added two tests for table_parsing.py

-testing number of parsed rectangles
-testing range of table coordinates (where to find a table)
This commit is contained in:
llocarnini 2022-02-28 16:12:30 +01:00
parent 6fb34735a2
commit 496957051c
2 changed files with 35 additions and 1 deletions

7
.gitignore vendored
View File

@ -8,4 +8,9 @@
/.idea/vcs.xml
/results/
/data
/table_parsing.egg-info
/table_parsing.egg-info
/tests/VV-313450.pdf
/vidocp.egg-info/dependency_links.txt
/vidocp.egg-info/PKG-INFO
/vidocp.egg-info/SOURCES.txt
/vidocp.egg-info/top_level.txt

View File

@ -0,0 +1,29 @@
import pytest
from vidocp.table_parsing import parse_table
import numpy as np
import pdf2image
@pytest.fixture()
def rects():
page_index = 0
pdf_path = "/home/lillian/vidocp/tests/VV-313450.pdf"
page = pdf2image.convert_from_path(pdf_path, first_page=page_index + 1, last_page=page_index + 1)[0]
page = np.array(page)
rectangles = parse_table(page)
return rectangles
def test_num_of_rects(rects):
assert len(rects) == 49
def test_range_of_rects(rects):
expected_range = ((210, 605), (1430, 1620))
topleft = min(rects)
x,y,w,h = max(rects)
bottomright = (x+w, y+h)
assert topleft >= expected_range[0]
assert bottomright <= expected_range[1]