Initial checkin for UcdXML

.gitignore

@@ -43,6 +43,7 @@ perf-*.xml
 test-*.xml
 
 # Directories
+.idea/
 .settings/
 .vs/
 .vscode/

unicodetools/src/main/java/org/unicode/xml/UCDDataResolver.java

@@ -0,0 +1,194 @@
+package org.unicode.xml;
+
+import com.ibm.icu.util.VersionInfo;
+import org.unicode.cldr.draft.FileUtilities;
+import org.unicode.props.IndexUnicodeProperties;
+import org.unicode.props.PropertyParsingInfo;
+import org.unicode.props.UcdLineParser;
+import org.unicode.props.UcdProperty;
+import org.xml.sax.SAXException;
+import org.xml.sax.helpers.AttributesImpl;
+
+import java.util.*;
+
+public class UCDDataResolver {
+
+    private final IndexUnicodeProperties indexUnicodeProperties;
+    private final String namespace;
+    private final UCDXMLWriter writer;
+
+    public UCDDataResolver(IndexUnicodeProperties iup, String namespace, UCDXMLWriter writer) {
+        indexUnicodeProperties = iup;
+        this.namespace = namespace;
+        this.writer = writer;
+    }
+
+    public void buildSection(UcdSectionDetail.UcdSection ucdSection) throws SAXException {
+        VersionInfo minVersion = ucdSection.getMinVersion();
+        VersionInfo maxVersion = ucdSection.getMaxVersion();
+        String tag = ucdSection.toString();
+        String childTag = ucdSection.getChildTag();
+        boolean parserWithRange = ucdSection.getParserWithRange();
+        boolean parserWithMissing = ucdSection.getParserWithMissing();
+        UcdSectionComponent[] ucdSectionComponents = ucdSection.getUcdSectionDetail().getUcdSectionComponents();
+
+        if (isCompatibleVersion(minVersion, maxVersion)) {
+            writer.startElement(tag); {
+                for (UcdSectionComponent ucdSectionComponent : ucdSectionComponents) {
+                    if (isCompatibleVersion(ucdSectionComponent.getMinVersion(), ucdSectionComponent.getMaxVersion())) {
+                        final PropertyParsingInfo fileInfoEVS = PropertyParsingInfo.getPropertyInfo(ucdSectionComponent.getUcdProperty());
+                        String fullFilename = fileInfoEVS.getFullFileName(indexUnicodeProperties.getUcdVersion());
+                        UcdLineParser parser = new UcdLineParser(FileUtilities.in("", fullFilename));
+                        parser.withRange(parserWithRange);
+                        parser.withMissing(parserWithMissing);
+                        switch (ucdSection) {
+                            case BLOCKS:
+                                for (UcdLineParser.UcdLine line : parser) {
+                                    if (!line.getOriginalLine().startsWith("#")) {
+                                        AttributesImpl attributes = getBlockAttributes(namespace, line);
+                                        writer.startElement(childTag, attributes); {
+                                            writer.endElement(childTag);
+                                        }
+                                    }
+                                }
+                                break;
+                            case NAMEDSEQUENCES:
+                                HashMap<String, String> namedSequences = new HashMap<>();
+                                for (UcdLineParser.UcdLine line : parser) {
+                                    String[] parts = line.getParts();
+                                    namedSequences.put(parts[0], parts[1]);
+                                }
+                                List<String> names = new ArrayList<>(namedSequences.keySet());
+                                Collections.sort(names);
+                                for (String name : names) {
+                                    AttributesImpl attributes = getNamedSequenceAttributes(namespace, name, namedSequences);
+                                    writer.startElement(childTag, attributes); {
+                                        writer.endElement(childTag);
+                                    }
+                                }
+                                break;
+                            default:
+                                for (UcdLineParser.UcdLine line : parser) {
+                                    AttributesImpl attributes = getAttributes(ucdSection, namespace, line);
+                                    writer.startElement(childTag, attributes);
+                                    {
+                                        writer.endElement(childTag);
+                                    }
+                                }
+                        }
+                    }
+                }
+                writer.endElement(tag);
+            }
+        }
+    }
+
+    private AttributesImpl getAttributes(UcdSectionDetail.UcdSection ucdSection, String namespace, UcdLineParser.UcdLine line) {
+        switch(ucdSection) {
+            case CJKRADICALS:
+                return getCJKRadicalAttributes(namespace, line);
+            case DONOTEMIT:
+                return getDoNotEmitAttributes(namespace, line);
+            case EMOJISOURCES:
+                return getEmojiSourceAttributes(namespace, line);
+            case NORMALIZATIONCORRECTIONS:
+                return getNCAttributes(namespace, line);
+            case STANDARDIZEDVARIANTS:
+                return getSVAttributes(namespace, line);
+            default:
+                throw new IllegalArgumentException("getAttributes failed on an unexpected UcdSection");
+        }
+    }
+
+    private static AttributesImpl getBlockAttributes(String namespace, UcdLineParser.UcdLine line) {
+        String[] parts = line.getParts();
+        String[] range = parts[0].split("\\.\\.");
+        AttributesImpl attributes = new AttributesImpl();
+        attributes.addAttribute(
+                namespace, "first-cp", "first-cp", "CDATA", range[0]);
+        attributes.addAttribute(
+                namespace, "last-cp", "last-cp", "CDATA", range[1]);
+        attributes.addAttribute(
+                namespace, "name", "name", "CDATA", parts[1]);
+        return attributes;
+    }
+
+    private static AttributesImpl getCJKRadicalAttributes(String namespace, UcdLineParser.UcdLine line) {
+        String[] parts = line.getParts();
+        AttributesImpl attributes = new AttributesImpl();
+        attributes.addAttribute(
+                namespace, "number", "number", "CDATA", parts[0]);
+        attributes.addAttribute(
+                namespace, "radical", "radical", "CDATA", parts[1]);
+        attributes.addAttribute(
+                namespace, "ideograph", "ideograph", "CDATA", parts[2]);
+        return attributes;
+    }
+
+    private static AttributesImpl getDoNotEmitAttributes(String namespace, UcdLineParser.UcdLine line) {
+        String[] parts = line.getParts();
+        AttributesImpl attributes = new AttributesImpl();
+        attributes.addAttribute(
+                namespace, "of", "of", "CDATA", parts[0]);
+        attributes.addAttribute(
+                namespace, "use", "use", "CDATA", parts[1]);
+        attributes.addAttribute(
+                namespace, "because", "because", "CDATA", parts[2]);
+        return attributes;
+    }
+
+    private static AttributesImpl getEmojiSourceAttributes(String namespace, UcdLineParser.UcdLine line) {
+        String[] parts = line.getParts();
+        AttributesImpl attributes = new AttributesImpl();
+        attributes.addAttribute(
+                namespace, "unicode", "unicode", "CDATA", parts[0]);
+        attributes.addAttribute(
+                namespace, "docomo", "docomo", "CDATA", parts[1]);
+        attributes.addAttribute(
+                namespace, "kddi", "kddi", "CDATA", parts[2]);
+        attributes.addAttribute(
+                namespace, "softbank", "softbank", "CDATA", parts[3]);
+        return attributes;
+    }
+
+    private static AttributesImpl getNamedSequenceAttributes(String namespace, String name, HashMap<String, String> namedSequences) {
+        AttributesImpl attributes = new AttributesImpl();
+        attributes.addAttribute(
+                namespace, "name", "name", "CDATA", name);
+        attributes.addAttribute(
+                namespace, "cps", "cps", "CDATA", namedSequences.get(name));
+        return attributes;
+    }
+
+    private static AttributesImpl getNCAttributes(String namespace, UcdLineParser.UcdLine line) {
+        String[] parts = line.getParts();
+        AttributesImpl attributes = new AttributesImpl();
+        attributes.addAttribute(
+                namespace, "cp", "cp", "CDATA", parts[0]);
+        attributes.addAttribute(
+                namespace, "old", "old", "CDATA", parts[1]);
+        attributes.addAttribute(
+                namespace, "new", "new", "CDATA", parts[2]);
+        attributes.addAttribute(
+                namespace, "version", "version", "CDATA", parts[3]);
+        return attributes;
+    }
+
+    private static AttributesImpl getSVAttributes(String namespace, UcdLineParser.UcdLine line) {
+        String[] parts = line.getParts();
+        AttributesImpl attributes = new AttributesImpl();
+        attributes.addAttribute(
+                namespace, "cps", "cps", "CDATA", parts[0]);
+        attributes.addAttribute(
+                namespace, "desc", "desc", "CDATA", parts[1]);
+        attributes.addAttribute(
+                namespace, "when", "when", "CDATA",
+                parts[2] != null ? parts[2] : "");
+        return attributes;
+    }
+
+    private boolean isCompatibleVersion(VersionInfo minVersion, VersionInfo maxVersion) {
+        return (indexUnicodeProperties.getUcdVersion().compareTo(minVersion) >= 0 && (
+                maxVersion == null || indexUnicodeProperties.getUcdVersion().compareTo(maxVersion) <= 0));
+    }
+}

unicodetools/src/main/java/org/unicode/xml/UCDXMLWriter.java

@@ -0,0 +1,78 @@
+package org.unicode.xml;
+
+import org.xml.sax.Attributes;
+import org.xml.sax.SAXException;
+import org.xml.sax.helpers.AttributesImpl;
+
+import javax.xml.transform.OutputKeys;
+import javax.xml.transform.Transformer;
+import javax.xml.transform.TransformerConfigurationException;
+import javax.xml.transform.TransformerFactory;
+import javax.xml.transform.sax.SAXTransformerFactory;
+import javax.xml.transform.sax.TransformerHandler;
+import javax.xml.transform.stream.StreamResult;
+import java.io.FileOutputStream;
+
+public class UCDXMLWriter {
+
+    public static final String NAMESPACE
+            = "http://www.unicode.org/ns/2003/ucd/1.0";
+
+    private final TransformerHandler transformerHandler;
+
+    public TransformerHandler getTransformerHandler() {
+        return transformerHandler;
+    }
+
+    public UCDXMLWriter(FileOutputStream f) throws TransformerConfigurationException {
+        TransformerFactory tfactory = TransformerFactory.newInstance();
+        SAXTransformerFactory sfactory = (SAXTransformerFactory) tfactory;
+        transformerHandler = sfactory.newTransformerHandler ();
+        Transformer transformer = transformerHandler.getTransformer ();
+        transformer.setOutputProperty(OutputKeys.ENCODING, "utf-8");
+        transformer.setOutputProperty(OutputKeys.METHOD, "xml");
+        transformer.setOutputProperty(OutputKeys.INDENT, "yes");
+        transformer.setOutputProperty(OutputKeys.STANDALONE, "yes");
+        transformer.setOutputProperty("{http://xml.apache.org/xslt}indent-amount", "3");
+        transformer.setOutputProperty("{http://xml.apache.org/xalan}indent-amount", "3");
+        transformerHandler.setResult (new StreamResult(f));
+    }
+
+    public void startFile() throws SAXException {
+        transformerHandler.startDocument ();
+        char[] c = "\n".toCharArray ();
+        transformerHandler.characters (c, 0, c.length);
+        c = " \u00A9 2023 Unicode\u00AE, Inc. ".toCharArray ();
+        transformerHandler.comment (c, 0, c.length);
+        c = "\n".toCharArray ();
+        transformerHandler.characters (c, 0, c.length);
+        c = " For terms of use, see http://www.unicode.org/terms_of_use.html ".toCharArray ();
+        transformerHandler.comment (c, 0, c.length);
+        c = "\n\n\n".toCharArray ();
+        transformerHandler.characters (c, 0, c.length);
+
+    }
+    public void endFile() throws SAXException {
+        transformerHandler.endDocument ();
+    }
+
+    public void startElement(String tagName) throws SAXException {
+        AttributesImpl attributes = new AttributesImpl ();
+        startElement(tagName, attributes);
+    }
+
+    public void startElement(String tagName, AttributesImpl attributes) throws SAXException {
+        transformerHandler.startElement (NAMESPACE, tagName, tagName, attributes);
+    }
+
+    public void addContent(String s) throws SAXException {
+        char[] d = s.toCharArray ();
+        transformerHandler.characters (d, 0, d.length);
+    }
+
+    public void endElement(String tagName) throws SAXException {
+        transformerHandler.endElement (NAMESPACE, tagName, tagName);
+    }
+}
+
+