def extract_plots_from_latex_and_pdf(url_tarball, url_pdf):
tarball = download_url(url_tarball)
path, name = os.path.split(url_tarball)
#run_shell_command('cp ' + tarball + ' ' + tarball + '_arxiv' + name)
tarball_with_arxiv_extension = tarball + '_arxiv' + name
os.rename(tarball, tarball_with_arxiv_extension)
run_shell_command(CFG_BINDIR + '/plotextractor -t ' +
tarball_with_arxiv_extension)
pdf = download_url(url_pdf)
run_shell_command(CFG_PDFPLOTEXTRACTOR_PATH + ' ' + pdf)
path, name = os.path.split(tarball_with_arxiv_extension)
plotextracted_xml_path = tarball_with_arxiv_extension + '_plots/' + name + '.xml'
plotextracted_pdf_path = pdf + '.extracted/extracted.json'
return plotextracted_xml_path, plotextracted_pdf_path
def extract_plots_from_latex(url_tarball):
tarball = download_url(url_tarball)
path, name = os.path.split(url_tarball)
tarball += name
run_shell_command(CFG_BINDIR + '/plotextractor -t ' + tarball)
path, name = os.path.split(tarball)
plotextracted_xml_path = tarball + '_plots/' + name + '.xml'
return plotextracted_xml_path
def download_fulltext(obj, eng):
"""
Will download the fulltext document
"""
from invenio.bibdocfile import download_url
obj.extra_data["last_task_name"] = 'Download Fulltext'
try:
eng.log_info("Starting download of %s" % (obj.data['url']))
url = download_url(obj.data['url'])
obj.extra_data['tasks_results']['fulltext_url'] = url
except KeyError:
# Log the error
obj.extra_data["error_msg"] = 'Record does not include url'
eng.log.error("Error: %s" % (obj.extra_data["error_msg"],))
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 CFG_BIBINDEX_SPLASH_PAGES.iteritems():
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, 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, 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 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 CFG_BIBINDEX_SPLASH_PAGES.iteritems():
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, 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, 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 extract_plots_from_pdf(url_pdf):
pdf = download_url(url_pdf)
run_shell_command(CFG_PDFPLOTEXTRACTOR_PATH + ' ' + pdf)
plotextracted_pdf_path = pdf + '.extracted/extracted.json'
return plotextracted_pdf_path