From ef95bfbdb6c651e334fe9043d79b8088c061d993 Mon Sep 17 00:00:00 2001
From: Alexander Schlemmer <a.schlemmer@indiscale.com>
Date: Thu, 14 Nov 2024 10:01:18 +0100
Subject: [PATCH] FIX(converters): added return values to match_properties
method
---
src/caoscrawler/converters/converters.py | 10 ++++++++--
src/caoscrawler/converters/rocrate.py | 3 ++-
src/caoscrawler/converters/xml_converter.py | 3 ++-
unittests/test_rocrate_converter.py | 9 ++++-----
4 files changed, 16 insertions(+), 9 deletions(-)
diff --git a/src/caoscrawler/converters/converters.py b/src/caoscrawler/converters/converters.py
index 40ddde92..1bc70a15 100644
--- a/src/caoscrawler/converters/converters.py
+++ b/src/caoscrawler/converters/converters.py
@@ -485,6 +485,11 @@ class Converter(object, metaclass=ABCMeta):
label: str
Default "match_properties". Can be used to change the name of the property in the definition. E.g. the
xml converter uses "match_attrib" which makes more sense in the context of xml trees.
+
+ Returns:
+ --------
+
+ Returns True when properties match and False otherwise. The vardict dictionary is updated in place.
"""
if label in self.definition:
# This matcher works analogously to the attributes matcher in the XMLConverter
@@ -499,7 +504,7 @@ class Converter(object, metaclass=ABCMeta):
matched_m_prop = m_prop
m_prop_value = re.match(prop_def_value, prop_value)
if m_prop_value is None:
- return None
+ return False
matched_m_prop_value = m_prop_value
# TODO: How to deal with multiple matches?
# There are multiple options:
@@ -509,11 +514,12 @@ class Converter(object, metaclass=ABCMeta):
# Currently the latter option is implemented.
# TODO: The ROCrateEntityConverter implements a very similar behavior.
if match_counter == 0:
- return None
+ return False
elif match_counter > 1:
raise RuntimeError("Multiple properties match the same {} entry.".format(label))
vardict.update(matched_m_prop.groupdict())
vardict.update(matched_m_prop_value.groupdict())
+ return True
def apply_transformers(self, values: GeneralStore, transformer_functions: dict):
"""
diff --git a/src/caoscrawler/converters/rocrate.py b/src/caoscrawler/converters/rocrate.py
index 7fd8378e..8a45af75 100644
--- a/src/caoscrawler/converters/rocrate.py
+++ b/src/caoscrawler/converters/rocrate.py
@@ -183,7 +183,8 @@ class ROCrateEntityConverter(Converter):
return None
vardict.update(m_type.groupdict())
- self.match_properties(element.entity.properties(), vardict)
+ if not self.match_properties(element.entity.properties(), vardict):
+ return None
return vardict
diff --git a/src/caoscrawler/converters/xml_converter.py b/src/caoscrawler/converters/xml_converter.py
index 472e4051..60d7b494 100644
--- a/src/caoscrawler/converters/xml_converter.py
+++ b/src/caoscrawler/converters/xml_converter.py
@@ -162,7 +162,8 @@ class XMLTagConverter(Converter):
return None
vardict.update(m_text.groupdict())
- self.match_properties(element.tag.attrib, vardict, "match_attrib")
+ if not self.match_properties(element.tag.attrib, vardict, "match_attrib"):
+ return None
return vardict
diff --git a/unittests/test_rocrate_converter.py b/unittests/test_rocrate_converter.py
index ef59a37c..06ce187e 100644
--- a/unittests/test_rocrate_converter.py
+++ b/unittests/test_rocrate_converter.py
@@ -32,17 +32,16 @@ import linkahead as db
import pytest
import rocrate
import yaml
-from linkahead.high_level_api import convert_to_python_object
-from lxml.etree import fromstring
-from rocrate.model.entity import Entity
-from rocrate.rocrate import ROCrate
-
from caoscrawler import scanner
from caoscrawler.converters import ELNFileConverter, ROCrateEntityConverter
from caoscrawler.scanner import load_definition
from caoscrawler.stores import GeneralStore
from caoscrawler.structure_elements import (DictElement, File, ROCrateEntity,
TextElement)
+from linkahead.high_level_api import convert_to_python_object
+from lxml.etree import fromstring
+from rocrate.model.entity import Entity
+from rocrate.rocrate import ROCrate
UNITTESTDIR = Path(__file__).parent
--
GitLab