Merge branch 'rpb-243-variantName' of https://github.com/hbz/rpb into…

… main See #121
hbz · Jan 21, 2025 · 647a7cb · 647a7cb
2 parents 08f8792 + 14ee5d2
commit 647a7cb
Show file tree

Hide file tree

Showing 15 changed files with 192,454 additions and 192,394 deletions.
diff --git a/conf/RPB-Export_HBZ_SW.tsv b/conf/RPB-Export_HBZ_SW.tsv
diff --git a/conf/output/test-output-32.json b/conf/output/test-output-32.json
@@ -57,6 +57,7 @@
     "componentList" : [ {
       "id" : "http://rpb.lobid.org/sw/107n220353",
       "label" : "Wilhelm-von-Humboldt-Gymnasium / Ludwigshafen am Rhein / Medien-AG",
+      "altLabel" : [ "WHG-Medien-AG" ],
       "source" : {
         "id" : "http://rpb.lobid.org/sw",
         "label" : "RPB-Sachsystematik"

diff --git a/conf/output/test-output-52.json b/conf/output/test-output-52.json
@@ -30,6 +30,7 @@
     "componentList" : [ {
       "id" : "http://rpb.lobid.org/sw/929n110300",
       "label" : "Volleyballclub Neuwied 77 / Neuwied",
+      "altLabel" : [ "VC Neuwied 77 / Neuwied" ],
       "source" : {
         "id" : "http://rpb.lobid.org/sw",
         "label" : "RPB-Sachsystematik"

diff --git a/conf/output/test-output-54.json b/conf/output/test-output-54.json
@@ -61,6 +61,7 @@
     "agent" : {
       "id" : "http://rpb.lobid.org/sw/00Pn08k0685884a",
       "label" : "Ruland, Wolfgang",
+      "altLabel" : [ "WR" ],
       "type" : [ "Person" ]
     },
     "role" : {

diff --git a/conf/output/test-output-55.json b/conf/output/test-output-55.json
@@ -36,6 +36,7 @@
     "componentList" : [ {
       "id" : "http://rpb.lobid.org/sw/929n090287",
       "label" : "Deutscher Orden / Kommende Waldbreitbach",
+      "altLabel" : [ "Kommende Waldbreitbach", "Deutschordenskommende Waldbreitbach", "Deutschherren-Ordenshaus Waldbreitbach", "Commende Waldbreitbach", "Deutsches Huis zu Brettbach" ],
       "source" : {
         "id" : "http://rpb.lobid.org/sw",
         "label" : "RPB-Sachsystematik"
@@ -83,6 +84,7 @@
     "agent" : {
       "id" : "http://rpb.lobid.org/sw/00Pn08k0685884a",
       "label" : "Ruland, Wolfgang",
+      "altLabel" : [ "WR" ],
       "type" : [ "Person" ]
     },
     "role" : {

diff --git a/conf/output/test-output-56.json b/conf/output/test-output-56.json
@@ -54,6 +54,7 @@
     "agent" : {
       "id" : "http://rpb.lobid.org/sw/929n991002",
       "label" : "Weger, Joachim",
+      "altLabel" : [ "jw" ],
       "type" : [ "Person" ]
     },
     "role" : {

diff --git a/conf/output/test-output-6.json b/conf/output/test-output-6.json
@@ -36,13 +36,15 @@
     "componentList" : [ {
       "id" : "http://rpb.lobid.org/sw/929n110302",
       "label" : "Unkeler Sonnenberg",
+      "altLabel" : [ "Unkeler Sonneberg", "Sonnenberg / Unkel", "Sonneberg / Unkel", "Unkel / Sonneberg" ],
       "source" : {
         "id" : "http://rpb.lobid.org/sw",
         "label" : "RPB-Sachsystematik"
       }
     }, {
       "id" : "http://rpb.lobid.org/sw/929n110303",
       "label" : "Weingut Krupp",
+      "altLabel" : [ "Krupp, Oliver / Firma" ],
       "source" : {
         "id" : "http://rpb.lobid.org/sw",
         "label" : "RPB-Sachsystematik"
@@ -70,6 +72,7 @@
     "agent" : {
       "id" : "http://rpb.lobid.org/sw/00Pn08k0685884a",
       "label" : "Ruland, Wolfgang",
+      "altLabel" : [ "WR" ],
       "type" : [ "Person" ]
     },
     "role" : {

diff --git a/conf/output/test-output-72.json b/conf/output/test-output-72.json
@@ -67,6 +67,7 @@
     "agent" : {
       "id" : "http://rpb.lobid.org/sw/929n981242",
       "label" : "Strauß, Torsten",
+      "altLabel" : [ "TS", "T.S." ],
       "type" : [ "Person" ]
     },
     "role" : {

diff --git a/conf/output/test-output-81.json b/conf/output/test-output-81.json
@@ -40,6 +40,7 @@
     }, {
       "id" : "http://rpb.lobid.org/sw/z48",
       "label" : "Geschichte 1800-",
+      "altLabel" : [ "Neuzeit 1800-Gegenwart" ],
       "source" : {
         "id" : "http://rpb.lobid.org/sw",
         "label" : "RPB-Sachsystematik"
@@ -58,6 +59,7 @@
     }, {
       "id" : "http://rpb.lobid.org/sw/z48",
       "label" : "Geschichte 1800-",
+      "altLabel" : [ "Neuzeit 1800-Gegenwart" ],
       "source" : {
         "id" : "http://rpb.lobid.org/sw",
         "label" : "RPB-Sachsystematik"

diff --git a/conf/output/test-output-88.json b/conf/output/test-output-88.json
@@ -28,6 +28,7 @@
     "componentList" : [ {
       "id" : "http://rpb.lobid.org/sw/00Sn01s167281807a",
       "label" : "Verein für Pfälzische Kirchengeschichte / Hauptversammlung / 1925",
+      "altLabel" : [ "Hauptversammlung / 1925 / Verein für Pfälzische Kirchengeschichte" ],
       "source" : {
         "id" : "http://rpb.lobid.org/sw",
         "label" : "RPB-Sachsystematik"

diff --git a/conf/output/test-output-rppd-lobid-19.json b/conf/output/test-output-rppd-lobid-19.json
@@ -14,7 +14,8 @@
   "variantName" : [ "Altgelt, Johann Philipp" ],
   "professionOrOccupation" : [ {
     "label" : "Pfarrer / Katholische Kirche",
-    "id" : "https://rpb.lobid.org/sw/929n170632"
+    "id" : "https://rpb.lobid.org/sw/929n170632",
+    "altLabel" : [ "Katholische Kirche / Pfarrer" ]
   }, {
     "label" : "Lehrer",
     "id" : "https://d-nb.info/gnd/4035088-5"

diff --git a/conf/output/test-output-rppd-lobid-23.json b/conf/output/test-output-rppd-lobid-23.json
@@ -13,7 +13,8 @@
   "preferredName" : "Ziegler, Jakob Anton",
   "professionOrOccupation" : [ {
     "label" : "Pfarrer / Katholische Kirche",
-    "id" : "https://rpb.lobid.org/sw/929n170632"
+    "id" : "https://rpb.lobid.org/sw/929n170632",
+    "altLabel" : [ "Katholische Kirche / Pfarrer" ]
   } ],
   "dateOfBirth" : [ "1893-06-15" ],
   "placeOfBirth" : [ {

diff --git a/conf/rpb-sw.flux b/conf/rpb-sw.flux
@@ -3,7 +3,10 @@ FLUX_DIR + "output/output-strapi-sw.ndjson"
 | as-lines
 | decode-json
 | fix("
-paste('row', 'data.rpbId', 'data.preferredName', join_char : '\t')
+set_array('variantNames')
+copy_field('data.variantName[].*.value', 'variantNames.$append')
+join_field('variantNames', '; ')
+paste('row', 'data.rpbId', 'data.preferredName', 'variantNames', join_char : '\t')
 retain(row)
 ")
 | stream-to-triples

diff --git a/conf/rpb-titel-to-lobid.fix b/conf/rpb-titel-to-lobid.fix
@@ -7,7 +7,8 @@ do once("map")
   put_rdfmap("https://raw.githubusercontent.com/hbz/lbz-vocabs/main/rpb-spatial.ttl", "spatial_map", target:"skos:prefLabel", select_language:"de")
   put_rdfmap("https://raw.githubusercontent.com/hbz/lbz-vocabs/main/rpb.ttl", "subject_map", target:"skos:prefLabel", select_language:"de")
 
-  put_filemap("./RPB-Export_HBZ_SW.tsv", "sw-rpb_map", "sep_char": "\t")
+  put_filemap("./RPB-Export_HBZ_SW.tsv", "sw-rpb_map", sep_char: "\t", allow_empty_values: "true", expected_columns:"3", key_column:"0", value_column:"1")
+  put_filemap("./RPB-Export_HBZ_SW.tsv", "sw-rpb-variants_map", sep_char: "\t", allow_empty_values: "true", expected_columns:"3", key_column:"0", value_column:"2")
   put_filemap("./RPB-Export_HBZ_SW_GND.tsv", "sw-gnd_map", "sep_char": "\t")
 end
 
@@ -16,6 +17,24 @@ do put_macro("move_here")
   move_field("_temp", "$[field]")
 end
 
+do put_macro("normalize")
+  replace_all("$[field]","^http.+[/#](.+)$","$1")
+  replace_all("$[field]","^(\\d+)-(\\d+)$","$1n$2")
+end
+
+do put_macro("add_labels")
+  unless exists("$[field].label")
+    copy_field("$[field].id", "$[field].label")
+    call_macro("normalize", field: "$[field].label")
+    lookup("$[field].label", "sw-rpb_map")
+    lookup("$[field].label", "sw-gnd_map")
+  end
+  copy_field("$[field].id", "$[field].altLabel[]")
+  call_macro("normalize", field: "$[field].altLabel[]")
+  lookup("$[field].altLabel[]", "sw-rpb-variants_map", delete: "true")
+  split_field("$[field].altLabel[]", "; ")
+end
+
 if all_equal("doNotIndex", "true")
   reject()
 end
@@ -243,13 +262,7 @@ end
 do
  list (path:"temp_subject.*.componentList[]", "var": "$i")
   move_field("$i.value", "$i.id")
-  unless exists("$i.label")
-    copy_field("$i.id", "$i.label")
-    replace_all("$i.label","^http.+[/#](.+)$","$1")
-    replace_all("$i.label","^(\\d+)-(\\d+)$","$1n$2")
-    lookup("$i.label", "sw-rpb_map")
-    lookup("$i.label", "sw-gnd_map")
-  end
+  call_macro("add_labels", field: "$i")
   if all_match("$i.id","^http://rpb.lobid.org/sw/.*$")
     add_field("$i.source.id", "http://rpb.lobid.org/sw")
     add_field("$i.source.label", "RPB-Sachsystematik")
@@ -327,13 +340,7 @@ do put_macro("contributions")
   copy_field("$[from][].*.value", "_temp.$append.agent.id")
   do list_as(f: "$[from][]", c: "_temp")
     copy_field("f.label", "c.agent.label")
-    unless exists("c.agent.label")
-      copy_field("c.agent.id", "c.agent.label")
-      replace_all("c.agent.label","^http.+[/#](.+)$","$1")
-      replace_all("c.agent.label","^(\\d+)-(\\d+)$","$1n$2")
-      lookup("c.agent.label", "sw-rpb_map")
-      lookup("c.agent.label", "sw-gnd_map")
-    end
+    call_macro("add_labels", field: "c.agent")
     set_array("c.agent.type[]", "Person")
     copy_field("f.type", "c.role.id")
     copy_field("f.type", "c.role.label")

diff --git a/conf/rppd-to-lobid.fix b/conf/rppd-to-lobid.fix
@@ -21,7 +21,8 @@ do put_macro("move_here")
 end
 
 do once("map")
-  put_filemap("conf/RPB-Export_HBZ_SW.tsv", "SW_map", sep_char: "\t")
+  put_filemap("./RPB-Export_HBZ_SW.tsv", "SW_map", sep_char: "\t", allow_empty_values: "true", expected_columns:"3", key_column:"0", value_column:"1")
+  put_filemap("./RPB-Export_HBZ_SW.tsv", "SW_variants-map", sep_char: "\t", allow_empty_values: "true", expected_columns:"3", key_column:"0", value_column:"2")
   put_filemap("conf/RPB-Export_HBZ_SW_GND.tsv", "SW_GND_map", sep_char: "\t")
   put_filemap("conf/RPB-Export_HBZ_SWN.tsv", "SWN_map", sep_char: "\t")
   put_rdfmap("https://raw.githubusercontent.com/hbz/lbz-vocabs/main/rpb-spatial.ttl", "spatial_map", target:"skos:prefLabel", select_language:"de")
@@ -171,8 +172,11 @@ end
 do list(path:"_temp", "var": "$i")
   replace_all("$i.label", "^http.+/(\\d.+)", "$1")
   replace_all("$i.label", "^(\\d*)-(\\d)", "$1n$2")
+  copy_field("$i.label", "$i.altLabel[]")
   lookup("$i.label", "SW_map", delete: "false")
   lookup("$i.label", "SW_GND_map", delete: "false")
+  lookup("$i.altLabel[]", "SW_variants-map", delete: "true")
+  split_field("$i.altLabel[]", "; ")
 end
 
 move_field("_temp", "professionOrOccupation[]")