def test_tei():
'''test TEI-related functions'''
# open local resources to avoid redownloading at each run
resources_dir = os.path.join(TEST_DIR, 'resources')
with open(os.path.join(resources_dir, 'httpbin_sample.html')) as f:
teststring = f.read()
# download, parse and validate simple html file
result = extract(teststring,
"mocked",
no_fallback=True,
tei_output=True,
tei_validation=False)
assert result is not None
assert xml.validate_tei(etree.fromstring(result)) is True
assert xml.validate_tei(etree.fromstring(teststring)) is False
# test with another file
with open(os.path.join(resources_dir, 'http_sample.html')) as f:
teststring = f.read()
# download, parse and validate simple html file
result = extract(teststring,
"mocked",
no_fallback=True,
tei_output=True,
tei_validation=False)
assert result is not None
assert xml.validate_tei(etree.fromstring(result)) is True
# include ID in metadata
result = extract(teststring,
"mocked",
no_fallback=True,
tei_output=True,
tei_validation=False,
record_id='0001')
assert result is not None
assert xml.validate_tei(etree.fromstring(result)) is True
def test_tei():
'''test TEI-related functions'''
# open local resources to avoid redownloading at each run
with open(os.path.join(RESOURCES_DIR, 'httpbin_sample.html')) as f:
teststring = f.read()
# download, parse and validate simple html file
result1 = extract(teststring,
"mocked",
no_fallback=True,
output_format='xmltei',
tei_validation=False)
result2 = extract(teststring,
"mocked",
no_fallback=True,
output_format='xmltei',
tei_validation=True)
assert result1 is not None and result1 == result2
assert xml.validate_tei(etree.fromstring(result1)) is True
assert xml.validate_tei(etree.fromstring(teststring)) is False
# test with another file
with open(os.path.join(RESOURCES_DIR, 'http_sample.html')) as f:
teststring = f.read()
# download, parse and validate simple html file
result = extract(teststring,
"mocked",
no_fallback=True,
output_format='xmltei',
tei_validation=False)
assert result is not None # and '<p>license</p>' in result
assert xml.validate_tei(etree.fromstring(result)) is True
# include ID in metadata
result = extract(teststring,
"mocked",
no_fallback=True,
output_format='xmltei',
tei_validation=False,
record_id='0001')
assert result is not None
assert xml.validate_tei(etree.fromstring(result)) is True
# test header + metadata
tei = etree.Element('TEI', xmlns='http://www.tei-c.org/ns/1.0')
header = etree.SubElement(tei, 'teiHeader')
docmeta = dict.fromkeys(METADATA_LIST)
docmeta['categories'], docmeta['tags'] = [], []
docmeta['title'] = 'Title'
assert xml.write_fullheader(header, docmeta) is not None
docmeta['sitename'] = 'Site Name'
assert xml.write_fullheader(header, docmeta) is not None
docmeta['hostname'], docmeta['sitename'] = 'hostname', None
assert xml.write_fullheader(header, docmeta) is not None
def test_tei():
'''test TEI-related functions'''
# open local resources to avoid redownloading at each run
with open(os.path.join(RESOURCES_DIR, 'httpbin_sample.html')) as f:
teststring = f.read()
# download, parse and validate simple html file
result1 = extract(teststring, "mocked", no_fallback=True, output_format='xmltei', tei_validation=False)
result2 = extract(teststring, "mocked", no_fallback=True, output_format='xmltei', tei_validation=True)
assert result1 is not None and result1 == result2
assert xml.validate_tei(etree.fromstring(result1)) is True
assert xml.validate_tei(etree.fromstring(teststring)) is False
# test with another file
with open(os.path.join(RESOURCES_DIR, 'http_sample.html')) as f:
teststring = f.read()
# download, parse and validate simple html file
result = extract(teststring, "mocked", no_fallback=True, include_comments=True, output_format='xmltei', tei_validation=False)
assert result is not None # and '<p>license</p>' in result
assert xml.validate_tei(etree.fromstring(result)) is True
result = extract(teststring, "mocked", no_fallback=True, include_comments=False, output_format='xmltei', tei_validation=False)
assert result is not None # and '<p>license</p>' in result
assert xml.validate_tei(etree.fromstring(result)) is True
# include ID in metadata
result = extract(teststring, "mocked", no_fallback=True, output_format='xmltei', tei_validation=False, record_id='0001')
assert result is not None
assert xml.validate_tei(etree.fromstring(result)) is True
# test header + metadata
tei = etree.Element('TEI', xmlns='http://www.tei-c.org/ns/1.0')
header = etree.SubElement(tei, 'teiHeader')
docmeta = Document()
docmeta.categories, docmeta.tags = [], []
docmeta.title = 'Title'
assert xml.write_fullheader(header, docmeta) is not None
docmeta.sitename = 'Site Name'
docmeta.date = '2021-01-01'
assert xml.write_fullheader(header, docmeta) is not None
docmeta.date = None
assert xml.write_fullheader(header, docmeta) is not None
docmeta.hostname = 'hostname'
assert xml.write_fullheader(header, docmeta) is not None
docmeta.sitename = None
docmeta.license = 'CC BY-SA'
docmeta.url = 'https://test.org/'
docmeta.categories = ['cat1', 'cat2']
assert xml.write_fullheader(header, docmeta) is not None
docmeta.date = '2021-01-01'
assert xml.write_fullheader(header, docmeta) is not None
docmeta.title, docmeta.sitename = None, None
assert xml.write_fullheader(header, docmeta) is not None
# Versuch, die Keyword-Argumente url und target_language zu aendern;
# result9 und result11 liefern None (liegt anscheinend am Aendern der Target-language),
# das Eingeben einer url funktioniert aber:
url1='https://github.blog/2019-03-29-leader-spotlight-erin-spiceland/'
result9 = trafilatura.extract(downloaded, target_language='de')
print(result9)
result10 = trafilatura.extract(downloaded, url=url1)
print(result10)
result11 = trafilatura.extract(downloaded, url=url1, target_language='de')
print(result11)
# Ausprobieren der Validierung bzgl. TEI (siehe https://trafilatura.readthedocs.io/en/latest/validation.html)
print("test TEI-Validation:")
mytree2=etree.parse("DU_sense.xml")
print("Correct file: ",validate_tei(mytree2))
mytree3=etree.parse("DU_sense_illformed.xml")
print("Illformed file: ")
print(validate_tei(mytree3))
# Ausprobieren der Funktion xmltotxt() (siehe https://trafilatura.readthedocs.io/en/latest/corefunctions.html)
#f=open("DU_sense.xml","r")
#xml_text=f.read()
#print(trafilatura.xml.xmltotxt(xml_text))
# --> fuehrt zu TypeError: Invalid input object: str
#f=open("DU_sense.xml","r")
#print(trafilatura.xml.xmltotxt(f))
# --> fuehrt zu TypeError: Invalid input object: _io.TextIOWrapper
#print(trafilatura.xml.xmltotxt("DU_sense.xml"))
def extract(filecontent,
url=None,
record_id='0001',
no_fallback=False,
include_comments=False,
csv_output=False,
xml_output=False,
tei_output=False,
tei_validation=False,
target_language=None,
include_tables=True,
include_formatting=False):
'''Main process for text extraction'''
# init
tree = load_html(filecontent)
if tree is None:
return None
# Metadata here
if csv_output is True or xml_output is True or tei_output is True:
docmeta = extract_metadata(tree, default_url=url)
else:
docmeta = None
# backup (or not) for further processing
if no_fallback is False:
backup_tree = deepcopy(tree)
else:
backup_tree = None
# clean
cleaned_tree = manual_cleaning(tree, include_tables)
# save space and processing time
cleaned_tree = prune_html(cleaned_tree)
# use LXML cleaner
cleaned_tree = HTML_CLEANER.clean_html(cleaned_tree)
# tree_cache[cleaned_tree] = list(cleaned_tree.iter())
# convert tags, the rest does not work without conversion
cleaned_tree = convert_tags(cleaned_tree)
# remove hi-element to avoid tail bug
if (xml_output is False
and tei_output is False) or include_formatting is False:
etree.strip_tags(cleaned_tree, 'hi')
# comments first, then remove
if include_comments is True:
commentsbody, temp_comments, len_comments, cleaned_tree = extract_comments(
cleaned_tree)
else:
commentsbody, temp_comments, len_comments = None, '', 0
# extract content
postbody, temp_text, len_text, sure_thing = extract_content(
cleaned_tree, include_tables)
# compare if necessary
if no_fallback is False: # and sure_thing is False:
postbody, temp_text, len_text = compare_extraction(
backup_tree, url, postbody, temp_text, len_text)
# try with justext
if len_text < MIN_EXTRACTED_SIZE:
LOGGER.error('not enough text %s %s', record_id, url)
postbody, len_text, temp_text = justext_rescue(
tree, url, target_language, postbody, len_text, temp_text)
LOGGER.error('justext length %s', len_text)
# second backup
# if len_text < MIN_EXTRACTED_SIZE:
# postbody, len_text, temp_text = baseline(filecontent)
else:
# rescue: try to use original/dirty tree
if sure_thing is False and len_text < MIN_EXTRACTED_SIZE:
postbody, len_text, temp_text = baseline(filecontent)
#tree = load_html(filecontent)
#tree = convert_tags(tree)
#postbody, temp_text, len_text, sure_thing = extract_content(tree)
LOGGER.debug('non-clean extracted length: %s (extraction)',
len_text)
if len_comments < MIN_EXTRACTED_COMM_SIZE:
LOGGER.info('not enough comments %s %s', record_id, url)
if len_text < MIN_OUTPUT_SIZE and len_comments < MIN_OUTPUT_COMM_SIZE:
LOGGER.info('text and comments not long enough: %s %s', len_text,
len_comments)
return None
# sanity check on language
if language_filter(temp_text, temp_comments, target_language, record_id,
url) is True:
return None
# check duplicates at body level
if duplicate_test(postbody) is True:
return None
# cache elements
put_in_cache(postbody)
if commentsbody is not None:
put_in_cache(commentsbody)
# XML (TEI) steps
if xml_output is True or tei_output is True:
if xml_output is True:
output = build_xml_output(postbody, commentsbody)
output = add_xml_meta(output, docmeta)
elif tei_output is True:
output = build_tei_output(postbody, commentsbody, docmeta)
# can be improved
control_string = etree.tostring(output, encoding='unicode')
control_string = sanitize(control_string)
# necessary for cleaning
control_parser = etree.XMLParser(remove_blank_text=True)
output_tree = etree.fromstring(control_string, control_parser)
# validate
if tei_output is True and tei_validation is True:
result = validate_tei(output_tree)
LOGGER.info('TEI validation result: %s %s %s', result, record_id,
docmeta.url)
returnstring = etree.tostring(output_tree,
pretty_print=True,
encoding='unicode').strip()
# CSV + TXT output
else:
if csv_output is True:
posttext = xmltotxt(postbody)
commentstext = xmltotxt(commentsbody)
returnstring = txttocsv(posttext, commentstext, docmeta)
else:
output = build_xml_output(postbody, commentsbody)
returnstring = xmltotxt(output)
return returnstring
