making IPTC data searchable

Question

I have a question about IPTC metadata. Is it possible to search images that aren\'t in a database by their IPTC metadata (keywords) and show them and how would I go about do

Answer 1

It is not clear what in particular is giving you problems, but perhaps this will give you some ideas:

<?php
# Images we're searching
$images = array('/path/to/image.jpg', 'another-image.jpg');

# IPTC keywords to values (from exiv2, see below)
$query = array('Byline' => 'Some Author');

# Perform the search
$result = select_jpgs_by_iptc_fields($images, $query);

# Display the results
foreach ($result as $path) {
    echo '<img src="', htmlspecialchars($path), '">';
}

function select_jpgs_by_iptc_fields($jpgs, $query) {
    $matches = array();
    foreach ($jpgs as $path) {
        $iptc = get_jpg_iptc_metadata($path);
        foreach ($query as $name => $values) {
            if (!is_array($values))
                $values = array($values);
            if (count(array_intersect($iptc[$name], $values)) != count($values))
                continue 2;
        }
        $matches[] = $path;
    }
    return $matches;
}

function get_jpg_iptc_metadata($path) {
    $size = getimagesize($path, $info);
    if(isset($info['APP13']))
    {
        return human_readable_iptc(iptcparse($info['APP13']));
    }
    else {
        return null;
    }
}

function human_readable_iptc($iptc) {
# From the exiv2 sources
static $iptc_codes_to_names =
array(    
// IPTC.Envelope-->
"1#000" => 'ModelVersion',
"1#005" => 'Destination',
"1#020" => 'FileFormat',
"1#022" => 'FileVersion',
"1#030" => 'ServiceId',
"1#040" => 'EnvelopeNumber',
"1#050" => 'ProductId',
"1#060" => 'EnvelopePriority',
"1#070" => 'DateSent',
"1#080" => 'TimeSent',
"1#090" => 'CharacterSet',
"1#100" => 'UNO',
"1#120" => 'ARMId',
"1#122" => 'ARMVersion',
// <-- IPTC.Envelope
// IPTC.Application2 -->
"2#000" => 'RecordVersion',
"2#003" => 'ObjectType',
"2#004" => 'ObjectAttribute',
"2#005" => 'ObjectName',
"2#007" => 'EditStatus',
"2#008" => 'EditorialUpdate',
"2#010" => 'Urgency',
"2#012" => 'Subject',
"2#015" => 'Category',
"2#020" => 'SuppCategory',
"2#022" => 'FixtureId',
"2#025" => 'Keywords',
"2#026" => 'LocationCode',
"2#027" => 'LocationName',
"2#030" => 'ReleaseDate',
"2#035" => 'ReleaseTime',
"2#037" => 'ExpirationDate',
"2#038" => 'ExpirationTime',
"2#040" => 'SpecialInstructions',
"2#042" => 'ActionAdvised',
"2#045" => 'ReferenceService',
"2#047" => 'ReferenceDate',
"2#050" => 'ReferenceNumber',
"2#055" => 'DateCreated',
"2#060" => 'TimeCreated',
"2#062" => 'DigitizationDate',
"2#063" => 'DigitizationTime',
"2#065" => 'Program',
"2#070" => 'ProgramVersion',
"2#075" => 'ObjectCycle',
"2#080" => 'Byline',
"2#085" => 'BylineTitle',
"2#090" => 'City',
"2#092" => 'SubLocation',
"2#095" => 'ProvinceState',
"2#100" => 'CountryCode',
"2#101" => 'CountryName',
"2#103" => 'TransmissionReference',
"2#105" => 'Headline',
"2#110" => 'Credit',
"2#115" => 'Source',
"2#116" => 'Copyright',
"2#118" => 'Contact',
"2#120" => 'Caption',
"2#122" => 'Writer',
"2#125" => 'RasterizedCaption',
"2#130" => 'ImageType',
"2#131" => 'ImageOrientation',
"2#135" => 'Language',
"2#150" => 'AudioType',
"2#151" => 'AudioRate',
"2#152" => 'AudioResolution',
"2#153" => 'AudioDuration',
"2#154" => 'AudioOutcue',
"2#200" => 'PreviewFormat',
"2#201" => 'PreviewVersion',
"2#202" => 'Preview',
// <--IPTC.Application2
      );
   $human_readable = array();
   foreach ($iptc as $code => $field_value) {
       $human_readable[$iptc_codes_to_names[$code]] = $field_value;
   }
   return $human_readable;
}

Answer 2

If you don't have extracted those IPTC data from your images, each time someone will search, you'll have to :

• loop on every images
• for each image, extract the IPTC data
• see if the IPTC data for the current image matches

If you have more than a couple image, this will be really bad for performances, I'd say.

So, in my opinion, it would be far better to :

• add a couple of fields in your database
• extract the relevant IPTC data when the image is uploaded / stored
• store the IPTC data in those DB fields
• search in those DB fields
  • Or use some search engine like Lucene or Sphinx -- but that is another problem.

It'll mean a bit more work for you right now : you have more code to write...

... But it also means your website will have better chances to survive when there are several images and many users doing searches.