Greenstone tutorial exercise

Back to wiki
Back to index
Prerequisite: Looking at a multimedia collection
Sample files:
Devised for Greenstone version: 2.60|3.06
Modified for Greenstone version: 2.87|3.08

Building a multimedia collection

We will proceed to reconstruct from scratch the Beatles collection that you have just looked at. We develop the collection using a small subset of the material, purely to speed up the repeated rebuilding that is involved.

  1. Start a new collection (FileNew...) called small beatles, basing it on the default -- New Collection --. (Basing it on the existing Advanced Beatles collection would make your life far easier, but we want you to learn how to build it from scratch!)

  1. Copy the files and folders provided in

    sample_files → beatles → advbeat_small

    into your new collection. Do this by opening up advbeat_small, selecting the eight items within it (from discography to, and dragging them across. Because some of these files are in MP3 and MARC formats you will be asked whether to include MP3Plugin and MARCPlugin in your collection. Click <Add Plugin>.

    A window may pop up explaining that the import documents contain css files, which none of Greenstone's plugins are expected to process directly. CSS files normally belong to a web page and we don't need to process them directly. Click <OK> button.

  1. Change to the Enrich panel and browse around the files. There is no metadata—yet. Recall that you can double-click files to view them.

    (There are no MIDI files in the collection: these require more advanced customisation because there is no MIDI plugin. We will deal with them later.)

  1. Change to the Create panel and build the collection.

  1. Preview the result.

Manually correcting metadata

  1. You might want to correct some of the metadata—for example, the atrocious misspelling in the titles "MAGICAL MISTERY TOUR." These documents are in the discography section, with filenames that contain the same misspelling. Locate one of them in the Enrich panel. Notice that the extracted metadata element ex.Title is now filled in, and misspelt. You cannot correct this element, for it is extracted from the file and will be re-extracted every time the collection is re-built.

  1. Instead, add dc.Title metadata for these two files: "Magical Mystery Tour." In the Enrich panel, open the discography folder and drill down to the individual files. Set the dc.Title value for the two offending items.

  1. Build the collection again, and preview it.

    Extracted metadata is unreliable. But it is very cheap! On the other hand, manually assigned metadata is reliable, but expensive. The previous section of this exercise has shown how to aim for the best of both worlds by using extracted metadata but correcting it when it is wrong.

Browsing by media type

  1. First let's remove the List classifier for filenames, which isn't very useful, and replace it with a browsing structure that groups documents by category (discography, lyrics, audio etc.). Categories are defined by manually assigned metadata.

    Build the collection again and preview it.

Note how we assigned dc.Format metadata to all documents in the collection with a minimum of labour. We did this by capitalizing on the folder structure of the original information. Even though we complained earlier about how messy this folder structure is, you can still take advantage of it when assigning metadata.

Suppressing dummy text

  1. Alongside the Audio files there is an MP3 icon, which plays the audio when you click it, and also a text document that contains some dummy text. Image files also have dummy documents. These dummy documents aren't supposed to be seen, but to suppress them you have to fiddle with a format statement.

    To make this easier for you we have prepared a plain text file that contains the new text. In WordPad open the following file:

    sample_files → beatles → format_tweaks → audio_tweak.txt

    (Be sure to use WordPad rather than Notepad, because Notepad does not display the line breaks correctly.) Place it in the copy buffer by highlighting the text in WordPad and selecting Edit → Copy. Now move back to the Librarian Interface, highlight all the text that makes up the current VList format statement, and use EditPaste (ctrl-v) to transform the old statement to the new one.

    Preview the result. You may need to click the browser's <Reload> button to force it to re-load the page.

  1. While we're at it, let's remove the source filename from where it appears after each document.

    Preview the result (you don't need to rebuild the collection.)

Using AZCompactList rather than List

  1. There are sometimes several documents with the same title. For example, All My Loving appears both as lyrics and tablature (under ALL MY LOVING). The Titles browser might be improved by grouping these together under a bookshelf icon. This is a job for an AZCompactList. In a previous tutorial we showed how to use the bookshelf_type option in List classifier to group documents with the same metadata value (dc.Format in that case) in one bookshelf. Here we use AZCompactList instead.

    Build the collection again and preview it. Both items for All My Loving now appear under the same bookshelf. However, many entries haven't been amalgamated because of non-uniform titles: for example A Hard Day's Night appears as several different variants. We will learn below how to amalgamate these.

Making bookshelves show how many items they contain

  1. Make the bookshelves show how many documents they contain by inserting a line in the VList format statement in the Format Features section of the Format panel. The added line is shown highlighted below. The complete format statement can be copied from sample_files → beatles → format_tweaks → show_num_docs.txt.

    <td valign=top>
    {If}{[dc.Format] eq 'Audio',
    {If}{[dc.Format] eq 'Images',
    {If}{[dc.Format] eq 'Supplementary', [srclink][srcicon][/srclink] [link][icon][/link], [link][icon][/link]}}}</td>
    <td valign=top>[highlight]

    Preview the result (you don't need to build the collection.) Bookshelves in the titles and browse classifiers should show how many documents they contain.

Adding a Phind phrase browser

  1. In the Browsing Classifiers section on the Design panel, add a Phind classifier. Leave the settings at their defaults: this generates a phrase browsing classifier that sources its phrases from Title and text.

    Build the collection again and preview it. Select the new Phrases option from the navigation bar. Enter a single word in the text box, such as band. The phrase browser will present you with phrases found in the collection containing the search term. This can provide a useful way of browsing a very large collection. Note that even though it is called a phrase browser, only single terms can be used as the starting point for browsing.

Branding the collection with an image

  1. To complete the collection, lets give it a new image for the top left corner of the page. Go to the General section of the Format panel. Use the browse button of URL to 'about page' image: to select the following image:

    sample_files → beatles → advbeat_large → images → beatlesmm.png

    Preview the collection, and make sure the new image appears.

Using UnknownPlugin

In this section we incorporate the MIDI files. Greenstone has no MIDI plugin (yet). But that doesn't mean you can't use MIDI files!

  1. UnknownPlugin is a useful generic plugin. It knows nothing about any given format but can be tailored to process particular document types—like MIDI—based on their filename extension, and set basic metadata.

    In the Document Plugins section of the Design panel:

    In this collection, all MIDI files are contained in the file ZIPPlugin (already in the list of default plugins) is used to unpack the files and pass them down the list of plugins until they reach UnknownPlugin.

  1. Build the collection and preview it. Unfortunately, the MIDI files don't appear as Audio under the browse button. That's because they haven't been assigned dc.Format metadata.

Cleaning up a title browser using regular expressions

We now clean up the Titles browser.

  1. We are going to use the removesuffix classifier option. The aim is to amalgamate variants of titles by stripping away extraneous text. For example, we would like to treat "ANTHOLOGY 1", "ANTHOLOGY 2" and "ANTHOLOGY 3" the same for grouping purposes. To achieve this:

    Build the collection and preview the result. Observe how many more times similar titles have been amalgamated under the same bookshelf. Test your understanding of regular expressions by trying to rationalize the amalgamations. (Note: [[:punct:]] stands for any punctuation character.) The icons beside the Word and PDF documents are not the correct ones, but that will be fixed in the next format statement.

One powerful use of regular expressions in the exercise was to clean up the Titles browser. Perhaps the best way of doing this would be to have proper title metadata. The metadata extracted from HTML files is messy and inconsistent, and this was reflected in the original Titles browser. Defining proper title metadata would be simple but rather laborious. Instead, we have opted to use regular expressions in the AZCompactList classifier to clean up the title metadata. This is difficult to understand, and a bit fiddly to do, but if you can cope with its idiosyncrasies it provides a quick way to clean up the extracted metadata and avoid having to enter a large amount of metadata.

Using non-standard macro files

To put finishing touches to our collection, we add some decorative features

  1. Close the collection in the Librarian Interface (FileClose).

  1. Using your file browser outside Greenstone, locate the folder

    sample_files → beatles → advbeat_large

  1. Open up another file browser, and locate the small beatles collection in your Greenstone installation:

    Greenstone → collect → smallbea

    smallbea is the folder name generated by Greenstone for this collection. You can determine what the folder name is for a collection by looking at the title bar of the Librarian Interface: the folder name is displayed in brackets after the collection name.

  1. Using the file browser, copy the images and macros folders from the advbeat_large folder into the smallbea folder. (It's OK to overwrite the existing images folder: the image in it is included in the folder being copied.) The images folder includes some useful icons, and the macros folder defines some macro names that use these images.

    To see the macro definitions, open the collection in the Librarian Interface (FileOpen...) and view the Collection Specific Macros section in the Format panel.

Using different icons for different media types

  1. Open the collection in GLI again and update VList your format statement (in Format Features on the Format panel) to be the following. You can copy this text from the file sample_files → beatles → format_tweaks → multi_icons.txt.

    <td valign=top>
      {If}{[dc.Format] eq 'Lyrics',[link]_iconlyrics_[/link]}
      {If}{[dc.Format] eq 'Discography',[link]_icondisc_[/link]}
      {If}{[dc.Format] eq 'Tablature',[link]_icontab_[/link]}
      {If}{[dc.Format] eq 'MARC',[link]_iconmarc_[/link]}
      {If}{[dc.Format] eq 'Images',[srclink][thumbicon][/srclink]}
      {If}{[dc.Format] eq 'Supplementary',[srclink][srcicon][/srclink]}
      {If}{[dc.Format] eq 'Audio',[srclink]{If}{[FileFormat] eq 'MIDI',_iconmidi_,_iconmp3_}[/srclink]}
    <td valign=top>

  1. Preview your collection as before. Now different icons are used for discography, lyrics, tablature, and MARC metadata. Even MP3 and MIDI audio file types are distinguished. If you let the mouse hover over one of these images a "tool tip" appears explaining what file type the icon represents in the current interface language (note: only defines English and French).

Changing the collection's background image

  1. Go to the Collection Specific Macros section in the Format panel.

  1. The content is fairly brief, specifying only what needs to be overridden from the default behaviour for this collection. Near the top you should see:

    _collectionspecificstyle_ {
    body.bgimage \{ background-image: url("_httpcimages_/beat_margin.gif"); \}
    \#page \{ margin-left: 120px; \}

    Replace the text beat_margin.gif with tile.jpg.

    This line relates to the background image used. The new image tile.jpg was in the images folder that was copied across previously.

  1. Preview the collection's home page. The page background is now the new graphic.

    Other features can be altered by editing the macros—for example, the headers and footers used on each page, and the highlighting style used for search terms (specify a different colour, use bold etc.).

Building a full-size version of the collection

  1. To finish, let's now build a larger version of the collection. To do this:

Adding an image collage browser

  1. Switch to the Design panel and select the Browsing Classifiers section. Pull down the Select classifier to add menu and select Collage. Click <Add Classifier...>. There is no need to customize the options, so click <OK> at the bottom of the resulting popup.

  1. Now change to the Create panel and build and preview the collection. Try out the collage browsing classifier. You can click on any image during the collage display and the image will be opened up.

Copyright © 2005-2016 by the New Zealand Digital Library Project at the University of Waikato, New Zealand
Permission is granted to copy, distribute and/or modify this document under the terms of the GNU Free Documentation License, Version 1.2 or any later version published by the Free Software Foundation; with no Invariant Sections, no Front-Cover Texts, and no Back-Cover Texts. A copy of the license is included in the section entitled “GNU Free Documentation License.”