public function testMultipleEmptySections()
{
$r = new RecipeParser_Recipe();
// Empty values for ingredients or instructions should be ingored.
$r->addIngredientsSection('Pasta');
$r->appendIngredient('1 lb Spaghetti');
$r->addIngredientsSection('');
$r->addIngredientsSection(' ');
$r->appendIngredient('1 C Water');
$r->addInstructionsSection('');
$r->addInstructionsSection(' ');
$r->addInstructionsSection(' ');
$r->appendInstruction('Heat water in large pot.');
$this->assertEquals(2, count($r->ingredients));
$this->assertEquals(1, count($r->instructions));
}
public static function parse($html, $url)
{
$recipe = new RecipeParser_Recipe();
// Turn off libxml errors to prevent mismatched tag warnings.
libxml_use_internal_errors(true);
$html = mb_convert_encoding($html, 'HTML-ENTITIES', "UTF-8");
$doc = new DOMDocument();
$doc->loadHTML('<?xml encoding="UTF-8">' . $html);
$xpath = new DOMXPath($doc);
// Title
$node_list = $doc->getElementsByTagName('title');
if ($node_list->length) {
$value = $node_list->item(0)->nodeValue;
$value = trim(str_replace("Cooks.com - Recipe - ", "", $value));
$value = trim(str_replace(" - Recipe - Cooks.com", "", $value));
$recipe->title = $value;
}
// This node contains all ingredients, section titles, and instructions
$node_list = $xpath->query('//table[@class="hrecipe"]//td/div');
foreach ($node_list as $node) {
// Can determine each piece of content by the "style" attributes.
$style = $node->getAttribute("style");
// Ingredients found in a div, black text
if (stripos($style, "color: BLACK;") !== false) {
$ing_nodes = $xpath->query('./span[@class = "ingredient"]', $node);
foreach ($ing_nodes as $ing_node) {
$recipe->appendIngredient($ing_node->nodeValue);
}
// Instructions node
} else {
if ($node->getAttribute('class') == "instructions") {
foreach ($node->childNodes as $child) {
$line = $child->nodeValue;
$line = RecipeParser_Text::formatAsOneLine($line);
$recipe->appendInstruction($line);
}
// Section title
} else {
if ($node->getAttribute("class") == "section") {
$title = RecipeParser_Text::formatSectionName($node->nodeValue);
$recipe->addIngredientsSection($title);
if (count($recipe->instructions) > 0) {
$recipe->addInstructionsSection($title);
}
}
}
}
}
return $recipe;
}
public static function parse($html, $url)
{
$recipe = new RecipeParser_Recipe();
libxml_use_internal_errors(true);
$html = mb_convert_encoding($html, 'HTML-ENTITIES', "UTF-8");
$doc = new DOMDocument();
$doc->loadHTML('<?xml encoding="UTF-8">' . $html);
$xpath = new DOMXPath($doc);
// Title
$nodes = $xpath->query('//h1[@itemprop="name"]');
if ($nodes->length) {
$value = trim($nodes->item(0)->nodeValue);
$recipe->title = $value;
}
// Times and yield
// <time datetime="PT35M" itemprop="prepTime">
$nodes = $xpath->query('//time[@itemprop="prepTime"]');
if ($nodes->length) {
if ($value = $nodes->item(0)->textContent) {
$value = RecipeParser_Text::mixedTimeToMinutes($value);
$recipe->time['total'] = $value;
}
}
$nodes = $xpath->query('//*[@itemprop="recipeYield"]');
if ($nodes->length) {
$value = $nodes->item(0)->nodeValue;
$recipe->yield = RecipeParser_Text::formatYield($value);
}
// Ingredients
$nodes = $xpath->query('//*[@itemprop="ingredients"]');
foreach ($nodes as $node) {
$value = trim($node->nodeValue);
if ($value != "Ingredients") {
$recipe->appendIngredient($value);
}
}
// Instructions
$nodes = $xpath->query('//span[@class = "steps-list__item__text"]');
foreach ($nodes as $node) {
$value = trim($node->nodeValue);
$value = RecipeParser_Text::stripLeadingNumbers($value);
$parts = self::splitDirections($value);
if ($parts['section']) {
$parts['section'] = RecipeParser_Text::formatSectionName($parts['section']);
$recipe->addInstructionsSection($parts['section']);
}
$recipe->appendInstruction($parts['direction']);
}
// Notes
$nodes = $xpath->query('//div[@class = "recipe-notes__content"]/div/p');
$notes = array();
if ($nodes->length) {
foreach ($nodes as $node) {
$value = trim($node->nodeValue);
array_push($notes, $value);
}
$recipe->notes = implode(' | ', $notes);
}
// Photo
$nodes = $xpath->query('//img[@class = "recipe-carousel__recipe__img"]');
if ($nodes && $nodes->item(1)) {
$photo_url = $nodes->item(1)->getAttribute('src');
if (strpos($photo_url, 'default-recipe-image.gif') === false && strpos($photo_url, 'placeholder.gif') === false) {
$recipe->photo_url = RecipeParser_Text::relativeToAbsolute($photo_url, $url);
}
}
return $recipe;
}
public static function parse($html, $url)
{
$recipe = new RecipeParser_Recipe();
libxml_use_internal_errors(true);
$doc = new DOMDocument();
$html = mb_convert_encoding($html, 'HTML-ENTITIES', "UTF-8");
$doc->loadHTML('<?xml encoding="UTF-8">' . $html);
$xpath = new DOMXPath($doc);
// Title
$nodes = $xpath->query('//h3[@class = "title"]');
if ($nodes->length) {
$value = trim($nodes->item(0)->nodeValue);
$recipe->title = $value;
}
// Cook times
$nodes = $xpath->query('//div[@class = "recipe-metadata"]/ul/li');
foreach ($nodes as $node) {
$sub_nodes = $node->childNodes;
$key = null;
$value = null;
foreach ($sub_nodes as $sub_node) {
if ($sub_node->nodeName == 'h5') {
$key = trim($sub_node->nodeValue);
}
if ($sub_node->nodeName == 'p') {
$value = trim($sub_node->nodeValue);
}
}
// Inspect keys/values we've found.
if ($key == 'Total Time:') {
$value = self::cleanupTime($value);
$recipe->time['total'] = RecipeParser_Times::toMinutes($value);
}
if ($key == 'Prep Time:') {
$value = self::cleanupTime($value);
$recipe->time['prep'] = RecipeParser_Times::toMinutes($value);
}
}
$node_list = $xpath->query('//dd[@class = "preptime"]');
if ($node_list->length) {
$value = $node_list->item(0)->nodeValue;
$recipe->time['prep'] = RecipeParser_Times::toMinutes($value);
}
$node_list = $xpath->query('//dd[@class = "cooktime"]');
if ($node_list->length) {
$value = $node_list->item(0)->nodeValue;
$recipe->time['cook'] = RecipeParser_Times::toMinutes($value);
}
$node_list = $xpath->query('//dd[@class = "duration totaltime special"]');
if ($node_list->length) {
$value = $node_list->item(0)->nodeValue;
$recipe->time['total'] = RecipeParser_Times::toMinutes($value);
}
// Ingredients, Yield, Description, Notes, etc.
$nodes = $xpath->query('//div[@class = "recipe-body"]/*');
$section_title = null;
foreach ($nodes as $node) {
// Section titles
if ($node->nodeName == 'h4') {
$value = $node->nodeValue;
$value = trim(strtolower($value));
$section_title = $value;
continue;
}
$in_section = false;
if ($node->nodeName == 'div') {
// Ensure that we're in a <div class="section"> node.
foreach ($node->attributes as $attr_name => $attr_node) {
if ($attr_name == 'class' && $attr_node->value == 'section') {
$in_section = true;
}
}
if (!$in_section) {
continue;
}
// Description should be first text, before any section titles.
if (!$section_title) {
$value = $node->nodeValue;
$value = preg_replace("/^(Drink\\:|Top Chef).*\$/m", '', $value);
$value = str_replace("\n\n", "\n", $value);
$value = trim($value);
$recipe->description = $value;
// Yield
} else {
if ($section_title == 'yield') {
$value = trim($node->nodeValue);
$recipe->yield = $value;
// Notes
} else {
if ($section_title == 'notes') {
$value = trim($node->nodeValue);
$value = str_replace("\n\n", "\n", $value);
$recipe->notes = $value;
// Ingredients
} else {
if ($section_title == 'ingredients') {
$sub_nodes = $node->childNodes;
foreach ($sub_nodes as $sub_node) {
if ($sub_node->nodeName == 'h5') {
$value = RecipeParser_Text::formatSectionName($sub_node->nodeValue);
$recipe->addIngredientsSection($value);
} else {
if ($sub_node->nodeName == 'ul') {
$li_nodes = $sub_node->childNodes;
foreach ($li_nodes as $li_node) {
$value = trim($li_node->nodeValue);
$recipe->appendIngredient($value);
}
}
}
}
// Instructions
} else {
if ($section_title == 'directions') {
$sub_nodes = $node->childNodes;
foreach ($sub_nodes as $sub_node) {
$value = trim($sub_node->nodeValue);
// Section titles appear in all-caps.
if ($value && ($value == strtoupper($value) || preg_match('/:$/', $value))) {
$value = RecipeParser_Text::formatSectionName($value);
$recipe->addInstructionsSection($value);
} else {
$value = RecipeParser_Text::stripLeadingNumbers($value);
$recipe->appendInstruction($value);
}
}
}
}
}
}
}
}
}
// Source / Chef
$nodes = $xpath->query('//div[@class = "recipe-sidebar"]/div/*');
$section_title = null;
$chef_name = null;
$show_name = 'Bravo TV';
foreach ($nodes as $node) {
if ($node->nodeName == 'h4') {
$value = trim($node->nodeValue);
$section_title = strtolower($value);
continue;
}
if ($node->nodeName == 'small') {
if ($section_title == 'chef' || $section_title == 'author') {
$value = trim($node->nodeValue);
$chef_name = $value;
break;
}
}
}
$nodes = $xpath->query('//div[@class = "section"]/p[1]');
if ($nodes->length) {
$value = $nodes->item(0)->nodeValue;
if (strpos($value, 'Top Chef Masters') !== false) {
$show_name = 'Top Chef Masters';
} else {
if (strpos($value, 'Top Chef') !== false) {
$show_name = 'Top Chef';
}
}
}
$recipe->credits = $chef_name . ', ' . $show_name;
$nodes = $xpath->query('//div[@class = "recipe-header clearfix"]//img');
if ($nodes->length) {
$photo_url = $nodes->item(0)->getAttribute('src');
$photo_url = str_replace('/medium/', '/original/', $photo_url);
$recipe->photo_url = RecipeParser_Text::formatPhotoUrl($photo_url, $url);
}
return $recipe;
}
public static function parse($html, $url)
{
$recipe = new RecipeParser_Recipe();
libxml_use_internal_errors(true);
$html = mb_convert_encoding($html, 'HTML-ENTITIES', "UTF-8");
$doc = new DOMDocument();
$doc->loadHTML('<?xml encoding="UTF-8">' . $html);
$xpath = new DOMXPath($doc);
// Title
$nodes = $xpath->query('//h1[@itemprop="name"]');
if ($nodes->length) {
$value = trim($nodes->item(0)->nodeValue);
$recipe->title = $value;
}
// Times and yield
// <meta content="PT3H30M" itemprop="totalTime">
$nodes = $xpath->query('//meta[@itemprop="totalTime"]');
if ($nodes->length) {
if ($value = $nodes->item(0)->getAttribute('content')) {
$value = RecipeParser_Text::iso8601ToMinutes($value);
$recipe->time['total'] = $value;
}
}
$nodes = $xpath->query('//*[@itemprop="recipeYield"]');
if ($nodes->length) {
$value = $nodes->item(0)->nodeValue;
$recipe->yield = RecipeParser_Text::formatYield($value);
}
// Ingredients
$nodes = $xpath->query('//div[@id = "ingredients"]/*');
foreach ($nodes as $node) {
if ($node->nodeName == 'h2') {
$value = trim($node->nodeValue);
$value = RecipeParser_Text::formatSectionName($value);
if ($value != "Ingredients") {
$recipe->addIngredientsSection($value);
}
} else {
if ($node->nodeName == 'ol') {
$subnodes = $xpath->query('./li/span', $node);
foreach ($subnodes as $subnode) {
$value = trim($subnode->nodeValue);
$recipe->appendIngredient($value);
}
}
}
}
// Instructions
$nodes = $xpath->query('//div[@id = "directions"]/ol/li');
foreach ($nodes as $node) {
$value = trim($node->nodeValue);
$value = RecipeParser_Text::stripLeadingNumbers($value);
$parts = self::splitDirections($value);
if ($parts['section']) {
$parts['section'] = RecipeParser_Text::formatSectionName($parts['section']);
$recipe->addInstructionsSection($parts['section']);
}
$recipe->appendInstruction($parts['direction']);
}
// Notes
$nodes = $xpath->query('//div[@id = "directions"]/div[@id = "endnotes"]');
if ($nodes->length) {
$value = trim($nodes->item(0)->nodeValue);
$recipe->notes = $value;
}
// Photo
$nodes = $xpath->query('//img[@itemprop="image"]');
if ($nodes && $nodes->item(0)) {
$photo_url = $nodes->item(0)->getAttribute('src');
if (strpos($photo_url, 'default-recipe-image.gif') === false && strpos($photo_url, 'placeholder.gif') === false) {
$recipe->photo_url = RecipeParser_Text::formatPhotoUrl($photo_url, $url);
}
}
return $recipe;
}
