Main.java Source code

Introduction

Here is the source code for Main.java

Source

//package com.java2s;
//License from project: Open Source License 

import java.io.Writer;

public class Main {
    /**
     * Unescape (and normalize) the given XML string.
     * 
     * Only element content (excluding markup) and attributes (values) need to be unescaped.
     * Markup, comments, CDATA sections, and prosessing instructions need to be left unchanged.
     * 
     *
     * @param writer         The writer where the escaped string is written to.
     * @param str            The given XML string (character text or attribute value; unescaped).
     */
    private static void unescape(Writer writer, String str) throws Exception {
        // First, normalize end-of-line delimiters
        str = str.replaceAll("(?i)
", "
"); // case-insensitive matching
        str = str.replaceAll("(?i)
…", "
"); // case-insensitive matching
        str = str.replaceAll("(?i)…", "
"); // case-insensitive matching
        str = str.replaceAll("(?i)
", "
"); // case-insensitive matching
        str = str.replaceAll("(?i)
", "
"); // case-insensitive matching

        // Finally, unescape characters
        for (int i = 0; i < str.length(); i++) {
            char c = str.charAt(i);

            switch (c) {
            // ampersand:
            case '&':
                int nextIdx = i + 1;
                int semiColonIdx = str.indexOf(';', nextIdx);
                // Found ampersand did not start an entity, no following semi-colon exists
                if (semiColonIdx == -1) {
                    // The the text looks like &...
                    writer.write(c);
                    continue;
                }
                int amphersandIdx = str.indexOf('&', nextIdx);
                // Found ampersand did not start an entity, no following semi-colon before next ampersand
                if (amphersandIdx != -1 && amphersandIdx < semiColonIdx) {
                    // Then the text looks like &...&...;
                    writer.write(c);
                    continue;
                }

                String entityContent = str.substring(nextIdx, semiColonIdx);
                int entityValue = -1;
                if (entityContent.length() > 0) {
                    // Escaped value content is an integer (decimal or hexadecimal)
                    if (entityContent.charAt(0) == '#') {
                        if (entityContent.length() > 1) {
                            char isHexChar = entityContent.charAt(1);

                            try {
                                switch (isHexChar) {
                                // hexadecimal, 16 base
                                case 'X':
                                case 'x':
                                    entityValue = Integer.parseInt(entityContent.substring(2), 16);
                                    break;
                                // decimal, 10 base
                                default:
                                    entityValue = Integer.parseInt(entityContent.substring(1), 10);
                                }
                                if (entityValue > 0xFFFF) {
                                    entityValue = -1;
                                }
                            } catch (NumberFormatException nfex) {
                                entityValue = -1;
                            }
                        }
                    }
                    // Escaped value content is an entity name
                    else {
                        // less-than:
                        if ("lt".equals(entityContent) == true) {
                            entityValue = 60;
                        }
                        // ampersand:
                        else if ("amp".equals(entityContent) == true) {
                            entityValue = 38;
                        }
                        // greater-than:
                        else if ("gt".equals(entityContent) == true) {
                            entityValue = 62;
                        }
                        // single-quote:
                        else if ("apos".equals(entityContent) == true) {
                            entityValue = 39;
                        }
                        // double-quote:
                        else if ("quot".equals(entityContent) == true) {
                            entityValue = 34;
                        }
                    }
                }

                // Entity value was not found (not an XML entity value)
                if (entityValue == -1) {
                    writer.write('&');
                    writer.write(entityContent);
                    writer.write(';');
                }
                // Entity value was found
                else {
                    writer.write(entityValue);
                }
                i = semiColonIdx; // move index up to the semi-colon
                break;
            // else, default print char
            default:
                writer.write(c);
            }
        }
    }
}