Spaces:
Running
Running
fix paths
Browse files
tests/resources/__init__.py
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
import os
|
| 2 |
+
|
| 3 |
+
TEST_DATA_PATH = os.path.dirname(__file__)
|
tests/test_grobid_processors.py
CHANGED
|
@@ -1,9 +1,12 @@
|
|
|
|
|
|
|
|
| 1 |
from bs4 import BeautifulSoup
|
| 2 |
from document_qa.grobid_processors import get_xml_nodes_body, get_xml_nodes_figures, get_xml_nodes_header
|
|
|
|
| 3 |
|
| 4 |
|
| 5 |
def test_get_xml_nodes_body_paragraphs():
|
| 6 |
-
with open("
|
| 7 |
soup = BeautifulSoup(fo, 'xml')
|
| 8 |
|
| 9 |
nodes = get_xml_nodes_body(soup, use_paragraphs=True)
|
|
@@ -12,7 +15,7 @@ def test_get_xml_nodes_body_paragraphs():
|
|
| 12 |
|
| 13 |
|
| 14 |
def test_get_xml_nodes_body_sentences():
|
| 15 |
-
with open("
|
| 16 |
soup = BeautifulSoup(fo, 'xml')
|
| 17 |
|
| 18 |
children = get_xml_nodes_body(soup, use_paragraphs=False)
|
|
@@ -21,7 +24,7 @@ def test_get_xml_nodes_body_sentences():
|
|
| 21 |
|
| 22 |
|
| 23 |
def test_get_xml_nodes_figures():
|
| 24 |
-
with open("
|
| 25 |
soup = BeautifulSoup(fo, 'xml')
|
| 26 |
|
| 27 |
children = get_xml_nodes_figures(soup)
|
|
@@ -30,15 +33,16 @@ def test_get_xml_nodes_figures():
|
|
| 30 |
|
| 31 |
|
| 32 |
def test_get_xml_nodes_header_paragraphs():
|
| 33 |
-
with open("
|
| 34 |
soup = BeautifulSoup(fo, 'xml')
|
| 35 |
|
| 36 |
children = get_xml_nodes_header(soup)
|
| 37 |
|
| 38 |
assert len(children) == 8
|
| 39 |
|
|
|
|
| 40 |
def test_get_xml_nodes_header_sentences():
|
| 41 |
-
with open("
|
| 42 |
soup = BeautifulSoup(fo, 'xml')
|
| 43 |
|
| 44 |
children = get_xml_nodes_header(soup, use_paragraphs=False)
|
|
|
|
| 1 |
+
import os
|
| 2 |
+
|
| 3 |
from bs4 import BeautifulSoup
|
| 4 |
from document_qa.grobid_processors import get_xml_nodes_body, get_xml_nodes_figures, get_xml_nodes_header
|
| 5 |
+
from tests.resources import TEST_DATA_PATH
|
| 6 |
|
| 7 |
|
| 8 |
def test_get_xml_nodes_body_paragraphs():
|
| 9 |
+
with open(os.path.join(TEST_DATA_PATH, "2312.07559.paragraphs.tei.xml"), 'r') as fo:
|
| 10 |
soup = BeautifulSoup(fo, 'xml')
|
| 11 |
|
| 12 |
nodes = get_xml_nodes_body(soup, use_paragraphs=True)
|
|
|
|
| 15 |
|
| 16 |
|
| 17 |
def test_get_xml_nodes_body_sentences():
|
| 18 |
+
with open(os.path.join(TEST_DATA_PATH, "2312.07559.sentences.tei.xml"), 'r') as fo:
|
| 19 |
soup = BeautifulSoup(fo, 'xml')
|
| 20 |
|
| 21 |
children = get_xml_nodes_body(soup, use_paragraphs=False)
|
|
|
|
| 24 |
|
| 25 |
|
| 26 |
def test_get_xml_nodes_figures():
|
| 27 |
+
with open(os.path.join(TEST_DATA_PATH, "2312.07559.paragraphs.tei.xml"), 'r') as fo:
|
| 28 |
soup = BeautifulSoup(fo, 'xml')
|
| 29 |
|
| 30 |
children = get_xml_nodes_figures(soup)
|
|
|
|
| 33 |
|
| 34 |
|
| 35 |
def test_get_xml_nodes_header_paragraphs():
|
| 36 |
+
with open(os.path.join(TEST_DATA_PATH, "2312.07559.paragraphs.tei.xml"), 'r') as fo:
|
| 37 |
soup = BeautifulSoup(fo, 'xml')
|
| 38 |
|
| 39 |
children = get_xml_nodes_header(soup)
|
| 40 |
|
| 41 |
assert len(children) == 8
|
| 42 |
|
| 43 |
+
|
| 44 |
def test_get_xml_nodes_header_sentences():
|
| 45 |
+
with open(os.path.join(TEST_DATA_PATH, "2312.07559.sentences.tei.xml"), 'r') as fo:
|
| 46 |
soup = BeautifulSoup(fo, 'xml')
|
| 47 |
|
| 48 |
children = get_xml_nodes_header(soup, use_paragraphs=False)
|