def test_get_AllTextLine(self):
with open(assets.path_to('gutachten/data/TEMP1/PAGE_TEMP1.xml'),
'r') as f:
page = parseString(f.read().encode('utf8'),
silence=True).get_Page()
assert len(page.get_AllTextLines()) == 55
def setUp(self):
with open(assets.path_to('glyph-consistency/data/OCR-D-GT-PAGE/FAULTY_GLYPHS.xml'), 'rb') as f:
self.xml_as_str = f.read()
self.pcgts = parseString(self.xml_as_str, silence=True)
# -*- coding: utf-8 -*-
from os.path import join as pjoin
from pathlib import Path
from tempfile import TemporaryDirectory
from tests.base import TestCase, assets, main, copy_of_directory
from ocrd.resolver import Resolver
from ocrd_utils import pushd_popd, initLogging
METS_HEROLD = assets.url_of('SBB0000F29300010000/data/mets.xml')
FOLDER_KANT = assets.path_to('kant_aufklaerung_1784')
# pylint: disable=redundant-unittest-assert, broad-except, deprecated-method, too-many-public-methods
class TestResolver(TestCase):
def setUp(self):
initLogging()
self.resolver = Resolver()
def test_workspace_from_url_bad(self):
with self.assertRaisesRegex(Exception, "Must pass 'mets_url'"):
self.resolver.workspace_from_url(None)
def test_workspace_from_url_tempdir(self):
self.resolver.workspace_from_url(
mets_basename='foo.xml',
mets_url='https://raw.githubusercontent.com/OCR-D/assets/master/data/kant_aufklaerung_1784/data/mets.xml')
def _fixture_kant_complex(tmp_path):
copytree(assets.path_to('kant_aufklaerung_1784-complex/data'),
str(tmp_path))
yield Workspace(Resolver, directory=tmp_path)
def test_str(self):
with Image.open(assets.path_to('SBB0000F29300010000/data/OCR-D-IMG/FILE_0001_IMAGE.tif')) as img:
exif = OcrdExif(img)
print(str(exif.to_xml()))
from tests.base import TestCase, main, assets, create_ocrd_file, create_ocrd_file_with_defaults
from ocrd_utils import MIMETYPE_PAGE
from ocrd_models import OcrdMets
from ocrd_modelfactory import (exif_from_filename, page_from_image,
page_from_file)
SAMPLE_IMG = assets.path_to(
'kant_aufklaerung_1784/data/OCR-D-IMG/INPUT_0017.tif')
SAMPLE_PAGE = assets.path_to(
'kant_aufklaerung_1784/data/OCR-D-GT-PAGE/PAGE_0017_PAGE.xml')
class TestModelFactory(TestCase):
def test_exif_from_filename(self):
exif_from_filename(SAMPLE_IMG)
with self.assertRaisesRegex(
Exception,
"Must pass 'image_filename' to 'exif_from_filename'"):
exif_from_filename(None)
def test_page_from_file(self):
f = create_ocrd_file_with_defaults(mimetype='image/tiff',
local_filename=SAMPLE_IMG,
ID='file1')
self.assertEqual(f.mimetype, 'image/tiff')
p = page_from_file(f)
self.assertEqual(p.pcGtsId, f.ID)
self.assertEqual(p.get_Page().imageWidth, 1457)
def test_page_from_file_page(self):
def _fixture_sbb_data_tmp(tmp_path):
copytree(assets.path_to('SBB0000F29300010000/data'), str(tmp_path))
yield str(tmp_path)
def test_validate_filename_off(self):
report = PageValidator.validate(filename=assets.path_to(
'glyph-consistency/data/OCR-D-GT-PAGE/FAULTY_GLYPHS.xml'),
strictness='off')
self.assertEqual(len(report.errors), 0, 'no errors')
from tests.base import TestCase, assets, main # pylint: disable=import-error,no-name-in-module
from ocrd.resolver import Resolver
from ocrd_validators import PageValidator
from ocrd_validators.page_validator import get_text, set_text, ConsistencyError
from ocrd_models.ocrd_page import parse, TextEquivType
from ocrd_utils import pushd_popd
FAULTY_GLYPH_PAGE_FILENAME = filename = assets.path_to(
'glyph-consistency/data/OCR-D-GT-PAGE/FAULTY_GLYPHS.xml')
class TestPageValidator(TestCase):
def setUp(self):
pass
def test_validate_err(self):
with self.assertRaisesRegex(
Exception,
'At least one of ocrd_page, ocrd_file or filename must be set'
):
PageValidator.validate()
with self.assertRaisesRegex(
Exception, 'page_textequiv_strategy best not implemented'):
PageValidator.validate(filename=FAULTY_GLYPH_PAGE_FILENAME,
page_textequiv_strategy='best')
# test with deprecated name
with self.assertRaisesRegex(
Exception, 'page_textequiv_strategy best not implemented'):
PageValidator.validate(filename=FAULTY_GLYPH_PAGE_FILENAME,
strategy='best')
with self.assertRaisesRegex(
def test_resolve_image_exif(self):
with pushd_popd(assets.path_to('kant_aufklaerung_1784/data/')):
ws = self.resolver.workspace_from_url('mets.xml')
exif = ws.resolve_image_exif('OCR-D-IMG/INPUT_0017.tif')
self.assertEqual(exif.compression, 'jpeg')
self.assertEqual(exif.width, 1457)
def test_validate_filename(self):
report = PageValidator.validate(filename=assets.path_to(
'glyph-consistency/data/OCR-D-GT-PAGE/FAULTY_GLYPHS.xml'))
self.assertEqual(len(report.errors), 17, '17 errors')
def test_validate_page(self):
page_path = assets.path_to(
'glyph-consistency/data/OCR-D-GT-PAGE/FAULTY_GLYPHS.xml')
result = self.runner.invoke(validate_cli, ['page', page_path])
self.assertEqual(result.exit_code, 1)
self.assertIn('<report valid="false">', result.stdout)
def _fixture_sbb(tmp_path):
src_path = assets.path_to('SBB0000F29300010000/data')
dst_path = tmp_path / 'SBB_directory'
shutil.copytree(src_path, dst_path)
mets_path = str(join(dst_path, 'mets.xml'))
yield OcrdMets(filename=mets_path)
def test_merge(sbb_sample_01):
assert len(sbb_sample_01.file_groups) == 17
other_mets = OcrdMets(filename=assets.path_to('kant_aufklaerung_1784/data/mets.xml'))
sbb_sample_01.merge(other_mets, fileGrp_mapping={'OCR-D-IMG': 'FOO'})
assert len(sbb_sample_01.file_groups) == 18
def test_validate_page(self):
page_path = assets.path_to(
'glyph-consistency/data/OCR-D-GT-PAGE/FAULTY_GLYPHS.xml')
code, out, _ = self.invoke_cli(validate_cli, ['page', page_path])
self.assertEqual(code, 1)
self.assertIn('<report valid="false">', out)
import numpy as np
import pytest
from tests.base import (assets, main, FIFOIO)
from ocrd_models import (OcrdFile, OcrdMets)
from ocrd_models.ocrd_page import parseString
from ocrd_models.ocrd_page import TextRegionType, CoordsType, AlternativeImageType
from ocrd_utils import polygon_mask, xywh_from_polygon, bbox_from_polygon, points_from_polygon
from ocrd_modelfactory import page_from_file
from ocrd.resolver import Resolver
from ocrd.workspace import Workspace
TMP_FOLDER = '/tmp/test-core-workspace'
SRC_METS = assets.path_to('kant_aufklaerung_1784/data/mets.xml')
SAMPLE_FILE_FILEGRP = 'OCR-D-IMG'
SAMPLE_FILE_ID = 'INPUT_0017'
SAMPLE_FILE_URL = join(SAMPLE_FILE_FILEGRP, '%s.tif' % SAMPLE_FILE_ID)
def copytree(src, dst, *args, **kwargs):
rmtree(dst)
copytree_(src, dst, *args, **kwargs)
def count_files(d):
return sum(len(files) for _, _, files in walk(d))
def test_remove_page_after_remove_file(self):
with copy_of_directory(assets.path_to('SBB0000F29300010000/data')) as tempdir:
mets = OcrdMets(filename=join(tempdir, 'mets.xml'))
self.assertEqual(mets.physical_pages, ['PHYS_0001', 'PHYS_0002', 'PHYS_0005'])
mets.remove_one_file('FILE_0005_IMAGE')
self.assertEqual(mets.physical_pages, ['PHYS_0001', 'PHYS_0002'])
def test_remove_file_group_rmdir(self):
with copy_of_directory(assets.path_to('SBB0000F29300010000/data')) as tempdir:
workspace = Workspace(self.resolver, directory=tempdir)
self.assertTrue(exists(join(tempdir, 'OCR-D-IMG')))
workspace.remove_file_group('OCR-D-IMG', recursive=True)
self.assertFalse(exists(join(tempdir, 'OCR-D-IMG')))
