[app][dep] comply next digiflow vers

ulb-sachsen-anhalt · Sep 27, 2024 · 9728836 · 9728836
1 parent 5a0eab8
commit 9728836
Show file tree

Hide file tree

Showing 3 changed files with 18 additions and 8 deletions.
diff --git a/cli_record_local.py b/cli_record_local.py
@@ -165,16 +165,17 @@ def wrap_save_record_state(status: str, urn, **kwargs):
         if odem_process.record.info != 'n.a.':
             try:
                 if isinstance(odem_process.record.info, str):
-                    _info = dict(ast.literal_eval(odem_process.record.info))
-                odem_process.record.info.update(_kwargs)
-                _info = f"{odem_process.record.info}"
+                    the_info = dict(ast.literal_eval(odem_process.record.info))
+                elif isinstance(odem_process.record.info, tuple):
+                    odem_process.record.info[-1].update(_kwargs)
+                    the_info = odem_process.record.info[-1]
             except SyntaxError:
                 odem_process.logger.warning("Can't parse '%s', store info str",
                                             odem_process.record.info)
-                _info = f"{_kwargs}"
+                the_info = f"{_kwargs}"
         else:
-            _info = f"{_kwargs}"
-        handler.save_record_state(record.identifier, MARK_OCR_DONE, INFO=_info)
+            the_info = f"{_kwargs}"
+        handler.save_record_state(record.identifier, MARK_OCR_DONE, INFO=str(the_info))
         odem_process.logger.info("[%s] duration: %s/%s (%s)", odem_process.process_identifier,
                                  odem_process.statistics['timedelta'], EXECUTORS,
                                  odem_process.statistics)

diff --git a/lib/odem/odem_commons.py b/lib/odem/odem_commons.py
@@ -41,7 +41,8 @@ class ExportFormat(str, Enum):
 #
 # ODEM configuration keys
 CFG_SEC_WORKFLOW = 'workflow'
-CFG_SEC_WORKFLOW_OPT_URL = 'oai_base_url'
+CFG_SEC_WORKFLOW_OPT_URL = "base_url"
+CFG_SEC_WORKFLOW_OPT_URL_KWARGS = "base_url_requests_kwargs"
 CFG_SEC_WORKFLOW_REM_RES = 'remove_resources'
 CFG_SEC_MONITOR = 'monitoring'
 CFG_SEC_OCR = 'ocr'

diff --git a/lib/odem/odem_process.py b/lib/odem/odem_process.py
@@ -21,6 +21,7 @@
 import digiflow as df
 import digiflow.digiflow_export as dfx
 import digiflow.digiflow_metadata as dfm
+import digiflow.digiflow_io as dfo
 import digiflow.record as df_r
 
 import lib.odem.odem_commons as odem_c
@@ -95,7 +96,14 @@ def load(self):
         self.logger.debug("[%s] download %s to %s",
                           self.process_identifier, request_identifier, req_dst)
         try:
-            loader = df.OAILoader(req_dst_dir, base_url=oai_base_url, post_oai=dfm.extract_mets)
+            req_kwargs = {}
+            if self.configuration.has_option(odem_c.CFG_SEC_WORKFLOW,
+                                             odem_c.CFG_SEC_WORKFLOW_OPT_URL_KWARGS):
+                kwargs_conf = self.configuration.get(odem_c.CFG_SEC_WORKFLOW,
+                                                    odem_c.CFG_SEC_WORKFLOW_OPT_URL_KWARGS)
+                req_kwargs = {dfo.OAI_KWARG_REQUESTS: kwargs_conf}
+            loader = df.OAILoader(req_dst_dir, base_url=oai_base_url,
+                                  post_oai=dfm.extract_mets, **req_kwargs)
             loader.store = self.store
             loader.load(request_identifier, local_dst=req_dst)
         except df.ClientError as load_err: