Update script to force CDS harvest of larger batch

michamos · michamos · commit 4ba67012f059 · 2023-01-30T11:53:39.000+01:00
diff --git a/scripts/force-cds-harvest/script.py b/scripts/force-cds-harvest/script.py
@@ -7,7 +7,7 @@ class ForceCDSHarvest(SearchCheckDo):
 
     query = (
         "(_oai.sets:CERN:arXiv or _oai.sets:ForCDS) "
-        "-external_system_identifiers.schema:CDS"
+        "and du > 2021-06-26 and arxiv_eprints.value:* and de >= 2021"
     )
 
     @staticmethod
@@ -19,6 +19,7 @@ def check(record, logger, state):
                 id_["schema"] == "CDS"
                 for id_ in record.get("external_system_identifiers", [])
             )
+            and "arxiv_eprints" in record
         )
 
     @staticmethod

Original file line number	Diff line number	Diff line change
`@@ -7,7 +7,7 @@ class ForceCDSHarvest(SearchCheckDo):`
`7`	`7`
`8`	`8`	`query = (`
`9`	`9`	`"(_oai.sets:CERN:arXiv or _oai.sets:ForCDS) "`
`10`		`- "-external_system_identifiers.schema:CDS"`
	`10`	`+ "and du > 2021-06-26 and arxiv_eprints.value:* and de >= 2021"`
`11`	`11`	`)`
`12`	`12`
`13`	`13`	`@staticmethod`
`@@ -19,6 +19,7 @@ def check(record, logger, state):`
`19`	`19`	`id_["schema"] == "CDS"`
`20`	`20`	`for id_ in record.get("external_system_identifiers", [])`
`21`	`21`	`)`
	`22`	`+ and "arxiv_eprints" in record`
`22`	`23`	`)`
`23`	`24`
`24`	`25`	`@staticmethod`