add yamlrecordconverter

src/newcrawler/converters.py

@@ -34,6 +34,8 @@ from typing import Optional, Union
 from abc import abstractmethod
 import yaml_header_tools
 
+from caosdb.high_level_api import (CaosDBPythonEntity,
+                                   create_entity_container)
 import yaml
 
 # These are special properties which are (currently) treated differently
@@ -315,6 +317,57 @@ class SimpleFileConverter(Converter):
             return None
         return m.groupdict()
 
+class YamlFileCaosDBRecord(Converter):
+    """
+    Load a file using pylib high level API and convert the contained
+    record into caosdb records.
+    """
+
+    def typecheck(self, element: StructureElement):
+        return isinstance(element, File)
+
+    def create_children(self, generalStore: GeneralStore,
+                        element: StructureElement):
+        return list()
+
+    def match(self, element: StructureElement):
+        if not isinstance(element, File):
+            raise RuntimeError("Element must be a file.")
+        m = re.match(self.definition["match"], element.name)
+        if m is None:
+            return None
+        return m.groupdict()
+
+    def create_records(self, values: GeneralStore,
+                       records: RecordStore,
+                       element: StructureElement):
+        if not isinstance(element, File):
+            raise RuntimeError("A yaml file is needed to create children.")
+        
+        keys_modified = []
+
+        with open(element.path, "r") as f:
+            entries = yaml.safe_load(f)
+
+        entity = CaosDBPythonEntity.deserialize(entries)
+        entities = create_entity_container(entity)
+
+        for n, ent in enumerate(entities):
+            name = ent.name
+            if name is None:
+                name = "YamlRecord_{}".format(n + 1)
+            records[name] = ent
+            values[name] = ent
+
+            for propname in ent.properties:
+                keys_modified.append((name, propname.name))
+
+        # Process the records section of the yaml definition:
+        keys_modified.extend(
+            super().create_records(values, records, element))
+        
+        return keys_modified
+    
 
 class MarkdownFileConverter(Converter):
     def __init__(self, definition: dict, name: str,

src/newcrawler/crawl.py

@@ -49,13 +49,13 @@ from caosdb.apiutils import compare_entities, merge_entities
 from copy import deepcopy
 from jsonschema import validate
 
+from caosdb.high_level_api import convert_to_python_object
 
 import importlib
 
 SPECIAL_PROPERTIES_STRICT = ("description", "name", "id", "path")
 SPECIAL_PROPERTIES_NOT_STRICT = ("file", "checksum", "size")
 
-
 def check_identical(record1: db.Entity, record2: db.Entity, ignore_id=False):
     """
     This function uses compare_entities to check whether to entities are identical
@@ -222,6 +222,9 @@ class Crawler(object):
             "SimpleFile": {
                 "converter": "SimpleFileConverter",
                 "package": "newcrawler.converters"},
+            "YamlFileCaosDBRecord": {
+                "converter": "YamlFileCaosDBRecord",
+                "package": "newcrawler.converters"},
             "MarkdownFile": {
                 "converter": "MarkdownFileConverter",
                 "package": "newcrawler.converters"},
@@ -842,8 +845,8 @@ def main():
         
     if args.dry_sync:
         ins, upd = crawler.synchronize(commit_changes=False)
-        inserts = [str(i) for i in ins]
-        updates = [str(i) for i in upd]
+        inserts = [convert_to_python_object(i).serialize() for i in ins]
+        updates = [convert_to_python_object(i).serialize() for i in upd]
         with open("dry.yml", "w") as f:
             f.write(yaml.dump({
                 "insert": inserts,