eyeling 1.27.9 → 1.28.1

package/examples/alma-rdf-messages.n3

@@ -0,0 +1,202 @@
+# Alma MARC extraction over a remote RDF Message Log.
+#
+# This example is intended to be run with the UGent Alma RDF Message Log URL.
+# The log is larger than 9 GB, so it deliberately remains remote and is not
+# checked into examples/input/.
+#
+# CLI:
+#   eyeling -r --stream-messages examples/alma-rdf-messages.n3 https://ugent-lib-opendata-prd.s3.ugent.be/alma-rdf/rdf-messages.20260404.nt
+#
+# Playground:
+#   https://eyereasoner.github.io/eyeling/playground?state=g.H4sIAAAAAAAC_02N3QqDMAxG36XXtpEpu_BtUhur0EZJWrYx9u7Tjf1chBz44Jy7ITMY05i0v7mUTQeAGomLTYu360YcsKDdJDjt3GtxngBTRithgv1sJlWMpO7Uns5t3_aOy66sf0rBi4tLmauvSjKuXA7RuGagGwmhrkxycFo4gtCkMBMGhYwLA10xb4n0m_0ludtL3gxFKjVGPpDf8HgCVZUvyeAAAAA
+#
+# Each message is one bibliographic record shaped as a nested list of MARC
+# fields: <record> :record (("245" "1" "0" "a" "Title…") ("650" …) …), where an
+# inner list is (tag ind1 ind2 code value code value …). These backward helper
+# rules (marc:field / marc:subf / marc:map / marc:join / marc:id) walk that
+# structure to pull a subfield's value by (tag, subfield-code); the forward
+# rules at the bottom emit :title / :subject / :type per record.
+#
+
+@prefix : <http://example.org/ns#> .
+@prefix list: <http://www.w3.org/2000/10/swap/list#> .
+@prefix log: <http://www.w3.org/2000/10/swap/log#> .
+@prefix eymsg: <https://eyereasoner.github.io/eyeling/vocab/message#> .
+@prefix string: <http://www.w3.org/2000/10/swap/string#> .
+@prefix math: <http://www.w3.org/2000/10/swap/math#> .
+@prefix marc: <https://codeberg.org/phochste/marcattacks#> .
+
+## ---- Helper functions (backward rules) ----
+
+# marc:splice - drop the first ?Idx members of a list (here: the tag + indicators)
+{ (?List ?Idx) marc:splice ?Result }
+<=
+{
+  ( ?X {
+      ?List list:iterate (?Num ?X).
+      ?Num math:notLessThan ?Idx.
+      } ?Result ) log:collectAllIn _:x.
+}.
+
+# marc:join - join a list with a separator
+{ (?List ?Sep) marc:join ?Result }
+<=
+{
+  (?List ?Sep "") marc:join ?Result.
+}.
+
+{ ( () ?Sep ?Acc ) marc:join ?Acc }
+<= true.
+
+{ ( ?List ?Sep ?Acc ) marc:join ?Result }
+<=
+{
+  ?List list:firstRest (?H ?T).
+  ?Acc log:equalTo "".
+  ( ?T ?Sep ?H ) marc:join ?Result .
+}.
+
+{ ( ?List ?Sep ?Acc ) marc:join ?Result }
+<=
+{
+  ?List list:firstRest (?H ?T).
+  ?Acc log:notEqualTo "".
+  ( ?Acc ?Sep ?H) string:concatenation ?AccNew.
+  ( ?T ?Sep ?AccNew ) marc:join ?Result .
+}.
+
+# marc:id - return the record id as an IRI (from the 001 control field)
+{ ?Record marc:id ?Result }
+<=
+{
+  (?Record "001") marc:field0 ?F001.
+  ?F001 marc:ctrl ?ID.
+  ( "http://lib.ugent.be/record/" ?ID ) string:concatenation ?IRI_ID.
+  ?Result log:uri ?IRI_ID.
+}.
+
+# marc:ctrl - return the control value of a field
+{ ?Field marc:ctrl ?Result }
+<=
+{
+  (?Field 3) list:memberAt "_" .
+  (?Field 4) list:memberAt ?Result.
+}.
+
+# marc:subf - return all values matching a subfield regex
+{ ( ?Field ?Regex) marc:subf ?Result }
+<=
+{
+  ( ?Field 3) marc:splice ?FieldData.
+  ( ?FieldData ?Regex ()) marc:subf ?Result.
+} .
+
+{ ( () ?Regex ?Acc ) marc:subf ?Acc }
+<= true.
+
+{ ( ?FieldData ?Regex ?Acc ) marc:subf ?Result }
+<=
+{
+    ?FieldData list:firstRest (?Subf ?Rest).
+    ?Rest list:firstRest (?Value ?Tail).
+    ?Subf string:matches ?Regex.
+    ( ?Acc (?Value)) list:append ?Acc2.
+    ( ?Tail ?Regex ?Acc2 ) marc:subf ?Result.
+}.
+
+{ ( ?FieldData ?Regex ?Acc ) marc:subf ?Result }
+<=
+{
+    ?FieldData list:firstRest (?Subf ?Rest).
+    ?Rest list:firstRest (?Value ?Tail).
+    ?Subf string:notMatches ?Regex.
+    ( ?Tail ?Regex ?Acc ) marc:subf ?Result.
+}.
+
+# marc:field0 - collect the first row of a marc field
+{ ( ?Record ?Field) marc:field0 ?Result }
+<=
+{
+  ( ?Record ?Field) marc:field ?F.
+  ?F list:first ?Result.
+}.
+
+# marc:field - collect all data for a marc field
+{ ( ?Record ?Field) marc:field ?Result}
+<=
+{
+  ( ?Record ?Field ()) marc:field ?Result.
+}.
+
+{ ( () ?Field ?Acc ) marc:field ?Acc}
+<= true.
+
+{ ( ?L ?Field ?Acc ) marc:field ?Result }
+<=
+{
+  ?L list:firstRest (?H ?T).
+  ( ?H 0 ) list:memberAt ?Field.
+  ( ?Acc (?H) ) list:append ?AccNew.
+  (?T ?Field ?AccNew) marc:field ?Result.
+}.
+
+{ (?L ?Field ?Acc) marc:field ?Result }
+<=
+{
+  ?L list:firstRest (?H ?T).
+  ( ?H 0 ) list:memberAt ?X.
+  ?Field log:notEqualTo ?X.
+  (?T ?Field ?Acc) marc:field ?Result.
+}.
+
+# marc:map - join all values of (tag, subfield) into one string
+{ ( ?Record ?Tag ?Subfield ) marc:map ?Result }
+<=
+{
+  (?Record ?Tag) marc:field ?FL.
+  ?FL list:member ?F.
+  (?F ?Subfield) marc:subf ?T .
+  (?T " ") marc:join ?Result.
+}.
+
+## ---- Extraction rules (forward) ----
+
+# In --stream-messages mode the current RDF Message payload is exposed as a
+# quoted graph via eymsg:payloadGraph/log:nameOf. Match :record inside that
+# payload graph, then reuse the ordinary MARC helper rules on the bound list.
+
+{
+  ?Envelope eymsg:payloadGraph ?Payload.
+  ?Payload log:nameOf ?PayloadContext.
+  ?PayloadContext log:includes { ?X :record ?Record. }.
+  ?Record marc:id ?ID.
+  (?Record "245" "a") marc:map ?Val.
+}
+=>
+{
+  ?ID :title ?Val.
+}.
+
+{
+  ?Envelope eymsg:payloadGraph ?Payload.
+  ?Payload log:nameOf ?PayloadContext.
+  ?PayloadContext log:includes { ?X :record ?Record. }.
+  ?Record marc:id ?ID.
+  (?Record "650" "a") marc:map ?Val.
+}
+=>
+{
+  ?ID :subject ?Val.
+}.
+
+{
+  ?Envelope eymsg:payloadGraph ?Payload.
+  ?Payload log:nameOf ?PayloadContext.
+  ?PayloadContext log:includes { ?X :record ?Record. }.
+  ?Record marc:id ?ID.
+  (?Record "920" "a") marc:map ?Val.
+}
+=>
+{
+  ?ID :type ?Val.
+}.