# This patch file was generated by NetBeans IDE
# This patch can be applied using context Tools: Apply Diff Patch action on respective folder.
# It uses platform neutral UTF-8 encoding.
# Above lines and this line are ignored by the patching process.
Index: contributions/modules/millennium/millennium.admin.inc
--- contributions/modules/millennium/millennium.admin.inc Base (1.1.2.21)
+++ contributions/modules/millennium/millennium.admin.inc Locally Modified (Based On 1.1.2.21)
@@ -298,6 +298,7 @@
'#options' => array(
'list' => t('Record number listing'),
'range' => t('Record number range'),
+ 'query' => t('Enter a keyword query'),
'existing' => t('All already-imported records (@num items)', array("@num" => $tot_imported_nodes)),
'test' => t('Run predefined tests against some known OPACs'),
),
@@ -361,6 +362,25 @@
'#size' => 7,
'#title' => t('Ending number'),
);
+ $form['query'] = array(
+ '#type' => 'fieldset',
+ '#title' => t('Keyword query'),
+ '#attributes' => array('id' => 'query-fieldset'),
+ );
+ $form['query']['kw'] = array(
+ '#type' => 'textfield',
+ '#default_value' => variable_get('millennium_import_form_kw', ''),
+ '#size' => 80,
+ '#title' => t('Keywords to search for'),
+ '#description' => t('Type in keywords as you would type them in the WebOPAC keyword search. Some lesser-known available fields are: mattype:mattype[letter] branch:branch[locationcode]')
+ );
+ $form['query']['num_items'] = array(
+ '#type' => 'textfield',
+ '#default_value' => variable_get('millennium_import_form_num_items', 10),
+ '#size' => 5,
+ '#title' => t('Maximum number of items to import'),
+ '#description' => t('Enter the number of items to import.')
+ );
$form['force_update'] = array(
'#type' => 'radios',
'#title' => t('Import action'),
@@ -402,7 +422,8 @@
function millennium_admin_queue_validate($form, &$form_state) {
if ($form_state["values"]["source"] == "range" ||
- $form_state["values"]["source"] == "range") {
+ $form_state["values"]["source"] == "list" ||
+ $form_state["values"]["source"] == "query") {
$base_url = $form_state['values']['millennium_webopac_baseurl'];
if (!$base_url) {
form_set_error('millennium_webopac_baseurl', t("Not a valid base URL."));
@@ -448,7 +469,16 @@
);
}
}
+
+ if ($form_state["values"]["source"] == "query") {
+ if (!$form_state['values']['kw']) {
+ form_set_error('kw', t("You must enter a query."));
}
+ if (($form_state['values']['num_items']+0) <= 0) {
+ form_set_error('num_items', t("You must enter a number greater than 0."));
+ }
+ }
+}
/**
* Submit function for millennium_admin_queue() form
@@ -467,6 +497,8 @@
variable_set('millennium_range_form_type', $form_state["values"]["type"]);
variable_set('millennium_range_form_start', $form_state["values"]["start"]);
variable_set('millennium_range_form_end', $form_state["values"]["end"]);
+ variable_set('millennium_import_form_kw', $form_state["values"]["kw"]);
+ variable_set('millennium_import_form_num_items', $form_state["values"]["num_items"]);
if ($form_state["clicked_button"]["#value"] == t("Process these records during cron runs")) {
@@ -569,6 +601,19 @@
}
}
+ if ($form_state["values"]["source"] == "query") {
+ require_once 'millennium.import.inc';
+ $base_url = $form_state['values']['millennium_webopac_baseurl'];
+ $records = millennium_query_scrape(
+ $form_state["values"]["kw"],
+ $base_url,
+ $form_state["values"]["num_items"]
+ );
+ foreach ($records as $recnum) {
+ $records_to_crawl[$base_url][$recnum] = $recnum;
+ }
+ }
+
if ($form_state["values"]["source"] == "existing") {
$result = db_query("SELECT bib_recnum, base_url FROM {millennium_node_bib} ORDER BY base_url");
while ($data = db_fetch_object($result)) {
Index: contributions/modules/millennium/millennium.import.inc
--- contributions/modules/millennium/millennium.import.inc Base (1.1.2.14)
+++ contributions/modules/millennium/millennium.import.inc Locally Modified (Based On 1.1.2.14)
@@ -2,6 +2,85 @@
// $Id: millennium.import.inc,v 1.1.2.14 2010/02/17 20:36:06 janusman Exp $
/**
+ * Recieves a keyword query, and returns scraped record numbers to later harvest
+ * via other functions like millennium_mass_fetch().
+ *
+ * @param $query
+ * a string with a keyword query, same as that typed on the /search/X URL on
+ * the WebOPAC.
+ * @param $base_url
+ * @param $limit
+ * Maximum number of records to return.
+ */
+function millennium_query_scrape($query, $base_url = false, $limit = 100) {
+ $records = array();
+ $headers = array();
+
+ if ($base_url == false) {
+ $base_url = millennium_get_real_baseurl();
+ } else {
+ $base_url = millennium_get_real_baseurl($base_url);
+ }
+
+ // TODO: Cache the session cookie!
+ // Build query URL
+ $path = "/search/X?" . drupal_urlencode($query);
+
+ // This will loop while more pages of results are found
+ while (TRUE) {
+ $url = $base_url . $path;
+
+ // Launch
+ $result = drupal_http_request($url, $headers);
+ if ($result->code != 200) {
+ break;
+ }
+
+ // If no sesion cookie yet, store it for next calls
+ if (empty($headers['Cookie'])) {
+ if (preg_match_all('/(SESSION_SCOPE=[0-9]+|III_SESSION_ID=[a-zA-Z0-9\.]+)/', $result->headers["Set-Cookie"], $matches)) {
+ $headers['Cookie'] = implode("; ", $matches[1]);
+ }
+ }
+
+ // Look for
+ $ok = preg_match_all(
+ '/name="save" value="(b[0-9]+)"[^>]+>/si',
+ $result->data,
+ $matches
+ );
+ if (!$ok) {
+ // Try second regexp: cart button
+ $ok = preg_match_all(
+ '/\?save=(b[0-9]+)/si',
+ $result->data,
+ $matches
+ );
+ }
+ if (!$ok) {
+ break;
+ }
+ $records = array_merge($records, $matches[1]);
+ if (sizeof($records) >= $limit) {
+ break;
+ }
+
+ // TODO: Cycle thru pages!
+ #dpm($result);
+
+ // Look for "next" pager
+ $ok = preg_match('/| [0-9]+<\/strong>[^<]*([0-9]+)/si', $result->data, $pager_link_matches);
+ #dpm($pager_link_matches);
+ if (!$ok) {
+ break;
+ }
+ // Set path for next call from "Next" pager
+ $path = $pager_link_matches[1];
+ }
+ return array_splice($records, 0, $limit);
+}
+
+/**
* Gets a sequential number of records obeying PHP's max_execution_time setting.
* @param $recnums
* Array of item or bib numbers to fetch.
Index: contributions/modules/millennium/millennium_admin.js
--- contributions/modules/millennium/millennium_admin.js Base (1.1.2.2)
+++ contributions/modules/millennium/millennium_admin.js Locally Modified (Based On 1.1.2.2)
@@ -10,6 +10,7 @@
$("#edit-offline-refresh-wrapper").hide();
$("input[@name=\'offline_refresh\']").change();
$("#baseurl-table").show();
+ $("#query-fieldset").hide();
$("#import-options").show();
$("#edit-using-cron").show();
}
@@ -20,14 +21,25 @@
$("#options-fieldset").show();
$("#edit-using-cron").show();
$("#baseurl-table").show();
+ $("#query-fieldset").hide();
$("#import-options").show();
$("#edit-using-cron").show();
}
+ if (choice == "query") {
+ $("#range-fieldset").hide();
+ $("#list-fieldset").hide();
+ $("#edit-offline-refresh-wrapper").hide();
+ $("#baseurl-table").show();
+ $("#query-fieldset").show();
+ $("#import-options").show();
+ $("#edit-using-cron").hide();
+ }
if (choice == "existing") {
$("#range-fieldset").hide();
$("#list-fieldset").hide();
$("#edit-offline-refresh-wrapper").show();
$("#baseurl-table").hide();
+ $("#query-fieldset").hide();
$("#import-options").show();
$("#edit-using-cron").show();
}
@@ -36,6 +48,7 @@
$("#list-fieldset").hide();
$("#edit-offline-refresh-wrapper").hide();
$("#baseurl-table").hide();
+ $("#query-fieldset").hide();
$("#import-options").hide();
$("#edit-using-cron").hide();
}
@@ -44,6 +57,7 @@
$("#list-fieldset").hide();
$("#edit-offline-refresh-wrapper").show();
$("#baseurl-table").hide();
+ $("#query-fieldset").hide();
\ No newline at end of file
$("#import-options").show();
$("#edit-using-cron").hide();
}
|