[jsvc.git] / src / dolda / jsvc / next / Parser.java

package dolda.jsvc.next;

import java.io.*;
import java.util.*;
import org.w3c.dom.*;
import org.w3c.dom.bootstrap.*;

public class Parser {
    private static final DOMImplementation domimp;
    
    static {
	DOMImplementationRegistry reg;
	try {
	    reg = DOMImplementationRegistry.newInstance();
	} catch(Exception e) {
	    throw(new Error(e));
	}
	DOMImplementation di = reg.getDOMImplementation("");
	if(di == null)
	    throw(new RuntimeException("Could not get a DOM implemenation"));
	domimp = di;
    }

    public class State {
	public final Document doc = domimp.createDocument(null, "dummy", null);
	public final PeekReader in;
	
	private State(Reader in) {
	    this.in = new PeekReader(in);
	}
    }

    private static boolean namechar(char c) {
	return((c == ':') || (c == '_') || (c == '$') || (c == '.') || (c == '-') || ((c >= '0') && (c <= '9')) || ((c >= 'A') && (c <= 'Z')) || ((c >= 'a') && (c <= 'z')));
    }

    protected String entity(String name) {
	if(name.equals("amp"))
	    return("&");
	if(name.equals("lt"))
	    return("<");
	if(name.equals("gt"))
	    return(">");
	if(name.equals("apos"))
	    return("'");
	if(name.equals("quot"))
	    return("\"");
	return(null);
    }
    
    protected Element makenode(Document doc, String name) {
	return(doc.createElementNS(null, name));
    }
    
    protected Attr makeattr(Document doc, Element el, String name, String val) {
	Attr a = doc.createAttributeNS(el.getNamespaceURI(), name);
	a.setValue(val);
	return(a);
    }

    protected Attr makeattr(Document doc, Element el, String name) {
	return(doc.createAttributeNS(el.getNamespaceURI(), name));
    }

    protected String name(State s) throws IOException {
	StringBuilder buf = new StringBuilder();
	while(true) {
	    int c = s.in.peek();
	    if(c < 0) {
		break;
	    } else if(namechar((char)c)) {
		buf.append((char)s.in.read());
	    } else {
		break;
	    }
	}
	if(buf.length() == 0)
	    throw(new ParseException("Expected name, got `" + printable(s.in.peek()) + "'"));
	return(buf.toString());
    }
    
    protected String entity(State s) throws IOException {
	int c = s.in.read();
	if(c != '&')
	    throw(new ParseException("Expected `&' while reading entity, got `" + printable(c) + "'"));
	String nm = name(s);
	c = s.in.read();
	if(c != ';')
	    throw(new ParseException("Expected `;' while reading entity, got `" + printable(c) + "'"));
	return(entity(nm));
    }

    protected Attr attribute(State s, Element el) throws IOException {
	String nm = name(s);
	s.in.peek(true);
	int c = s.in.read();
	if(c != '=')
	    throw(new ParseException("Expected `=' while reading attribute, got `" + printable(c) + "'"));
	s.in.peek(true);
	int qt = s.in.read();
	if((qt != '"') && (qt != '\''))
	    throw(new ParseException("Expected double or single quote while reading attribute, got `" + printable(qt) + "'"));
	StringBuilder buf = new StringBuilder();
	while(true) {
	    c = s.in.peek();
	    if(c < 0) {
		throw(new ParseException("Unexpected end-of-file while reading attribute value"));
	    } else if(c == qt) {
		s.in.read();
		break;
	    } else if(c == '&') {
		buf.append(entity(s));
	    } else {
		buf.append((char)s.in.read());
	    }
	}
	return(makeattr(s.doc, el, nm, buf.toString()));
    }
    
    protected Element element(State s) throws IOException {
	Element n = makenode(s.doc, name(s));
	while(true) {
	    int c = s.in.peek(true);
	    if(c < 0) {
		throw(new ParseException("Unexpected end-of-file while parsing start tag"));
	    } else if(c == '>') {
		s.in.read();
		break;
	    } else if(c == '/') {
		s.in.read();
		s.in.peek(true);
		c = s.in.read();
		if(c != '>')
		    throw(new ParseException("Unexpected character `" + printable(c) + "' encountered in end of empty tag"));
		return(n);
	    } else if(namechar((char)c)) {
		n.setAttributeNodeNS(attribute(s, n));
	    } else {
		throw(new ParseException("Unexpected character `" + printable(c) + "' encountered in start tag"));
	    }
	}
	while(true) {
	    int c = s.in.peek();
	    if(c < 0) {
		break;
	    } else if(c == '<') {
		s.in.read();
		c = s.in.peek(true);
		if(c == '/') {
		    s.in.read();
		    s.in.peek(true);
		    String nm = name(s);
		    if(!nm.equals(n.getTagName()))
			throw(new ParseException("Unexpected end tag for `" + nm + "' while parsing `" + n.getTagName() + "'"));
		    if(s.in.peek(true) != '>')
			throw(new ParseException("Expected `>' while reading end tag, got `" + printable(c) + "'"));
		    s.in.read();
		    break;
		} else {
		    n.appendChild(stag(s));
		}
	    } else {
		n.appendChild(text(s));
	    }
	}
	return(n);
    }
    
    protected Comment comment(State s) throws IOException {
	if((s.in.read() != '!') ||
	   (s.in.read() != '-') ||
	   (s.in.read() != '-'))
	    throw(new ParseException("Illegal start of comment"));
	StringBuilder buf = new StringBuilder();
	while(true) {
	    int c = s.in.peek();
	    if(c < 0) {
		throw(new ParseException("Unexpected end-of-file while parsing comment"));
	    } else if(c == '-') {
		s.in.read();
		if(s.in.peek() == '-') {
		    s.in.read();
		    if(s.in.peek() == '>') {
			s.in.read();
			break;
		    } else {
			buf.append("--");
		    }
		} else {
		    buf.append("-");
		}
	    } else {
		buf.append((char)s.in.read());
	    }
	}
	return(s.doc.createComment(buf.toString()));
    }

    protected Node stag(State s) throws IOException {
	int c = s.in.peek(true);
	if(c < 0) {
	    throw(new ParseException("Unexpected end-of-file while parsing tag type"));
	} else if(c == '!') {
	    return(comment(s));
	} else {
	    return(element(s));
	}
    }

    protected Text text(State s) throws IOException {
	StringBuilder buf = new StringBuilder();
	while(true) {
	    int c = s.in.peek();
	    if(c < 0) {
		break;
	    } else if(c == '<') {
		break;
	    } else if(c == '&') {
		buf.append(entity(s));
	    } else {
		buf.append((char)s.in.read());
	    }
	}
	return(s.doc.createTextNode(buf.toString()));
    }

    public DocumentFragment parse(Reader in) throws IOException {
	State s = new State(in);
	DocumentFragment frag = s.doc.createDocumentFragment();
	while(true) {
	    int c = s.in.peek();
	    if(c < 0) {
		return(frag);
	    } else if(c == '<') {
		s.in.read();
		frag.appendChild(stag(s));
	    } else {
		frag.appendChild(text(s));
	    }
	}
    }

    private static String printable(int c) {
	if(c < 0)
	    return("EOF");
	if(c < 32)
	    return(String.format("\\%03o", (int)c));
	return(Character.toString((char)c));
    }

    public static void main(String[] args) throws Exception {
	Parser p = new Parser();
	DocumentFragment f = p.parse(new FileReader(args[0]));
	javax.xml.transform.TransformerFactory fac = javax.xml.transform.TransformerFactory.newInstance();
	fac.setAttribute("indent-number", 2);
	javax.xml.transform.Transformer t = fac.newTransformer();
	t.setOutputProperty(javax.xml.transform.OutputKeys.INDENT, "yes");
	t.transform(new javax.xml.transform.dom.DOMSource(f), new javax.xml.transform.stream.StreamResult(System.out));
	System.out.println(t.getClass());
    }
}
Commit	Line	Data
a5e6bd24 FT	1	package dolda.jsvc.next;
	2
	3	import java.io.*;
	4	import java.util.*;
	5	import org.w3c.dom.*;
	6	import org.w3c.dom.bootstrap.*;
	7
	8	public class Parser {
	9	private static final DOMImplementation domimp;
	10
	11	static {
	12	DOMImplementationRegistry reg;
	13	try {
	14	reg = DOMImplementationRegistry.newInstance();
	15	} catch(Exception e) {
	16	throw(new Error(e));
	17	}
	18	DOMImplementation di = reg.getDOMImplementation("");
	19	if(di == null)
	20	throw(new RuntimeException("Could not get a DOM implemenation"));
	21	domimp = di;
	22	}
	23
7c0e72ac FT	24	public class State {
	25	public final Document doc = domimp.createDocument(null, "dummy", null);
	26	public final PeekReader in;
	27
	28	private State(Reader in) {
	29	this.in = new PeekReader(in);
	30	}
	31	}
	32
a5e6bd24 FT	33	private static boolean namechar(char c) {
	34	return((c == ':') \|\| (c == '_') \|\| (c == '$') \|\| (c == '.') \|\| (c == '-') \|\| ((c >= '0') && (c <= '9')) \|\| ((c >= 'A') && (c <= 'Z')) \|\| ((c >= 'a') && (c <= 'z')));
	35	}
	36
	37	protected String entity(String name) {
	38	if(name.equals("amp"))
	39	return("&");
	40	if(name.equals("lt"))
	41	return("<");
	42	if(name.equals("gt"))
	43	return(">");
	44	if(name.equals("apos"))
	45	return("'");
	46	if(name.equals("quot"))
	47	return("\"");
	48	return(null);
	49	}
	50
	51	protected Element makenode(Document doc, String name) {
	52	return(doc.createElementNS(null, name));
	53	}
	54
7c0e72ac FT	55	protected Attr makeattr(Document doc, Element el, String name, String val) {
	56	Attr a = doc.createAttributeNS(el.getNamespaceURI(), name);
	57	a.setValue(val);
	58	return(a);
	59	}
	60
a5e6bd24 FT	61	protected Attr makeattr(Document doc, Element el, String name) {
	62	return(doc.createAttributeNS(el.getNamespaceURI(), name));
	63	}
	64
7c0e72ac	65	protected String name(State s) throws IOException {
a5e6bd24	66	StringBuilder buf = new StringBuilder();
a5e6bd24	67	while(true) {
7c0e72ac FT	68	int c = s.in.peek();
	69	if(c < 0) {
	70	break;
	71	} else if(namechar((char)c)) {
	72	buf.append((char)s.in.read());
	73	} else {
	74	break;
	75	}
	76	}
	77	if(buf.length() == 0)
	78	throw(new ParseException("Expected name, got `" + printable(s.in.peek()) + "'"));
	79	return(buf.toString());
	80	}
	81
	82	protected String entity(State s) throws IOException {
	83	int c = s.in.read();
	84	if(c != '&')
	85	throw(new ParseException("Expected `&' while reading entity, got `" + printable(c) + "'"));
	86	String nm = name(s);
	87	c = s.in.read();
	88	if(c != ';')
	89	throw(new ParseException("Expected `;' while reading entity, got `" + printable(c) + "'"));
	90	return(entity(nm));
	91	}
	92
	93	protected Attr attribute(State s, Element el) throws IOException {
	94	String nm = name(s);
	95	s.in.peek(true);
	96	int c = s.in.read();
	97	if(c != '=')
	98	throw(new ParseException("Expected `=' while reading attribute, got `" + printable(c) + "'"));
	99	s.in.peek(true);
	100	int qt = s.in.read();
	101	if((qt != '"') && (qt != '\''))
	102	throw(new ParseException("Expected double or single quote while reading attribute, got `" + printable(qt) + "'"));
	103	StringBuilder buf = new StringBuilder();
	104	while(true) {
	105	c = s.in.peek();
	106	if(c < 0) {
	107	throw(new ParseException("Unexpected end-of-file while reading attribute value"));
	108	} else if(c == qt) {
	109	s.in.read();
	110	break;
	111	} else if(c == '&') {
	112	buf.append(entity(s));
	113	} else {
	114	buf.append((char)s.in.read());
	115	}
	116	}
	117	return(makeattr(s.doc, el, nm, buf.toString()));
	118	}
	119
	120	protected Element element(State s) throws IOException {
	121	Element n = makenode(s.doc, name(s));
	122	while(true) {
	123	int c = s.in.peek(true);
	124	if(c < 0) {
	125	throw(new ParseException("Unexpected end-of-file while parsing start tag"));
	126	} else if(c == '>') {
	127	s.in.read();
	128	break;
	129	} else if(c == '/') {
	130	s.in.read();
	131	s.in.peek(true);
132	c = s.in.read();
133	if(c != '>')
134	throw(new ParseException("Unexpected character `" + printable(c) + "' encountered in end of empty tag"));
135	return(n);
136	} else if(namechar((char)c)) {
137	n.setAttributeNodeNS(attribute(s, n));
138	} else {
139	throw(new ParseException("Unexpected character `" + printable(c) + "' encountered in start tag"));
140	}
141	}
142	while(true) {
143	int c = s.in.peek();
144	if(c < 0) {
145	break;
146	} else if(c == '<') {
147	s.in.read();
148	c = s.in.peek(true);
149	if(c == '/') {
150	s.in.read();
151	s.in.peek(true);
152	String nm = name(s);
153	if(!nm.equals(n.getTagName()))
154	throw(new ParseException("Unexpected end tag for `" + nm + "' while parsing `" + n.getTagName() + "'"));
155	if(s.in.peek(true) != '>')
156	throw(new ParseException("Expected `>' while reading end tag, got `" + printable(c) + "'"));
157	s.in.read();
158	break;
a5e6bd24	159	} else {
7c0e72ac	160	n.appendChild(stag(s));
a5e6bd24	161	}
7c0e72ac FT	162	} else {
	163	n.appendChild(text(s));
	164	}
	165	}
	166	return(n);
	167	}
	168
	169	protected Comment comment(State s) throws IOException {
	170	if((s.in.read() != '!') \|\|
	171	(s.in.read() != '-') \|\|
	172	(s.in.read() != '-'))
	173	throw(new ParseException("Illegal start of comment"));
	174	StringBuilder buf = new StringBuilder();
	175	while(true) {
	176	int c = s.in.peek();
	177	if(c < 0) {
	178	throw(new ParseException("Unexpected end-of-file while parsing comment"));
	179	} else if(c == '-') {
	180	s.in.read();
	181	if(s.in.peek() == '-') {
	182	s.in.read();
	183	if(s.in.peek() == '>') {
	184	s.in.read();
	185	break;
	186	} else {
	187	buf.append("--");
	188	}
a5e6bd24	189	} else {
7c0e72ac	190	buf.append("-");
a5e6bd24 FT	191	}
a5e6bd24 FT	192	} else {
7c0e72ac	193	buf.append((char)s.in.read());
a5e6bd24 FT	194	}
a5e6bd24 FT	195	}
7c0e72ac	196	return(s.doc.createComment(buf.toString()));
a5e6bd24	197	}
7c0e72ac FT	198
	199	protected Node stag(State s) throws IOException {
	200	int c = s.in.peek(true);
	201	if(c < 0) {
	202	throw(new ParseException("Unexpected end-of-file while parsing tag type"));
	203	} else if(c == '!') {
	204	return(comment(s));
	205	} else {
	206	return(element(s));
	207	}
	208	}
	209
	210	protected Text text(State s) throws IOException {
	211	StringBuilder buf = new StringBuilder();
	212	while(true) {
	213	int c = s.in.peek();
	214	if(c < 0) {
	215	break;
	216	} else if(c == '<') {
	217	break;
	218	} else if(c == '&') {
	219	buf.append(entity(s));
	220	} else {
	221	buf.append((char)s.in.read());
	222	}
	223	}
	224	return(s.doc.createTextNode(buf.toString()));
	225	}
	226
	227	public DocumentFragment parse(Reader in) throws IOException {
	228	State s = new State(in);
	229	DocumentFragment frag = s.doc.createDocumentFragment();
	230	while(true) {
	231	int c = s.in.peek();
	232	if(c < 0) {
	233	return(frag);
	234	} else if(c == '<') {
	235	s.in.read();
	236	frag.appendChild(stag(s));
	237	} else {
	238	frag.appendChild(text(s));
	239	}
	240	}
	241	}
	242
	243	private static String printable(int c) {
	244	if(c < 0)
	245	return("EOF");
a5e6bd24 FT	246	if(c < 32)
a5e6bd24 FT	247	return(String.format("\\%03o", (int)c));
7c0e72ac	248	return(Character.toString((char)c));
a5e6bd24 FT	249	}
	250
	251	public static void main(String[] args) throws Exception {
	252	Parser p = new Parser();
	253	DocumentFragment f = p.parse(new FileReader(args[0]));
	254	javax.xml.transform.TransformerFactory fac = javax.xml.transform.TransformerFactory.newInstance();
	255	fac.setAttribute("indent-number", 2);
	256	javax.xml.transform.Transformer t = fac.newTransformer();
	257	t.setOutputProperty(javax.xml.transform.OutputKeys.INDENT, "yes");
	258	t.transform(new javax.xml.transform.dom.DOMSource(f), new javax.xml.transform.stream.StreamResult(System.out));
	259	System.out.println(t.getClass());
	260	}
	261	}