class TestDedocManager(TestCase):
path = os.path.abspath(
os.path.join(os.path.dirname(__file__), "..", "data"))
config = get_config()
manager_config = get_manager_config(config=config)
dedoc_manager = DedocManager.from_config(version="test",
manager_config=manager_config,
config=config)
def test_parse_file(self):
filename = "csv_tab.tsv"
result = self.dedoc_manager.parse_file(
os.path.join(self.path, "csv_tab.tsv"), {})
self.assertEqual(filename, result.metadata.file_name)
self.assertEqual(filename, result.metadata.file_name)
self.assertLessEqual(["1", "2", "3"],
result.content.tables[0].cells[0])
self.assertLessEqual(["2", "1", "5"],
result.content.tables[0].cells[1])
self.assertLessEqual(["5", "3", "1"],
result.content.tables[0].cells[2])
def test_version(self):
self.assertEqual("test", self.dedoc_manager.version)
def test_file_not_exists(self):
with self.assertRaises(FileNotFoundError):
self.dedoc_manager.parse_file("afagahcr", {})
def get_api_dict(api: Api,
depth: int = 0,
name: str = 'TreeNode') -> Model:
return api.model(
name, {
'node_id':
fields.String(
description=
"Document element identifier. It is unique within one tree (i.e. "
"there will be no other such node_id in this tree, but in attachment "
"it may occur) The identifier has the form 0.2.1 where each number "
"means a serial number at the corresponding level of the hierarchy.",
required=True,
example="0.2.1"),
'text':
fields.String(description="text of node",
required=True,
example="Закон"),
'annotations':
fields.List(
fields.Nested(Annotation.get_api_dict(api),
description="Text annotations "
"(font, size, bold, italic and etc)")),
'metadata':
fields.Nested(ParagraphMetadata.get_api_dict(api),
skip_none=True,
allow_null=False,
description="Paragraph meta information"),
'subparagraphs':
fields.List(
fields.Nested(api.model('others_TreeNode', {})),
description=
"Node childes (with type 'TreeNode') of structure tree")
if depth == get_config()['recursion_deep_subparagraphs'] else
fields.List(
fields.Nested(
TreeNode.get_api_dict(
api,
depth=depth +
1,
name='refTreeNode' + str(depth))),
description=
"Node childes (with type 'TreeNode') of structure tree")
})
def get_api_dict(api: Api,
depth: int = 0,
name: str = 'ParsedDocument') -> Model:
return api.model(
name, {
'content':
fields.Nested(DocumentContent.get_api_dict(api),
description='Document content structure'),
'metadata':
fields.Nested(DocumentMetadata.get_api_dict(api),
allow_null=False,
skip_none=True,
description='Document meta information'),
'version':
fields.String(
description=
'the version of the program that parsed this document',
example="2020.07.11"),
'warnings':
fields.List(
fields.String(
description='list of warnings and possible errors',
example="DOCX: seems that document corrupted")),
'attachments':
fields.List(fields.Nested(
api.model('others_ParsedDocument', {})),
description='structure of attachments',
required=False) if depth
== get_config()['recursion_deep_attachments'] else fields.List(
fields.Nested(ParsedDocument.get_api_dict(
api,
depth=depth + 1,
name='refParsedDocument' + str(depth)),
description='Attachment structure',
required=False))
})
from dedoc.config import get_config
config = get_config()
def get_command_keep_models():
"""
Ignore some private models in swagger api documentations
"""
none_display = "{display: none !important}"
discription = "API <style> " + ' '.join(["div#model-refTreeNode{depth} {none_display}"
.format(depth=i, none_display=none_display)
for i in range(config['recursion_deep_subparagraphs'])]) \
+ ' '.join(["div#model-refParsedDocument{depth} {none_display}"
.format(depth=i, none_display=none_display)
for i in range(config['recursion_deep_attachments'])]) \
+ " div#model-others_ParsedDocument {none_display} ".format(none_display=none_display) \
+ " div#model-others_TreeNode {none_display}".format(none_display=none_display)\
+ " div#model-Predicts {none_display}".format(none_display=none_display)\
+ " </style>"
return discription