def test_add_title_with_tags():
svg_root = etree.Element(SVG_DOC, nsmap=SVG_NSMAP)
svg_root.append(_create_tagged_text(TAG1))
result_svg = visualize_svg_annotations(svg_root)
text_node = result_svg.find(SVG_TEXT + '/title')
assert text_node is not None
assert text_node.text == TAG1
def test_add_style_block_for_multiple_tags_on_same_node():
svg_root = etree.Element(SVG_DOC, nsmap=SVG_NSMAP)
svg_root.append(_create_tagged_text(' '.join([TAG1, TAG2])))
result_svg = visualize_svg_annotations(svg_root)
style_block = result_svg.find('style')
assert style_block is not None
assert style_block.text == style_block_for_tags([TAG1, TAG2])
def convert(args):
logger = get_logger()
svg_filename_pattern = args.svg_path
if not svg_filename_pattern:
svg_filename_pattern = svg_pattern_for_lxml_path(args.lxml_path)
logger.debug('svg_filename_pattern: %s', svg_filename_pattern)
lxml_root = etree.parse(args.lxml_path).getroot()
match_detail_reporter = None
if args.annotate:
annotators = DEFAULT_ANNOTATORS
if args.debug_match:
match_detail_reporter = CsvMatchDetailReporter(
open_csv_output(args.debug_match), args.debug_match)
if args.xml_path:
xml_mapping = parse_xml_mapping(args.xml_mapping_path)
target_annotations = xml_root_to_target_annotations(
etree.parse(args.xml_path).getroot(), xml_mapping)
annotators = annotators + [
MatchingAnnotator(target_annotations,
match_detail_reporter=match_detail_reporter,
use_tag_begin_prefix=True)
]
annotator = Annotator(annotators)
else:
annotator = None
if annotator:
svg_roots = list(iter_svg_pages_for_lxml(lxml_root))
annotator.annotate(SvgStructuredDocument(svg_roots))
else:
svg_roots = iter_svg_pages_for_lxml(lxml_root)
for page_index, svg_root in enumerate(svg_roots):
if annotator:
svg_root = visualize_svg_annotations(svg_root)
svg_filename = svg_filename_pattern.format(1 + page_index)
logger.info('writing to: %s', svg_filename)
with open(svg_filename, 'wb') as f:
etree.ElementTree(svg_root).write(f, pretty_print=True)
if annotator:
tagging_evaluation_results = evaluate_document_by_page(
SvgStructuredDocument(svg_roots))
logger.info(
'tagging evaluation:\n%s', '\n'.join([
'page{}: {}'.format(1 + i, r)
for i, r in enumerate(tagging_evaluation_results)
]))
if args.annotation_evaluation_csv:
write_dict_csv(
args.annotation_evaluation_csv, DEFAULT_EVALUATION_COLUMNS,
to_annotation_evaluation_csv_dict_rows(
tagging_evaluation_results,
document=os.path.basename(args.lxml_path)))
if match_detail_reporter:
match_detail_reporter.close()
def convert_and_annotate_lxml_content(lxml_content, xml_content, xml_mapping, name=None):
stop_watch_recorder = StopWatchRecorder()
stop_watch_recorder.start('parse lxml')
lxml_root = etree.fromstring(lxml_content)
# use a more lenient way to parse xml as xml errors are not uncomment
stop_watch_recorder.start('parse xml')
xml_root = xml_from_string_with_recover(xml_content)
stop_watch_recorder.start('extract target annotations')
target_annotations = xml_root_to_target_annotations(
xml_root,
xml_mapping
)
stop_watch_recorder.stop()
annotators = DEFAULT_ANNOTATORS + [MatchingAnnotator(
target_annotations,
use_tag_begin_prefix=True
)]
annotator = Annotator(annotators)
stop_watch_recorder.start('convert to svg')
svg_roots = list(iter_svg_pages_for_lxml(lxml_root))
stop_watch_recorder.start('annotate svg')
annotator.annotate(SvgStructuredDocument(svg_roots))
stop_watch_recorder.start('add visualisation')
svg_roots = [
visualize_svg_annotations(svg_root)
for svg_root in svg_roots
]
stop_watch_recorder.stop()
get_logger().info(
'processed: name=%s, lxml size=%s, xml size=%s, timings=[%s] (native align impl=%s)',
name, format(len(lxml_content), ','), format(len(xml_content), ','),
stop_watch_recorder, align_native_enabled
)
return svg_roots