Merge branch 'rpb-154-missingLabels' of https://github.com/hbz/rpb in…

…to main Resolves https://jira.hbz-nrw.de/browse/RPB-154 See https://rppd.lobid.org/121384462
hbz · Apr 17, 2024 · 4763f0d · 4763f0d
2 parents 55fb38b + 61f1be6
commit 4763f0d
Show file tree

Hide file tree

Showing 18 changed files with 18,756 additions and 16,116 deletions.
diff --git a/.gitignore b/.gitignore
@@ -22,4 +22,5 @@ RPB-Export_HBZ_SWN.txt
 RPB-Export_HBZ_Syst.txt
 RPB-Export_HBZ_ZSS.txt
 conf/RPBEXP/*.ZIP
+conf/strapi-export.tar.gz
 nohup.out*
diff --git a/conf/map-gnd-person-to-label.fix b/conf/map-gnd-person-to-label.fix
@@ -0,0 +1,8 @@
+if any_contain("gndIdentifier", "Keine")
+	reject()
+end
+
+replace_all("preferredName", "\\/\\s(ca\\.|um)?-?\\s?\\d.+$", "")
+trim("preferredName")
+
+retain("gndIdentifier", "preferredName")
diff --git a/conf/map-gnd-person-to-label.flux b/conf/map-gnd-person-to-label.flux
@@ -0,0 +1,8 @@
+FLUX_DIR + "output/rppd-export.jsonl"
+| open-file
+| as-lines
+| decode-json(recordPath="data")
+| fix(FLUX_DIR + "map-gnd-person-to-label.fix")
+| encode-csv(includeheader="true", noquotes="true",separator="\t")
+| write(FLUX_DIR + "maps/gndId-to-label.tsv")
+;
diff --git a/conf/map-rppd-to-label.fix b/conf/map-rppd-to-label.fix
@@ -0,0 +1,8 @@
+unless any_contain("gndIdentifier", "Keine")
+	reject()
+end
+
+replace_all("preferredName", "\\/\\s(ca\\.|um)?-?\\s?\\d.+$", "")
+trim("preferredName")
+
+retain("rppdId", "preferredName")
diff --git a/conf/map-rppd-to-label.flux b/conf/map-rppd-to-label.flux
@@ -0,0 +1,8 @@
+FLUX_DIR + "output/rppd-export.jsonl"
+| open-file
+| as-lines
+| decode-json(recordPath="data")
+| fix(FLUX_DIR + "map-rppd-to-label.fix")
+| encode-csv(includeheader="true", noquotes="true",separator="\t")
+| write(FLUX_DIR + "maps/rppdId-to-label.tsv")
+;