source: main/trunk/gli/src/org/greenstone/gatherer/metadata/FilenameEncoding.java@ 33728

/**
 *############################################################################
 * A component of the Greenstone Librarian Interface, part of the Greenstone
 * digital library suite from the New Zealand Digital Library Project at the
 * University of Waikato, New Zealand.
 *
 * Author: Michael Dewsnip, NZDL Project, University of Waikato, NZ
 *
 * Copyright (C) 2010 Greenstone Digital Library Project
 *
 * This program is free software; you can redistribute it and/or modify
 * it under the terms of the GNU General Public License as published by
 * the Free Software Foundation; either version 2 of the License, or
 * (at your option) any later version.
 *
 * This program is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 * GNU General Public License for more details.
 *
 * You should have received a copy of the GNU General Public License
 * along with this program; if not, write to the Free Software
 * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
 *############################################################################
 */
 
package org.greenstone.gatherer.metadata;

import java.io.File;
import java.net.*;
import java.nio.charset.*;
import java.util.*;
import org.greenstone.gatherer.collection.CollectionManager;
import org.greenstone.gatherer.DebugStream;

import java.util.regex.Matcher;
import java.util.regex.Pattern;



/** Static access class that contains many of the methods used to work with filename encodings. 
* Works closely with classes FileNode, CollectionTreeNode, MetadataXMLFile, MetadataXMLFileManager
* to maintain a map of URLEncodedFilenames to their filename encodings.
* The process of filename encoding further affects the CollectionManager which refreshes its CollectionTree, 
* FileManager (move, delete, rename actions), MetadataValueTableModel, EnrichPane. */

public class FilenameEncoding {
    /** Display of filenames in the trees are in URL encoding, if debugging */
    public static boolean DEBUGGING = false;

    /** Set to false by Gatherer if the locale is UTF-8, as Java's handling is 
    * such that non-UTF8 filename encodings on a UTF-8 locale are destructively 
    * converted so that the bytecodes in the filename are not preserved. */
    public static boolean MULTIPLE_FILENAME_ENCODINGS_SUPPORTED = false;

    /** Also set by Gatherer. 
    * If the OS supports multiple filename encodings, we will be working with URL strings
    * and the applicable separators are always the forward slash ("/") not File.separator. 
    * If  multiple filename encodings are not supported, we're dealing with File.separator. */
    public static String URL_FILE_SEPARATOR = File.separator;
    

    /** gs.filenameEncoding is a special sort of metadata that is not merely to be stored along 
     * with a file, but is to be applied in real-time on the file's name in the CollectionTree 
     * display. Since FileNodes are constantly destroyed and reconstructed by that Tree when 
     * its nodes are expanded and contracted, storing the filename encodings of each file along
     * with the file in a FileNode doesn't help because it doesn't last. Instead of rediscovering
     * the encoding at every stage by querying the metadataXML file, we store the encodings for
     * fast access: in a map of (URLEncodedFilePath, filename-encoding) pairs. 
     * The current design of the map is to only store any active filename metadata assigned
     * directly at that file/folder's level, and if there is none discovered at that level, then
     * storing the empty string for it. Therefore, if the hashmap contains no entry for 
     * a file, it means this still needs to be retrieved. */
    public static Map map = new HashMap();

//*********************** BUSY REFRESHING / REQUIRING  REFRESH *********************

    /** Set to true if filename encoding metadata was changed. Called by the enter keyPress 
    * event in gui.EnrichPane and when the gs.FilenameEncoding field loses focus. */
    private static boolean refreshRequired = false;

    synchronized public static boolean isRefreshRequired() { 
        return refreshRequired; 
    }
    
    synchronized public static void setRefreshRequired(boolean state) { 
        if(MULTIPLE_FILENAME_ENCODINGS_SUPPORTED) {
            refreshRequired = state;
        } else {
            refreshRequired = false; 
        }       
    }

//************************** MAP RETRIEVAL METHODS ******************************   

    /** Returns the cumulative gs.filenameEncoding metadata 
    * assigned to a file inside the collection. */    
    public static String findFilenameEncoding(
            File file, String urlEncodedFilePath, boolean bruteForceLookup) 
    {
        //if(bruteForceLookup) {
        //  return findFilenameEncodingBruteForce(file, urlEncodedFilePath, bruteForceLookup);
        //}     
    
        String encoding = "";
        
        // Check any assigned encoding at this level, starting with the map first
        // and else retrieving the filename encoding from the metadata file
        if(!map.containsKey(urlEncodedFilePath)) {
        
            // Check for filename encoding metadata *directly* associated with the file
            // Now don't need to get any inherited encoding metadata here, because of 
            // the way we're storing and retrieving encoding information from the map.          
            ArrayList list = MetadataXMLFileManager.getMetadataAssignedDirectlyToFile(file, true); // true: gets gs.filenameEncoding only
            if(!list.isEmpty()) {
                MetadataValue metavalue = (MetadataValue)list.get(0); // get(list.size()-1); 
                encoding = metavalue.getValue();                
            } // else no filename encoding set yet at this level
        
            // Now we've done a lookup at this level cache the result in the map,
            // including empty strings, to indicate that we've done a full lookup
            map.put(urlEncodedFilePath, encoding);
        } 
        else { // an entry exists in the map, get it from there
            encoding = (String)map.get(urlEncodedFilePath);
        }   
        
        // if no meta was specified at at the file level, look for any inherited metadata
        if(encoding.equals("")) {
            encoding = getInheritedFilenameEncoding(urlEncodedFilePath, file);
        }
        
        //System.err.println("\n@@@@Looked for: " + urlEncodedFilePath + " | found: " + encoding);
        return encoding; // found something in map, may still be "", but it's what was stored
    }
    
    /** Checks the file-to-encoding map for all the superfolders of the given 
     * filename in sequence for an applicable encoding. Note that the file/folder
     * at the level of urlFoldername (and dir) has already been inspected. */
    static public String getInheritedFilenameEncoding(String urlFoldername, File dir) 
    {
        String encoding = "";
        boolean done = false;
        
        // don't want to search past import folder which is as 
        // far as we need to go to determine inherited encodings
        File importDir = new File(CollectionManager.getLoadedCollectionImportDirectoryPath());
        if(dir.equals(importDir)) { // if the top-level dir was already checked, we're done
            done = true;
        }

        // For directories, first remove trailing file separator in order to start checking from higher level folders
        int lastIndex = urlFoldername.length()-1;
        char urlFileSeparatorChar = URL_FILE_SEPARATOR.charAt(0);
        if(urlFoldername.charAt(lastIndex) == urlFileSeparatorChar) { 
            urlFoldername = urlFoldername.substring(0, lastIndex);
        }

        while(!done) {
            // get the folder that's one level up
            dir = dir.getParentFile();      
            
            int index = urlFoldername.lastIndexOf(URL_FILE_SEPARATOR);      
            if(index == -1) { // no more slashes
                done = true;
            } else { 
                urlFoldername = urlFoldername.substring(0, index);
            }
            
            // now look in the map to see whether there's an encoding for this folder
            String folder = urlFoldername + URL_FILE_SEPARATOR;
            if(map.containsKey(folder)) {
                encoding = (String)map.get(folder); // may be ""                
            } else { // no entry in map, so look in the metadata.xml at this folder level
                ArrayList list = MetadataXMLFileManager.getMetadataAssignedDirectlyToFile(
                        dir, true); // true: gets gs.filenameEncoding only
                if(!list.isEmpty()) {
                    MetadataValue metavalue = (MetadataValue)list.get(0); // get(list.size()-1); 
                    encoding = metavalue.getValue();                
                }
                map.put(folder, encoding); // may be ""
            }
            
            if(!encoding.equals("")){
                done = true;
            } // else if "", loop to check next folder up
            else if(dir.equals(importDir)) { // don't iterate past the import folder, which we've now checked
                done = true;
            }
        }

        return encoding;
    }

    /** Called by GUIManager when a collection is closed. This then empties the 
    * file-to-encoding map which is applicable only on a per-collection basis */
    static public void closeCollection() {  
        //printFilenameMap("Closing collection. Clearing file-to-encoding map of entries:");
        map.clear();
    }

    // Useful for debugging: prints contents of file-to-encoding map
    static public void printFilenameMap(String heading) {
        System.err.println("\n********************************************");
        System.err.println(heading.toUpperCase());
        Iterator entries = map.entrySet().iterator();
        while(entries.hasNext()) {
            Map.Entry entry = (Map.Entry)entries.next();
            System.err.println("+ " + (String)entry.getKey() + ": " + (String)entry.getValue());
        }
        System.err.println("********************************************\n");
    }
    
    // UNUSED at present. Brute force version of the findFilenameEncoding() method
    // Doesn't use the map, but gets *all* the metadata assigned to a file/folder to 
    // work out the encoding applicable to a file/folder.
    public static String findFilenameEncodingBruteForce(File file, String urlEncodedFilename, 
                         boolean bruteForceLookup) 
    {
        System.err.println("\n***** BRUTE FORCE getFilenameEncoding() called\n");
    
    
        String encoding = "";

        // Check for filename encoding metadata *directly* associated with the file
        // Now don't need to get any inherited encoding metadata here, because of 
        // the way we're storing and retrieving encoding information from the map.

        ArrayList list = MetadataXMLFileManager.getMetadataAssignedToFile(file, true); // true: gets gs.filenameEncoding only
        if(!list.isEmpty()) {      
            // try to get the filename encoding meta that was assigned last to this 
            // file, even though it makes no sense to have multiple values for it
            MetadataValue metavalue = (MetadataValue)list.get(list.size()-1); 
            encoding = metavalue.getValue();
            
            if(encoding == null) { // unlikely ???
                System.err.println("**** ERROR: encoding for " 
                        + urlEncodedFilename + " is NULL!");    
                encoding = "";
            }
        } // else no filename encoding set yet, perhaps
        //System.err.println("**** Found encoding for " + urlEncodedFilename + " " + encoding);
        return encoding;
    }

//****************************** APPLYING ENCODINGS TO FILENAMES *****************************
    
    /** URL encoded version of the byte codes of the given file's name */
    public static String calcURLEncodedFilePath(File file) {    
        if(!MULTIPLE_FILENAME_ENCODINGS_SUPPORTED) {
            return file.getAbsolutePath();
        }
        else {
            String filename = fileToURLEncoding(file);
            return filename;
        }
    }

    /** URL encoded version of the byte codes of this file's name */
    public static String calcURLEncodedFileName(String urlfilepath) {
        String filename = urlfilepath;
        if(filename.endsWith(URL_FILE_SEPARATOR)) { // directory, remove trailing slash
            filename = filename.substring(0, filename.length() - 1);
        }       
        
        // remove the directory prefix (if any) to get the filename
        int index = filename.lastIndexOf(URL_FILE_SEPARATOR);
        if(index != -1) {
            filename = filename.substring(index+1); // skip separator
        }
        
        return filename;    
    }

    /** Given a string representing an alias to an official encoding (and unofficial ones 
    * starting with "Latin-"), attempts to work out what the canonical encoding for that is. 
    * If the given encoding is unrecognised, it is returned as is. */
    public static String canonicalEncodingName(String encoding) {
        String canonicalEncoding = encoding;
        try {
            // Latin-1 -> ISO-8859-1
            String alias = canonicalEncoding.toLowerCase();
            if(alias.startsWith("latin")){                              
                canonicalEncoding = "ISO-8859" + alias.substring("latin".length());
            }

            // canonical encoding for official aliases
            canonicalEncoding = Charset.forName(canonicalEncoding).name();          
            return canonicalEncoding;
        } catch (Exception e) {
            System.err.println("(Could not recognise encoding (alias): " 
                    + encoding + ".)");
            return encoding; // no alias could be found, return the original parameter
        }
    }

//************************* GETTING THE URL ENCODING OF FILENAMES *********************************

    /** 
     * Given a String containing hexentities, will convert back into the unicode version of the String.
     * e.g. A string like "02 Tēnā Koutou\.mp3" will be returned as "02 Tena Koutou\.mp3" with macrons on e and a
     * I've tested this in a separate file that imports java.util.regex.Matcher and java.util.regex.Pattern
     * and contains a copy of Utility.debugUnicodeString(String) with the following main function:
        public static void main(String args[]) {
            String str = "02 Tēnā Koutou\\.mp3"; // or more basic case: String str = "mmmmānnnnēpppp\\.txt";        
            System.err.println("About to decode hex string: " + str);
            String result = decodeStringContainingHexEntities(str);
            System.err.println("Decoded hex string: " + result + " - debug unicode form: " + debugUnicodeString(result));
        }
    */
    public static String decodeStringContainingHexEntities(String str) {
        String result = "";
        boolean done = false;       
        Pattern hexPattern = Pattern.compile("(&#x[0-9a-zA-Z]{1,4}+;)");
        Matcher matcher = hexPattern.matcher(str);
        
        int searchFromIndex = 0;
        int endMatchIndex = -1;
        
        while(matcher.find(searchFromIndex)) {
            String hexPart = matcher.group();
            //System.err.println("Found hexpart match: " + hexPart);
            
            int startMatchIndex = matcher.start();
            endMatchIndex = matcher.end();
            result += str.substring(searchFromIndex, startMatchIndex);          
            
            String hexNumberStr = hexPart.substring(3, hexPart.length()-1); // lose the "&#x" prefix and the ";" suffix to get just the hex number portion of the match
            // https://stackoverflow.com/questions/16625865/java-unicode-to-hex-string
            // https://stackoverflow.com/questions/11194513/convert-hex-string-to-int
            
            //System.err.println("hexNumberStr so far: " + hexNumberStr);           
            int tmpDigit = Integer.parseInt(hexNumberStr);
            //System.err.println("As digit: " + tmpDigit);
            hexNumberStr = String.format("%04d", tmpDigit); 
            //System.err.println("2 hexNumberStr so far: " + hexNumberStr);
            hexNumberStr = "0x" + hexNumberStr; // e.g "0xDDDD"
            //int hexNumber = Integer.parseInt(hexNumberStr);
            int hexNumber = Integer.decode(hexNumberStr);
            String hexNumberAsChar = Character.toString((char) hexNumber);
            result += hexNumberAsChar;
             
            searchFromIndex = endMatchIndex;
            
        }
        
        if(endMatchIndex != -1) {
            result += str.substring(endMatchIndex);
            //System.err.println("suffix: " + str.substring(endMatchIndex));
        }
        
        return result;
    }   

    /** Attempting to produce the equivalent method fileToURLEncoding() above, but taking a String as input parameter */
     public static String fileNameToHex(String filename) {
        /*String filename_url_encoded = "";
        try {
            URI filename_uri = new URI(filename);
            String filename_ascii = filename_uri.toASCIIString();
            String filename_raw_bytes = URLDecoder.decode(filename_ascii,"ISO-8859-1");
            filename_url_encoded = iso_8859_1_filename_to_url_encoded(filename_raw_bytes); 
            return filename_url_encoded;
        } catch (Exception e) {
            e.printStackTrace();
            // Give up trying to convert
            filename_url_encoded = filename; 
        }
        return filename_url_encoded;
        */
        
        String hexFilename = "";
        for(int i = 0; i < filename.length(); i++) {
            int charCode = filename.codePointAt(i); // unicode codepoint / ASCII code
            
            // ASCII table: https://cdn.sparkfun.com/assets/home_page_posts/2/1/2/1/ascii_table_black.png
            // If the unicode character code pt is less than the ASCII code for space and greater than for tilda, let's display the char in hex (x0000 format)
            if((charCode >= 20 && charCode <= 126) || charCode == 9 || charCode == 10 || charCode == 13) { // space, tilda, TAB, LF, CR are printable, leave them in for XML element printing
                hexFilename += filename.charAt(i);
            } else {
                hexFilename += "&#x" + String.format("%x", charCode).toUpperCase() + ";"; // looks like: "&#x[up-to-4-hexdigits-in-UPPERCASE];"
            }
        }
          
        return hexFilename;
     }

    // Dr Bainbridge's methods
    /* On Linux machines that are set to using an ISO-8859 (Latin) type encoding, 
    * we can work with URL-encoded filenames in Java. Java works with whatever 
    * encoding the filesystem uses. Unlike systems working with UTF-8, where Java 
    * interprets filenames as UTF-8 (a destructive process since characters invalid
    * for UTF-8 are replaced with the invalid character, which means the original
    * character's byte codes can not be regained), working with an ISO-8859-1 
    * system means the original byte codes of the characters are preserved, 
    * regardless of whether the characters represent ISO-8859-1 or not. Such byte
    * codes are converted by the following method to the correct URL versions of 
    * the strings that the filenames represent (that is, the correct URL representations
    * of the filenames in their original encodings). This is useful for interactions with
    * Perl as Java and Perl can use URL-encoded filenames to talk about the same files
    * on the file system, instead of having to work out what encoding they are in. */
    
    public static String fileToURLEncoding(File file) {
        if(!MULTIPLE_FILENAME_ENCODINGS_SUPPORTED) {
            return file.getAbsolutePath();
        }

        String filename_url_encoded = "";
        
        // The following test for whether the file exists or not is a problem
        // when a File object--whose actual file is in the process of being moved 
        // and therefore temporarily does not 'exist' on the actual system--can't
        // be URL encoded: the following would return "" when a file doesn't exist.
        // So commenting out the test.
        /* 
        if(!file.getName().equals("recycle")) {
            if(!file.isFile() && !file.isDirectory()) {
            System.err.println("*** ERROR. Java can't see file: " + file.getAbsolutePath());
            return "";
            }
            
            if(!file.exists()) {
            System.err.println("*** NOTE: File doesn't exist: " + file.getAbsolutePath());
            return ""; //file.getName();
            }
        }
        */

        URI filename_uri = file.toURI();
        try {
            // The trick:
            //  1. toASCIIString() will %xx encode values > 127
            //  2. Decode the result to "ISO-8859-1" 
            //  3. URL encode the bytes to string
            
            // Step 2 forces the string to be 8-bit values.  It
            // doesn't matter if the starting raw filename was *not*
            // in the ISO-8859-1 encoding, the effect is to ensure
            // we have an 8-bit byte string that (numerically)
            // captures the right value.  These numerical values are
            // then used to determine how to URL encode it
            
            String filename_ascii = filename_uri.toASCIIString();
            String filename_raw_bytes = URLDecoder.decode(filename_ascii,"ISO-8859-1");
            filename_url_encoded = iso_8859_1_filename_to_url_encoded(filename_raw_bytes);
            
        }
        catch (Exception e) {
            e.printStackTrace();
            // Give up trying to convert
            filename_url_encoded = file.getAbsolutePath(); 
        }
        return filename_url_encoded;
    }
    
    // For unicode codepoints see:
    // http://unicode.org/Public/MAPPINGS/ISO8859/8859-1.TXT for ISO8859-1 (Latin-1)
    // where 0xE2 maps to codepoint 0x00E2 and is defined as "Latin small letter a with circumflex"
    // http://unicode.org/Public/MAPPINGS/ISO8859/8859-7.TXT for ISO8859-7 (Greek)
    // where 0xE2 maps to codepoint 0x03B2 and is defined as "Greek small letter beta"    
    public static String iso_8859_1_filename_to_url_encoded(String raw_bytes_filename) 
        throws Exception
    {
        String urlEncoded = "";

        try {
            // By this point we have a UTF-8 encoded string that captures
            // what the ISO-8859-1 (Latin-1) character is that corresponded to the
            // 8-bit numeric value for that character in the filename
            // on the file system
            
            // For example:
            //   File system char:             <lower-case beta char in Latin-7> = %E2
            //   Equivalent Latin 1 char:      <lower-case a with circumflex>    = %E2
            //   Mapped to UTF-8:              <lower-case a with circumflex>    = <C3><A2>

            // Our task is to take the string the contains <C3><A2> and ensure that
            // we "see" it as <E2>

            byte [] raw_bytes = raw_bytes_filename.getBytes("ISO-8859-1");
            String unicode_filename = new String(raw_bytes,"UTF-8");
            
            for(int i = 0; i < unicode_filename.length(); i++) {
            char charVal = unicode_filename.charAt(i);
            if ((int)charVal > 255) {
                urlEncoded += String.format("&#x%02X;", (int)charVal);
            }
            else if((int)charVal > 127) {
                urlEncoded += String.format("%%%02X", (int)charVal);
            } else {
                urlEncoded += String.format("%c", (char)charVal);
            }
            }
        }
        catch (Exception e) {
            //e.printStackTrace();
            throw(e);
        }

        return urlEncoded;
    }

    // unused for now
    public static String raw_filename_to_url_encoded(String fileName)
        throws Exception
    {
        String urlEncoded = "";
        try {
            byte[] bytes = fileName.getBytes();
            
            for(int i = 0; i < bytes.length; i++) {
            // mask each byte (by applying & 0xFF) to make the signed
            // byte (in the range -128 to 127) unsigned (in the range
            // 0 to 255).

            int byteVal = (int)(bytes[i] & 0xFF);
            
            if(byteVal > 127) {
                urlEncoded += String.format("%%%02X", (int)byteVal);
            } else {
                urlEncoded += String.format("%c",(char)byteVal);
            }
            }
        }
        catch (Exception e) {
            //e.printStackTrace();
            throw(e);
        }

        return urlEncoded;
    }
   
}