Merge pull request #36 from Peyman-N/openminds_persons

creating person

Author: lzehl

bids2openminds/main.py

@@ -1,8 +1,10 @@
+import re
 import os
 import pathlib
 from warnings import warn
 
 import pandas as pd
+from nameparser import HumanName
 
 import openminds.latest.core as omcore
 import openminds.latest.controlled_terms as controlled_terms
@@ -12,6 +14,64 @@
 from .mapping import bids2openminds_instance
 
 
+def create_openminds_person(full_name):
+    # Regex for detecting any unwanted characters.
+    name_regex = re.compile(
+        "^[\w'\-, .][^0-9_!¡?÷?¿/\\+=@#$%^&*(){}|~<>;:[\]]{1,}$")
+    alternate_names = []
+    person = HumanName(full_name)
+    given_name = person.first
+    family_name = person.last
+
+    # Handle the situation in which there is no given name or the given name consists of unwanted characters.
+    if not (given_name and name_regex.match(given_name)):
+        return None
+
+    # Handle the situation in which the family name consists of unwanted characters.
+    if not (name_regex.match(family_name)):
+        family_name = None
+
+    if person.middle:
+        given_name = f"{given_name} {person.middle}"
+
+    if person.nickname:
+        alternate_names.append(person.nickname)
+
+    if not alternate_names:
+        alternate_names = None
+
+    openminds_person = omcore.Person(
+        alternate_names=alternate_names, given_name=given_name, family_name=family_name)
+
+    return openminds_person
+
+
+def create_persons(dataset_description, collection):
+
+    if "Authors" in dataset_description:
+        person_list = dataset_description["Authors"]
+    else:
+        return None
+
+    if not (isinstance(person_list, list)):
+        # handel's only one name
+        if isinstance(person_list, str):
+            openminds_person = create_openminds_person(person_list)
+            if openminds_person is not None:
+                collection.add(openminds_person)
+            return openminds_person
+        else:
+            return None
+
+    openminds_list = []
+    for person in person_list:
+        openminds_person = create_openminds_person(person)
+        openminds_list.append(openminds_person)
+        collection.add(openminds_person)
+
+    return openminds_list
+
+
 def create_techniques(layout_df):
     suffixs = layout_df["suffix"].unique().tolist()
     techniques = []
@@ -56,8 +116,7 @@ def create_dataset_version(bids_layout, dataset_description, layout_df, studied_
     else:
         digital_identifier = None
 
-    # TODO extract person
-    # author=person_create(dataset_description["Authors"])
+    authors = create_persons(dataset_description, collection)
 
     if "Acknowledgements" in dataset_description:
         other_contribution = dataset_description["Acknowledgements"]
@@ -92,6 +151,7 @@ def create_dataset_version(bids_layout, dataset_description, layout_df, studied_
         experimental_approaches=experimental_approaches,
         short_name=dataset_description["Name"],
         studied_specimens=studied_specimens,
+        authors=authors,
         techniques=techniques,
         how_to_cite=how_to_cite,
         repository=file_repository,

pyproject.toml

@@ -6,6 +6,7 @@ dependencies = [
   "openminds >= 0.2.2",
   "click>=8.1",
   "pandas",
+  "nameparser >= 1.1.3"
 ]
 requires-python = ">=3.9"
 authors = [

test/test_example_datasets_click.py

@@ -3,7 +3,7 @@
 from bids2openminds.converter import convert_click
 from click.testing import CliRunner
 
-(test_data_set, number_of_openminds_files) = ("ds003", 98)
+(test_data_set, number_of_openminds_files) = ("ds003", 100)
 
 
 def test_example_datasets_click():

test/test_person.py

@@ -0,0 +1,33 @@
+# test for create_openminds_person function in the main
+import pytest
+from bids2openminds.main import create_openminds_person
+import openminds.latest.core as omcore
+
+# Test data: (full_name, given_name, family_name)
+example_names = [("John Ronald Reuel Tolkien", "John Ronald Reuel", "Tolkien"),
+                 ("Bilbo Baggins", "Bilbo", "Baggins"),
+                 ("Xue, G.", "G.", "Xue"),
+                 ("Arndís Þórarinsdóttir", "Arndís", "Þórarinsdóttir"),
+                 ("Loïc Le Clézio", "Loïc", "Le Clézio")]
+
+example_not_names = ["42", "#", "", "A34 hajb"]
+
+
+@pytest.mark.parametrize("full_name, given_name, family_name", example_names)
+def test_create_openminds_person(full_name, given_name, family_name):
+    openminds_person_object = omcore.Person(given_name=given_name,
+                                            family_name=family_name)
+    bids2openminds_person_object = create_openminds_person(full_name)
+    assert openminds_person_object.given_name == bids2openminds_person_object.given_name, \
+        f"Given names don't match for input '{full_name}'"
+    assert openminds_person_object.family_name == bids2openminds_person_object.family_name, \
+        f"Family names don't match for input '{full_name}'"
+    assert openminds_person_object.type_ == bids2openminds_person_object.type_, \
+        f"Person types don't match for input '{full_name}'"
+    # assert openminds_person_object == bids2openminds_person_object
+
+
+@pytest.mark.parametrize("full_not_name", example_not_names)
+def test_create_openminds_person_not_names(full_not_name):
+    bids2openminds_person_object = create_openminds_person(full_not_name)
+    assert bids2openminds_person_object is None