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

package dolda.jsvc.next;

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

public class Parser {
    public class State {
	public final Document doc = DomUtil.document(null, "dummy");
	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 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 {
	Attr a = s.doc.createAttributeNS(null, 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());
	    }
	}
	a.setValue(buf.toString());
	return(a);
    }
    
    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.*;
a5e6bd24 FT	6
a5e6bd24 FT	7	public class Parser {
7c0e72ac	8	public class State {
816cbb00	9	public final Document doc = DomUtil.document(null, "dummy");
7c0e72ac FT	10	public final PeekReader in;
	11
	12	private State(Reader in) {
	13	this.in = new PeekReader(in);
	14	}
	15	}
	16
a5e6bd24 FT	17	private static boolean namechar(char c) {
	18	return((c == ':') \|\| (c == '_') \|\| (c == '$') \|\| (c == '.') \|\| (c == '-') \|\| ((c >= '0') && (c <= '9')) \|\| ((c >= 'A') && (c <= 'Z')) \|\| ((c >= 'a') && (c <= 'z')));
	19	}
	20
	21	protected String entity(String name) {
	22	if(name.equals("amp"))
	23	return("&");
	24	if(name.equals("lt"))
	25	return("<");
	26	if(name.equals("gt"))
	27	return(">");
	28	if(name.equals("apos"))
	29	return("'");
	30	if(name.equals("quot"))
	31	return("\"");
	32	return(null);
	33	}
	34
	35	protected Element makenode(Document doc, String name) {
	36	return(doc.createElementNS(null, name));
	37	}
	38
7c0e72ac	39	protected String name(State s) throws IOException {
a5e6bd24	40	StringBuilder buf = new StringBuilder();
a5e6bd24	41	while(true) {
7c0e72ac FT	42	int c = s.in.peek();
	43	if(c < 0) {
	44	break;
	45	} else if(namechar((char)c)) {
	46	buf.append((char)s.in.read());
	47	} else {
	48	break;
	49	}
	50	}
	51	if(buf.length() == 0)
	52	throw(new ParseException("Expected name, got `" + printable(s.in.peek()) + "'"));
	53	return(buf.toString());
	54	}
	55
	56	protected String entity(State s) throws IOException {
	57	int c = s.in.read();
	58	if(c != '&')
	59	throw(new ParseException("Expected `&' while reading entity, got `" + printable(c) + "'"));
	60	String nm = name(s);
	61	c = s.in.read();
	62	if(c != ';')
	63	throw(new ParseException("Expected `;' while reading entity, got `" + printable(c) + "'"));
	64	return(entity(nm));
	65	}
	66
	67	protected Attr attribute(State s, Element el) throws IOException {
3de0fa23	68	Attr a = s.doc.createAttributeNS(null, name(s));
7c0e72ac FT	69	s.in.peek(true);
	70	int c = s.in.read();
	71	if(c != '=')
	72	throw(new ParseException("Expected `=' while reading attribute, got `" + printable(c) + "'"));
	73	s.in.peek(true);
	74	int qt = s.in.read();
	75	if((qt != '"') && (qt != '\''))
	76	throw(new ParseException("Expected double or single quote while reading attribute, got `" + printable(qt) + "'"));
	77	StringBuilder buf = new StringBuilder();
	78	while(true) {
	79	c = s.in.peek();
	80	if(c < 0) {
	81	throw(new ParseException("Unexpected end-of-file while reading attribute value"));
	82	} else if(c == qt) {
	83	s.in.read();
	84	break;
	85	} else if(c == '&') {
	86	buf.append(entity(s));
	87	} else {
	88	buf.append((char)s.in.read());
	89	}
	90	}
3de0fa23 FT	91	a.setValue(buf.toString());
3de0fa23 FT	92	return(a);
7c0e72ac FT	93	}
	94
	95	protected Element element(State s) throws IOException {
	96	Element n = makenode(s.doc, name(s));
	97	while(true) {
	98	int c = s.in.peek(true);
	99	if(c < 0) {
	100	throw(new ParseException("Unexpected end-of-file while parsing start tag"));
	101	} else if(c == '>') {
	102	s.in.read();
	103	break;
	104	} else if(c == '/') {
	105	s.in.read();
	106	s.in.peek(true);
	107	c = s.in.read();
	108	if(c != '>')
	109	throw(new ParseException("Unexpected character `" + printable(c) + "' encountered in end of empty tag"));
	110	return(n);
	111	} else if(namechar((char)c)) {
	112	n.setAttributeNodeNS(attribute(s, n));
	113	} else {
	114	throw(new ParseException("Unexpected character `" + printable(c) + "' encountered in start tag"));
	115	}
	116	}
	117	while(true) {
	118	int c = s.in.peek();
	119	if(c < 0) {
	120	break;
	121	} else if(c == '<') {
	122	s.in.read();
	123	c = s.in.peek(true);
	124	if(c == '/') {
	125	s.in.read();
	126	s.in.peek(true);
	127	String nm = name(s);
	128	if(!nm.equals(n.getTagName()))
	129	throw(new ParseException("Unexpected end tag for `" + nm + "' while parsing `" + n.getTagName() + "'"));
	130	if(s.in.peek(true) != '>')
	131	throw(new ParseException("Expected `>' while reading end tag, got `" + printable(c) + "'"));
	132	s.in.read();
	133	break;
a5e6bd24	134	} else {
7c0e72ac	135	n.appendChild(stag(s));
a5e6bd24	136	}
7c0e72ac FT	137	} else {
	138	n.appendChild(text(s));
	139	}
	140	}
	141	return(n);
	142	}
	143
	144	protected Comment comment(State s) throws IOException {
	145	if((s.in.read() != '!') \|\|
	146	(s.in.read() != '-') \|\|
	147	(s.in.read() != '-'))
	148	throw(new ParseException("Illegal start of comment"));
	149	StringBuilder buf = new StringBuilder();
	150	while(true) {
	151	int c = s.in.peek();
	152	if(c < 0) {
	153	throw(new ParseException("Unexpected end-of-file while parsing comment"));
	154	} else if(c == '-') {
	155	s.in.read();
	156	if(s.in.peek() == '-') {
	157	s.in.read();
	158	if(s.in.peek() == '>') {
	159	s.in.read();
	160	break;
	161	} else {
	162	buf.append("--");
	163	}
a5e6bd24	164	} else {
7c0e72ac	165	buf.append("-");
a5e6bd24 FT	166	}
a5e6bd24 FT	167	} else {
7c0e72ac	168	buf.append((char)s.in.read());
a5e6bd24 FT	169	}
a5e6bd24 FT	170	}
7c0e72ac	171	return(s.doc.createComment(buf.toString()));
a5e6bd24	172	}
7c0e72ac FT	173
	174	protected Node stag(State s) throws IOException {
	175	int c = s.in.peek(true);
	176	if(c < 0) {
	177	throw(new ParseException("Unexpected end-of-file while parsing tag type"));
	178	} else if(c == '!') {
	179	return(comment(s));
	180	} else {
	181	return(element(s));
	182	}
	183	}
	184
	185	protected Text text(State s) throws IOException {
	186	StringBuilder buf = new StringBuilder();
	187	while(true) {
	188	int c = s.in.peek();
	189	if(c < 0) {
	190	break;
	191	} else if(c == '<') {
	192	break;
	193	} else if(c == '&') {
	194	buf.append(entity(s));
	195	} else {
	196	buf.append((char)s.in.read());
	197	}
	198	}
	199	return(s.doc.createTextNode(buf.toString()));
	200	}
	201
	202	public DocumentFragment parse(Reader in) throws IOException {
	203	State s = new State(in);
	204	DocumentFragment frag = s.doc.createDocumentFragment();
	205	while(true) {
	206	int c = s.in.peek();
	207	if(c < 0) {
	208	return(frag);
	209	} else if(c == '<') {
	210	s.in.read();
	211	frag.appendChild(stag(s));
	212	} else {
	213	frag.appendChild(text(s));
	214	}
	215	}
	216	}
	217
	218	private static String printable(int c) {
	219	if(c < 0)
	220	return("EOF");
a5e6bd24 FT	221	if(c < 32)
a5e6bd24 FT	222	return(String.format("\\%03o", (int)c));
7c0e72ac	223	return(Character.toString((char)c));
a5e6bd24 FT	224	}
	225
	226	public static void main(String[] args) throws Exception {
	227	Parser p = new Parser();
	228	DocumentFragment f = p.parse(new FileReader(args[0]));
	229	javax.xml.transform.TransformerFactory fac = javax.xml.transform.TransformerFactory.newInstance();
	230	fac.setAttribute("indent-number", 2);
	231	javax.xml.transform.Transformer t = fac.newTransformer();
	232	t.setOutputProperty(javax.xml.transform.OutputKeys.INDENT, "yes");
	233	t.transform(new javax.xml.transform.dom.DOMSource(f), new javax.xml.transform.stream.StreamResult(System.out));
	234	System.out.println(t.getClass());
	235	}
	236	}