Integrate SearchWP with WP Document Revisions by indexing Revision content via SearchWP's Xpdf Integration extension (required) as a Custom Field named 'wp_document_revision'

gistfile1.php

<?php

// BEGIN WP Document Revisions/SearchWP integration (REQUIRES XPDF INTEGRATION EXTENSION)

// NOTE: after this snippet has been added to functions.php you will need to customize
// the Document weight on SearchWP's settings screen, adding a Custom Field with the
// name of 'wp_document_revision' and giving it the weight; that's where document
// content will be stored (as Custom Field content)

/**
 * This was taken directly from WPDR's main class because we can't instantiate a new object
 * Retrieves all revisions for a given post (including the current post)
 * Workaround for #16215 to ensure revision author is accurate
 * http://core.trac.wordpress.org/ticket/16215
 * @since 1.0
 * @param int $postID the post ID
 * @return array array of post objects
 */
function mywpdr_get_revisions( $postID ) {

	// Revision authors are actually shifted by one
	// This moves each revision author up one, and then uses the post_author as the initial revision

	//get the actual post
	$post = get_post( $postID );

	if ( !$post )
		return false;

	if ( $cache = wp_cache_get( $postID, 'document_revisions' ) )
		return $cache;

	//correct the modified date
	$post->post_date = date( 'Y-m-d H:i:s', (int) get_post_modified_time( 'U', null, $postID ) );

	//grab the post author
	$post_author = $post->post_author;

	//fix for Quotes in the most recent post because it comes from get_post
	$post->post_excerpt = html_entity_decode( $post->post_excerpt );

	//get revisions, and prepend the post
	$revs = wp_get_post_revisions( $postID, array( 'order' => 'DESC' ) );
	array_unshift( $revs, $post );

	//loop through revisions
	foreach ( $revs as $ID => &$rev ) {

		//if this is anything other than the first revision, shift author 1
		if ( $ID < sizeof( $revs ) - 1)
			$rev->post_author = $revs[$ID+1]->post_author;

		//if last revision, get the post author
		else
			$rev->post_author = $post_author;

	}

	wp_cache_set( $postID, $revs, 'document_revisions' );

	return $revs;

}

/**
 * This was taken directly from WPDR's main class because we can't instantiate a new object
 * Given a post object, returns all attached uploads
 * @since 0.5
 * @param object $post (optional) post object
 * @return object all attached uploads
 */
function mywpdr_get_attachments( $post = '' ) {

	if ($post == '')
		global $post;

	//verify that it's an object
	if ( !is_object( $post ) )
		$post = get_post( $post );

	//check for revisions
	if ( $parent = wp_is_post_revision( $post ) )
		$post = get_post( $parent );

	//check for attachments
	if ( $post->post_type == 'attachment' )
		$post = get_post( $post->post_parent );

	if ( !isset( $post->ID ) )
		return array();

	$args = array(
		'post_parent' => $post->ID,
		'post_status' => 'inherit',
		'post_type'   => 'attachment',
		'order'       => 'DESC',
		'orderby'     => 'post_date',
	);

	$args = apply_filters( 'document_revision_query', $args );

	return get_children( $args );

}

/**
 * This was taken directly from WPDR's main class because we can't instantiate a new object
 * Given a post ID, returns the latest revision attachment
 * @param unknown $post
 * @return object latest revision object
 */
function mywpdr_get_latest_revision( $post ) {

	if ( is_object( $post ) ) {
		$post = $post->ID;
	}

	$revisions = mywpdr_get_revisions( $post );

	if ( !$revisions )
		return false;

	//verify that there's an upload ID in the content field
	//if there's no upload ID for some reason, default to latest attached upload
	if ( !is_numeric( $revisions[0]->post_content ) ) {

		$attachments = mywpdr_get_attachments( $post );

		if ( empty( $attachments ) )
			return false;

		$latest_attachment = reset( $attachments );
		$revisions[0]->post_content = $latest_attachment->ID;

	}

	return $revisions[0];

}


/**
 * Tap in to the SearchWP indexing process and check if a WP Document Revisions post is being indexed. If it is, we're 
 * going to retrieve the most recent revision and extract the PDF content of that file. We'll store that data as a
 * pseudo Custom Field called wp_document_revision, allowing SearchWP to search for that content
 * 
 * @param $post_metadata
 * @param $post_to_index
 *
 * @return bool
 */
function my_index_wp_document_revisions( $post_metadata, $post_to_index ) {

	global $searchwp;

	// make sure it's a WP Document Revisions Document
	if( 'document' !== $post_to_index->post_type ) {
		return $post_metadata;
	}

	// make sure Xpdf Integration is installed and activated
	include_once( ABSPATH . 'wp-admin/includes/plugin.php' );
	if ( ! is_plugin_active( 'searchwp-xpdf/searchwp-xpdf.php' ) || ! class_exists( 'SearchWPXpdf' ) ) {
		return false;
	}

	// grab Xpdf and Document Revisions
	$xpdf = new SearchWPXpdf();

	// get the latest Revision
	$rev_id   = mywpdr_get_latest_revision( $post_to_index );
	$rev_post = get_post ( $rev_id );
	$revision = get_post( $rev_post->post_content );
	$filename = get_attached_file( $revision->ID );

	// grab the PDF content from Xpdf
	$pdf_content = $xpdf->extractPdfContent( '', $filename );

	// add it to the pseudo-metadata array
	$post_metadata[ 'wp_document_revision' ] = $pdf_content;

	return $post_metadata;
}

add_filter( 'searchwp_extra_metadata', 'my_index_wp_document_revisions', 10, 2 );
// END WP Document Revisions/SearchWP integration