/*
 * [SortQuotes.java]
 *
 * Summary: Sort quotes. Does not deDup. Standalone.
 *
 * Copyright: (c) 2008-2017 Roedy Green, Canadian Mind Products, http://mindprod.com
 *
 * Licence: This software may be copied and used freely for any purpose but military.
 *          http://mindprod.com/contact/nonmil.html
 *
 * Requires: JDK 1.8+
 *
 * Created with: JetBrains IntelliJ IDEA IDE http://www.jetbrains.com/idea/
 *
 * Version History:
 *  1.0 2008-06-08
 */
package com.mindprod.htmlmacros;

import com.mindprod.common18.Misc;
import com.mindprod.fastcat.FastCat;
import com.mindprod.htmlmacros.macro.Global;
import com.mindprod.htmlmacros.support.ConfigurationForMindprod;
import com.mindprod.htmlmacros.support.QuoteFlock;
import com.mindprod.hunkio.HunkIO;

import java.io.File;
import java.io.IOException;
import java.util.ArrayList;
import java.util.Collections;
import java.util.regex.Matcher;
import java.util.regex.Pattern;

import static java.lang.System.*;

/**
 * Sort quotes. Does not deDup. Standalone.
 * <p/>
 * SortQuotes will accept either stripgenerated or expanded html.
 * However, it generates unexpanded headers.
 * The bodies of the quotes will be left stripgenerated or expanded as they were.
 * We used to run this before every macro expansion. Now it is does less often,
 * in tquotes.btm when the quotes are tidied and verified.
 * Gets its list of Flocks from the QuoteFlock enum.
 *
 * @author Roedy Green, Canadian Mind Products
 * @version 1.0 2008-06-08
 * @see com.mindprod.htmlmacros.support.QuoteFlock
 * @since 2008-06-08
 */
public class SortQuotes
    {
    /**
     * true if want extra debugging output
     */
    private static final boolean DEBUGGING = false;

    /**
     * what we look for surrounding text of interest
     */
    private static final String CLOSE_TAG = "</blockquote>";

    /**
     * what we look for surrounding text of interest
     */
    private static final String OPEN_TAG = "<blockquote";

    /**
     * count of haw many times each author referenced
     */
    private static final ArrayList<SortableQuote> quotes = new ArrayList<>( 10000 );

    /**
     * pattern of quote already handled, used to extract author
     * Should ignore multiple famous inside quote.
     */
    private static final Pattern AUTHOR = Pattern.compile( ".+<span class=\"said\">(.+?)</span>", Pattern.DOTALL );

    /**
     * pattern to check that quote ends with suitable HTML
     */
    private static final Pattern TEXT_TAIL = Pattern.compile( ".+(?:</p>|</ol>|</ul>|</table>|</dl>|</div>|<br>| -->)" +
                                                              "\\s*(<br>)?\\s*~\\s*<span " +
                                                              "class=\"said\">", Pattern.DOTALL );

    /**
     * true when sorting first batch of quotes
     */
    private static boolean first = true;

    /**
     * sort ALL quotations to be used for random quotes in all flocks.
     * Use in frequeuncy mode to pull quotes to top containing a magic
     * word.
     *
     * @param args Command line: [plain] [frequency word] [length] [weight]
     *             Sorts all quotes from configuration.
     *
     * @throws java.io.IOException if cannot read quotations
     */
    public static void main( String[] args ) throws IOException
        {
        Global.installConfiguration( new ConfigurationForMindprod() );
        int errorCount = 0;
        out.println( "Sorting " + QuoteFlock.values().length + " quotation libraries" );
        for ( QuoteFlock f : QuoteFlock.values() )
            {
            File file = f.getFile();
            try
                {
                final String big = HunkIO.readEntireFile( file );
                int start0 = 0;
                int start1;
                while ( ( start1 = big.indexOf( OPEN_TAG, start0 ) ) >= 0 )
                    {
                    final int start2 = start1 + OPEN_TAG.length();
                    final int end1 = big.indexOf( CLOSE_TAG, start2 );
                    if ( end1 < 0 )
                        {
                        throw new IllegalArgumentException( ">>>missing </blockquote>" );
                        }
                    final int end2 = end1 + CLOSE_TAG.length();
                    final String rawBlockQuote = big.substring( start1, end2 );
                    // process text <blockquote .. </blockquote>
                    // Test number of quotes in collected,
                    // duplicates
                    // and format of <!-- added yyyy-mm-dd later in htmlmacros.support.QuoteFlock
                    if ( rawBlockQuote.startsWith( "<blockquote>" ) )
                        {
                        out.println( ">>>classless quote: " + file.toString() + " " + rawBlockQuote );
                        errorCount++;
                        }
                    final Matcher t = TEXT_TAIL.matcher( rawBlockQuote );
                    if ( t.lookingAt() )
                        {
                        if ( t.groupCount() == 1 && t.group( 1 ) != null && t.group( 1 ).equals( "<br>" ) )
                            {
                            out.println( ">>>trailing <br>: " + file.toString() + " " + rawBlockQuote );
                            errorCount++;
                            }
                        }
                    else
                        {
                        out.println( ">>>no trailing </p> </ol> </ul> </table> </dl> </div> <br> --> followed by ~: " + file
                                .toString() + " " + rawBlockQuote );
                        errorCount++;
                        }
                    final Matcher a = AUTHOR.matcher( rawBlockQuote );
                    if ( a.lookingAt() && a.groupCount() == 1 )
                        {
                        quotes.add( new SortableQuote( a.group( 1 ), rawBlockQuote ) );
                        }
                    else
                        {
                        out.println( "authorless quote: " + file.toString() + " " + rawBlockQuote );
                        errorCount++;
                        // file it under blank for now
                        quotes.add( new SortableQuote( "", rawBlockQuote ) );
                        }
                    start0 = end2;
                    } // end while
                }
            catch ( IOException e )
                {
                e.printStackTrace( err );
                err.println( "Fatal error: unable to open " + file + " possibly locked in the editor." );
                System.exit( 2 );
                }
            {
            final String order = args.length == 0 ? "plain" : args[ 0 ].toLowerCase();
            switch ( order )
                {
                case "plain":
                case "alpha":
                {
                if ( first )
                    {
                    out.println( "Sort quotes in plain order" );
                    }
                Collections.sort( quotes );
                break;
                }
                // we have collected quotes for one file.
                case "frequency":
                {
                if ( first )
                    {
                    out.println( "Sort quotes by frequency " + args[ 1 ] );
                    }
                Collections.sort( quotes, new ByFrequency( args[ 1 ] ) );
                break;
                }
                case "length":
                {
                if ( first )
                    {
                    out.println( "Sort quotes in length order" );
                    }
                Collections.sort( quotes, new ByLength() );
                break;
                }
                case "date":
                {
                if ( first )
                    {
                    out.println( "Sort quotes in date order" );
                    }
                Collections.sort( quotes, new ByDate() );
                break;
                }
                case "weight":
                {
                if ( first )
                    {
                    out.println( "Sort quotes in weight order" );
                    }
                Collections.sort( quotes, new ByWeight() );
                break;
                }
                default:
                {
                if ( first )
                    {
                    err.println( "no keyword on command line: plain, frequency word, length, date or weight" );
                    for ( String arg : args )
                        {
                        err.println( arg );
                        }
                    }
                System.exit( 1 );
                break;
                }
                }
            first = false;
            }
            if ( DEBUGGING )
                {
                out.println( "Sorting " + quotes.size() + " " + f.getDescription() );
                }
            // regenerate head and tail
            final FastCat sb = new FastCat( quotes.size() * 2 + 6 );
            sb.append( "<!-- macro QuoteFlockHead {" );
            sb.append( f.getDescription() );
            sb.append( "} -->\n" );
            sb.append( "<p>Quotations are selected from this pool (and other quotation pools)\n"
                       + "in a pseudorandom way every hour\n"
                       + "and inserted at the top and bottom of some of the major pages on this website.\n"
                       + "Feel free to copy any of these quotes and paste them for whatever purpose you please, \n"
                       + "including on your own website, blog, social media page or forum debate posts.</p>\n" );
            for ( SortableQuote s : quotes )
                {
                sb.append( s.getQuotation() );
                sb.append( "\n" );
                }
            sb.append( "<!-- macro Foot -->\n" );
            HunkIO.writeEntireFile( file, sb.toString() );
            // prepare for next file of quotes
            quotes.clear();
            } // end for to process each file
        if ( errorCount != 0 )
            {
            out.println( errorCount + " errors in quotations" );
            }
        Misc.trackLastThread();
        System.exit( 0 );
        }
    }