589-fixRunnerExamples

metafacture-linkeddata/src/main/java/org/metafacture/linkeddata/RdfMacroPipe.java

@@ -29,7 +29,7 @@
  * @author Markus Michael Geipel
  *
  */
-@Description("Expands some macros for RDF/XML")
+@Description("Expands some macros for RDF/XML. When using Fix, configure `referenceMarker` to any char but the default `*`")
 @In(StreamReceiver.class)
 @Out(StreamReceiver.class)
 @FluxCommand("rdf-macros")
@@ -41,6 +41,8 @@ public final class RdfMacroPipe extends DefaultStreamPipe<StreamReceiver> {
     public static final String RDF_ABOUT = "~rdf:about";
     public static final String XML_LANG = "~xml:lang";
     private String autoAddedSubject = "";
+    private char referenceMarker = REFERENCE_MARKER;
+    private char languageMarker = LANGUAGE_MARKER;
 
     /**
      * Creates an instance of {@link RdfMacroPipe}.
@@ -57,6 +59,42 @@ public void setAutoAddedSubject(final String autoAddedSubject) {
         this.autoAddedSubject = autoAddedSubject;
     }
 
+    /**
+     * Sets the single char reference marker.
+     *
+     * @param referenceMarker the reference marker
+     */
+    public void setReferenceMarker(final String referenceMarker) {
+        this.referenceMarker = referenceMarker.charAt(0);
+    }
+
+    /**
+     * Gets the reference marker.
+     *
+     * @return the reference marker
+     */
+    public char getReferenceMarker() {
+        return referenceMarker;
+    }
+
+    /**
+     * Sets the single char language marker.
+     *
+     * @param languageMarker the language marker
+     */
+    public void setLanguageMarker(final String languageMarker) {
+        this.languageMarker = languageMarker.charAt(0);
+    }
+
+    /**
+     * Gets the language marker.
+     *
+     * @return the language marker
+     */
+    public char getLanguageMarker() {
+        return languageMarker;
+    }
+
     @Override
     public void startRecord(final String identifier) {
         getReceiver().startRecord(identifier);
@@ -87,8 +125,8 @@ public void endEntity() {
 
     @Override
     public void literal(final String name, final String value) {
-        final int index = name.indexOf(LANGUAGE_MARKER);
-        if (!name.isEmpty() && name.charAt(0) == REFERENCE_MARKER) {
+        final int index = name.indexOf(languageMarker);
+        if (!name.isEmpty() && name.charAt(0) == referenceMarker) {
             getReceiver().startEntity(name.substring(1));
             getReceiver().literal(RDF_REFERENCE, value);
             getReceiver().endEntity();

metafacture-runner/src/main/dist/examples/beacon/create/create.flux

@@ -18,11 +18,11 @@ as-lines|
 catch-object-exception|
 decode-pica|
 batch-log(batchsize="100000")|
-morph(FLUX_DIR + "extract.xml", *)|
+fix(FLUX_DIR + "extract.fix", *)|
 stream-to-triples(redirect="true")|
 sort-triples(by="subject")|
 collect-triples|
-morph(FLUX_DIR + "output.xml")|
+fix(FLUX_DIR + "output.fix")|
 batch-log("merged ${totalRecords}", batchsize="100000")|
 stream-to-triples|
 template("${s}")|

metafacture-runner/src/main/dist/examples/beacon/create/extract.fix

@@ -0,0 +1,50 @@
+# 002@ not repeatable
+
+if any_match("[email protected]", "^Tp.*$")
+    copy_field("[email protected]","ok")
+end
+
+#                <!-- DBSM: -->
+#                <!-- (006U $0 “04p01*”) or (017A $a “yy”) -->
+if any_match("006U.0","04p01.*")
+        add_field("@value","DBSM|ALL")
+elsif any_equal("017A.a","yy")
+        add_field("@value","DBSM|ALL")
+
+#                 <!-- DEA: -->
+#                 <!-- (001@ $a 2””) or (209A $f “Exilarchiv” or 209A $f “HB/EB”) -->
+elsif any_equal("[email protected]","2")
+        add_field("@value","DEA|ALL")
+elsif any_equal("209A.f","HB/EB")
+        add_field("@value","DEA|ALL")
+
+#                 <!-- DMA: -->
+#                 <!-- (002@ $0 “G*” or 002@ $0 “M*”) or (006U $0 “10,P01*”) -->
+
+elsif any_equal("[email protected]","^[GM].*")
+        add_field("@value","DMA|ALL")
+
+elsif any_equal("006U.0","^10,P01.*")
+        add_field("@value","DMA|ALL")
+else
+    add_field("@value","ALL")
+end
+
+# Test if type variable fits
+
+if any_contain("@value","$[type]")
+    add_field("@value","$[type]")
+else
+    remove_field("@value")
+end
+
+do list(path: "041A*|028A*|029B*|028C*|028Q*|028P*|028F*|028M*|028D*|028E*", "var":"$i")
+    trim("$i.9")
+    to_var("$i.9","ref")
+    if exists("$i.9")
+        copy_field("@value","{to:$[ref]}refed")
+    end
+end
+
+retain("{to*","ok")
+

metafacture-runner/src/main/dist/examples/beacon/create/output.fix

@@ -0,0 +1,8 @@
+unless exists("refed")
+    remove_field("ok","")
+end
+
+unless exists("ok")
+  remove_field("ok","")
+end
+

metafacture-runner/src/main/dist/examples/count/gnd/count-gnd-types.flux

@@ -4,7 +4,7 @@ fileName|
 open-file|
 as-lines|
 decode-pica|
-morph(FLUX_DIR + "gnd-type.xml")|
+fix(FLUX_DIR + "gnd-type.fix")|
 stream-to-triples|
 count-triples(countBy="object")|
 template("${s}\t${o}")|

metafacture-runner/src/main/dist/examples/count/gnd/gnd-type.fix

@@ -0,0 +1,6 @@
+if any_match("[email protected]","...*")
+  replace_all("[email protected]","^(..).*","$1") #only keep the first two letters
+  retain("[email protected]") # only keep the relevent element
+else
+  reject()
+end

metafacture-runner/src/main/dist/examples/count/subjects/references.fix

@@ -0,0 +1,8 @@
+do list(path:"041A*","var":"$i")
+  copy_field("$i.9","relevantField.$append")
+end
+
+trim("relevantField.*")
+uniq("relevantField")
+
+retain("relevantField")

metafacture-runner/src/main/dist/examples/count/subjects/references.flux

@@ -1,5 +1,5 @@
 
-default counts="myflux/counts.dat";
+default counts=FLUX_DIR + "counts.dat";
 default catalogue = FLUX_DIR + "10.pica";
 
 //count references
@@ -10,10 +10,9 @@ open-file|
 as-lines|
 catch-object-exception|
 decode-pica|
-morph(FLUX_DIR + "references.xml")|
+fix(FLUX_DIR + "references.fix")|
 stream-to-triples|
 count-triples(countBy="object")|
-
-write("subjects.dat");
+write(counts);
 
 

metafacture-runner/src/main/dist/examples/filter/fix/filter.fix

@@ -0,0 +1,9 @@
+if any_match("[email protected]","^Tp.*")
+  if any_match("041R.a",".*[Aa][Rr][Zz][Tt].*")
+    nothing()
+  else
+    reject()
+  end
+else
+  reject()
+end

metafacture-runner/src/main/dist/examples/filter/fix/filter.flux

@@ -0,0 +1,11 @@
+// opens file 'fileName', interprets the content as pica and filters the results
+
+default fileName = FLUX_DIR + "gnd-sample.pica";
+
+fileName|
+open-file|
+as-lines|
+decode-pica|
+fix(FLUX_DIR + "filter.fix")| // Fix does not use the filter function but has its own filter mechanism within fix.
+encode-formeta(style="verbose")|
+write("stdout");