Add incomplete xml parser

Doesn't adhere to any standards. Instead, it parses what I thought was
valid xml in a haphazard way. Might still be an interesting thing to
look at, to see how one would use the parser library.

src/main/java/de/plugh/compositeparse/parsers/QuotedString.java

@@ -0,0 +1,53 @@
+package de.plugh.compositeparse.parsers;
+
+import de.plugh.compositeparse.Block;
+import de.plugh.compositeparse.ParseException;
+import de.plugh.compositeparse.Parser;
+
+import java.util.List;
+import java.util.function.Function;
+
+public class QuotedString implements Parser<String> {
+
+    private final String quoteChar;
+
+    public QuotedString(String quoteChar) {
+        this.quoteChar = quoteChar;
+    }
+
+    public QuotedString() {
+        this("\"");
+    }
+
+    @Override
+    public Function<List<Block>, String> getNamingScheme() {
+        return Block.label("quoted string");
+    }
+
+    @Override
+    public String read(Block block) throws ParseException {
+        Literal.literally(quoteChar).parse(block);
+
+        StringBuilder result = new StringBuilder();
+        boolean escaped = false;
+        while (true) {
+            String s = block.getInput().read(1);
+
+            if (s.isEmpty()) {
+                throw new ParseException(block);
+            } else if (escaped) {
+                result.append(s);
+                escaped = false;
+            } else if (s.equals(quoteChar)) {
+                break;
+            } else if (s.equals("\\")) {
+                escaped = true;
+            } else {
+                result.append(s);
+            }
+        }
+
+        return result.toString();
+    }
+
+}

src/main/java/de/plugh/compositeparse/xml/AttributesParser.java

@@ -0,0 +1,49 @@
+package de.plugh.compositeparse.xml;
+
+import de.plugh.compositeparse.Block;
+import de.plugh.compositeparse.ParseException;
+import de.plugh.compositeparse.Parser;
+import de.plugh.compositeparse.parsers.Expression;
+import de.plugh.compositeparse.parsers.Literal;
+import de.plugh.compositeparse.parsers.QuotedString;
+import de.plugh.compositeparse.parsers.Repeat;
+
+import java.util.HashMap;
+import java.util.List;
+import java.util.Map;
+
+public class AttributesParser implements Parser<Map<String, String>> {
+
+    private static final String REGEX_NAME = "[:a-zA-Z_][:a-zA-Z0-9_.-]*";
+
+    @Override
+    public Map<String, String> read(Block block) throws ParseException {
+        List<Attribute> attributes = new Repeat<>(block1 -> {
+            Repeat.atLeast(1, Literal.literally(" ")).parse(block1);
+            
+            String name = new Expression(REGEX_NAME).parse(block1);
+            Literal.literally("=").parse(block1);
+            String value = new QuotedString().parse(block1);
+
+            return new Attribute(name, value);
+        }).parse(block);
+
+        Map<String, String> attributeMap = new HashMap<>();
+        attributes.forEach(attribute -> attributeMap.put(attribute.name, attribute.value));
+
+        return attributeMap;
+    }
+
+    private class Attribute {
+
+        final String name;
+        final String value;
+
+        Attribute(String name, String value) {
+            this.name = name;
+            this.value = value;
+        }
+
+    }
+
+}

src/main/java/de/plugh/compositeparse/xml/CommentNode.java

@@ -0,0 +1,27 @@
+package de.plugh.compositeparse.xml;
+
+public class CommentNode extends Node {
+
+    private final String content;
+
+    public CommentNode(String content) {
+        this.content = content;
+    }
+
+    @Override
+    public String prettyPrint(String indent, boolean newline) {
+        if (newline) {
+            return indent + "<!--" + content + "-->" + "\n";
+        } else {
+            return indent + "<!--" + content + "-->";
+        }
+    }
+
+    @Override
+    public String toString() {
+        return "CommentNode{" +
+                "content='" + content + '\'' +
+                '}';
+    }
+
+}

src/main/java/de/plugh/compositeparse/xml/CommentNodeParser.java

@@ -0,0 +1,20 @@
+package de.plugh.compositeparse.xml;
+
+import de.plugh.compositeparse.Block;
+import de.plugh.compositeparse.ParseException;
+import de.plugh.compositeparse.Parser;
+import de.plugh.compositeparse.parsers.Expression;
+import de.plugh.compositeparse.parsers.Literal;
+
+public class CommentNodeParser implements Parser<CommentNode> {
+
+    @Override
+    public CommentNode read(Block block) throws ParseException {
+        Literal.literally("<!--").parse(block);
+        String content = new Expression("^.*(?=--)").parse(block);
+        Literal.literally("-->").parse(block);
+
+        return new CommentNode(content);
+    }
+
+}

src/main/java/de/plugh/compositeparse/xml/ElementNode.java

@@ -0,0 +1,76 @@
+package de.plugh.compositeparse.xml;
+
+import java.util.List;
+import java.util.Map;
+
+public class ElementNode extends Node {
+
+    private final String name;
+    private final Map<String, String> attributes;
+    private final List<Node> subnodes;
+
+    public ElementNode(String name, Map<String, String> attributes,
+                       List<Node> subnodes) {
+        this.name = name;
+        this.attributes = attributes;
+        this.subnodes = subnodes;
+    }
+
+    @Override
+    public String prettyPrint(String indent, boolean newline) {
+        StringBuilder result = new StringBuilder()
+                .append(indent)
+                .append("<")
+                .append(name);
+
+        attributes.forEach((s, s2) -> {
+            result
+                    .append(" ")
+                    .append(s)
+                    .append("=\"")
+                    .append(s2) // TODO escape
+                    .append("\"");
+        });
+
+        if (subnodes.isEmpty()) {
+            result.append("/>");
+        } else {
+            result.append(">");
+
+            if (subnodes.size() == 1 && subnodes.get(0) instanceof TextNode) {
+                result.append(subnodes.get(0).prettyPrint("", false));
+
+                result
+                        .append("</")
+                        .append(name)
+                        .append(">");
+            } else {
+                result.append("\n");
+
+                subnodes.forEach(node -> result.append(node.prettyPrint(indent + "  ", true)));
+
+                result
+                        .append(indent)
+                        .append("</")
+                        .append(name)
+                        .append(">");
+            }
+        }
+
+        if (newline) {
+            result.append("\n");
+        }
+
+        return result.toString();
+    }
+
+    @Override
+    public String toString() {
+        return "ElementNode{" +
+                "name='" + name + '\'' +
+                ", attributes=" + attributes +
+                ", subnodes=" + subnodes +
+                '}';
+    }
+
+}

src/main/java/de/plugh/compositeparse/xml/ElementNodeParser.java

@@ -0,0 +1,51 @@
+package de.plugh.compositeparse.xml;
+
+import de.plugh.compositeparse.Block;
+import de.plugh.compositeparse.Pair;
+import de.plugh.compositeparse.ParseException;
+import de.plugh.compositeparse.Parser;
+import de.plugh.compositeparse.parsers.Decision;
+import de.plugh.compositeparse.parsers.Expression;
+import de.plugh.compositeparse.parsers.Literal;
+import de.plugh.compositeparse.parsers.Repeat;
+
+import java.util.ArrayList;
+import java.util.List;
+import java.util.Map;
+
+public class ElementNodeParser implements Parser<ElementNode> {
+
+    private static final String REGEX_NAME = "[:a-zA-Z_][:a-zA-Z0-9_.-]*";
+
+    @Override
+    public ElementNode read(Block block) throws ParseException {
+        Literal.literally("<").parse(block);
+        String name = new Expression(REGEX_NAME).parse(block);
+
+        Map<String, String> attributes = new AttributesParser().parse(block);
+
+        List<Node> subnodes = new Decision<List<Node>>(
+                // First, try to find a close tag
+                new Pair<>(
+                        block1 -> {
+                            new Repeat<>(Literal.literally(" ")).parse(block1);
+                            Literal.literally("/>").parse(block1);
+                            return null;
+                        },
+                        block1 -> new ArrayList<>()
+                ),
+                // If that fails, actually parse the contents and the rest
+                new Pair<>(
+                        Literal.literally(">"),
+                        block1 -> {
+                            List<Node> foundSubnodes = new Repeat<>(new NodeParser()).parse(block);
+                            Literal.literally("</" + name + ">").parse(block1);
+                            return foundSubnodes;
+                        }
+                )
+        ).parse(block);
+
+        return new ElementNode(name, attributes, subnodes);
+    }
+
+}

src/main/java/de/plugh/compositeparse/xml/Node.java

@@ -0,0 +1,11 @@
+package de.plugh.compositeparse.xml;
+
+public abstract class Node {
+
+    public abstract String prettyPrint(String indent, boolean newline);
+
+    public String prettyPrint() {
+        return prettyPrint("", false);
+    }
+
+}

src/main/java/de/plugh/compositeparse/xml/NodeParser.java

@@ -0,0 +1,19 @@
+package de.plugh.compositeparse.xml;
+
+import de.plugh.compositeparse.Block;
+import de.plugh.compositeparse.ParseException;
+import de.plugh.compositeparse.Parser;
+import de.plugh.compositeparse.parsers.Options;
+
+public class NodeParser implements Parser<Node> {
+
+    @Override
+    public Node read(Block block) throws ParseException {
+        return new Options<>(
+                block1 -> new CommentNodeParser().parse(block1),
+                block1 -> new ElementNodeParser().parse(block1),
+                block1 -> new TextNodeParser().parse(block1)
+        ).parse(block);
+    }
+
+}

src/main/java/de/plugh/compositeparse/xml/TextNode.java

@@ -0,0 +1,27 @@
+package de.plugh.compositeparse.xml;
+
+public class TextNode extends Node {
+
+    private final String text;
+
+    public TextNode(String text) {
+        this.text = text;
+    }
+
+    @Override
+    public String prettyPrint(String indent, boolean newline) {
+        if (newline) {
+            return indent + text.trim() + "\n";
+        } else {
+            return indent + text.trim();
+        }
+    }
+
+    @Override
+    public String toString() {
+        return "TextNode{" +
+                "text='" + text + '\'' +
+                '}';
+    }
+
+}

src/main/java/de/plugh/compositeparse/xml/TextNodeParser.java

@@ -0,0 +1,17 @@
+package de.plugh.compositeparse.xml;
+
+import de.plugh.compositeparse.Block;
+import de.plugh.compositeparse.ParseException;
+import de.plugh.compositeparse.Parser;
+import de.plugh.compositeparse.parsers.Expression;
+
+public class TextNodeParser implements Parser<TextNode> {
+
+    @Override
+    public TextNode read(Block block) throws ParseException {
+        String text = new Expression("^[^<]+").parse(block);
+
+        return new TextNode(text);
+    }
+
+}