SOLR' and CFG_SOLR_URL:
solr_add_fulltext(None, text) # FIXME: use real record ID
if indexer == 'XAPIAN' and CFG_XAPIAN_ENABLED:
#xapian_add(None, 'fulltext', text) # FIXME: use real record ID
pass
# we are relying on an external information retrieval system
# to provide full-text indexing, so dispatch text to it and
# return nothing here:
tmpwords = []
else:
tmpwords = self.tokenize_for_words_default(text)
words.update(dict(map(lambda x: (x, 1), tmpwords)))
except Exception as e:
message = 'ERROR: it\'s impossible to correctly extract words from %s referenced by %s: %s' % (url, url_direct_or_indirect, e)
register_exception(prefix=message, alert_admin=True)
write_message(message, stream=sys.stderr)
finally:
os.remove(tmpdoc)
if self.verbose > 3:
file_converter_logger.setLevel(old_logging_level)
return words.keys()
except Exception as e:
message = 'ERROR: it\'s impossible to correctly extract words from %s: %s' % (url_direct_or_indirect, e)
register_exception(prefix=message, alert_admin=True)
write_message(message, stream=sys.stderr)
return []
def tokenize_for_words(self, phrase):
return self.get_words_from_fulltext(phrase)
def tokenize_for_pairs(self, phrase):
return []
def tokenize_for_phrases(self, phrase):
return []

Event Timeline

c4science · Help