def test_merge_tables():
p = os.path.dirname(os.path.realpath(__file__))
f = open(os.path.join(p, "data/gib_multi_page_tables.json"))
j = json.load(f)
t_document: t2.TDocument = t2.TDocumentSchema().load(j)
tbl_id1 = 'fed02fb4-1996-4a15-98dc-29da193cc476'
tbl_id2 = '47c6097f-02d5-4432-8423-13c05fbfacbd'
pre_merge_tbl1_cells_no = len(t_document.get_block_by_id(tbl_id1).relationships[0].ids)
pre_merge_tbl2_cells_no = len(t_document.get_block_by_id(tbl_id2).relationships[0].ids)
pre_merge_tbl1_lastcell = t_document.get_block_by_id(tbl_id1).relationships[0].ids[-1]
pre_merge_tbl2_lastcell = t_document.get_block_by_id(tbl_id2).relationships[0].ids[-1]
pre_merge_tbl1_last_row = t_document.get_block_by_id(pre_merge_tbl1_lastcell).row_index
pre_merge_tbl2_last_row = t_document.get_block_by_id(pre_merge_tbl2_lastcell).row_index
t_document.merge_tables([[tbl_id1,tbl_id2]])
post_merge_tbl1_cells_no = len(t_document.get_block_by_id(tbl_id1).relationships[0].ids)
post_merge_tbl1_lastcell = t_document.get_block_by_id(tbl_id1).relationships[0].ids[-1]
post_merge_tbl1_last_row = t_document.get_block_by_id(post_merge_tbl1_lastcell).row_index
assert post_merge_tbl1_cells_no == pre_merge_tbl1_cells_no + pre_merge_tbl2_cells_no
assert pre_merge_tbl2_last_row
assert post_merge_tbl1_last_row == pre_merge_tbl1_last_row + pre_merge_tbl2_last_row
def test_get_relationships_for_type(caplog):
# existing relationships
p = os.path.dirname(os.path.realpath(__file__))
with open(os.path.join(p, "data/gib.json")) as f:
j = json.load(f)
t_document: t2.TDocument = t2.TDocumentSchema().load(j)
page = t_document.pages[0]
new_block = t2.TBlock(id=str(uuid4()))
t_document.add_block(new_block)
page.add_ids_to_relationships([new_block.id])
assert t_document.get_block_by_id(new_block.id) == new_block
#empty relationships
t_document: t2.TDocument = t2.TDocument()
t_document.add_block(t2.TBlock(id=str(uuid4()), block_type="PAGE"))
page = t_document.pages[0]
new_block = t2.TBlock(id=str(uuid4()))
t_document.add_block(new_block)
page.add_ids_to_relationships([new_block.id])
assert t_document.get_block_by_id(new_block.id) == new_block
def test_pipeline_merge_tables():
p = os.path.dirname(os.path.realpath(__file__))
f = open(os.path.join(p, "data/gib_multi_page_table_merge.json"))
j = json.load(f)
t_document: t2.TDocument = t2.TDocumentSchema().load(j)
tbl_id1 = '5685498d-d196-42a7-8b40-594d6d886ca9'
tbl_id2 = 'a9191a66-0d32-4d36-8fd6-58e6917f4ea6'
tbl_id3 = 'e0368543-c9c3-4616-bd6c-f25e66c859b2'
pre_merge_tbl1_cells_no = len(
t_document.get_block_by_id(
tbl_id1).relationships[0].ids) # type: ignore
pre_merge_tbl2_cells_no = len(
t_document.get_block_by_id(
tbl_id2).relationships[0].ids) # type: ignore
pre_merge_tbl3_cells_no = len(
t_document.get_block_by_id(
tbl_id3).relationships[0].ids) # type: ignore
t_document = pipeline_merge_tables(t_document, MergeOptions.MERGE, None,
HeaderFooterType.NONE)
post_merge_tbl1_cells_no = len(
t_document.get_block_by_id(
tbl_id1).relationships[0].ids) # type: ignore
assert post_merge_tbl1_cells_no == pre_merge_tbl1_cells_no + pre_merge_tbl2_cells_no + pre_merge_tbl3_cells_no
def test_custom_page_orientation(json_response):
doc = Document(json_response)
assert 1 == len(doc.pages)
lines = [line for line in doc.pages[0].lines]
assert 22 == len(lines)
words = [word for line in lines for word in line.words]
assert 53 == len(words)
t_document: t2.TDocument = t2.TDocumentSchema().load(json_response)
t_document.custom = {'orientation': 180}
new_t_doc_json = t2.TDocumentSchema().dump(t_document)
assert "Custom" in new_t_doc_json
assert "orientation" in new_t_doc_json["Custom"]
assert new_t_doc_json["Custom"]["orientation"] == 180
p = os.path.dirname(os.path.realpath(__file__))
f = open(os.path.join(p, "data/gib.json"))
j = json.load(f)
t_document: t2.TDocument = t2.TDocumentSchema().load(j)
t_document = add_page_orientation(t_document)
assert -1 < t_document.pages[0].custom['Orientation'] < 2
p = os.path.dirname(os.path.realpath(__file__))
f = open(os.path.join(p, "data/gib_10_degrees.json"))
j = json.load(f)
t_document: t2.TDocument = t2.TDocumentSchema().load(j)
t_document = add_page_orientation(t_document)
assert 5 < t_document.pages[0].custom['Orientation'] < 15
p = os.path.dirname(os.path.realpath(__file__))
f = open(os.path.join(p, "data/gib__15_degrees.json"))
j = json.load(f)
t_document: t2.TDocument = t2.TDocumentSchema().load(j)
t_document = add_page_orientation(t_document)
assert 10 < t_document.pages[0].custom['Orientation'] < 20
p = os.path.dirname(os.path.realpath(__file__))
f = open(os.path.join(p, "data/gib__25_degrees.json"))
j = json.load(f)
t_document: t2.TDocument = t2.TDocumentSchema().load(j)
t_document = add_page_orientation(t_document)
assert 17 < t_document.pages[0].custom['Orientation'] < 30
p = os.path.dirname(os.path.realpath(__file__))
f = open(os.path.join(p, "data/gib__180_degrees.json"))
j = json.load(f)
t_document: t2.TDocument = t2.TDocumentSchema().load(j)
t_document = add_page_orientation(t_document)
assert 170 < t_document.pages[0].custom['Orientation'] < 190
p = os.path.dirname(os.path.realpath(__file__))
f = open(os.path.join(p, "data/gib__270_degrees.json"))
j = json.load(f)
t_document: t2.TDocument = t2.TDocumentSchema().load(j)
t_document = add_page_orientation(t_document)
assert -100 < t_document.pages[0].custom['Orientation'] < -80
p = os.path.dirname(os.path.realpath(__file__))
f = open(os.path.join(p, "data/gib__90_degrees.json"))
j = json.load(f)
t_document: t2.TDocument = t2.TDocumentSchema().load(j)
t_document = add_page_orientation(t_document)
assert 80 < t_document.pages[0].custom['Orientation'] < 100
p = os.path.dirname(os.path.realpath(__file__))
f = open(os.path.join(p, "data/gib__minus_10_degrees.json"))
j = json.load(f)
t_document: t2.TDocument = t2.TDocumentSchema().load(j)
t_document = add_page_orientation(t_document)
assert -10 < t_document.pages[0].custom['Orientation'] < 5
doc = t1.Document(t2.TDocumentSchema().dump(t_document))
for page in doc.pages:
assert page.custom['Orientation']