/*
 * [FixCSSClasses.java]
 *
 * Summary: Correct HTML Class of a link to CMP conventions.
 *
 * Copyright: (c) 1999-2017 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 2010-11-21 initial version
 */
package com.mindprod.repair;

import com.mindprod.commandline.CommandLine;
import com.mindprod.common18.EIO;
import com.mindprod.fastcat.FastCat;
import com.mindprod.filter.OnlyDirectoriesFilter;
import com.mindprod.htmlmacros.macro.Global;
import com.mindprod.htmlmacros.support.AssignCSSClasses;
import com.mindprod.hunkio.HunkIO;

import java.io.File;
import java.io.FileNotFoundException;
import java.io.FileWriter;
import java.io.FilenameFilter;
import java.io.IOException;
import java.util.regex.Matcher;
import java.util.regex.Pattern;

import static java.lang.System.*;

/**
 * Correct HTML Class of a link to CMP conventions.
 * <p/>
 * Also puts links on single line with canonical spacing.
 * <p/>
 *
 * @author Roedy Green, Canadian Mind Products
 * @version 1.0 2010-11-21 initial version.
 * @noinspection WeakerAccess
 * @since 2010-11-21
 */
public final class FixCSSClasses
    {
    // declarations

    /**
     * true, updates classes in files.  false just displays potential changes.
     */
    private static final boolean CHANGE_FILES = true;

    /**
     * true if want extra debugging output
     */
    private static final boolean DEBUGGING = false;

    private static final int FIRST_COPYRIGHT_YEAR = 1999;

    /**
     * undisplayed copyright notice.
     *
     * @noinspection UnusedDeclaration
     */
    private static final String EMBEDDED_COPYRIGHT =
            "Copyright: (c) 1999-2017 Roedy Green, Canadian Mind Products, http://mindprod.com";

    /**
     * piece of pattern uses do find class="xxx"
     */
    private static final String FRAGMENT_CLASS = "class\\s*=\\s*\"([\\p{Lower}\\-]+)\"";

    /**
     * piece of pattern used to get href="url". extract just part in quotes. We want the whole thing, including http://www.mindprod.com/jgloss/jgloss.html
     */
    private static final String FRAGMENT_HREF =
            "href=\"([^\"]+)\"";

    /**
     * date this version released.
     *
     * @noinspection UnusedDeclaration
     */
    private static final String RELEASE_DATE = "2010-11-21";

    /**
     * embedded version string.
     *
     * @noinspection UnusedDeclaration
     */
    private static final String VERSION_STRING = "1.0";

    /**
     * Regex pattern to look for links without a class
     */
    private static final Pattern PATTERN_WITHOUT_CLASS = Pattern.compile( "<a\\s+" + FRAGMENT_HREF,
            Pattern.CASE_INSENSITIVE );

    /**
     * Regex pattern to look for links with a class
     */
    private static final Pattern PATTERN_WITH_CLASS = Pattern.compile( "<a\\s+" + FRAGMENT_CLASS + "\\s+" +
                                                                       FRAGMENT_HREF, Pattern.CASE_INSENSITIVE );
    // /declarations
    // methods

    /**
     * compose replacement class href with canonical spacing
     *
     * @param betterClass name of class, "none" for no class
     * @param url         link url
     *
     * @return HTML to link to that url with that class, not quoted
     */
    private static String composeReplacement( final String betterClass, final String url )
        {
        final FastCat sb = new FastCat( 7 );
        sb.append( "<a " );
        if ( betterClass != null )
            {
            sb.append( "class=\"" );
            sb.append( betterClass );
            sb.append( "\" " );
            }
        sb.append( "href=\"" );
        sb.append( url );
        sb.append( "\"" );
        // convert $ -> \$ and \ to \\
        return sb.toString();
        }      // /method

    /**
     * fix classes  in one file.
     *
     * @param fileBeingProcessed the file currently being processed.
     *
     * @throws java.io.IOException if can't read file
     * @noinspection SameParameterValue, WeakerAccess
     */
    private static void fixHTMLClassesInFile( File fileBeingProcessed ) throws IOException
        {
        String big = HunkIO.readEntireFile( fileBeingProcessed );
        String result = fixHTMLClassesInString( big, fileBeingProcessed );
        if ( result.equals( big ) )
            {
            // nothing changed. No need to write results.
            //  out.println( "- " + fileBeingProcessed.getDecoratedStoreName() );
            return;
            }
        // generate output into a temporary file until we are sure all is ok.
        // create a temp file in the same directory as filename
        // it changed
        out.println( "* " + EIO.getCanOrAbsPath( fileBeingProcessed ) );
        if ( CHANGE_FILES )
            {
            final File tempFile = HunkIO.createTempFile( "temp.html", ".tmp.html", fileBeingProcessed );
            FileWriter emit = new FileWriter( tempFile );
            emit.write( result );
            emit.close();
            HunkIO.deleteAndRename( tempFile, fileBeingProcessed );
            }
        }       // /method

    /**
     * fix classes  in contents of one file.
     *
     * @param contents           file contents to repair.
     * @param fileBeingProcessed file we are repairing. We don't touch file, just use its name to compute css classes.
     *
     * @return improved contents of one entire webpage, all link classes updated.
     * @noinspection SameParameterValue, WeakerAccess
     */
    private static String fixHTMLClassesInString( final String contents, File fileBeingProcessed )
        {
        // we don't modify the head or tail.
        final int headEndsAt = contents.indexOf( "<!-- /generated by" );
        final int footerStartsAt = contents.lastIndexOf( "<!-- macro Foot " );
        if ( headEndsAt >= 0 && footerStartsAt >= 0 && footerStartsAt > headEndsAt )
            {
            // at this point body needs fixes both for classless and previously classed links
            final FastCat result = new FastCat( 3 );
            result.append( contents.substring( 0, headEndsAt ) );
            result.append( replaceClassesInBody( contents.substring( headEndsAt, footerStartsAt ),
                    fileBeingProcessed ) );
            result.append( contents.substring( footerStartsAt ) );
            return result.toString();
            }
        else if ( fileBeingProcessed.getParent().endsWith( "include" ) )
            {
            // include has no head or tail
            return replaceClassesInBody( contents, fileBeingProcessed );
            }
        else
            {
            // who knows what this is. Leave it alone.
            return contents;
            }
        }        // /method

    /**
     * replace class in the body of this html file
     * * @param body body of the html file with head and tail stripped.
     *
     * @param body               contents ond <body..</body>
     * @param fileBeingProcessed file this text came from.
     *
     * @return body with cssclasses updated.
     */
    private static String replaceClassesInBody( String body, final File fileBeingProcessed )
        {
        // must be StringBuffer, not StringBuilder or FastCat
        final StringBuffer sb = new StringBuffer( body.length() * 120 / 100 ); // give 20% for growth.
        // do those with existing class first to avoid reprocessing.
        final Matcher mw = PATTERN_WITH_CLASS.matcher( body );  // Matchers are used both for matching and finding.
        // we search is two passes, first for links with a class already attached, then again for links with any class yet.
        // search for better css classes than one's already there.
        while ( mw.find() )
            {
            if ( DEBUGGING )
                {
                out.println( "found: class:" + mw.group( 1 ) + " url:" + mw.group( 2 ) );
                }
            assert mw.groupCount() == 2 : "bad match: " + mw.group( 0 );
            final String oldClass = mw.group( 1 );
            final String url = mw.group( 2 );
            String betterClass = oldClass;
            if ( Global.assignCSSClasses.isCSSClassModifiable( oldClass ) )
                {
                betterClass = Global.assignCSSClasses.assignCSSClass( url, fileBeingProcessed );
                if ( !oldClass.equals( betterClass ) )
                    {
                    out.println( oldClass + " --> " + ( betterClass != null ? betterClass : "none" ) + " " + url );
                    }
                }
            final String replacement = composeReplacement( betterClass, url );
            mw.appendReplacement( sb, Matcher.quoteReplacement( replacement ) ); // also appends junk between hits
            }
        mw.appendTail( sb );
        // |||||||||||||||||||||||||||||||||||||||||||||||||||||
        // do it all over, this time looking for links without any class.
        body = sb.toString();
        sb.setLength( 0 );
        final Matcher mc = PATTERN_WITHOUT_CLASS.matcher( body );  // Matchers are used both for matching and finding.
        while ( mc.find() )
            {
            if ( DEBUGGING )
                {
                out.println( "found: class:- url:" + mc.group( 1 ) );
                }
            assert mc.groupCount() == 1 : "bad match: " + mc.group( 0 );
            final String url = mc.group( 1 ); // might have http:// prefix and any extension.
            final String betterClass;
            betterClass = Global.assignCSSClasses.assignCSSClass( url, fileBeingProcessed );
            if ( betterClass != null )
                {
                out.println( " --> " + betterClass + " " + url );
                }
            // works with null betterClass too.
            final String replacement = composeReplacement( betterClass, url );
            mc.appendReplacement( sb, Matcher.quoteReplacement( replacement ) ); // also appends junk between hits.
            }
        mc.appendTail( sb );
        // at this point, the sb contains the body with all classless links repaired.
        body = sb.toString();
        return body;
        }       // /method

    /**
     * fixes classes in HTML files.
     *
     * @param args configuration then names of files to process, dirs, files, -s, *.*, no wildcards.
     */
    public static void main( final String[] args )
        {
        Global.installConfiguration( args );
        if ( DEBUGGING )
            {
            out.println( PATTERN_WITH_CLASS );
            out.println( PATTERN_WITHOUT_CLASS );
            }
        // gather all the files mentioned on the command line.
        // either directories, files, with -s and subdirs option.
        // warning. Windows expands any wildcards in a nasty way.
        // do not use wildcards.
        // See http://mindprod.com/jgloss/wildcard.html
        out.println( "Gathering html files to tidy the CSS classes on links..." );
        // command line lists dirs and files to process, pruned by following filters.
        CommandLine commandLine = new CommandLine( args,
                new OnlyDirectoriesFilter(),
                new ModifiableFilter( Global.assignCSSClasses ) );
        for ( File file : commandLine )
            {
            try
                {
                // -q gives no output at all, otherwise just files that changed.
                fixHTMLClassesInFile( file );
                }
            catch ( FileNotFoundException e )
                {
                err.println( "Error: "
                             + EIO.getCanOrAbsPath( file )
                             + " not found." );
                }
            catch ( Exception e )
                {
                err.println();
                e.printStackTrace( err );
                err.println( e.getClass().toString()
                             + e.getMessage()
                             + " in file "
                             + EIO.getCanOrAbsPath( file ) );
                err.println();
                }
            } // end for
        System.exit( 0 );
        } // end main
    }  // /method
// /methods

/**
 * Selects files we are permitted to touch, to upgrade their CSS classes.
 * Get info from the configuration
 *
 * @author Roedy Green, Canadian Mind Products
 * @version 1.0 2011-01-09 - initial version
 * @since 2011-01-09
 */
final class ModifiableFilter implements FilenameFilter
    {
    private final AssignCSSClasses assignCSSClasses;

    /**
     * constructor
     *
     * @param assignCSSClasses AssignCSSClasses configuration
     */
    ModifiableFilter( AssignCSSClasses assignCSSClasses )
        {
        this.assignCSSClasses = assignCSSClasses;
        }      // /method

    /**
     * Select only files with that pass muster
     *
     * @param dir  the directory in which the file was found.
     * @param name the name of the file
     *
     * @return true if and only if the name should be included in the file list; false otherwise.
     */
    public boolean accept( File dir, String name )
        {
        // URL we are considering disturbing, as it appears embedded in the html, e.g. ../jgloss/x.html x.html
        return assignCSSClasses.needsCSSRepair( new File( dir, name ) );
        }      // /method
    } // end ModifiableFilter