/*
 * [FindUndone.java]
 *
 * Summary: Find block quotes not yet done.
 *
 * 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.repair;

import com.mindprod.commandline.CommandLine;
import com.mindprod.entities.DeEntifyStrings;
import com.mindprod.filter.AllButSVNDirectoriesFilter;
import com.mindprod.filter.ExtensionListFilter;
import com.mindprod.hunkio.HunkIO;

import java.io.File;
import java.io.IOException;
import java.util.Arrays;
import java.util.HashMap;
import java.util.regex.Matcher;
import java.util.regex.Pattern;

import static java.lang.System.*;

/**
 * Find block quotes not yet done.
 *
 * @author Roedy Green, Canadian Mind Products
 * @version 1.0 2008-06-08
 * @since 2008-06-08
 */
public class FindUndone
    {
    // ------------------------------ CONSTANTS ------------------------------

    /**
     * estimated number of authors whose dates we collect, with some padding
     */
    private static final int EST_AUTHORS = 600;

    /**
     * count of haw many times each author referenced
     */
    private static final HashMap<String, RankedAuthor> authorLookup = new HashMap<String, RankedAuthor>( EST_AUTHORS );

    /**
     * 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";

    /**
     * pattern of quote already handled, used to extract author
     */
    private static final Pattern AUTHOR = Pattern.compile( ".+<span class=\"said\">(.+?)</span>.*", Pattern.DOTALL );
    // $ ( ) * + - . ? [ \ ] ^ { | } <- quotable regex chars

    /**
     * /**
     * pattern of quote already handled
     */
    private static final Pattern DONE1 = Pattern.compile( ".+<span class=\"said\">(.+)</span>[ \n\r]+<!\\-\\- macro Born .*", Pattern.DOTALL );

    private static final Pattern DONE2 = Pattern.compile( ".+<span class=\"quoth\">(.+)</span> ", Pattern.DOTALL );

    /**
     * Study all blockqkotes to see which done. List ones not complte.
     *
     * @param rawBlockQuote blockquote contents as they are now, including <blockquote .. </blockquote>
     */
    private static void processDoneBlockQuote( String rawBlockQuote )
        {
        if ( DONE1.matcher( rawBlockQuote ).matches() || DONE2.matcher( rawBlockQuote ).matches() || rawBlockQuote.contains( "macro KJV" ) )
            {
            // if was already done.
            }
        else
            {
            //  not yet done.
            final Matcher a = AUTHOR.matcher( rawBlockQuote );
            if ( a.matches() && a.groupCount() == 1 )
                {
                // unprocessed author
                String author = a.group( 1 );
                if ( author.length() > 150 )
                    {
                    author = author.substring( 0, 150 );
                    }
                author = DeEntifyStrings.flattenHTML( author, ' ' );
                RankedAuthor existing = authorLookup.get( author );
                if ( existing != null )
                    {
                    existing.inc();
                    }
                else
                    {
                    authorLookup.put( author, new RankedAuthor( author ) );
                    }
                }
            else
                {
                // block quote without parseable author.
                }
            }
        }

    // -------------------------- STATIC METHODS --------------------------
    // --------------------------- main() method ---------------------------
    @SuppressWarnings( { "ResultOfMethodCallIgnored" } )
    public static void main
    ( String[] args )
        {
        // get files to process from command line.
        out.println( "Gathering files to find non-standard <blockquotes>..." );
        CommandLine wantedFiles = new CommandLine( args,
                new AllButSVNDirectoriesFilter(),
                new ExtensionListFilter( "html" ) );
        for ( File file : wantedFiles )
            {
            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>
                    processDoneBlockQuote( rawBlockQuote );
                    start0 = end2;
                    } // end while
                }
            catch ( IOException e )
                {
                err.println();
                e.printStackTrace( err );
                err.println();
                }
            }// end for to process each file
        // export to array for easy sort.
        RankedAuthor[] authors = authorLookup.values().toArray( new RankedAuthor[ authorLookup.size() ] );
        Arrays.sort( authors );
        out.println();
        out.println( authors.length + " fixes yet to do" );
        for ( RankedAuthor author : authors )
            {
            out.println( author.count + "  " + author.author );
            }
        out.println( "done" );
        }
    }

/**
 * used to count hom many times each author is quoted
 */
class RankedAuthor implements Comparable<RankedAuthor>
    {
    // ------------------------------ FIELDS ------------------------------
    public final String author;

    /**
     * how many times this author was quoted on the website
     */
    public int count;

    // -------------------------- PUBLIC INSTANCE  METHODS --------------------------
    // --------------------------- CONSTRUCTORS ---------------------------
    RankedAuthor( String author )
        {
        this.author = author;
        this.count = 1;
        }

    // -------------------------- OTHER METHODS --------------------------
    void inc()
        {
        count++;
        }

    /**
     * sort by rank big to small, author tie breaker.
     * Defines default the sort order for RankedAuthor Objects.
     * Compare this RankedAuthor with another RankedAuthor.
     * Compares descending count then author.
     * Informally, returns (this-other) or +ve if this is more positive than other.
     *
     * @param other other RankedAuthor to compare with this one
     *
     * @return +ve if this&gt;other, 0 if this==other, -ve if this&lt;other
     */
    public final int compareTo( RankedAuthor other )
        {
        int diff = other.count - this.count;
        if ( diff != 0 )
            {
            return diff;
            }
        return this.author.compareToIgnoreCase( other.author );
        }
    }