parserutils.java

html 解析处理代码

 	
        StringBuffer output = new StringBuffer();

        boolean charFound=false;
        for (int index=0; index<input.length(); index++)
        {    
            charFound=false;
            for (int charsCount=0; charsCount<charsToBeRemoved.length(); charsCount++)
                if (charsToBeRemoved.charAt(charsCount)==input.charAt(index))
                    charFound=true;
            if (!((Character.isWhitespace(input.charAt(index))) || (Character.isSpaceChar(input.charAt(index))) || (charFound)))
                output.append(input.charAt(index));
        }

        return output.toString();

    }

    /**
     * Remove from the beginning and the end of the input string all the spaces and tabs like chars.
     * Remove also the chars specified in the input variable charsToBeRemoved.
     * <BR>The removal process removes only chars at the beginning and at the end of the string.
     * <BR>For example if you call trimSpacesBeginEnd(&quot;&lt;DIV&gt;  +12.5 &lt;/DIV&gt;&quot;, &quot;&lt;>DIV/&quot;),
     * <BR>you obtain a string &quot;+12.5&quot; as output (space chars and &lt;,&gt;,D,I,V,/ are chars that must be removed).
     * <BR>For example if you call trimSpacesBeginEnd(&quot;&lt;DIV&gt;  Trim all spaces but not the ones inside the string &lt;/DIV&gt;&quot;, &quot;&lt;>DIV/&quot;),
     * <BR>you obtain a string &quot;Trim all spaces but not the ones inside the string&quot; as output (all the spaces inside the string are preserved).
     * @param input The string in input.
     * @param charsToBeRemoved The chars to be removed.
     * @return The string as output.
    */
    public static String trimSpacesBeginEnd (String input, String charsToBeRemoved)
    {
 	
        String output = new String();

        int begin=0;
        int end=input.length()-1;
        boolean charFound=false;
        boolean ok=true;
        for (int index=begin; (index<input.length()) && ok; index++)
        {                
            charFound=false;
            for (int charsCount=0; charsCount<charsToBeRemoved.length(); charsCount++)
                if (charsToBeRemoved.charAt(charsCount)==input.charAt(index))
                    charFound=true;
            if (!( (Character.isWhitespace(input.charAt(index))) || (Character.isSpaceChar(input.charAt(index))) || (charFound) ))
            {
                begin=index;
                ok=false;
            }
        }
        ok=true;
        for (int index=end; (index>=0) && ok; index--)
        {
            charFound=false;
            for (int charsCount=0; charsCount<charsToBeRemoved.length(); charsCount++)
                if (charsToBeRemoved.charAt(charsCount)==input.charAt(index))
                    charFound=true;
            if (!( (Character.isWhitespace(input.charAt(index))) || (Character.isSpaceChar(input.charAt(index))) || (charFound) ))
            {
                end=index;
                ok=false;
            }
        }
        output=input.substring(begin,end+1);

        return output;
        
    }
    
    /**
     * Split the input string considering as string separator
     * all the characters
     * with the only exception of the characters specified in charsDoNotBeRemoved param.
     * <BR>For example if you call splitButChars(&quot;&lt;DIV&gt;  +12.5, +3.4 &lt;/DIV&gt;&quot;, &quot;+.1234567890&quot;),
     * <BR>you obtain an array of strings {&quot;+12.5&quot;, &quot;+3.4&quot;} as output (+,.,1,2,3,4,5,6,7,8,9,0 are chars that do not be removed).
     * @param input The string in input.
     * @param charsDoNotBeRemoved The chars that do not be removed.
     * @return The array of strings as output.
    */
    public static String[] splitButChars (String input, String charsDoNotBeRemoved)
    {
 	
        ArrayList output = new ArrayList();
        int minCapacity = 0;
        StringBuffer str = new StringBuffer();

        boolean charFound = false;
        boolean toBeAdd = false;
        for (int index=0; index<input.length(); index++)
        {    
            charFound=false;
            for (int charsCount=0; charsCount<charsDoNotBeRemoved.length(); charsCount++)
                if (charsDoNotBeRemoved.charAt(charsCount)==input.charAt(index))
                    charFound=true;
            if (charFound)
            {
                str.append(input.charAt(index));
                toBeAdd=false;
            }
            else
                if (!toBeAdd)
                    toBeAdd=true;
            // finished to parse one string
            if (toBeAdd && (str.length()!=0)) {
                minCapacity++;
                output.ensureCapacity(minCapacity);
                if (output.add(str.toString()))
                    str = new StringBuffer();
                else
                    minCapacity--;
            }
        }
        // add the last string
        if (str.length()!=0) {
            minCapacity++;
            output.ensureCapacity(minCapacity);
            if (output.add(str.toString()))
                str = new StringBuffer();
            else
                minCapacity--;
        }

        output.trimToSize();
        Object[] outputObj = output.toArray();
        String[] outputStr = new String[output.size()];
        for (int i=0; i<output.size(); i++)
            outputStr[i] = new String((String) outputObj[i]);
        return outputStr;
        
    }
    
    /**
     * Remove from the input string all the characters
     * with the only exception of the characters specified in charsDoNotBeRemoved param.
     * <BR>For example if you call trimButChars(&quot;&lt;DIV&gt;  +12.5 &lt;/DIV&gt;&quot;, &quot;+.1234567890&quot;),
     * <BR>you obtain a string &quot;+12.5&quot; as output (+,.,1,2,3,4,5,6,7,8,9,0 are chars that do not be removed).
     * <BR>For example if you call trimButChars(&quot;&lt;DIV&gt;  +1 2 . 5 &lt;/DIV&gt;&quot;, &quot;+.1234567890&quot;),
     * <BR>you obtain a string &quot;+12.5&quot; as output (the spaces between 1 and 2, 2 and ., . and 5 are removed).
     * @param input The string in input.
     * @param charsDoNotBeRemoved The chars that do not be removed.
     * @return The string as output.
    */
    public static String trimButChars (String input, String charsDoNotBeRemoved)
    {
 	
        StringBuffer output = new StringBuffer();

        boolean charFound=false;
        for (int index=0; index<input.length(); index++)
        {    
            charFound=false;
            for (int charsCount=0; charsCount<charsDoNotBeRemoved.length(); charsCount++)
                if (charsDoNotBeRemoved.charAt(charsCount)==input.charAt(index))
                    charFound=true;
            if (charFound)
                output.append(input.charAt(index));
        }
        
        return output.toString();
        
    }
    
    /**
     * Remove from the beginning and the end of the input string all the characters
     * with the only exception of the characters specified in charsDoNotBeRemoved param.
     * <BR>The removal process removes only chars at the beginning and at the end of the string.
     * <BR>For example if you call trimButCharsBeginEnd(&quot;&lt;DIV&gt;  +12.5 &lt;/DIV&gt;&quot;, &quot;+.1234567890&quot;),
     * <BR>you obtain a string &quot;+12.5&quot; as output (+,.,1,2,3,4,5,6,7,8,9,0 are chars that do not be removed).
     * <BR>For example if you call trimButCharsBeginEnd(&quot;&lt;DIV&gt;  +1 2 . 5 &lt;/DIV&gt;&quot;, &quot;+.1234567890&quot;),
     * <BR>you obtain a string &quot;+1 2 . 5&quot; as output (the spaces inside the string are not removed).
     * @param input The string in input.
     * @param charsDoNotBeRemoved The chars that do not be removed.
     * @return The string as output.
    */
    public static String trimButCharsBeginEnd (String input, String charsDoNotBeRemoved)
    {
 	
        String output = new String();

        int begin=0;
        int end=input.length()-1;
        boolean charFound=false;
        boolean ok=true;
        for (int index=begin; (index<input.length()) && ok; index++)
        {                
            charFound=false;
            for (int charsCount=0; charsCount<charsDoNotBeRemoved.length(); charsCount++)
                if (charsDoNotBeRemoved.charAt(charsCount)==input.charAt(index))
                    charFound=true;
            if (charFound)
            {
                begin=index;
                ok=false;
            }
        }
        ok=true;
        for (int index=end; (index>=0) && ok; index--)
        {
            charFound=false;
            for (int charsCount=0; charsCount<charsDoNotBeRemoved.length(); charsCount++)
                if (charsDoNotBeRemoved.charAt(charsCount)==input.charAt(index))
                    charFound=true;
            if (charFound)
            {
                end=index;
                ok=false;
            }
        }
        output=input.substring(begin,end+1);

        return output;
        
    }

    /**
     * Split the input string considering as string separator
     * the chars specified in the input variable charsToBeRemoved.
     * <BR>For example if you call splitChars(&quot;&lt;DIV&gt;  +12.5, +3.4 &lt;/DIV&gt;&quot;, &quot; <>DIV/,&quot;),
     * <BR>you obtain an array of strings {&quot;+12.5&quot;, &quot;+3.4&quot;} as output (space chars and &lt;,&gt;,D,I,V,/ and the comma are chars that must be removed).
     * @param input The string in input.
     * @param charsToBeRemoved The chars to be removed.
     * @return The array of strings as output.
    */
    public static String[] splitChars (String input, String charsToBeRemoved)
    {
 	
        ArrayList output = new ArrayList();
        int minCapacity = 0;
        StringBuffer str = new StringBuffer();

        boolean charFound = false;
        boolean toBeAdd = false;
        for (int index=0; index<input.length(); index++)
        {    
            charFound=false;
            for (int charsCount=0; charsCount<charsToBeRemoved.length(); charsCount++)
                if (charsToBeRemoved.charAt(charsCount)==input.charAt(index))
                    charFound=true;
            if (!(charFound))
            {
                str.append(input.charAt(index));
                toBeAdd=false;
            }
            else
                if (!toBeAdd)
                    toBeAdd=true;
            // finished to parse one string
            if (toBeAdd && (str.length()!=0)) {
                minCapacity++;
                output.ensureCapacity(minCapacity);
                if (output.add(str.toString()))
                    str = new StringBuffer();
                else
                    minCapacity--;
            }
        }
        // add the last string
        if (str.length()!=0) {
            minCapacity++;
            output.ensureCapacity(minCapacity);
            if (output.add(str.toString()))
                str = new StringBuffer();
            else
                minCapacity--;
        }

        output.trimToSize();
        Object[] outputObj = output.toArray();
        String[] outputStr = new String[output.size()];
        for (int i=0; i<output.size(); i++)
            outputStr[i] = new String((String) outputObj[i]);
        return outputStr;
        
    }

    /**
     * Remove from the input string all the chars specified in the input variable charsToBeRemoved.
     * <BR>For example if you call trimChars(&quot;&lt;DIV&gt;  +12.5 &lt;/DIV&gt;&quot;, &quot;<>DIV/ &quot;),
     * <BR>you obtain a string &quot;+12.5&quot; as output (&lt;,&gt;,D,I,V,/ and space char are chars that must be removed).
     * <BR>For example if you call trimChars(&quot;&lt;DIV&gt;  Trim All Chars Also The Ones Inside The String &lt;/DIV&gt;&quot;, &quot;<>DIV/ &quot;),
     * <BR>you obtain a string &quot;TrimAllCharsAlsoTheOnesInsideTheString&quot; as output (all the spaces inside the string are removed).
     * @param input The string in input.
     * @param charsToBeRemoved The chars to be removed.
     * @return The string as output.
    */
    public static String trimChars (String input, String charsToBeRemoved)
    {
 	
        StringBuffer output = new StringBuffer();

        boolean charFound=false;
        for (int index=0; index<input.length(); index++)
        {    
            charFound=false;
            for (int charsCount=0; charsCount<charsToBeRemoved.length(); charsCount++)
                if (charsToBeRemoved.charAt(charsCount)==input.charAt(index))
                    charFound=true;
            if (!(charFound))
                output.append(input.charAt(index));
        }

        return output.toString();

    }

    /**
     * Remove from the beginning and the end of the input string all the chars specified in the input variable charsToBeRemoved.
     * <BR>The removal process removes only chars at the beginning and at the end of the string.
     * <BR>For example if you call trimCharsBeginEnd(&quot;&lt;DIV&gt;  +12.5 &lt;/DIV&gt;&quot;, &quot;<>DIV/ &quot;),
     * <BR>you obtain a string &quot;+12.5&quot; as output (' ' is a space char and &lt;,&gt;,D,I,V,/ are chars that must be removed).
     * <BR>For example if you call trimCharsBeginEnd(&quot;&lt;DIV&gt;  Trim all spaces but not the ones inside the string &lt;/DIV&gt;&quot;, &quot;<>DIV/ &quot;),