📄 parser.java
字号:
/* * @(#)Parser.java 1.47 06/02/26 * * Copyright 2006 Sun Microsystems, Inc. All rights reserved. * SUN PROPRIETARY/CONFIDENTIAL. Use is subject to license terms. */package javax.swing.text.html.parser;import javax.swing.text.SimpleAttributeSet;import javax.swing.text.html.HTML;import javax.swing.text.ChangedCharSetException;import java.io.*;import java.util.Hashtable;import java.util.Properties;import java.util.Vector;import java.util.Enumeration;import java.net.URL;import sun.misc.MessageUtils;/** * A simple DTD-driven HTML parser. The parser reads an * HTML file from an InputStream and calls various methods * (which should be overridden in a subclass) when tags and * data are encountered. * <p> * Unfortunately there are many badly implemented HTML parsers * out there, and as a result there are many badly formatted * HTML files. This parser attempts to parse most HTML files. * This means that the implementation sometimes deviates from * the SGML specification in favor of HTML. * <p> * The parser treats \r and \r\n as \n. Newlines after starttags * and before end tags are ignored just as specified in the SGML/HTML * specification. * <p> * The html spec does not specify how spaces are to be coalesced very well. * Specifically, the following scenarios are not discussed (note that a * space should be used here, but I am using &nbsp to force the space to * be displayed): * <p> * '<b>blah <i> <strike> foo' which can be treated as: * '<b>blah <i><strike>foo' * <p>as well as: * '<p><a href="xx"> <em>Using</em></a></p>' * which appears to be treated as: * '<p><a href="xx"><em>Using</em></a></p>' * <p> * If <code>strict</code> is false, when a tag that breaks flow, * (<code>TagElement.breaksFlows</code>) or trailing whitespace is * encountered, all whitespace will be ignored until a non whitespace * character is encountered. This appears to give behavior closer to * the popular browsers. * * @see DTD * @see TagElement * @see SimpleAttributeSet * @version 1.47, 02/26/06 * @author Arthur van Hoff * @author Sunita Mani */publicclass Parser implements DTDConstants { private char text[] = new char[1024]; private int textpos = 0; private TagElement last; private boolean space; private char str[] = new char[128]; private int strpos = 0; protected DTD dtd = null; private int ch; private int ln; private Reader in; private Element recent; private TagStack stack; private boolean skipTag = false; private TagElement lastFormSent = null; private SimpleAttributeSet attributes = new SimpleAttributeSet(); // State for <html>, <head> and <body>. Since people like to slap // together HTML documents without thinking, occasionally they // have multiple instances of these tags. These booleans track // the first sightings of these tags so they can be safely ignored // by the parser if repeated. private boolean seenHtml = false; private boolean seenHead = false; private boolean seenBody = false; /** * The html spec does not specify how spaces are coalesced very well. * If strict == false, ignoreSpace is used to try and mimic the behavior * of the popular browsers. * <p> * The problematic scenarios are: * '<b>blah <i> <strike> foo' which can be treated as: * '<b>blah <i><strike>foo' * as well as: * '<p><a href="xx"> <em>Using</em></a></p>' * which appears to be treated as: * '<p><a href="xx"><em>Using</em></a></p>' * <p> * When a tag that breaks flow, or trailing whitespace is encountered * ignoreSpace is set to true. From then on, all whitespace will be * ignored. * ignoreSpace will be set back to false the first time a * non whitespace character is encountered. This appears to give * behavior closer to the popular browsers. */ private boolean ignoreSpace; /** * This flag determines whether or not the Parser will be strict * in enforcing SGML compatibility. If false, it will be lenient * with certain common classes of erroneous HTML constructs. * Strict or not, in either case an error will be recorded. * */ protected boolean strict = false; /** Number of \r\n's encountered. */ private int crlfCount; /** Number of \r's encountered. A \r\n will not increment this. */ private int crCount; /** Number of \n's encountered. A \r\n will not increment this. */ private int lfCount; // // To correctly identify the start of a tag/comment/text we need two // ivars. Two are needed as handleText isn't invoked until the tag // after the text has been parsed, that is the parser parses the text, // then a tag, then invokes handleText followed by handleStart. // /** The start position of the current block. Block is overloaded here, * it really means the current start position for the current comment, * tag, text. Use getBlockStartPosition to access this. */ private int currentBlockStartPos; /** Start position of the last block. */ private int lastBlockStartPos; /** * array for mapping numeric references in range * 130-159 to displayable Unicode characters. */ private static final char[] cp1252Map = { 8218, // ‚ 402, // ƒ 8222, // „ 8230, // … 8224, // † 8225, // ‡ 710, // ˆ 8240, // ‰ 352, // Š 8249, // ‹ 338, // Œ 141, //  142, // Ž 143, //  144, //  8216, // ‘ 8217, // ’ 8220, // “ 8221, // ” 8226, // • 8211, // – 8212, // — 732, // ˜ 8482, // ™ 353, // š 8250, // › 339, // œ 157, //  158, // ž 376 // Ÿ }; public Parser(DTD dtd) { this.dtd = dtd; } /** * @return the line number of the line currently being parsed */ protected int getCurrentLine() { return ln; } /** * Returns the start position of the current block. Block is * overloaded here, it really means the current start position for * the current comment tag, text, block.... This is provided for * subclassers that wish to know the start of the current block when * called with one of the handleXXX methods. */ int getBlockStartPosition() { return Math.max(0, lastBlockStartPos - 1); } /** * Makes a TagElement. */ protected TagElement makeTag(Element elem, boolean fictional) { return new TagElement(elem, fictional); } protected TagElement makeTag(Element elem) { return makeTag(elem, false); } protected SimpleAttributeSet getAttributes() { return attributes; } protected void flushAttributes() { attributes.removeAttributes(attributes); } /** * Called when PCDATA is encountered. */ protected void handleText(char text[]) { } /** * Called when an HTML title tag is encountered. */ protected void handleTitle(char text[]) { // default behavior is to call handleText. Subclasses // can override if necessary. handleText(text); } /** * Called when an HTML comment is encountered. */ protected void handleComment(char text[]) { } protected void handleEOFInComment() { // We've reached EOF. Our recovery strategy is to // see if we have more than one line in the comment; // if so, we pretend that the comment was an unterminated // single line comment, and reparse the lines after the // first line as normal HTML content. int commentEndPos = strIndexOf('\n'); if (commentEndPos >= 0) { handleComment(getChars(0, commentEndPos)); try { in.close(); in = new CharArrayReader(getChars(commentEndPos + 1)); ch = '>'; } catch (IOException e) { error("ioexception"); } resetStrBuffer(); } else { // no newline, so signal an error error("eof.comment"); } } /** * Called when an empty tag is encountered. */ protected void handleEmptyTag(TagElement tag) throws ChangedCharSetException { } /** * Called when a start tag is encountered. */ protected void handleStartTag(TagElement tag) { } /** * Called when an end tag is encountered. */ protected void handleEndTag(TagElement tag) { } /** * An error has occurred. */ protected void handleError(int ln, String msg) { /* Thread.dumpStack(); System.out.println("**** " + stack); System.out.println("line " + ln + ": error: " + msg); System.out.println(); */ } /** * Output text. */ void handleText(TagElement tag) { if (tag.breaksFlow()) { space = false; if (!strict) { ignoreSpace = true; } } if (textpos == 0) { if ((!space) || (stack == null) || last.breaksFlow() || !stack.advance(dtd.pcdata)) { last = tag; space = false; lastBlockStartPos = currentBlockStartPos; return; } } if (space) { if (!ignoreSpace) { // enlarge buffer if needed if (textpos + 1 > text.length) { char newtext[] = new char[text.length + 200]; System.arraycopy(text, 0, newtext, 0, text.length); text = newtext; } // output pending space text[textpos++] = ' '; if (!strict && !tag.getElement().isEmpty()) { ignoreSpace = true; } } space = false; } char newtext[] = new char[textpos]; System.arraycopy(text, 0, newtext, 0, textpos); // Handles cases of bad html where the title tag // was getting lost when we did error recovery. if (tag.getElement().getName().equals("title")) { handleTitle(newtext); } else { handleText(newtext); } lastBlockStartPos = currentBlockStartPos; textpos = 0; last = tag; space = false; } /** * Invoke the error handler. */ protected void error(String err, String arg1, String arg2, String arg3) { handleError(ln, err + " " + arg1 + " " + arg2 + " " + arg3); } protected void error(String err, String arg1, String arg2) { error(err, arg1, arg2, "?"); } protected void error(String err, String arg1) { error(err, arg1, "?", "?"); } protected void error(String err) { error(err, "?", "?", "?"); } /** * Handle a start tag. The new tag is pushed * onto the tag stack. The attribute list is * checked for required attributes. */ protected void startTag(TagElement tag) throws ChangedCharSetException { Element elem = tag.getElement(); // If the tag is an empty tag and texpos != 0 // this implies that there is text before the // start tag that needs to be processed before // handling the tag. // if (!elem.isEmpty() || textpos != 0) { handleText(tag); } else { // this variable gets updated in handleText(). // Since in this case we do not call handleText() // we need to update it here. // last = tag; // Note that we should really check last.breakFlows before // assuming this should be false. space = false; } lastBlockStartPos = currentBlockStartPos; // check required attributes for (AttributeList a = elem.atts ; a != null ; a = a.next) { if ((a.modifier == REQUIRED) && ((attributes.isEmpty()) || ((!attributes.isDefined(a.name)) && (!attributes.isDefined(HTML.getAttributeKey(a.name)))))) { error("req.att ", a.getName(), elem.getName()); } } if (elem.isEmpty()) { handleEmptyTag(tag); /* } else if (elem.getName().equals("form")) { handleStartTag(tag); */ } else { recent = elem; stack = new TagStack(tag, stack); handleStartTag(tag); } } /** * Handle an end tag. The end tag is popped * from the tag stack. */ protected void endTag(boolean omitted) { handleText(stack.tag); if (omitted && !stack.elem.omitEnd()) { error("end.missing", stack.elem.getName()); } else if (!stack.terminate()) { error("end.unexpected", stack.elem.getName()); } // handle the tag handleEndTag(stack.tag); stack = stack.next; recent = (stack != null) ? stack.elem : null; } boolean ignoreElement(Element elem) { String stackElement = stack.elem.getName(); String elemName = elem.getName(); /* We ignore all elements that are not valid in the context of a table except <td>, <th> (these we handle in legalElementContext()) and #pcdata. We also ignore the <font> tag in the context of <ul> and <ol> We additonally ignore the <meta> and the <style> tag if the body tag has been seen. **/ if ((elemName.equals("html") && seenHtml) || (elemName.equals("head") && seenHead) || (elemName.equals("body") && seenBody)) { return true; } if (elemName.equals("dt") || elemName.equals("dd")) { TagStack s = stack; while (s != null && !s.elem.getName().equals("dl")) { s = s.next; } if (s == null) { return true; } } if (((stackElement.equals("table")) && (!elemName.equals("#pcdata")) && (!elemName.equals("input"))) || ((elemName.equals("font")) && (stackElement.equals("ul") || stackElement.equals("ol"))) || (elemName.equals("meta") && stack != null) || (elemName.equals("style") && seenBody) || (stackElement.equals("table") && elemName.equals("a"))) { return true; } return false; } /** * Marks the first time a tag has been seen in a document */ protected void markFirstTime(Element elem) { String elemName = elem.getName(); if (elemName.equals("html")) { seenHtml = true; } else if (elemName.equals("head")) { seenHead = true; } else if (elemName.equals("body")) { if (buf.length == 1) { // Refer to note in definition of buf for details on this. char[] newBuf = new char[256]; newBuf[0] = buf[0]; buf = newBuf; } seenBody = true; } } /** * Create a legal content for an element. */ boolean legalElementContext(Element elem) throws ChangedCharSetException { // System.out.println("-- legalContext -- " + elem); // Deal with the empty stack if (stack == null) { // System.out.println("-- stack is empty"); if (elem != dtd.html) { // System.out.println("-- pushing html"); startTag(makeTag(dtd.html, true)); return legalElementContext(elem); } return true; } // Is it allowed in the current context if (stack.advance(elem)) { // System.out.println("-- legal context"); markFirstTime(elem); return true; } boolean insertTag = false; // The use of all error recovery strategies are contingent // on the value of the strict property. // // These are commonly occuring errors. if insertTag is true, // then we want to adopt an error recovery strategy that // involves attempting to insert an additional tag to // legalize the context. The two errors addressed here // are: // 1) when a <td> or <th> is seen soon after a <table> tag. // In this case we insert a <tr>. // 2) when any other tag apart from a <tr> is seen // in the context of a <tr>. In this case we would // like to add a <td>. If a <tr> is seen within a // <tr> context, then we will close out the current // <tr>. // // This insertion strategy is handled later in the method. // The reason for checking this now, is that in other cases // we would like to apply other error recovery strategies for example // ignoring tags. // // In certain cases it is better to ignore a tag than try to // fix the situation. So the first test is to see if this // is what we need to do. // String stackElemName = stack.elem.getName(); String elemName = elem.getName(); if (!strict && ((stackElemName.equals("table") && elemName.equals("td")) || (stackElemName.equals("table") && elemName.equals("th")) || (stackElemName.equals("tr") && !elemName.equals("tr")))){ insertTag = true; } if (!strict && !insertTag && (stack.elem.getName() != elem.getName() || elem.getName().equals("body"))) { if (skipTag = ignoreElement(elem)) { error("tag.ignore", elem.getName()); return skipTag; } } // Check for anything after the start of the table besides tr, td, th // or caption, and if those aren't there, insert the <tr> and call⌨️ 快捷键说明
复制代码
Ctrl + C
搜索代码
Ctrl + F
全屏模式
F11
切换主题
Ctrl + Shift + D
显示快捷键
?
增大字号
Ctrl + =
减小字号
Ctrl + -