function anno_validate_on_save($post_id, $post)
{
remove_action('save_post_article', 'anno_validate_on_save', 999, 2);
$error = false;
$schema = trailingslashit(get_template_directory()) . 'functions/schema/kipling-jp3-partial.rng';
$body_content = anno_validation_prep_body($post->post_content_filtered, $post_id);
$body_validation = anno_validate($body_content, $schema);
$abstract_content = anno_validation_prep_abstract($post->post_excerpt);
$abstract_validation = anno_validate($abstract_content, $schema);
if (isset($body_validation['status']) && $body_validation['status'] == 'error') {
$error = true;
}
if (isset($abstract_validation['status']) && $abstract_validation['status'] == 'error') {
$error = true;
}
if ($error && $post->post_status == 'publish') {
$post->post_status = 'draft';
if (anno_workflow_enabled()) {
$status = 'pending';
update_post_meta($post_id, '_post_state', 'approved');
} else {
$status = 'draft';
}
add_filter('redirect_post_location', 'anno_validation_redirect_post_location_message');
remove_action('post_updated', 'annowf_transistion_state', 10, 3);
remove_action('add_post_meta', 'anno_save_appendices_xml_as_html', 10, 3);
remove_filter('wp_insert_post_data', 'anno_insert_post_data', null, 2);
remove_filter('wp_insert_post_data', 'annowf_insert_post_data', 10, 2);
wp_update_post(array('ID' => $post_id, 'post_status' => 'draft'));
}
}
function parse($file)
{
global $wp_filesystem;
$authors = $posts = $attachments = $post = $author_snapshots = $authors_meta = array();
$file_content = $wp_filesystem->get_contents($file);
if (!$file_content) {
return new WP_Error('xml_parse_error', __('There was an error when reading this Kipling DTD file', 'anno'));
}
$schema = trailingslashit(get_template_directory()) . 'functions/schema/kipling-jp3.rng';
$validation_results = anno_validate($file_content, trailingslashit(get_template_directory()) . 'functions/schema/kipling-jp3.rng');
if ($validation_results['status'] == 'error') {
$validation_results['content'] = $file_content;
return $validation_results;
} else {
if (!class_exists('phpQueryObject')) {
require trailingslashit(TEMPLATEPATH) . 'functions/phpQuery/phpQuery.php';
}
phpQuery::newDocumentXML($file_content);
// Made up post IDs just for sanities sake, and parent relationship
$post_id = 1;
$articles = pq('article');
// Lets make sure we have article tags
$num_articles = $articles->length();
if (empty($num_articles)) {
return new WP_Error('xml_parse_article_error', __('This does not appear to be a Kipling DTD file, no articles could be found', 'anno'));
}
// Process articles, this contains all catergory, tag, author, term etc... processing.
foreach ($articles as $article) {
$article = pq('article');
$article_meta = pq('article-meta', $article);
$article_back = pq('back', $article);
$post['post_type'] = 'article';
$post['post_content_filtered'] = trim(pq('> body', $article)->html());
$post['post_title'] = trim(pq('article-title', $article_meta)->text());
$post['postmeta'][] = array('key' => '_anno_subtitle', 'value' => trim(pq('subtitle', $article_meta)->text()));
// Auto generated
$post['guid'] = '';
$abstract = pq('abstract', $article_meta);
// We don't want the title of the abstract
pq('title', $abstract)->remove();
// Just the text, wpautop is run on it later (excerpt)
$post['post_excerpt'] = trim($abstract->text());
// Post content gets generated by Annotum theme from the XML on wp_insert_post. We can leave it empty for now.
$post['post_content'] = '';
$post['post_id'] = $post_id;
// Generated from post title on insert
$post['post_name'] = '';
$pub_date = pq('pub-date', $article_meta);
$pub_date = $this->parse_date($pub_date);
$post['post_date'] = (string) $pub_date;
$post['post_date_gmt'] = (string) $pub_date;
$post['status'] = 'draft';
// Reflect in post_state meta as well.
$post['postmeta'][] = array('key' => '_post_state', 'value' => 'draft');
// Not used in Kipling DTD, but set for data structure integrity required by the importer
$post['post_parent'] = 0;
$post['menu_order'] = 0;
$post['post_password'] = '';
$post['is_sticky'] = 0;
$post['ping_status'] = '';
$post['comment_status'] = '';
// Grab the category(ies). Annotum DTD should contain only one, Kipling DTD does not contain this requirement.
foreach (pq('subject', $article_meta) as $category) {
$category = pq($category);
// We really don't care about the global categories, categories aren't defined outside of an article in the XML
$cat_name = trim($category->text());
if (!empty($cat_name)) {
$post['terms'][] = array('name' => $cat_name, 'slug' => sanitize_title($cat_name), 'domain' => 'article_category');
}
}
// Grab the tags.
foreach (pq('kwd', $article_meta) as $tag) {
$tag = pq($tag);
// We really don't care about the global tags, tags aren't defined outside of an article in the XML
$tag_name = trim($tag->text());
if (!empty($tag_name)) {
$post['terms'][] = array('name' => $tag_name, 'slug' => sanitize_title($tag_name), 'domain' => 'article_tag');
}
}
// First author is the primary author, possible @todo - look for primary-author contrib type
$first_author_check = true;
$default_author_id = $first_author_id = 1;
// Grab the author(s).
$authors = array();
foreach (pq('contrib', $article_meta) as $contributor) {
$contributor = pq($contributor);
$author_arr = $this->parse_author($contributor);
$author = $author_arr['author'];
$author_meta = $author_arr['author_meta'];
// Check for author_id existance, if not, assign one.
if (empty($author['author_id'])) {
$author['author_id'] = $default_author_id;
}
// Save in authors
$authors[$author['author_id']] = $author;
// Save in authors_meta, consistant with author_id to match on import of user
$authors_meta[$author['author_id']] = $author_meta;
if ($first_author_check) {
$post['post_author'] = $author['author_id'];
}
$author_snapshots[] = $this->author_snapshot($author, $author_meta);
if ($first_author_check) {
// Used in attachment assignment
$first_author_id = $author['author_id'];
}
// We'll convert this in the import process
$post['postmeta'][] = array('key' => '_anno_author_' . $author['author_id'], 'value' => $author['author_id']);
$first_author_check = false;
$default_author_id++;
}
// Acknowledgements
$ack = trim(pq('ack p', $article_back)->text());
if (!empty($ack)) {
$post['postmeta'][] = array('key' => '_anno_acknowledgements', 'value' => $ack);
}
// Funding
$funding = trim(pq('funding-statement', $article_meta)->text());
if (!empty($funding)) {
$post['postmeta'][] = array('key' => '_anno_funding', 'value' => $funding);
}
// Appendices
$appendices = pq('app', $article_back);
$appendix_array = array();
foreach ($appendices as $appendix) {
$appendix = trim(pq($appendix)->html());
if (!empty($appendix)) {
$appendix_array[] = $appendix;
}
}
if (!empty($appendix_array)) {
// Process to HTML on import
$post['postmeta'][] = array('key' => '_anno_appendices', 'value' => serialize($appendix_array));
}
// References
$references = pq('ref', $article_back);
$ref_array = array();
$single_ref = array('doi' => '', 'text' => '', 'pmid' => '', 'figures' => '', 'url' => '');
foreach ($references as $reference) {
$reference = pq($reference);
// For now, just support mixed-citations as text.
$ref_id = str_replace('ref', '', $reference->attr('id'));
// Only store numeric values
if (is_numeric($ref_id)) {
$ref_id = intval($ref_id) - 1;
} else {
$ref_id = null;
}
$ref_text = pq('mixed-citation', $reference);
$ref_data['text'] = trim($ref_text->text());
if (empty($ref_id)) {
$ref_array[] = $ref_data;
} else {
// Possibility that this key was already set programmatically, replace it and add old ref to end.
if (isset($ref_array[$ref_id])) {
$old_ref = $ref_array[$ref_id];
$ref_array[$ref_id] = $ref_data;
$ref_array[] = $old_ref;
} else {
$ref_array[$ref_id] = $ref_data;
}
}
}
if (!empty($ref_array)) {
$post['postmeta'][] = array('key' => '_anno_references', 'value' => serialize($ref_array));
}
// Attachments
// Modification for post_id
$attachment_id_mod = 0;
// $pub_date is the date gathered from the post data.
$attachment_template = array('upload_date' => (string) $pub_date, 'post_date' => (string) $pub_date, 'post_date_gmt' => (string) $pub_date, 'post_author' => $first_author_id, 'post_type' => 'attachment', 'post_parent' => $post_id, 'post_id' => '', 'post_content' => '', 'post_content_filtered' => '', 'postmeta' => '', 'guid' => '', 'attachment_url' => '', 'status' => 'inherit', 'post_title' => '', 'ping_status' => '', 'menu_order' => '', 'post_password' => '', 'terms' => '', 'comment_status' => '', 'is_sticky' => '', 'post_excerpt' => '', 'post_name' => '');
$inline_images = pq('> body inline-graphic', $article);
foreach ($inline_images as $img) {
$img = pq($img);
$img_url = $img->attr('xlink:href');
// Dont save chart api images (most likely formulas)
if (!empty($img_url) && strpos($img_url, 'googleapis.com/chart') === false) {
$post_meta = array();
$alt_text = pq('alt-text', $img)->html();
if (!empty($alt_text)) {
$post_meta[] = array('key' => '_wp_attachment_image_alt', 'value' => $alt_text);
}
$attachment_title = !empty($alt_text) ? $alt_text : end(explode('/', $img_url));
$attachments[] = array_merge($attachment_template, array('post_id' => $post_id . '.' . $attachment_id_mod, 'guid' => $img_url, 'attachment_url' => $img_url, 'post_parent' => $post_id, 'title' => trim($attachment_title), 'postmeta' => $post_meta, 'post_title' => $img_url));
$attachment_id_mod++;
}
}
// Find media and save as attachment
$media_images = pq('> body media', $article);
foreach ($media_images as $media_image) {
$media_image = pq($media_image);
// Parse Media will return an array with:
// attachment_url
// guid
// post_title
// post_content
// postmeta
$media_array = $this->parse_media($media_image);
if (is_array($media_array) && !empty($media_array['attachment_url'])) {
// Check if this is a figure image
$figure = $media_image->parent('fig');
$figure_html = trim($figure->html());
$caption = '';
if (!empty($figure_html)) {
$label = pq('label', $figure)->html();
$caption = pq('caption', $figure)->html();
$post_meta[] = array('key' => '_anno_attachment_image_label', 'value' => $label);
}
$attachment = array_merge($media_array, array('post_id' => $post_id . '.' . $attachment_id_mod, 'post_parent' => $post_id, 'postmeta' => array_merge($post_meta, $media_array['postmeta'])));
$attachments[] = array_merge($attachment_template, $attachment);
$attachment_id_mod++;
}
}
$comments = pq('response');
foreach ($comments as $comment) {
$comment = pq($comment);
$comment_content = pq('body', $comment)->html();
$comment_date = $this->parse_date(pq('pub-date', $comment));
$comment_author_arr = $this->parse_author(pq('contrib', $comment));
$comment_author = $comment_author_arr['author'];
$post['comments'][] = array('comment_id' => '', 'comment_author' => (string) $comment_author['author_display_name'], 'comment_author_email' => (string) $comment_author['author_email'], 'comment_author_IP' => '', 'comment_author_url' => (string) $comment_author['author_url'], 'comment_date' => (string) $comment_date, 'comment_date_gmt' => '', 'comment_content' => $comment_content, 'comment_approved' => 1, 'comment_type' => '', 'comment_parent' => '', 'comment_user_id' => 0, 'commentmeta' => array());
}
// Save our author snapshots
$post['postmeta'][] = array('key' => '_anno_author_snapshot', 'value' => serialize($author_snapshots));
$posts[] = $post;
// Concat, both indexed
$posts = array_merge($posts, $attachments);
}
return array('authors' => $authors, 'authors_meta' => $authors_meta, 'posts' => $posts, 'categories' => array(), 'tags' => array(), 'terms' => array(), 'base_url' => '', 'version' => 1.1);
}
}