[app][rfct] move xml validation

pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "digiflow"
-version = "3.9.2"
+version = "3.10.2"
 description = "Father's Little Digitization Workflow Helper"
 readme = "README.md"
 requires-python = ">=3.8"

src/digiflow/__init__.py

@@ -1,9 +1,10 @@
+
 from .digflow_identifier import *
 from .digiflow_io import *
 from .digiflow_metadata import *
 from .digiflow_generate import *
 from .digiflow_export import *
-from .digiflow_validate import (
+from .validate.metadata_xslt import (
     DDB_IGNORE_RULES_BASIC,
     DDB_IGNORE_RULES_MVW,
     DDB_IGNORE_RULES_NEWSPAPERS,
@@ -23,7 +24,6 @@
     LABEL_SCAN_VALIDATOR_COMPRESSION,
     LABEL_SCAN_VALIDATOR_RESOLUTION,
     LABEL_SCAN_VALIDATOR_FILEDATA,
-    UNSET_LABEL,
     UNSET_NUMBR,
     INVALID_LABEL_UNSET,
     FSReadException,
@@ -45,3 +45,5 @@
     group_can_write,
     validate_tiff,
 )
+
+from .common import UNSET_LABEL, XMLNS

src/digiflow/common.py

@@ -0,0 +1,25 @@
+"""common constants"""
+
+
+XMLNS = {
+    'alto': 'http://www.loc.gov/standards/alto/ns-v4#',
+    'dc': 'http://purl.org/dc/elements/1.1/',
+    'dv': 'http://dfg-viewer.de/',
+    'epicur': 'urn:nbn:de:1111-2004033116',
+    'marcxml': 'http://www.loc.gov/MARC21/slim',
+    'goobi': 'http://meta.goobi.org/v1.5.1/',
+    'mets': 'http://www.loc.gov/METS/',
+    'mix': 'http://www.loc.gov/mix/v20',
+    'mods': 'http://www.loc.gov/mods/v3',
+    'oai': 'http://www.openarchives.org/OAI/2.0/',
+    'oai_dc': 'http://www.openarchives.org/OAI/2.0/oai_dc/',
+    'ulb': 'https://bibliothek.uni-halle.de',
+    'vl': 'http://visuallibrary.net/vl',
+    'vlz': 'http://visuallibrary.net/vlz/1.0/',
+    'xlink': 'http://www.w3.org/1999/xlink',
+    'zvdd': 'http://zvdd.gdz-cms.de/',
+}
+
+
+UNSET_LABEL = 'n.a.'
+

src/digiflow/digflow_identifier.py

@@ -18,9 +18,10 @@
 
 from .digiflow_metadata import (
     MetsReader,
-    XMLNS,
 )
 
+import digiflow.common as dfc
+
 
 ##################
 # script constants
@@ -218,7 +219,7 @@ def enrich_urn_granular(xml_tree,
     # if no granular urns exist at all, then all must get new urn ...
     if not cnt_urn_fits:
         # identify insertion point
-        phys_divs = xml_tree.findall('.//mets:structMap[@TYPE="PHYSICAL"]/mets:div/mets:div', XMLNS)
+        phys_divs = xml_tree.findall('.//mets:structMap[@TYPE="PHYSICAL"]/mets:div/mets:div', dfc.XMLNS)
         # alert invalid data
         if len(phys_divs) == 0:
             raise GranularURNException(f"No phys pages in {xml_tree.base}!")
@@ -329,7 +330,7 @@ def insert_granular_urn(urn_type, phys_conts, main_urn, page_num=None, padd_left
 def _get_phys_containers(xml_tree):
     """Get all physical containers that contain @ORDER attribute"""
     return xml_tree.findall(
-        './/mets:structMap[@TYPE="PHYSICAL"]//mets:div[@ORDER]', XMLNS)
+        './/mets:structMap[@TYPE="PHYSICAL"]//mets:div[@ORDER]', dfc.XMLNS)
 
 
 def enrich_urn_kitodo2(process_path:'str|Path', collection='1', system_id=KITODO2_ID, exemplar='1',

src/digiflow/digiflow_io.py

@@ -26,11 +26,12 @@
 from lxml import etree as ET
 
 from .digiflow_metadata import (
-    XMLNS,
     MetsReader,
     write_xml_file,
 )
 
+import digiflow.common as dfc
+
 
 ####
 #
@@ -53,28 +54,6 @@
 STATETIME_FORMAT_ALT = '%Y-%m-%dT%H:%M:%SZ'
 
 
-def post_oai_extract_metsdata(xml_tree):
-    """Extract METS as new root from OAI envelope"""
-
-    namespace = xml_tree.xpath('namespace-uri(.)')
-    if namespace == 'http://www.loc.gov/METS/':
-        return xml_tree
-
-    if namespace == 'http://www.openarchives.org/OAI/2.0/':
-        mets_root_el = xml_tree.find('.//mets:mets', XMLNS)
-        if mets_root_el is not None:
-            return ET.ElementTree(mets_root_el).getroot()
-    return None
-
-
-def post_oai_extract_mets(the_self, the_data):
-    """Just extract METS from OAI body"""
-
-    xml_root = ET.fromstring(the_data)
-    mets_tree = post_oai_extract_metsdata(xml_root)
-    write_xml_file(mets_tree, the_self.path_mets, preamble=None)
-
-
 def post_oai_store_ocr(path_local, the_data):
     """
     Store OCR XML as it is
@@ -816,8 +795,8 @@ def load_resource(self, url, path_local, post_func):
                 # from test-data or *real* requests
                 if not isinstance(_snippet, str):
                     _snippet = _snippet.decode('utf-8')
-                if XMLNS['mets'] in _snippet or XMLNS['oai'] in _snippet:
-                    data = post_func(self, data)
+                if dfc.XMLNS['mets'] in _snippet or dfc.XMLNS['oai'] in _snippet:
+                    data = post_func(self.path_mets, data)
                 elif 'http://www.loc.gov/standards/alto' in _snippet:
                     data = post_func(local_path, data)
                 else: