[app][feat] validate if configured close #19

cli_oai_client.py

@@ -353,7 +353,7 @@ def _clear_sub_dirs(root_dir: str):
         process_resource_monitor.check_vmem()
         process_resource_monitor.monit_disk_space(PROCESS.load)
         if CFG.getboolean('mets', 'prevalidate', fallback=True):
-            PROCESS.validate_mets()
+            PROCESS.validate_metadata()
         PROCESS.inspect_metadata()
         PROCESS.clear_existing_entries()
         PROCESS.language_modelconfig()
@@ -372,7 +372,7 @@ def _clear_sub_dirs(root_dir: str):
             PROCESS.create_text_bundle_data()
         PROCESS.postprocess_mets()
         if CFG.getboolean('mets', 'postvalidate', fallback=True):
-            PROCESS.validate_mets()
+            PROCESS.validate_metadata()
         if not MUST_KEEP_RESOURCES:
             PROCESS.delete_before_export(LOCAL_DELETE_BEFORE_EXPORT)
         PROCESS.export_data()

cli_oai_local.py

@@ -198,7 +198,7 @@ def wrap_save_record_state(status: str, urn, **kwargs):
         process_resource_monitor.check_vmem()
         process_resource_monitor.monit_disk_space(PROCESS.load)
         if CFG.getboolean('mets','prevalidate', fallback=True):
-            PROCESS.validate_mets()
+            PROCESS.validate_metadata()
         PROCESS.inspect_metadata()
         PROCESS.clear_existing_entries()
         PROCESS.language_modelconfig()
@@ -214,7 +214,7 @@ def wrap_save_record_state(status: str, urn, **kwargs):
             PROCESS.create_text_bundle_data()
         PROCESS.postprocess_mets()
         if CFG.getboolean('mets','postvalidate', fallback=True):
-            PROCESS.validate_mets()
+            PROCESS.validate_metadata()
         if not MUST_KEEP_RESOURCES:
             PROCESS.delete_before_export(LOCAL_DELETE_BEVOR_EXPORT)
         PROCESS.export_data()

lib/ocrd3_odem/ocrd3_odem.py

@@ -49,10 +49,9 @@
     CATALOG_ULB,
     ODEMMetadataInspecteur,
     ODEMMetadataMetsException,
-    # extract_mets_data,
     integrate_ocr_file,
     postprocess_mets,
-    validate_mets,
+    validate,
 )
 from .processing_ocrd import (
     run_ocr_page,
@@ -480,14 +479,18 @@ def postprocess_mets(self):
 
         postprocess_mets(self.mets_file, self.cfg.get('ocr', 'ocrd_baseimage'))
 
-    def validate_mets(self):
-        """Forward METS-schema validation"""
-        try:
-            validate_mets(self.mets_file)
-        except dfv.InvalidXMLException as err:
-            if len(err.args) > 0 and ('SCHEMASV' in str(err.args[0])):
-                raise ODEMException(str(err.args[0])) from err
-            raise err
+    def validate_metadata(self):
+        """Forward (optional) validation concerning
+        METS/MODS XML-schema and/or current DDB-schematron
+        validation for 'digitalisierte medien'
+        """
+        check_ddb = False
+        if self.cfg.has_option('mets', 'ddb_validation'):
+            check_ddb = self.cfg.getboolean('mets', 'ddb_validation', fallback=False)
+        dtype = 'Aa'
+        if 'pica' in self.record.info:
+            dtype = self.record.info['pica']
+        return validate(self.mets_file, validate_ddb=check_ddb, digi_type=dtype)
 
     def export_data(self):
         """re-do metadata and transform into output format"""

lib/ocrd3_odem/processing_mets.py

@@ -1,10 +1,7 @@
 """Encapsulate Implementations concerning METS/MODS handling"""
 
 import os
-
-from typing import (
-    List,
-)
+import typing
 
 import lxml.etree as ET
 import digiflow as df
@@ -13,6 +10,7 @@
 from .odem_commons import (
     FILEGROUP_IMG,
     FILEGROUP_OCR,
+    ODEMException,
 )
 
 TYPE_PRINTS_PICA = ['a', 'f', 'F', 'Z', 'B']
@@ -29,16 +27,6 @@
 IMAGE_GROUP_DEFAULT = 'DEFAULT'
 
 
-# def extract_mets_data(the_self, the_data):
-#     """
-#     Migration Post-recive OAI METS/MODS callback
-#     """
-
-#     xml_root = ET.fromstring(the_data)
-#     mets_tree = df.post_oai_extract_metsdata(xml_root)
-#     df.write_xml_file(mets_tree, the_self.path_mets)
-
-
 class ODEMMetadataMetsException(Exception):
     """Mark state when inconsistencies exist
     between linkings of physical and logical
@@ -261,7 +249,7 @@ def clear_filegroups(xml_file, removals):
     proc.write()
 
 
-def integrate_ocr_file(xml_tree, ocr_files: List) -> int:
+def integrate_ocr_file(xml_tree, ocr_files: typing.List) -> int:
     """Enrich given OCR-Files into XML tree
     
     Returns number of linked files
@@ -319,7 +307,7 @@ def _link_fulltext(file_ident, xml_tree):
     return 0
 
 
-def is_in(tokens: List[str], label):
+def is_in(tokens: typing.List[str], label):
     """label contained somewhere in a list of tokens?"""
 
     return any(t in label for t in tokens)
@@ -370,8 +358,18 @@ def _clear_provenance_links(mproc):
         parent.remove(old_dv)
 
 
-def validate_mets(mets_file:str):
+def validate(mets_file:str, validate_ddb=False, digi_type='Aa'):
     """Forward METS-schema validation"""
 
     xml_root = ET.parse(mets_file).getroot()
-    dfv.validate_xml(xml_root)
+    try:
+        dfv.validate_xml(xml_root)
+        if validate_ddb:
+            df.ddb_validation(path_mets=mets_file, digi_type=digi_type)
+    except dfv.InvalidXMLException as err:
+            if len(err.args) > 0 and ('SCHEMASV' in str(err.args[0])):
+                raise ODEMException(str(err.args[0])) from err
+            raise err
+    except df.DigiflowDDBException as ddb_err:
+        raise ODEMException(ddb_err.args[0]) from ddb_err
+    return True