589-fixRunnerExamples

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

@@ -1,12 +1,12 @@
 default basedir = FLUX_DIR;
 
-basedir|
-read-dir|
-filter-strings(".*\\.beacon", passmatches="true")|
-open-file|
-read-beacon(metadatafilter="name|isil")|
-stream-to-triples|
-sort-triples(by="subject")|
-collect-triples|
-encode-formeta(style="multiline")|
-write("stdout");
+basedir
+| read-dir
+| filter-strings(".*\\.beacon", passmatches="true")
+| open-file
+| read-beacon(metadatafilter="name|isil")
+| stream-to-triples
+| sort-triples(by="subject")
+| collect-triples
+| encode-formeta(style="multiline")
+| write("stdout");

metafacture-runner/src/main/dist/examples/beacon/combine/read-beacon.flux

@@ -1,7 +1,7 @@
 default fileName = FLUX_DIR + "austriaforum.beacon";
 
-fileName|
-open-file(encoding="UTF-8")|
-read-beacon(metadatafilter="name|isil")|
-encode-formeta(style="multiline")|
-write("stdout");
+fileName
+| open-file(encoding="UTF-8")
+| read-beacon(metadatafilter="name|isil")
+| encode-formeta(style="multiline")
+| write("stdout");

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

@@ -12,25 +12,25 @@ header|open-file|as-lines|@Y;
 //count references
 "counting references in " + dump | write("stdout");
 
-dump|
-open-file|
-as-lines|
-catch-object-exception|
-decode-pica|
-batch-log(batchsize="100000")|
-morph(FLUX_DIR + "extract.xml", *)|
-stream-to-triples(redirect="true")|
-sort-triples(by="subject")|
-collect-triples|
-morph(FLUX_DIR + "output.xml")|
-batch-log("merged ${totalRecords}", batchsize="100000")|
-stream-to-triples|
-template("${s}")|
-@Y;
-
-@Y|
-wait-for-inputs("2")|
-write(out);
+dump
+| open-file
+| as-lines
+| catch-object-exception
+| decode-pica
+| batch-log(batchsize="100000")
+| fix(FLUX_DIR + "extract.fix", *)
+| stream-to-triples(redirect="true")
+| sort-triples(by="subject")
+| collect-triples
+| fix(FLUX_DIR + "output.fix")
+| batch-log("merged ${totalRecords}", batchsize="100000")
+| stream-to-triples
+| template("${s}")
+| @Y;
+
+@Y
+| wait-for-inputs("2")
+| write(out);
 
 
 

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

@@ -0,0 +1,50 @@
+# 002@ not repeatable
+
+if any_match("002@.0", "^Tp.*$")
+    copy_field("002@.0","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("001@.a","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 $0 “10,P01*”) -->
+
+elsif any_equal("002@.0","^[GM].*")
+        add_field("@value","DMA|ALL")
+
+elsif any_equal("006U.0","^10,P01.*")
+        add_field("@value","DMA|ALL")
+else
+    add_field("@value","ALL")
+end
+
+# Test if type variable fits
+
+if any_contain("@value","$[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

@@ -1,11 +1,11 @@
 default fileName = FLUX_DIR + "gnd-sample.pica";
 
-fileName|
-open-file|
-as-lines|
-decode-pica|
-morph(FLUX_DIR + "gnd-type.xml")|
-stream-to-triples|
-count-triples(countBy="object")|
-template("${s}\t${o}")|
-write("stdout");
+fileName
+| open-file
+| as-lines
+| decode-pica
+| fix(FLUX_DIR + "gnd-type.fix")
+| stream-to-triples
+| count-triples(countBy="object")
+| template("${s}\t${o}")
+| write("stdout");

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

@@ -0,0 +1,6 @@
+if any_match("002@.0","...*")
+  replace_all("002@.0","^(..).*","$1") #only keep the first two letters
+  retain("002@.0") # only keep the relevent element
+else
+  reject()
+end

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

@@ -1,14 +1,13 @@
-
-catalogue|
-open-file|
-as-lines|
-catch-object-exception|
-decode-pica|
-batch-log(batchsize="100000")|
-morph(FLUX_DIR + "subject-cooccurrence.xml")|
-stream-to-triples|
-count-triples(countBy="object")|
-calculate-metrics("X2")|
-template("${s} ${o}")|
-//write("stdout");
+catalogue
+| open-file
+| as-lines
+| catch-object-exception
+| decode-pica
+| batch-log(batchsize="100000")
+| morph(FLUX_DIR + "subject-cooccurrence.xml")
+| stream-to-triples
+| count-triples(countBy="object")
+| calculate-metrics("X2")
+| template("${s} ${o}")
+| //write("stdout");
 write(FLUX_DIR+"x2.dat");

metafacture-runner/src/main/dist/examples/count/subjects/counts.dat

@@ -0,0 +1,9 @@
+040118908:count=1 (STRING)
+040221539:count=1 (STRING)
+040356337:count=1 (STRING)
+040511766:count=1 (STRING)
+040600874:count=1 (STRING)
+040667243:count=1 (STRING)
+040674886:count=1 (STRING)
+042194814:count=2 (STRING)
+942342860:count=1 (STRING)

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

@@ -0,0 +1,8 @@
+do list(path:"041A*|041A","var":"$i") # Until https://github.com/metafacture/metafacture-core/issues/651 is fixed one hass to add "041A"
+  copy_field("$i.9","relevantField.$append")
+end
+
+trim("relevantField.*")
+uniq("relevantField")
+
+retain("relevantField")

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

@@ -1,19 +1,18 @@
 
-default counts="myflux/counts.dat";
+default counts=FLUX_DIR + "counts.dat";
 default catalogue = FLUX_DIR + "10.pica";
 
 //count references
 "counting references in " + catalogue | write("stdout");
 
-catalogue|
-open-file|
-as-lines|
-catch-object-exception|
-decode-pica|
-morph(FLUX_DIR + "references.xml")|
-stream-to-triples|
-count-triples(countBy="object")|
-
-write("subjects.dat");
+catalogue
+| open-file
+| as-lines
+| catch-object-exception
+| decode-pica
+| fix(FLUX_DIR + "references.fix")
+| stream-to-triples
+| count-triples(countBy="object")
+| write(counts);
 
 

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

@@ -0,0 +1,9 @@
+if any_match("002@.0","^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");