From 21c84faf603a19a991edc595bab850dde7098487 Mon Sep 17 00:00:00 2001
From: Jens Wille <jens.wille@gmail.com>
Date: Fri, 15 Oct 2021 20:20:38 +0200
Subject: [PATCH 1/2] Make emitting MARCXML namespace configurable.

Fixes #403.
---
 .../biblio/marc21/MarcXmlEncoder.java         | 65 ++++++++++++-------
 .../biblio/marc21/MarcXmlEncoderTest.java     | 11 ++++
 2 files changed, 53 insertions(+), 23 deletions(-)

diff --git a/metafacture-biblio/src/main/java/org/metafacture/biblio/marc21/MarcXmlEncoder.java b/metafacture-biblio/src/main/java/org/metafacture/biblio/marc21/MarcXmlEncoder.java
index 1935569b7..919dac144 100644
--- a/metafacture-biblio/src/main/java/org/metafacture/biblio/marc21/MarcXmlEncoder.java
+++ b/metafacture-biblio/src/main/java/org/metafacture/biblio/marc21/MarcXmlEncoder.java
@@ -40,28 +40,35 @@
 @FluxCommand("encode-marcxml")
 public final class MarcXmlEncoder extends DefaultStreamPipe<ObjectReceiver<String>> {
 
-    private static final String ROOT_OPEN = "<marc:collection xmlns:marc=\"http://www.loc.gov/MARC21/slim\" xmlns:xsi=\"http://www.w3.org/2001/XMLSchema-instance\" xsi:schemaLocation=\"http://www.loc.gov/MARC21/slim http://www.loc.gov/standards/marcxml/schema/MARC21slim.xsd\">";
-    private static final String ROOT_CLOSE = "</marc:collection>";
+    private static final String NAMESPACE = "http://www.loc.gov/MARC21/slim";
+    private static final String NAMESPACE_NAME = "marc";
+    /*package-private*/ static final String NAMESPACE_PREFIX = NAMESPACE_NAME + ":";
+    private static final String NAMESPACE_SUFFIX = ":" + NAMESPACE_NAME;
 
-    private static final String RECORD_OPEN = "<marc:record>";
-    private static final String RECORD_CLOSE = "</marc:record>";
+    private static final String SCHEMA_ATTRIBUTES = " xmlns:xsi=\"http://www.w3.org/2001/XMLSchema-instance\" xsi:schemaLocation=\"" + NAMESPACE + " http://www.loc.gov/standards/marcxml/schema/MARC21slim.xsd\"";
+
+    /*package-private*/ static final String ROOT_OPEN_TEMPLATE = "<%scollection xmlns%s=\"" + NAMESPACE + "\"%s>";
+    private static final String ROOT_CLOSE_TEMPLATE = "</%scollection>";
+
+    private static final String RECORD_OPEN_TEMPLATE = "<%srecord>";
+    private static final String RECORD_CLOSE_TEMPLATE = "</%srecord>";
 
     private static final String ATTRIBUTE_TEMPLATE = " %s=\"%s\"";
 
-    private static final String CONTROLFIELD_OPEN_TEMPLATE = "<marc:controlfield tag=\"%s\">";
-    private static final String CONTROLFIELD_CLOSE = "</marc:controlfield>";
+    private static final String CONTROLFIELD_OPEN_TEMPLATE = "<%scontrolfield tag=\"%s\">";
+    private static final String CONTROLFIELD_CLOSE_TEMPLATE = "</%scontrolfield>";
 
-    private static final String DATAFIELD_OPEN_TEMPLATE = "<marc:datafield tag=\"%s\" ind1=\"%s\" ind2=\"%s\">";
-    private static final String DATAFIELD_CLOSE = "</marc:datafield>";
+    private static final String DATAFIELD_OPEN_TEMPLATE = "<%sdatafield tag=\"%s\" ind1=\"%s\" ind2=\"%s\">";
+    private static final String DATAFIELD_CLOSE_TEMPLATE = "</%sdatafield>";
 
-    private static final String SUBFIELD_OPEN_TEMPLATE = "<marc:subfield code=\"%s\">";
-    private static final String SUBFIELD_CLOSE = "</marc:subfield>";
+    private static final String SUBFIELD_OPEN_TEMPLATE = "<%ssubfield code=\"%s\">";
+    private static final String SUBFIELD_CLOSE_TEMPLATE = "</%ssubfield>";
 
-    private static final String LEADER_OPEN_TEMPLATE = "<marc:leader>";
-    private static final String LEADER_CLOSE_TEMPLATE = "</marc:leader>";
+    private static final String LEADER_TEMPLATE = "<%sleader>%s</%sleader>";
 
     private static final String NEW_LINE = "\n";
     private static final String INDENT = "\t";
+    private static final String EMPTY = "";
 
     private static final String XML_DECLARATION_TEMPLATE = "<?xml version=\"%s\" encoding=\"%s\"?>";
 
@@ -82,6 +89,9 @@ public final class MarcXmlEncoder extends DefaultStreamPipe<ObjectReceiver<Strin
     private String xmlVersion;
     private String xmlEncoding;
 
+    private boolean emitNamespace;
+    private String namespacePrefix;
+
     private String currentEntity;
     private int indentationLevel;
     private int recordAttributeOffset;
@@ -99,6 +109,13 @@ public MarcXmlEncoder() {
 
         this.indentationLevel = 0;
         this.formatted = true;
+
+        setEmitNamespace(true);
+    }
+
+    public void setEmitNamespace(final boolean emitNamespace) {
+        this.emitNamespace = emitNamespace;
+        namespacePrefix = emitNamespace ? NAMESPACE_PREFIX : EMPTY;
     }
 
     public void omitXmlDeclaration(final boolean currentOmitXmlDeclaration) {
@@ -130,14 +147,16 @@ public void startRecord(final String identifier) {
                 writeHeader();
                 prettyPrintNewLine();
             }
-            writeRaw(ROOT_OPEN);
+            writeRaw(String.format(ROOT_OPEN_TEMPLATE, namespacePrefix,
+                        emitNamespace ? NAMESPACE_SUFFIX : EMPTY,
+                        emitNamespace ? SCHEMA_ATTRIBUTES : EMPTY));
             prettyPrintNewLine();
             incrementIndentationLevel();
         }
         atStreamStart = false;
 
         prettyPrintIndentation();
-        writeRaw(RECORD_OPEN);
+        writeRaw(String.format(RECORD_OPEN_TEMPLATE, namespacePrefix));
         recordAttributeOffset = builder.length() - 1;
         prettyPrintNewLine();
 
@@ -148,7 +167,7 @@ public void startRecord(final String identifier) {
     public void endRecord() {
         decrementIndentationLevel();
         prettyPrintIndentation();
-        writeRaw(RECORD_CLOSE);
+        writeRaw(String.format(RECORD_CLOSE_TEMPLATE, namespacePrefix));
         prettyPrintNewLine();
         sendAndClearData();
     }
@@ -167,7 +186,7 @@ public void startEntity(final String name) {
             final String ind1 = name.substring(IND1_BEGIN, IND1_END);
             final String ind2 = name.substring(IND2_BEGIN, IND2_END);
             prettyPrintIndentation();
-            writeRaw(String.format(DATAFIELD_OPEN_TEMPLATE, tag, ind1, ind2));
+            writeRaw(String.format(DATAFIELD_OPEN_TEMPLATE, namespacePrefix, tag, ind1, ind2));
             prettyPrintNewLine();
             incrementIndentationLevel();
         }
@@ -178,7 +197,7 @@ public void endEntity() {
         if (!currentEntity.equals(Marc21EventNames.LEADER_ENTITY)) {
             decrementIndentationLevel();
             prettyPrintIndentation();
-            writeRaw(DATAFIELD_CLOSE);
+            writeRaw(String.format(DATAFIELD_CLOSE_TEMPLATE, namespacePrefix));
             prettyPrintNewLine();
         }
         currentEntity = "";
@@ -194,19 +213,19 @@ public void literal(final String name, final String value) {
             }
             else if (!writeLeader(name, value)) {
                 prettyPrintIndentation();
-                writeRaw(String.format(CONTROLFIELD_OPEN_TEMPLATE, name));
+                writeRaw(String.format(CONTROLFIELD_OPEN_TEMPLATE, namespacePrefix, name));
                 if (value != null) {
                     writeEscaped(value.trim());
                 }
-                writeRaw(CONTROLFIELD_CLOSE);
+                writeRaw(String.format(CONTROLFIELD_CLOSE_TEMPLATE, namespacePrefix));
                 prettyPrintNewLine();
             }
         }
         else if (!writeLeader(currentEntity, value)) {
             prettyPrintIndentation();
-            writeRaw(String.format(SUBFIELD_OPEN_TEMPLATE, name));
+            writeRaw(String.format(SUBFIELD_OPEN_TEMPLATE, namespacePrefix, name));
             writeEscaped(value.trim());
-            writeRaw(SUBFIELD_CLOSE);
+            writeRaw(String.format(SUBFIELD_CLOSE_TEMPLATE, namespacePrefix));
             prettyPrintNewLine();
         }
     }
@@ -243,7 +262,7 @@ private void writeHeader() {
 
     /** Closes the root tag */
     private void writeFooter() {
-        writeRaw(ROOT_CLOSE);
+        writeRaw(String.format(ROOT_CLOSE_TEMPLATE, namespacePrefix));
     }
 
     /** Writes a unescaped sequence */
@@ -259,7 +278,7 @@ private void writeEscaped(final String str) {
     private boolean writeLeader(final String name, final String value) {
         if (name.equals(Marc21EventNames.LEADER_ENTITY)) {
             prettyPrintIndentation();
-            writeRaw(LEADER_OPEN_TEMPLATE + value + LEADER_CLOSE_TEMPLATE);
+            writeRaw(String.format(LEADER_TEMPLATE, namespacePrefix, value, namespacePrefix));
             prettyPrintNewLine();
 
             return true;
diff --git a/metafacture-biblio/src/test/java/org/metafacture/biblio/marc21/MarcXmlEncoderTest.java b/metafacture-biblio/src/test/java/org/metafacture/biblio/marc21/MarcXmlEncoderTest.java
index 478d55ce7..566ec13d1 100644
--- a/metafacture-biblio/src/test/java/org/metafacture/biblio/marc21/MarcXmlEncoderTest.java
+++ b/metafacture-biblio/src/test/java/org/metafacture/biblio/marc21/MarcXmlEncoderTest.java
@@ -168,6 +168,17 @@ public void createTwoRecordsInOneCollection() {
         assertEquals(expected, actual);
     }
 
+    @Test
+    public void issue403_shouldNotEmitNamespaceIfDisabled() {
+        encoder.setEmitNamespace(false);
+        addOneRecord(encoder);
+        addOneRecord(encoder);
+        encoder.closeStream();
+        String expected = XML_DECLARATION + String.format(MarcXmlEncoder.ROOT_OPEN_TEMPLATE, "", "", "") + XML_RECORD + XML_RECORD + XML_MARC_COLLECTION_END_TAG;
+        String actual = resultCollector.toString();
+        assertEquals(expected.replace(MarcXmlEncoder.NAMESPACE_PREFIX, ""), actual);
+    }
+
     @Test(expected = MetafactureException.class)
     public void emitExceptionWhenEntityLengthNot5() {
         encoder.startRecord(RECORD_ID);

From ef9bbb1b76f7687e99b1674c0decad34be6a6db7 Mon Sep 17 00:00:00 2001
From: Jens Wille <jens.wille@gmail.com>
Date: Mon, 18 Oct 2021 17:44:12 +0200
Subject: [PATCH 2/2] Simplify MARCXML tag definition.

- Open and close tag are derived from a single definition.
- Writing call sites are simplified (`writeRaw` -> `writeTag`).
---
 .../biblio/marc21/MarcXmlEncoder.java         | 85 ++++++++++++-------
 .../biblio/marc21/MarcXmlEncoderTest.java     |  5 +-
 2 files changed, 56 insertions(+), 34 deletions(-)

diff --git a/metafacture-biblio/src/main/java/org/metafacture/biblio/marc21/MarcXmlEncoder.java b/metafacture-biblio/src/main/java/org/metafacture/biblio/marc21/MarcXmlEncoder.java
index 919dac144..9a1e6e9f6 100644
--- a/metafacture-biblio/src/main/java/org/metafacture/biblio/marc21/MarcXmlEncoder.java
+++ b/metafacture-biblio/src/main/java/org/metafacture/biblio/marc21/MarcXmlEncoder.java
@@ -25,7 +25,9 @@
 import org.metafacture.framework.annotations.Out;
 import org.metafacture.framework.helpers.DefaultStreamPipe;
 
+import java.util.Arrays;
 import java.util.Collections;
+import java.util.function.Function;
 
 /**
  * Encodes a stream into MARCXML.
@@ -40,31 +42,44 @@
 @FluxCommand("encode-marcxml")
 public final class MarcXmlEncoder extends DefaultStreamPipe<ObjectReceiver<String>> {
 
-    private static final String NAMESPACE = "http://www.loc.gov/MARC21/slim";
-    private static final String NAMESPACE_NAME = "marc";
-    /*package-private*/ static final String NAMESPACE_PREFIX = NAMESPACE_NAME + ":";
-    private static final String NAMESPACE_SUFFIX = ":" + NAMESPACE_NAME;
+    private enum Tag {
 
-    private static final String SCHEMA_ATTRIBUTES = " xmlns:xsi=\"http://www.w3.org/2001/XMLSchema-instance\" xsi:schemaLocation=\"" + NAMESPACE + " http://www.loc.gov/standards/marcxml/schema/MARC21slim.xsd\"";
+        collection(" xmlns%s=\"" + NAMESPACE + "\"%s"),
+        controlfield(" tag=\"%s\""),
+        datafield(" tag=\"%s\" ind1=\"%s\" ind2=\"%s\""),
+        leader(""),
+        record(""),
+        subfield(" code=\"%s\"");
 
-    /*package-private*/ static final String ROOT_OPEN_TEMPLATE = "<%scollection xmlns%s=\"" + NAMESPACE + "\"%s>";
-    private static final String ROOT_CLOSE_TEMPLATE = "</%scollection>";
+        private static final String OPEN_TEMPLATE = "<%%s%s%s>";
+        private static final String CLOSE_TEMPLATE = "</%%s%s>";
 
-    private static final String RECORD_OPEN_TEMPLATE = "<%srecord>";
-    private static final String RECORD_CLOSE_TEMPLATE = "</%srecord>";
+        private final String openTemplate;
+        private final String closeTemplate;
 
-    private static final String ATTRIBUTE_TEMPLATE = " %s=\"%s\"";
+        Tag(final String template) {
+            openTemplate = String.format(OPEN_TEMPLATE, name(), template);
+            closeTemplate = String.format(CLOSE_TEMPLATE, name());
+        }
+
+        public String open(final Object[] args) {
+            return String.format(openTemplate, args);
+        }
+
+        public String close(final Object[] args) {
+            return String.format(closeTemplate, args);
+        }
 
-    private static final String CONTROLFIELD_OPEN_TEMPLATE = "<%scontrolfield tag=\"%s\">";
-    private static final String CONTROLFIELD_CLOSE_TEMPLATE = "</%scontrolfield>";
+    }
 
-    private static final String DATAFIELD_OPEN_TEMPLATE = "<%sdatafield tag=\"%s\" ind1=\"%s\" ind2=\"%s\">";
-    private static final String DATAFIELD_CLOSE_TEMPLATE = "</%sdatafield>";
+    private static final String NAMESPACE = "http://www.loc.gov/MARC21/slim";
+    private static final String NAMESPACE_NAME = "marc";
+    private static final String NAMESPACE_PREFIX = NAMESPACE_NAME + ":";
+    private static final String NAMESPACE_SUFFIX = ":" + NAMESPACE_NAME;
 
-    private static final String SUBFIELD_OPEN_TEMPLATE = "<%ssubfield code=\"%s\">";
-    private static final String SUBFIELD_CLOSE_TEMPLATE = "</%ssubfield>";
+    private static final String SCHEMA_ATTRIBUTES = " xmlns:xsi=\"http://www.w3.org/2001/XMLSchema-instance\" xsi:schemaLocation=\"" + NAMESPACE + " http://www.loc.gov/standards/marcxml/schema/MARC21slim.xsd\"";
 
-    private static final String LEADER_TEMPLATE = "<%sleader>%s</%sleader>";
+    private static final String ATTRIBUTE_TEMPLATE = " %s=\"%s\"";
 
     private static final String NEW_LINE = "\n";
     private static final String INDENT = "\t";
@@ -90,7 +105,7 @@ public final class MarcXmlEncoder extends DefaultStreamPipe<ObjectReceiver<Strin
     private String xmlEncoding;
 
     private boolean emitNamespace;
-    private String namespacePrefix;
+    private Object[] namespacePrefix;
 
     private String currentEntity;
     private int indentationLevel;
@@ -115,7 +130,7 @@ public MarcXmlEncoder() {
 
     public void setEmitNamespace(final boolean emitNamespace) {
         this.emitNamespace = emitNamespace;
-        namespacePrefix = emitNamespace ? NAMESPACE_PREFIX : EMPTY;
+        namespacePrefix = new Object[]{emitNamespace ? NAMESPACE_PREFIX : EMPTY};
     }
 
     public void omitXmlDeclaration(final boolean currentOmitXmlDeclaration) {
@@ -147,16 +162,14 @@ public void startRecord(final String identifier) {
                 writeHeader();
                 prettyPrintNewLine();
             }
-            writeRaw(String.format(ROOT_OPEN_TEMPLATE, namespacePrefix,
-                        emitNamespace ? NAMESPACE_SUFFIX : EMPTY,
-                        emitNamespace ? SCHEMA_ATTRIBUTES : EMPTY));
+            writeTag(Tag.collection::open, emitNamespace ? NAMESPACE_SUFFIX : EMPTY, emitNamespace ? SCHEMA_ATTRIBUTES : EMPTY);
             prettyPrintNewLine();
             incrementIndentationLevel();
         }
         atStreamStart = false;
 
         prettyPrintIndentation();
-        writeRaw(String.format(RECORD_OPEN_TEMPLATE, namespacePrefix));
+        writeTag(Tag.record::open);
         recordAttributeOffset = builder.length() - 1;
         prettyPrintNewLine();
 
@@ -167,7 +180,7 @@ public void startRecord(final String identifier) {
     public void endRecord() {
         decrementIndentationLevel();
         prettyPrintIndentation();
-        writeRaw(String.format(RECORD_CLOSE_TEMPLATE, namespacePrefix));
+        writeTag(Tag.record::close);
         prettyPrintNewLine();
         sendAndClearData();
     }
@@ -186,7 +199,7 @@ public void startEntity(final String name) {
             final String ind1 = name.substring(IND1_BEGIN, IND1_END);
             final String ind2 = name.substring(IND2_BEGIN, IND2_END);
             prettyPrintIndentation();
-            writeRaw(String.format(DATAFIELD_OPEN_TEMPLATE, namespacePrefix, tag, ind1, ind2));
+            writeTag(Tag.datafield::open, tag, ind1, ind2);
             prettyPrintNewLine();
             incrementIndentationLevel();
         }
@@ -197,7 +210,7 @@ public void endEntity() {
         if (!currentEntity.equals(Marc21EventNames.LEADER_ENTITY)) {
             decrementIndentationLevel();
             prettyPrintIndentation();
-            writeRaw(String.format(DATAFIELD_CLOSE_TEMPLATE, namespacePrefix));
+            writeTag(Tag.datafield::close);
             prettyPrintNewLine();
         }
         currentEntity = "";
@@ -213,19 +226,19 @@ public void literal(final String name, final String value) {
             }
             else if (!writeLeader(name, value)) {
                 prettyPrintIndentation();
-                writeRaw(String.format(CONTROLFIELD_OPEN_TEMPLATE, namespacePrefix, name));
+                writeTag(Tag.controlfield::open, name);
                 if (value != null) {
                     writeEscaped(value.trim());
                 }
-                writeRaw(String.format(CONTROLFIELD_CLOSE_TEMPLATE, namespacePrefix));
+                writeTag(Tag.controlfield::close);
                 prettyPrintNewLine();
             }
         }
         else if (!writeLeader(currentEntity, value)) {
             prettyPrintIndentation();
-            writeRaw(String.format(SUBFIELD_OPEN_TEMPLATE, namespacePrefix, name));
+            writeTag(Tag.subfield::open, name);
             writeEscaped(value.trim());
-            writeRaw(String.format(SUBFIELD_CLOSE_TEMPLATE, namespacePrefix));
+            writeTag(Tag.subfield::close);
             prettyPrintNewLine();
         }
     }
@@ -262,7 +275,7 @@ private void writeHeader() {
 
     /** Closes the root tag */
     private void writeFooter() {
-        writeRaw(String.format(ROOT_CLOSE_TEMPLATE, namespacePrefix));
+        writeTag(Tag.collection::close);
     }
 
     /** Writes a unescaped sequence */
@@ -278,7 +291,9 @@ private void writeEscaped(final String str) {
     private boolean writeLeader(final String name, final String value) {
         if (name.equals(Marc21EventNames.LEADER_ENTITY)) {
             prettyPrintIndentation();
-            writeRaw(String.format(LEADER_TEMPLATE, namespacePrefix, value, namespacePrefix));
+            writeTag(Tag.leader::open);
+            writeRaw(value);
+            writeTag(Tag.leader::close);
             prettyPrintNewLine();
 
             return true;
@@ -288,6 +303,12 @@ private boolean writeLeader(final String name, final String value) {
         }
     }
 
+    private void writeTag(final Function<Object[], String> function, final Object... args) {
+        final Object[] allArgs = Arrays.copyOf(namespacePrefix, namespacePrefix.length + args.length);
+        System.arraycopy(args, 0, allArgs, namespacePrefix.length, args.length);
+        writeRaw(function.apply(allArgs));
+    }
+
     private void prettyPrintIndentation() {
         if (formatted) {
             final String prefix = String.join("", Collections.nCopies(indentationLevel, INDENT));
diff --git a/metafacture-biblio/src/test/java/org/metafacture/biblio/marc21/MarcXmlEncoderTest.java b/metafacture-biblio/src/test/java/org/metafacture/biblio/marc21/MarcXmlEncoderTest.java
index 566ec13d1..bc6fb0d49 100644
--- a/metafacture-biblio/src/test/java/org/metafacture/biblio/marc21/MarcXmlEncoderTest.java
+++ b/metafacture-biblio/src/test/java/org/metafacture/biblio/marc21/MarcXmlEncoderTest.java
@@ -174,9 +174,10 @@ public void issue403_shouldNotEmitNamespaceIfDisabled() {
         addOneRecord(encoder);
         addOneRecord(encoder);
         encoder.closeStream();
-        String expected = XML_DECLARATION + String.format(MarcXmlEncoder.ROOT_OPEN_TEMPLATE, "", "", "") + XML_RECORD + XML_RECORD + XML_MARC_COLLECTION_END_TAG;
+        String expected = XML_DECLARATION + "<collection xmlns=\"http://www.loc.gov/MARC21/slim\">"
+            + XML_RECORD + XML_RECORD + XML_MARC_COLLECTION_END_TAG;
         String actual = resultCollector.toString();
-        assertEquals(expected.replace(MarcXmlEncoder.NAMESPACE_PREFIX, ""), actual);
+        assertEquals(expected.replace("marc:", ""), actual);
     }
 
     @Test(expected = MetafactureException.class)