diff --git a/src/library/CrawlDaemon.php b/src/library/CrawlDaemon.php index f1f2fab8d..6ac0ce9d1 100644 --- a/src/library/CrawlDaemon.php +++ b/src/library/CrawlDaemon.php @@ -252,7 +252,7 @@ class CrawlDaemon implements CrawlConstants $process_user_info = posix_getpwuid(posix_getuid()); $process_home = $process_user_info['dir']; if (C\nsdefined("FORCE_HHVM") || ( - stristr(phpversion(), "hhvm") !==false && + stristr(phpversion(), "hhvm") !== false && posix_access($process_home, POSIX_W_OK))) { $php = 'hhvm -f '; if (C\nsdefined("HHVM_PATH") ) { diff --git a/src/library/indexing_plugins/AddressesPlugin.php b/src/library/indexing_plugins/AddressesPlugin.php index def10ed56..3339059b3 100644 --- a/src/library/indexing_plugins/AddressesPlugin.php +++ b/src/library/indexing_plugins/AddressesPlugin.php @@ -142,6 +142,7 @@ class AddressesPlugin extends IndexingPlugin implements CrawlConstants "TANZANIA, UNITED REPUBLIC OF" => "TZ", "UKRAINE" => "UA","UGANDA" => "UG", "UNITED STATES MINOR OUTLYING ISLANDS" => "UM", + "UNITED STATES OF AMERICA" => "USA", "UNITED STATES" => "US","URUGUAY" => "UY","UZBEKISTAN" => "UZ", "VATICAN CITY" => "VA","SAINT VINCENT AND THE GRENADINES" => "VC", "VENEZUELA, BOLIVARIAN REPUBLIC OF" => "VE", @@ -395,7 +396,7 @@ class AddressesPlugin extends IndexingPlugin implements CrawlConstants } break; case "maybe": - if ($len_about_right){ + if ($len_about_right) { if ($num_lines < $max_lines) { $current_candidate[] = $line; $num_lines++; @@ -406,7 +407,8 @@ class AddressesPlugin extends IndexingPlugin implements CrawlConstants } } else { $state = "dont"; - if ($num_lines <= $max_lines&&$num_lines >= $min_lines){ + if ($num_lines <= $max_lines && + $num_lines >= $min_lines) { $current_candidate = $this->checkCandidate( $current_candidate); if ($current_candidate) { @@ -443,8 +445,8 @@ class AddressesPlugin extends IndexingPlugin implements CrawlConstants $address = false; $found_count = 0; $num_lines = count($pre_address); - $check_array = ["checkCountry"=>"checkCountry", - "checkStreet"=>"checkStreet", + $check_array = ["checkCountry" => "checkCountry", + "checkStreet" => "checkStreet", "checkPhoneOrEmail" => "checkPhoneOrEmail", "checkRegion" => "checkRegion", "checkZipPostalCodeWords" => "checkZipPostalCodeWords"]; @@ -563,7 +565,6 @@ class AddressesPlugin extends IndexingPlugin implements CrawlConstants $num_parts = count($line_parts); $line = mb_strtoupper(trim($line_parts[$num_parts - 1])); $countries = $this->countries; - $country_codes = array_flip($countries); if (strlen($line) == 2) { $line = substr($line, 0, 2); diff --git a/src/library/indexing_plugins/RecipePlugin.php b/src/library/indexing_plugins/RecipePlugin.php index 30e84a4fe..b351be772 100644 --- a/src/library/indexing_plugins/RecipePlugin.php +++ b/src/library/indexing_plugins/RecipePlugin.php @@ -58,11 +58,10 @@ require_once C\BASE_DIR . "/library/LocaleFunctions.php"; * looking at what was needed to screen scrape recipes from the * following sites: * - * http://allrecipes.com/ - * http://www.food.com/ + * https://allrecipes.com/ + * http://www.geniuskitchen.com/ * http://www.betterrecipes.com/ - * http://www.foodnetwork.com/ - * http://www.bettycrocker.com/ + * https://www.bettycrocker.com/ * * * @author Priya Gangaraju, Chris Pollett (re-organized, added documentation, @@ -130,25 +129,28 @@ class RecipePlugin extends IndexingPlugin implements CrawlConstants //detect recipes $recipes_per_page = $xpath->evaluate( /*allr, f.com, brec, fnet*/ - "/html//header[@class='recipe']/h1| - /html//*[@id='recipe_title']| - /html//*[@itemtype='http://schema.org/Recipe']"); + "//header[contains(@class, 'recipe-header')]/h1|" . + "/html//*[@id='recipe_title']|" . + "/html//*[@itemtype='http://schema.org/Recipe']"); $recipe = []; $subdocs_description = []; if (is_object($recipes_per_page) && $recipes_per_page->length != 0) { $recipes_count = $recipes_per_page->length; $titles = $xpath->evaluate( /* allr, f.com, brec, fnet */ - "/html//*[@id = 'recipe_title']| - /html//header[@class = 'recipe']/h1| - /html//*[@itemprop='name']"); + "/html//*[@id = 'recipe_title']|" . + "//header[contains(@class, 'recipe-header')]/h1|" . + "/html//*[@itemprop = 'name']"); + if ($titles->length == 0) { + return $subdocs_description; + } for ($i = 0; $i < $recipes_count; $i++) { $ingredients = $xpath->evaluate( /*allr*, fcomm, brec, fnet*/ - "/html//ul[@class = 'ingredient-wrap']/li | - /html//li[@data-ingredient]| - /html//*[@itemprop ='ingredient']| - /html//*[@itemprop='ingredients']"); + "/html//ul[@class = 'ingredient-wrap']/li|" . + "/html//li[@data-ingredient]|" . + "/html//*[@itemprop ='ingredient']|" . + "/html//*[@itemprop='ingredients']"); $ingredients_result = ""; if (is_object($ingredients) && $ingredients->length != 0){ $lastIngredient = end($ingredients); diff --git a/src/library/media_jobs/FeedsUpdateJob.php b/src/library/media_jobs/FeedsUpdateJob.php index 28f77214d..0f664d1d1 100644 --- a/src/library/media_jobs/FeedsUpdateJob.php +++ b/src/library/media_jobs/FeedsUpdateJob.php @@ -263,10 +263,16 @@ class FeedsUpdateJob extends MediaJob //strip namespaces $page = preg_replace('@<(/?)(\w+\s*)\:@u', '<$1', $feed[CrawlConstants::PAGE]); + if (empty($page)) { + $page = $feed[CrawlConstants::PAGE]; + } } if (isset($feed['IMAGE_XPATH']) && !$is_regex) { $feed['IMAGE_XPATH'] = preg_replace('@/(\s*\w+\s*)\:@u', '/', $feed['IMAGE_XPATH']); + if (empty($feed['IMAGE_XPATH'])) { + $feed['IMAGE_XPATH'] = $feed['IMAGE_XPATH']; + } } if ($is_html) { @$dom->loadHTML($page); diff --git a/src/locale/ar/configure.ini b/src/locale/ar/configure.ini index 6c3965dca..7afb07372 100755 --- a/src/locale/ar/configure.ini +++ b/src/locale/ar/configure.ini @@ -866,7 +866,6 @@ manageaccount_element_username = "" manageaccount_element_firstname = "" manageaccount_element_lastname = "" manageaccount_element_email = "" -manageaccount_element_enable_store = "" manageaccount_element_is_bot = "" manageaccount_element_bot_unique_token = "" manageaccount_element_bot_callback_url = "" diff --git a/src/locale/bn/configure.ini b/src/locale/bn/configure.ini index 68a7468fb..22371535a 100755 --- a/src/locale/bn/configure.ini +++ b/src/locale/bn/configure.ini @@ -866,7 +866,6 @@ manageaccount_element_username = "" manageaccount_element_firstname = "" manageaccount_element_lastname = "" manageaccount_element_email = "" -manageaccount_element_enable_store = "" manageaccount_element_is_bot = "" manageaccount_element_bot_unique_token = "" manageaccount_element_bot_callback_url = "" diff --git a/src/locale/de/configure.ini b/src/locale/de/configure.ini index 441748a3b..44abbd47d 100755 --- a/src/locale/de/configure.ini +++ b/src/locale/de/configure.ini @@ -866,7 +866,6 @@ manageaccount_element_username = "" manageaccount_element_firstname = "" manageaccount_element_lastname = "" manageaccount_element_email = "" -manageaccount_element_enable_store = "" manageaccount_element_is_bot = "" manageaccount_element_bot_unique_token = "" manageaccount_element_bot_callback_url = "" diff --git a/src/locale/en_US/configure.ini b/src/locale/en_US/configure.ini index 884460e7f..4d8ecbda9 100644 --- a/src/locale/en_US/configure.ini +++ b/src/locale/en_US/configure.ini @@ -832,17 +832,17 @@ botstory_element_delete_operation = "Do you really want to delete this pattern?" botstory_element_deletepattern = "Delete" ; ; ManagecreditsElement.php -managecredits_element_purchase_credits = "Purchase Ad Credits" +managecredits_element_purchase_credits = "Purchase Credits" managecredit_element_num_credits = "Quantity" managecredit_element_card_number = "Credit Card Number" managecredit_element_cvc = "CVC" managecredit_element_expiration = "Expiration" managecredits_element_charge_warning = "Using the Purchase button charges the above card the Quantity field's amount in US dollars and agrees to the " -managecredits_element_program_terms = "Ad Program Terms" +managecredits_element_program_terms = "Program Terms" managecredits_element_purchase = "Purchase" managecredits_element_script_failure = "No Working Credit Card Script Found!" managecredits_element_balance = "Balance: %s credits" -managecredits_element_transactions = "Ad Credit Transactions" +managecredits_element_transactions = "Credit Transactions" managecredits_element_type = "Type" managecredits_element_amount = "Amount" managecredits_element_date = "Date" @@ -866,7 +866,6 @@ manageaccount_element_username = "Username" manageaccount_element_firstname = "First Name" manageaccount_element_lastname = "Last Name" manageaccount_element_email = "Email" -manageaccount_element_enable_store = "Enable Store:" manageaccount_element_is_bot = "Bot User:" manageaccount_element_bot_unique_token = "Bot Unique Token" manageaccount_element_bot_callback_url = "Bot Callback URL" @@ -1441,7 +1440,7 @@ manageadvertisements_element_keyword_bid_amount = "Minimum Bid Required" manageadvertisements_element_expensive_word = "Expensive word" manageadvertisements_element_calculate_bid = "Calculate Bid" manageadvertisements_element_update = "Update" -manageadvertisements_element_buy_info = "Using the Purchase button deducts the Budget field amount from your %s available ad credits." +manageadvertisements_element_buy_info = "Using the Purchase button deducts the Budget field amount from your %s available credits." manageadvertisements_element_buy_credits = "Buy more ad credits" manageadvertisements_element_edit_ad = "Edit Ad" manageadvertisements_element_purchase = "Purchase" diff --git a/src/locale/es/configure.ini b/src/locale/es/configure.ini index 8c59fc6c9..57efd1d02 100755 --- a/src/locale/es/configure.ini +++ b/src/locale/es/configure.ini @@ -866,7 +866,6 @@ manageaccount_element_username = "" manageaccount_element_firstname = "" manageaccount_element_lastname = "" manageaccount_element_email = "" -manageaccount_element_enable_store = "" manageaccount_element_is_bot = "" manageaccount_element_bot_unique_token = "" manageaccount_element_bot_callback_url = "" diff --git a/src/locale/fa/configure.ini b/src/locale/fa/configure.ini index 962eca19c..c73c7a95b 100755 --- a/src/locale/fa/configure.ini +++ b/src/locale/fa/configure.ini @@ -866,7 +866,6 @@ manageaccount_element_username = "" manageaccount_element_firstname = "" manageaccount_element_lastname = "" manageaccount_element_email = "" -manageaccount_element_enable_store = "" manageaccount_element_is_bot = "" manageaccount_element_bot_unique_token = "" manageaccount_element_bot_callback_url = "" diff --git a/src/locale/fr_FR/configure.ini b/src/locale/fr_FR/configure.ini index d9e7f4133..09574c152 100755 --- a/src/locale/fr_FR/configure.ini +++ b/src/locale/fr_FR/configure.ini @@ -866,7 +866,6 @@ manageaccount_element_username = "" manageaccount_element_firstname = "" manageaccount_element_lastname = "" manageaccount_element_email = "" -manageaccount_element_enable_store = "" manageaccount_element_is_bot = "" manageaccount_element_bot_unique_token = "" manageaccount_element_bot_callback_url = "" diff --git a/src/locale/he/configure.ini b/src/locale/he/configure.ini index cc6c7c4ea..c8fa6b01b 100755 --- a/src/locale/he/configure.ini +++ b/src/locale/he/configure.ini @@ -866,7 +866,6 @@ manageaccount_element_username = "" manageaccount_element_firstname = "" manageaccount_element_lastname = "" manageaccount_element_email = "" -manageaccount_element_enable_store = "" manageaccount_element_is_bot = "" manageaccount_element_bot_unique_token = "" manageaccount_element_bot_callback_url = "" diff --git a/src/locale/hi/configure.ini b/src/locale/hi/configure.ini index edb69231b..5d9b1dba4 100755 --- a/src/locale/hi/configure.ini +++ b/src/locale/hi/configure.ini @@ -866,7 +866,6 @@ manageaccount_element_username = "" manageaccount_element_firstname = "" manageaccount_element_lastname = "" manageaccount_element_email = "" -manageaccount_element_enable_store = "" manageaccount_element_is_bot = "" manageaccount_element_bot_unique_token = "" manageaccount_element_bot_callback_url = "" diff --git a/src/locale/in_ID/configure.ini b/src/locale/in_ID/configure.ini index 92876bebf..2acc2dd67 100755 --- a/src/locale/in_ID/configure.ini +++ b/src/locale/in_ID/configure.ini @@ -866,7 +866,6 @@ manageaccount_element_username = "" manageaccount_element_firstname = "" manageaccount_element_lastname = "" manageaccount_element_email = "" -manageaccount_element_enable_store = "" manageaccount_element_is_bot = "" manageaccount_element_bot_unique_token = "" manageaccount_element_bot_callback_url = "" diff --git a/src/locale/it/configure.ini b/src/locale/it/configure.ini index bd17e7756..2004095e2 100755 --- a/src/locale/it/configure.ini +++ b/src/locale/it/configure.ini @@ -866,7 +866,6 @@ manageaccount_element_username = "" manageaccount_element_firstname = "" manageaccount_element_lastname = "" manageaccount_element_email = "" -manageaccount_element_enable_store = "" manageaccount_element_is_bot = "" manageaccount_element_bot_unique_token = "" manageaccount_element_bot_callback_url = "" diff --git a/src/locale/ja/configure.ini b/src/locale/ja/configure.ini index e17739e95..c07c80463 100755 --- a/src/locale/ja/configure.ini +++ b/src/locale/ja/configure.ini @@ -866,7 +866,6 @@ manageaccount_element_username = "" manageaccount_element_firstname = "" manageaccount_element_lastname = "" manageaccount_element_email = "" -manageaccount_element_enable_store = "" manageaccount_element_is_bot = "" manageaccount_element_bot_unique_token = "" manageaccount_element_bot_callback_url = "" diff --git a/src/locale/kn/configure.ini b/src/locale/kn/configure.ini index 0fba115e2..66abefcb7 100755 --- a/src/locale/kn/configure.ini +++ b/src/locale/kn/configure.ini @@ -866,7 +866,6 @@ manageaccount_element_username = "" manageaccount_element_firstname = "" manageaccount_element_lastname = "" manageaccount_element_email = "" -manageaccount_element_enable_store = "" manageaccount_element_is_bot = "" manageaccount_element_bot_unique_token = "" manageaccount_element_bot_callback_url = "" diff --git a/src/locale/ko/configure.ini b/src/locale/ko/configure.ini index ba58e6672..1734b5a6b 100755 --- a/src/locale/ko/configure.ini +++ b/src/locale/ko/configure.ini @@ -866,7 +866,6 @@ manageaccount_element_username = "" manageaccount_element_firstname = "" manageaccount_element_lastname = "" manageaccount_element_email = "" -manageaccount_element_enable_store = "" manageaccount_element_is_bot = "" manageaccount_element_bot_unique_token = "" manageaccount_element_bot_callback_url = "" diff --git a/src/locale/nl/configure.ini b/src/locale/nl/configure.ini index f7656da22..bf4a694da 100644 --- a/src/locale/nl/configure.ini +++ b/src/locale/nl/configure.ini @@ -866,7 +866,6 @@ manageaccount_element_username = "" manageaccount_element_firstname = "" manageaccount_element_lastname = "" manageaccount_element_email = "" -manageaccount_element_enable_store = "" manageaccount_element_is_bot = "" manageaccount_element_bot_unique_token = "" manageaccount_element_bot_callback_url = "" diff --git a/src/locale/pl/configure.ini b/src/locale/pl/configure.ini index b9bfaf355..42144e947 100755 --- a/src/locale/pl/configure.ini +++ b/src/locale/pl/configure.ini @@ -866,7 +866,6 @@ manageaccount_element_username = "" manageaccount_element_firstname = "" manageaccount_element_lastname = "" manageaccount_element_email = "" -manageaccount_element_enable_store = "" manageaccount_element_is_bot = "" manageaccount_element_bot_unique_token = "" manageaccount_element_bot_callback_url = "" diff --git a/src/locale/pt/configure.ini b/src/locale/pt/configure.ini index 234c4de17..a703975b6 100755 --- a/src/locale/pt/configure.ini +++ b/src/locale/pt/configure.ini @@ -866,7 +866,6 @@ manageaccount_element_username = "" manageaccount_element_firstname = "" manageaccount_element_lastname = "" manageaccount_element_email = "" -manageaccount_element_enable_store = "" manageaccount_element_is_bot = "" manageaccount_element_bot_unique_token = "" manageaccount_element_bot_callback_url = "" diff --git a/src/locale/ru/configure.ini b/src/locale/ru/configure.ini index c42d70bf5..345a881a0 100755 --- a/src/locale/ru/configure.ini +++ b/src/locale/ru/configure.ini @@ -866,7 +866,6 @@ manageaccount_element_username = "" manageaccount_element_firstname = "" manageaccount_element_lastname = "" manageaccount_element_email = "" -manageaccount_element_enable_store = "" manageaccount_element_is_bot = "" manageaccount_element_bot_unique_token = "" manageaccount_element_bot_callback_url = "" diff --git a/src/locale/te/configure.ini b/src/locale/te/configure.ini index 9334339f7..f7f7d2c51 100644 --- a/src/locale/te/configure.ini +++ b/src/locale/te/configure.ini @@ -866,7 +866,6 @@ manageaccount_element_username = "" manageaccount_element_firstname = "" manageaccount_element_lastname = "" manageaccount_element_email = "" -manageaccount_element_enable_store = "" manageaccount_element_is_bot = "" manageaccount_element_bot_unique_token = "" manageaccount_element_bot_callback_url = "" diff --git a/src/locale/th/configure.ini b/src/locale/th/configure.ini index b25150cce..bb94f8fbb 100755 --- a/src/locale/th/configure.ini +++ b/src/locale/th/configure.ini @@ -866,7 +866,6 @@ manageaccount_element_username = "" manageaccount_element_firstname = "" manageaccount_element_lastname = "" manageaccount_element_email = "" -manageaccount_element_enable_store = "" manageaccount_element_is_bot = "" manageaccount_element_bot_unique_token = "" manageaccount_element_bot_callback_url = "" diff --git a/src/locale/tr/configure.ini b/src/locale/tr/configure.ini index 322d8ac58..7488d07d0 100755 --- a/src/locale/tr/configure.ini +++ b/src/locale/tr/configure.ini @@ -866,7 +866,6 @@ manageaccount_element_username = "" manageaccount_element_firstname = "" manageaccount_element_lastname = "" manageaccount_element_email = "" -manageaccount_element_enable_store = "" manageaccount_element_is_bot = "" manageaccount_element_bot_unique_token = "" manageaccount_element_bot_callback_url = "" diff --git a/src/locale/vi_VN/configure.ini b/src/locale/vi_VN/configure.ini index c71f56873..5e46a5366 100755 --- a/src/locale/vi_VN/configure.ini +++ b/src/locale/vi_VN/configure.ini @@ -866,7 +866,6 @@ manageaccount_element_username = "" manageaccount_element_firstname = "" manageaccount_element_lastname = "" manageaccount_element_email = "" -manageaccount_element_enable_store = "" manageaccount_element_is_bot = "" manageaccount_element_bot_unique_token = "" manageaccount_element_bot_callback_url = "" diff --git a/src/locale/zh_CN/configure.ini b/src/locale/zh_CN/configure.ini index de9e9d386..1bf771e54 100755 --- a/src/locale/zh_CN/configure.ini +++ b/src/locale/zh_CN/configure.ini @@ -866,7 +866,6 @@ manageaccount_element_username = "" manageaccount_element_firstname = "" manageaccount_element_lastname = "" manageaccount_element_email = "" -manageaccount_element_enable_store = "" manageaccount_element_is_bot = "" manageaccount_element_bot_unique_token = "" manageaccount_element_bot_callback_url = ""