websphinx
Class HTMLParser

java.lang.Object
  |
  +--websphinx.HTMLParser

public class HTMLParser

extends java.lang.Object

HTML parser. Parses an input stream or String and converts it to a sequence of Tags and a tree of Elements. HTMLParser is used by Page to parse pages.

Field Summary

private static int	AFTEREQ
private static int	ATTR
(package private) java.lang.StringBuffer	attrName
private static int	ATTRNAME
(package private) java.util.Vector	attrs
(package private) java.lang.StringBuffer	attrVal
private static int	ATTRVAL
private static int	ATTRVAL_DQ
private static int	ATTRVAL_SQ
private static int	BANG
private static int	BANG_DASH
(package private) static java.util.Hashtable	blocktag
(package private) char[]	buf
(package private) static int	BUFFER_SIZE
private static int	CMT
private static int	CMT_DASH
private static int	CMT_DASHDASH
(package private) java.lang.StringBuffer	contentBuf
(package private) static java.util.Hashtable	context
(package private) Form	currentForm
private static int	DIRECTIVE
private static int	DONE
(package private) java.util.Stack	elems
private static java.util.Hashtable	empty
private static java.util.Hashtable	entities
(package private) java.lang.StringBuffer	entity
private static int	ENTITY
private static int	ENTNUM
private static int	ENTREF
private static int	EQ
private static int	ETAG
private static java.util.Hashtable	forcesClosed
(package private) static java.util.Hashtable	headtag
private static int	INWORD
(package private) static java.util.Hashtable	linktag
(package private) static java.util.Hashtable	literal
private static int	LT
(package private) int	maxBytes
(package private) int[]	openElems
(package private) int	openPtr
(package private) static java.util.Hashtable	savetext
private static int	STAG
private static int	START
(package private) java.lang.StringBuffer	tagName
(package private) java.lang.StringBuffer	text
(package private) java.util.Vector	vElements
(package private) java.util.Vector	vLinks
(package private) java.lang.StringBuffer	wordBuf

Constructor Summary

HTMLParser()
Make an HTMLParser.

HTMLParser(DownloadParameters dp)
Make an HTMLParser which retrieves pages using the specified download parameters.

Method Summary

(package private) static void	()
(package private) static void	()
private void	buildParseTree(Page page)
private void	close(Element elem, int end)
private void	close(Element elem, Tag tag)
private void	closeAll(int end)
void	dontParse(Page page, java.io.InputStream stream) Download an input stream without parsing it.
void	dontParse(Page page, java.io.Reader stream) Download an input stream without parsing it.
private Element	findOpenElement(java.util.Hashtable tags)
private Element	findOpenElement(java.lang.String tagname)
private static java.lang.String	indentation(int indent)
private static boolean	isWhitespace(char c)
private static java.lang.Character	lookupEntityRef(java.lang.String name)
static void	main(java.lang.String[] args)
private Element	makeElement(java.net.URL base, Tag tag)
private void	open(Element e)
void	parse(Page page, java.io.InputStream stream) Parse an input stream.
void	parse(Page page, java.io.Reader stream) Parse an input stream.
void	parse(Page page, java.lang.String content) Parse a string.
private static void	printout(Element element, int indent)
private static void	printout(Link[] elements, int indent)
private void	tokenize(Page page, java.io.Reader stream, boolean saveContent)
private static void	union(java.util.Hashtable map, java.lang.Object tagname, java.util.Hashtable tagset)

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, registerNatives, toString, wait, wait, wait

Field Detail

BUFFER_SIZE

static final int BUFFER_SIZE

maxBytes

int maxBytes

START

private static final int START

INWORD

private static final int INWORD

ENTITY

private static final int ENTITY

LT

private static final int LT

BANG

private static final int BANG

BANG_DASH

private static final int BANG_DASH

CMT

private static final int CMT

CMT_DASH

private static final int CMT_DASH

CMT_DASHDASH

private static final int CMT_DASHDASH

DIRECTIVE

private static final int DIRECTIVE

STAG

private static final int STAG

ETAG

private static final int ETAG

ATTR

private static final int ATTR

ATTRNAME

private static final int ATTRNAME

EQ

private static final int EQ

AFTEREQ

private static final int AFTEREQ

ATTRVAL

private static final int ATTRVAL

ATTRVAL_SQ

private static final int ATTRVAL_SQ

ATTRVAL_DQ

private static final int ATTRVAL_DQ

DONE

private static final int DONE

ENTNUM

private static final int ENTNUM

ENTREF

private static final int ENTREF

buf

char[] buf

contentBuf

java.lang.StringBuffer contentBuf

wordBuf

java.lang.StringBuffer wordBuf

tagName

java.lang.StringBuffer tagName

attrName

java.lang.StringBuffer attrName

attrVal

java.lang.StringBuffer attrVal

attrs

java.util.Vector attrs

entity

java.lang.StringBuffer entity

entities

private static java.util.Hashtable entities

vElements

java.util.Vector vElements

vLinks

java.util.Vector vLinks

text

java.lang.StringBuffer text

empty

private static java.util.Hashtable empty

blocktag

static java.util.Hashtable blocktag

forcesClosed

private static java.util.Hashtable forcesClosed

context

static java.util.Hashtable context

literal

static java.util.Hashtable literal

linktag

static java.util.Hashtable linktag

savetext

static java.util.Hashtable savetext

headtag

static java.util.Hashtable headtag

elems

java.util.Stack elems

openElems

int[] openElems

openPtr

int openPtr

currentForm

Form currentForm

Constructor Detail

HTMLParser

public HTMLParser()

Make an HTMLParser.

HTMLParser

public HTMLParser(DownloadParameters dp)

Make an HTMLParser which retrieves pages using the specified download parameters. Pages larger than dp.getMaxPageSize() are rejected by parse() with an IOException.

Parameters:

dp - download parameters used during parsing

Method Detail

parse

public void parse(Page page,
                  java.io.InputStream stream)
           throws java.io.IOException

Parse an input stream.

Parameters:

page - Page to receive parsed HTML

input - stream containing HTML

parse

public void parse(Page page,
                  java.io.Reader stream)
           throws java.io.IOException

Parse an input stream.

Parameters:

page - Page to receive parsed HTML

input - stream containing HTML

parse

public void parse(Page page,
                  java.lang.String content)
           throws java.io.IOException

Parse a string.

Parameters:

page - Page to receive parsed HTML

content - String containing HTML

dontParse

public void dontParse(Page page,
                      java.io.InputStream stream)
               throws java.io.IOException

Download an input stream without parsing it.

Parameters:

page - Page to receive the downloaded content

input - stream containing content

dontParse

public void dontParse(Page page,
                      java.io.Reader stream)
               throws java.io.IOException

Download an input stream without parsing it.

Parameters:

page - Page to receive the downloaded content

r - stream containing content

tokenize

private void tokenize(Page page,
                      java.io.Reader stream,
                      boolean saveContent)
               throws java.io.IOException

isWhitespace

private static boolean isWhitespace(char c)

lookupEntityRef

private static java.lang.Character lookupEntityRef(java.lang.String name)

static void ()

static void ()

union

private static void union(java.util.Hashtable map,
                          java.lang.Object tagname,
                          java.util.Hashtable tagset)

buildParseTree

private void buildParseTree(Page page)

makeElement

private Element makeElement(java.net.URL base,
                            Tag tag)

open

private void open(Element e)

findOpenElement

private Element findOpenElement(java.lang.String tagname)

findOpenElement

private Element findOpenElement(java.util.Hashtable tags)

close

private void close(Element elem,
                   Tag tag)

close

private void close(Element elem,
                   int end)

closeAll

private void closeAll(int end)

main

public static void main(java.lang.String[] args)
                 throws java.lang.Exception

indentation

private static java.lang.String indentation(int indent)

printout

private static void printout(Element element,
                             int indent)

printout

private static void printout(Link[] elements,
                             int indent)