inspirehep · github-actions · Jan 29, 2025
diff --git a/.kargo-render/metadata.yaml b/.kargo-render/metadata.yaml
@@ -0,0 +1 @@
+sourceCommit: 05299924ac8e38f2deef61026000c57af77088c8
diff --git a/curation-scripts/hep-script-add-to-fermilab-collection-cronjob.yaml b/curation-scripts/hep-script-add-to-fermilab-collection-cronjob.yaml
@@ -0,0 +1,82 @@
+apiVersion: batch/v1
+kind: CronJob
+metadata:
+  labels:
+    app.kubernetes.io/component: job
+  name: hep-script-add-to-fermilab-collection
+spec:
+  jobTemplate:
+    metadata:
+      labels:
+        app.kubernetes.io/component: job
+    spec:
+      completionMode: Indexed
+      completions: 10
+      parallelism: 10
+      template:
+        metadata:
+          labels:
+            app.kubernetes.io/component: job
+        spec:
+          containers:
+          - args:
+            - shell
+            - /usr/local/src/script.py
+            command:
+            - inspirehep
+            env:
+            - name: SENTRY_ENVIRONMENT
+              valueFrom:
+                fieldRef:
+                  fieldPath: metadata.namespace
+            - name: POSTGRESQL_HOST
+              value: inspire-qa-db-cluster-pooler-rw.inspire-qa.svc
+            - name: POSTGRESQL_PORT
+              value: "5432"
+            - name: POSTGRESQL_USER
+              valueFrom:
+                secretKeyRef:
+                  key: user
+                  name: postgres-inspire-pguser-hep
+            - name: POSTGRESQL_PASSWORD
+              valueFrom:
+                secretKeyRef:
+                  key: password
+                  name: postgres-inspire-pguser-hep
+            - name: JOB_COMPLETIONS
+              value: "10"
+            envFrom:
+            - configMapRef:
+                name: hep-defaults
+            - configMapRef:
+                name: hep-globals
+            - configMapRef:
+                name: hep-feature-flags
+            - secretRef:
+                name: hep-creds
+            image: registry.cern.ch/docker.io/inspirehep/hep
+            name: hep
+            volumeMounts:
+            - mountPath: /usr/local/var/instance/inspirehep_api.cfg
+              name: hep-cfg
+              subPath: inspirehep.cfg
+            - mountPath: /usr/local/var/instance/inspirehep.cfg
+              name: hep-cfg
+              subPath: inspirehep.cfg
+            - mountPath: /home/invenio
+              name: invenio-home
+            - mountPath: /usr/local/src/script.py
+              name: hep-script
+              subPath: script.py
+          restartPolicy: Never
+          volumes:
+          - configMap:
+              name: hep-cfg
+            name: hep-cfg
+          - emptyDir: {}
+            name: invenio-home
+          - configMap:
+              name: hep-script-add-to-fermilab-collection-ff8f9mmk6t
+            name: hep-script
+  schedule: '@yearly'
+  suspend: true
diff --git a/curation-scripts/hep-script-add-to-fermilab-collection-ff8f9mmk6t-configmap.yaml b/curation-scripts/hep-script-add-to-fermilab-collection-ff8f9mmk6t-configmap.yaml
@@ -0,0 +1,28 @@
+apiVersion: v1
+data:
+  script.py: |
+    from inspirehep.curation.search_check_do import SearchCheckDo
+
+
+    class AddToFermilabCollection(SearchCheckDo):
+        """Add records with Fermilab report numbers to the Fermilab collection."""
+
+        query = "r FERMILAB* -_collections:Fermilab"
+
+        @staticmethod
+        def check(record, logger, state):
+            reports = record.get_value("report_numbers.value", [])
+            logger.info("Report numbers in record", reports=reports)
+            if "Fermilab" in record["_collections"]:
+                return False
+            return any(report.lower().startswith("fermilab") for report in reports)
+
+        @staticmethod
+        def do(record, logger, state):
+            record["_collections"].append("Fermilab")
+
+
+    AddToFermilabCollection()
+kind: ConfigMap
+metadata:
+  name: hep-script-add-to-fermilab-collection-ff8f9mmk6t
diff --git a/curation-scripts/hep-script-amend-pbn-cronjob.yaml b/curation-scripts/hep-script-amend-pbn-cronjob.yaml
@@ -0,0 +1,82 @@
+apiVersion: batch/v1
+kind: CronJob
+metadata:
+  labels:
+    app.kubernetes.io/component: job
+  name: hep-script-amend-pbn
+spec:
+  jobTemplate:
+    metadata:
+      labels:
+        app.kubernetes.io/component: job
+    spec:
+      completionMode: Indexed
+      completions: 10
+      parallelism: 10
+      template:
+        metadata:
+          labels:
+            app.kubernetes.io/component: job
+        spec:
+          containers:
+          - args:
+            - shell
+            - /usr/local/src/script.py
+            command:
+            - inspirehep
+            env:
+            - name: SENTRY_ENVIRONMENT
+              valueFrom:
+                fieldRef:
+                  fieldPath: metadata.namespace
+            - name: POSTGRESQL_HOST
+              value: inspire-qa-db-cluster-pooler-rw.inspire-qa.svc
+            - name: POSTGRESQL_PORT
+              value: "5432"
+            - name: POSTGRESQL_USER
+              valueFrom:
+                secretKeyRef:
+                  key: user
+                  name: postgres-inspire-pguser-hep
+            - name: POSTGRESQL_PASSWORD
+              valueFrom:
+                secretKeyRef:
+                  key: password
+                  name: postgres-inspire-pguser-hep
+            - name: JOB_COMPLETIONS
+              value: "10"
+            envFrom:
+            - configMapRef:
+                name: hep-defaults
+            - configMapRef:
+                name: hep-globals
+            - configMapRef:
+                name: hep-feature-flags
+            - secretRef:
+                name: hep-creds
+            image: registry.cern.ch/docker.io/inspirehep/hep
+            name: hep
+            volumeMounts:
+            - mountPath: /usr/local/var/instance/inspirehep_api.cfg
+              name: hep-cfg
+              subPath: inspirehep.cfg
+            - mountPath: /usr/local/var/instance/inspirehep.cfg
+              name: hep-cfg
+              subPath: inspirehep.cfg
+            - mountPath: /home/invenio
+              name: invenio-home
+            - mountPath: /usr/local/src/script.py
+              name: hep-script
+              subPath: script.py
+          restartPolicy: Never
+          volumes:
+          - configMap:
+              name: hep-cfg
+            name: hep-cfg
+          - emptyDir: {}
+            name: invenio-home
+          - configMap:
+              name: hep-script-amend-pbn-f84mgc2959
+            name: hep-script
+  schedule: '@yearly'
+  suspend: true
diff --git a/curation-scripts/hep-script-amend-pbn-f84mgc2959-configmap.yaml b/curation-scripts/hep-script-amend-pbn-f84mgc2959-configmap.yaml
@@ -0,0 +1,44 @@
+apiVersion: v1
+data:
+  script.py: |
+    from inspirehep.curation.search_check_do import SearchCheckDo
+
+    tag_info = ("parent_isbn", "9789811947506")
+
+    missing_info = {
+        "journal_title": "Springer Proc.Math.Stat.",
+        "journal_volume": "396",
+        "parent_record": {"$ref": "https://inspirehep.net/api/literature/2628642"},
+    }
+
+
+    class AmendPBN(SearchCheckDo):
+        """Add missing info to PBNs with tag"""
+
+        query = "publication_info.%s:%s" % tag_info
+
+        @staticmethod
+        def check(record, logger, state):
+            # flag PBN containing tag info
+
+            state["pos_tag"] = []
+            for npbn, pbn in enumerate(record.get("publication_info", [])):
+                tag = pbn.get(tag_info[0], "")
+                if tag == tag_info[1]:
+                    state["pos_tag"].append(npbn)
+            if state["pos_tag"]:
+                return True
+            return False
+
+        @staticmethod
+        def do(record, logger, state):
+            # append missing info
+
+            for npbn in state["pos_tag"]:
+                record["publication_info"][npbn].update(missing_info)
+
+
+    AmendPBN()
+kind: ConfigMap
+metadata:
+  name: hep-script-amend-pbn-f84mgc2959
diff --git a/curation-scripts/hep-script-change-cnum-cronjob.yaml b/curation-scripts/hep-script-change-cnum-cronjob.yaml
@@ -0,0 +1,82 @@
+apiVersion: batch/v1
+kind: CronJob
+metadata:
+  labels:
+    app.kubernetes.io/component: job
+  name: hep-script-change-cnum
+spec:
+  jobTemplate:
+    metadata:
+      labels:
+        app.kubernetes.io/component: job
+    spec:
+      completionMode: Indexed
+      completions: 10
+      parallelism: 10
+      template:
+        metadata:
+          labels:
+            app.kubernetes.io/component: job
+        spec:
+          containers:
+          - args:
+            - shell
+            - /usr/local/src/script.py
+            command:
+            - inspirehep
+            env:
+            - name: SENTRY_ENVIRONMENT
+              valueFrom:
+                fieldRef:
+                  fieldPath: metadata.namespace
+            - name: POSTGRESQL_HOST
+              value: inspire-qa-db-cluster-pooler-rw.inspire-qa.svc
+            - name: POSTGRESQL_PORT
+              value: "5432"
+            - name: POSTGRESQL_USER
+              valueFrom:
+                secretKeyRef:
+                  key: user
+                  name: postgres-inspire-pguser-hep
+            - name: POSTGRESQL_PASSWORD
+              valueFrom:
+                secretKeyRef:
+                  key: password
+                  name: postgres-inspire-pguser-hep
+            - name: JOB_COMPLETIONS
+              value: "10"
+            envFrom:
+            - configMapRef:
+                name: hep-defaults
+            - configMapRef:
+                name: hep-globals
+            - configMapRef:
+                name: hep-feature-flags
+            - secretRef:
+                name: hep-creds
+            image: registry.cern.ch/docker.io/inspirehep/hep
+            name: hep
+            volumeMounts:
+            - mountPath: /usr/local/var/instance/inspirehep_api.cfg
+              name: hep-cfg
+              subPath: inspirehep.cfg
+            - mountPath: /usr/local/var/instance/inspirehep.cfg
+              name: hep-cfg
+              subPath: inspirehep.cfg
+            - mountPath: /home/invenio
+              name: invenio-home
+            - mountPath: /usr/local/src/script.py
+              name: hep-script
+              subPath: script.py
+          restartPolicy: Never
+          volumes:
+          - configMap:
+              name: hep-cfg
+            name: hep-cfg
+          - emptyDir: {}
+            name: invenio-home
+          - configMap:
+              name: hep-script-change-cnum-th6km57527
+            name: hep-script
+  schedule: '@yearly'
+  suspend: true
diff --git a/curation-scripts/hep-script-change-cnum-th6km57527-configmap.yaml b/curation-scripts/hep-script-change-cnum-th6km57527-configmap.yaml
@@ -0,0 +1,43 @@
+apiVersion: v1
+data:
+  script.py: |
+    from inspirehep.curation.search_check_do import SearchCheckDo
+
+    wrong_cnum = "C20-05-18.1"
+    new_cnum = "C21-05-31"
+    new_conf_record = "https://inspirehep.net/api/conferences/1812458"
+
+
+    class ChangeCNUM(SearchCheckDo):
+        """Wrong CNUM assigned - replace by correct information"""
+
+        query = "publication_info.cnum:%s" % wrong_cnum
+
+        @staticmethod
+        def check(record, logger, state):
+            # flag PBN with wrong CNUM
+
+            state["pos_cnum"] = []
+            for npbn, pbn in enumerate(record.get("publication_info", [])):
+                cnum = pbn.get("cnum", "")
+                if cnum == wrong_cnum:
+                    state["pos_cnum"].append(npbn)
+            if state["pos_cnum"]:
+                return True
+            return False
+
+        @staticmethod
+        def do(record, logger, state):
+            # replace CNUM and conference record
+
+            for npbn in state["pos_cnum"]:
+                record["publication_info"][npbn]["cnum"] = new_cnum
+                record["publication_info"][npbn]["conference_record"] = (
+                    "{'$ref': '%s'}" % new_conf_record
+                )
+
+
+    ChangeCNUM()
+kind: ConfigMap
+metadata:
+  name: hep-script-change-cnum-th6km57527
diff --git a/curation-scripts/hep-script-change-internal-cdf-collection-527b6845t2-configmap.yaml b/curation-scripts/hep-script-change-internal-cdf-collection-527b6845t2-configmap.yaml
@@ -0,0 +1,27 @@
+apiVersion: v1
+data:
+  script.py: |
+    from inspirehep.curation.search_check_do import SearchCheckDo
+
+
+    class ChangeInternalCDFCollection(SearchCheckDo):
+        """Ensure all CDF Internal Notes are really private."""
+
+        query = '_collections:"CDF Internal Notes"'
+
+        @staticmethod
+        def check(record, logger, state):
+            return (
+                len(record["_collections"]) > 1
+                and "CDF Internal Notes" in record["_collections"]
+            )
+
+        @staticmethod
+        def do(record, logger, state):
+            record["_collections"] = ["CDF Internal Notes"]
+
+
+    ChangeInternalCDFCollection()
+kind: ConfigMap
+metadata:
+  name: hep-script-change-internal-cdf-collection-527b6845t2
Original file line number	Diff line number	Diff line change
		@@ -0,0 +1 @@
		sourceCommit: 05299924ac8e38f2deef61026000c57af77088c8