source: main/trunk/gli/src/org/greenstone/gatherer/remote/ZipCollectionArchives.java@ 21787

package org.greenstone.gatherer.remote;

import java.io.*;
import java.util.zip.*;


/**
 */
public class ZipCollectionArchives
{
    static public void main(String[] args)
    {
    if (args.length != 3) {
        System.err.println("Usage: ZipCollectionArchives <zip-file> <collect-directory-path> <collection-name>");
        return;
    }

    String zip_file_path = args[0];
    String collect_directory_path = args[1];
    String collection_name = args[2];

    if (!collect_directory_path.endsWith(File.separator)) {
        collect_directory_path += File.separator;
    }

    try {
        ZipOutputStream zos = new ZipOutputStream(new FileOutputStream(zip_file_path));

        // Add archives directory, with doc.xml files only
        String archives_relative_path = collection_name + File.separator + "archives";
        ZipTools.ZipFilter collection_archives_zip_filter = new CollectionArchivesZipFilter();
        ZipTools.addFileToZip(zos, collect_directory_path, archives_relative_path, collection_archives_zip_filter);

        zos.close();
    }
    catch (Exception exception) {
        exception.printStackTrace();
    }
    }


    static public class CollectionArchivesZipFilter
    extends ZipTools.NullZipFilter
    {
    private byte[] bytes_remaining = new byte[0];
    private int section_depth = 0;
    private boolean in_section_content = false;


    public boolean shouldIncludeFile(String relative_file_path)
    {
        // Only doc.xml files are strictly required, but we include archives.inf as well to prevent
        //   errors when zipping up the archives of a collection where no files were imported
        //   (in this case the zip file would contain nothing at all)
        return (relative_file_path.equals("archives.inf") || relative_file_path.endsWith(File.separator + "archives.inf") || relative_file_path.equals("doc.xml") || relative_file_path.endsWith(File.separator + "doc.xml"));
    }


    public boolean shouldIncludeFileContent(String relative_file_path)
    {
        // Only content for doc.xml files is included
        return (relative_file_path.equals("doc.xml") || relative_file_path.endsWith(File.separator + "doc.xml"));
    }


    public void filterFileContent(String relative_file_path, BufferedInputStream bis, ZipOutputStream zos)
    {
        // Reset the status in case there were errors in previous doc.xml files
        section_depth = 0;
        in_section_content = false;

        // Filter out the <Content>...</Content> of the doc.xml files
        try {
        BufferedReader reader = new BufferedReader(new InputStreamReader(bis, "UTF-8"));
        String line = null;
        while ((line = reader.readLine()) != null) {
            // If this line isn't filtered, write it out to zos
            if (!filterFileLine(line)) {
            line += "\n";
            byte[] bytes = line.getBytes("UTF-8");
            zos.write(bytes, 0, bytes.length);
            }
        }
        }
        catch (Exception exception) {
        exception.printStackTrace();
        }
    }


    private boolean filterFileLine(String line)
    {
        boolean filter_line = false;

        if (line.indexOf("<Section>") != -1) {
        section_depth++;
        }
        if (line.indexOf("<Content>") != -1) {
        in_section_content = true;
        }

        // If we're in a subsection or in a content element, filter this line
        if (section_depth > 1 || in_section_content) {
        filter_line = true;
        }

        if (line.indexOf("</Content>") != -1) {
        in_section_content = false;
        }
        if (line.indexOf("</Section>") != -1) {
        section_depth--;
        }

        return filter_line;
    }
    }
}