Merge branch 'dev' into f-integrationtest

src/newcrawler/crawl.py

@@ -122,6 +122,23 @@ def check_identical(record1: db.Entity, record2: db.Entity, ignore_id=False):
     return True
 
 
+def _resolve_datatype(prop: db.Property, remote_entity: db.Entity):
+
+    if remote_entity.role == "Property":
+        datatype = remote_entity.datatype
+    elif remote_entity.role == "RecordType":
+        datatype = remote_entity.name
+    else:
+        raise RuntimeError("Cannot set datatype.")
+
+    # Treat lists separately
+    if isinstance(prop.value, list) and not datatype.startswith("LIST"):
+        datatype = db.LIST(datatype)
+
+    prop.datatype = datatype
+    return prop
+
+
 class Crawler(object):
     """
     Crawler class that encapsulates crawling functions.
@@ -465,8 +482,8 @@ class Crawler(object):
         """
         for p in record.properties:
             if (isinstance(p.value, list)):
-                for el in p.value:
                 lst = []
+                for el in p.value:
                     if (isinstance(el, db.Entity) and el.id is None):
                         cached = self.get_identified_record_from_local_cache(
                             el)
@@ -703,6 +720,10 @@ class Crawler(object):
 
     @staticmethod
     def execute_inserts_in_list(to_be_inserted):
+        for record in to_be_inserted:
+            for prop in record.properties:
+                entity = db.Entity(name=prop.name).retrieve()
+                prop = _resolve_datatype(prop, entity)
         print("INSERT")
         print(to_be_inserted)
         if len(to_be_inserted) > 0:
@@ -719,12 +740,7 @@ class Crawler(object):
                 if prop.id is None:
                     entity = db.Entity(name=prop.name).retrieve()
                     prop.id = entity.id
-                    if entity.role == "Property":
-                        prop.datatype = entity.datatype
-                    elif entity.role == "RecordType":
-                        prop.datatype = entity.name
-                    else:
-                        raise RuntimeError("Cannot set datatype.")
+                    prop = _resolve_datatype(prop, entity)
         print("UPDATE")
         print(to_be_updated)
         if len(to_be_updated) > 0:
@@ -996,6 +1012,7 @@ def parse_args():
 
     return parser.parse_args()
 
+
 def main():
     args = parse_args()
     return crawler_main(
@@ -1009,5 +1026,6 @@ def main():
         args.prefix
     )
 
+
 if __name__ == "__main__":
     sys.exit(main())

unittests/records.xml

@@ -78,6 +78,7 @@
     <Property id="248" name="identifier" description="identifier of the experiment" datatype="TEXT" importance="FIX" flag="inheritance:FIX">TimeOfFlight</Property>
     <Property id="250" name="project" datatype="Project" importance="FIX" flag="inheritance:FIX">287</Property>
     <Property id="249" name="responsible" datatype="LIST&lt;Person&gt;" importance="FIX" flag="inheritance:FIX">
+      <Value>289</Value>
       <Value>288</Value>
     </Property>
   </Record>

unittests/test_converters.py

@@ -203,9 +203,10 @@ def test_json_converter(converter_registry):
     assert children[2].name == "archived"
     assert children[2].value.__class__ == bool
 
-    assert children[3].__class__ == DictDictElement
-    assert children[3].name == "coordinator"
-    assert children[3].value.__class__ == dict
+    assert children[3].__class__ == DictListElement
+    assert children[3].name == "Person"
+    assert children[3].value.__class__ == list
+    assert len(children[3].value) == 2
 
     assert children[4].__class__ == DictTextElement
     assert children[4].name == "start_date"

unittests/test_directories/examples_json/jsontest_cfood.yml

@@ -3,13 +3,56 @@ JSONTest:  # name of the converter
   type: JSONFile
   match: '(.*)'
   validate: ./testjson.schema.json
+  records:
+    Project:  # this is an identifiable in this case
+      parents:
+        - Project  # not needed as the name is equivalent
   subtree:
-    element:  # name of the first subtree element which is a converter
+    name_element:
+      type: DictTextElement
+      match_name: "name"
+      match_value: "(?P<name>.*)"
+      records:
+        Project:
+          name: $name
+    url_element:  # name of the first subtree element which is a converter
       type: DictTextElement
       match_value: "(?P<url>.*)"
       match_name: "url"
       records:
-        Project:  # this is an identifiable in this case
-          parents:
-            - Project  # not needed as the name is equivalent
+        Project:
           url: $url
+    persons_element:
+      type: DictListElement
+      match_name: "Person"
+      subtree:
+        person_element:
+          type: Dict
+          records:
+            Person:
+              parents:
+                - Person
+            Project:
+              Person: +$Person
+          subtree:
+            firstname_element:
+              type: DictTextElement
+              match_name: "firstname"
+              match_value: "(?P<firstname>.*)"
+              records:
+                Person:
+                  firstname: $firstname
+            lastname_element:
+              type: DictTextElement
+              match_name: "lastname"
+              match_value: "(?P<lastname>.*)"
+              records:
+                Person:
+                  lastname: $lastname
+            email_element:
+              type: DictTextElement
+              match_name: "email"
+              match_value: "(?P<email>.*)"
+              records:
+                Person:
+                  email: $email

unittests/test_directories/examples_json/testjson.json

@@ -2,11 +2,18 @@
 	"name": "DEMO",
 	"projectId": 10002,
 	"archived": false,
-	"coordinator": {
+	"Person": [
+        {
 		    "firstname": "Miri",
 		    "lastname": "Mueller",
 		    "email": "miri.mueller@science.de"
 	    },
+        {
+            "firstname": "Mara",
+            "lastname": "Mueller",
+		    "email": "mara.mueller@science.de"
+        }
+    ],
 	"start_date": "2022-03-01",
 	"candidates": ["Mouse", "Penguine"],
 	"rvalue": 0.4444,

unittests/test_directories/examples_json/testjson.schema.json

@@ -11,7 +11,9 @@
     "archived": {
       "type": "boolean"
     },
-    "coordinator": {
+    "Person": {
+        "type": "array",
+        "items": {
             "type": "object",
             "properties": {
                 "firstname": {
@@ -30,6 +32,7 @@
                 "email"
             ],
             "additionalProperties": true
+        }
     },
     "start_date": {
       "type": "string",
@@ -51,7 +54,7 @@
   "required": [
     "name",
     "projectId",
-    "coordinator"
+    "Person"
   ],
   "additionalProperties": false
 }

unittests/test_json.py

@@ -31,6 +31,8 @@ import os
 
 from pytest import raises
 
+import caosdb as db
+
 from newcrawler.converters import JSONFileConverter, DictConverter
 from newcrawler.crawl import Crawler
 from newcrawler.structure_elements import File, JSONFile
@@ -47,19 +49,27 @@ def test_json():
     # Load and register converter packages:
     converter_registry = crawler.load_converters(crawler_definition)
 
-    crawler.start_crawling(
+    records = crawler.start_crawling(
         JSONFile(os.path.basename(json_file_path), json_file_path),
         crawler_definition,
         converter_registry
     )
-    subd = crawler.debug_tree
-    subc = crawler.debug_metadata
-    #print(json.dumps(subd, indent=3))
-    print(subd)
-    print(subc)
+
+    rec = [r for r in records if r.name == "DEMO"]
+    assert len(rec) == 1
+    rec = rec[0]
+    assert len(rec.parents) == 1
+    assert rec.parents[0].name == "Project"
+    assert rec.get_property("url") is not None
+    assert rec.get_property("url").value == "https://site.de/index.php/"
+    assert rec.get_property("Person") is not None
+    assert isinstance(rec.get_property("Person").value, list)
+    assert len(rec.get_property("Person").value) == 2
+
 
 def test_broken_validation():
-    crawler_definition_path = rfp("broken_cfoods", "broken_validation_path.yml")
+    crawler_definition_path = rfp(
+        "broken_cfoods", "broken_validation_path.yml")
     crawler = Crawler()
     with raises(FileNotFoundError) as err:
         crawler_definition = crawler.load_definition(crawler_definition_path)