Add Wikipedia fallback for SMILES lookup

composer.json

@@ -15,7 +15,8 @@
         "mustangostang/spyc": "^0.5.1",
         "intervention/image": "^2.4",
         "guzzlehttp/guzzle": "^6.3",
-        "lambdacasserole/condense": "^1.1"
+        "lambdacasserole/condense": "^1.1",
+        "tburry/pquery": "^1.1"
     },
     "autoload": {
         "psr-4": {

composer.lock

@@ -4,7 +4,7 @@
         "Read more about it at https://getcomposer.org/doc/01-basic-usage.md#composer-lock-the-lock-file",
         "This file is @generated automatically"
     ],
-    "content-hash": "414181a9b439757b16b4938d109a80bd",
+    "content-hash": "18fab38459b21165a1075c73d836e92c",
     "packages": [
         {
             "name": "defuse/php-encryption",
@@ -1056,6 +1056,58 @@
             ],
             "time": "2016-06-29T05:41:56+00:00"
         },
+        {
+            "name": "tburry/pquery",
+            "version": "v1.1.1",
+            "source": {
+                "type": "git",
+                "url": "https://github.com/tburry/pquery.git",
+                "reference": "872339ffd38d261c4417ea1855428b1b4ff9abf1"
+            },
+            "dist": {
+                "type": "zip",
+                "url": "https://api.github.com/repos/tburry/pquery/zipball/872339ffd38d261c4417ea1855428b1b4ff9abf1",
+                "reference": "872339ffd38d261c4417ea1855428b1b4ff9abf1",
+                "shasum": ""
+            },
+            "require": {
+                "php": ">=5.3.0"
+            },
+            "require-dev": {
+                "htmlawed/htmlawed": "dev-master"
+            },
+            "type": "library",
+            "autoload": {
+                "classmap": [
+                    "IQuery.php",
+                    "gan_formatter.php",
+                    "gan_node_html.php",
+                    "gan_parser_html.php",
+                    "gan_selector_html.php",
+                    "gan_tokenizer.php",
+                    "gan_xml2array.php",
+                    "pQuery.php"
+                ]
+            },
+            "notification-url": "https://packagist.org/downloads/",
+            "license": [
+                "LGPL-2.1"
+            ],
+            "authors": [
+                {
+                    "name": "Todd Burry",
+                    "email": "todd@vanillaforums.com",
+                    "role": "developer"
+                }
+            ],
+            "description": "A jQuery like html dom parser written in php.",
+            "keywords": [
+                "dom",
+                "ganon",
+                "php"
+            ],
+            "time": "2016-01-14T20:55:00+00:00"
+        },
         {
             "name": "twig/twig",
             "version": "v1.24.1",

src/CactusSmilesConverter.php

@@ -6,55 +6,26 @@ use Condense\Database;
 use \GuzzleHttp\Client;
 
 /**
- * Provides a conversion service between compound names and SMILES strings.
+ * Provides a conversion service between compound names and SMILES strings that uses a dedicated lookup service.
  *
  * @author Saul Johnson <saul.a.johnson@gmail.com>
  * @since 06/08/2017
  * @package Avogadrio
  */
-class CactusSmilesConverter
+class CactusSmilesConverter extends SmilesConverter
 {
-    /**
-     * The compound name/SMILES cache database.
-     *
-     * @var Database
-     */
-    private $db;
-
     /**
      * Initializes a new instance of a conversion service between compound names and SMILES strings.
      *
-     * @param Database $db  the database in which to cache lookup results
+     * @param Database $db              the database in which to cache lookup results
+     * @param SmilesConverter $fallback the conversion service to fall back to in case of error
      */
-    public function __construct($db = null) {
+    public function __construct($db = null, $fallback = null) {
-        $this->db = $db;
+        parent::__construct($db, $fallback);
     }
 
     /**
-     * Gets whether or not the API result cache is enabled.
+     * @inheritdoc
-     *
-     * @return bool true if the cache is enabled, otherwise false
-     */
-    public function isCacheEnabled() {
-        return $this->db !== null;
-    }
-    
-    /**
-     * Attempts to fix any errors in a SMILES string and returns the result.
-     *
-     * @param string $smiles    the SMILES string to fix
-     * @return string           the fixed SMILES string
-     */
-    private static function fixSmiles($smiles) {
-        $output = str_replace('|', '', $smiles); // Zap vertical bars.
-        return $output;
-    }
-
-    /**
-     * Converts a compound name to SMILES notation.
-     *
-     * @param string $name  the name of the compound
-     * @return null|string  the SMILES notation for the named compound or null if not found
      */
     public function nameToSmiles($name)
     {
@@ -62,11 +33,9 @@ class CactusSmilesConverter
         $encoded = rawurlencode($name);
 
         // Check if we've got the name cached already.
-        if ($this->isCacheEnabled()) {
+        $cached = $this->getIfCachedAndValid($encoded);
-            $cachedSmiles = $this->db->get('smiles', 'name', $encoded);
+        if ($cached !== null) {
-            if ($cachedSmiles !== null) {
+            return $cached;
-                return self::fixSmiles($cachedSmiles);
-            }
         }
 
         // Convert chemical name to SMILES if we can using API.
@@ -76,13 +45,12 @@ class CactusSmilesConverter
         // If request was successful.
         if ($response->getStatusCode() === 200) {
             $smiles = $response->getBody()->getContents();
-            if ($this->isCacheEnabled()) {
+            if (self::isValidSmiles($smiles)) {
-                $this->db->insert(['name' => $encoded, 'smiles' => $smiles]); // Cache name for future.
+                $this->cache($encoded, $smiles);
+                return $smiles;
             }
-            return self::fixSmiles($smiles);
         }
-        
+
-        // Request failed.
+        return $this->fallback($name); // Fall back.
-        return null;
     }
 }

src/SmilesConverter.php

@@ -0,0 +1,111 @@
+<?php
+
+namespace Avogadrio;
+
+use Condense\Database;
+
+/**
+ * Provides a conversion service between compound names and SMILES strings.
+ *
+ * @author Saul Johnson <saul.a.johnson@gmail.com>
+ * @since 27/08/2017
+ * @package Avogadrio
+ */
+abstract class SmilesConverter
+{
+    /**
+     * The compound name/SMILES cache database.
+     *
+     * @var Database
+     */
+    private $db;
+
+    /**
+     * The conversion service to fall back to in case of error.
+     *
+     * @var SmilesConverter
+     */
+    private $fallback;
+
+    /**
+     * Initializes a new instance of a conversion service between compound names and SMILES strings.
+     *
+     * @param Database $db              the database in which to cache lookup results
+     * @param SmilesConverter $fallback the conversion service to fall back to in case of error
+     */
+    protected function __construct($db, $fallback) {
+        $this->db = $db;
+        $this->fallback = $fallback;
+    }
+
+    /**
+     * Gets whether or not the API result cache is enabled.
+     *
+     * @return bool true if the cache is enabled, otherwise false
+     */
+    public function isCacheEnabled() {
+        return $this->db !== null;
+    }
+
+    /**
+     * Attempts to fix any errors in a SMILES string and returns the result.
+     *
+     * @param string $smiles    the SMILES string to fix
+     * @return string           the fixed SMILES string
+     */
+    protected static function fixSmiles($smiles) {
+        $output = str_replace('|', '', $smiles); // Zap vertical bars.
+        return $output;
+    }
+
+    /**
+     * Returns true if a given SMILES string is valid, otherwise returns false.
+     *
+     * @param string $smiles    the SMILES string to check
+     * @return bool             true if the string was valid, otherwise false
+     */
+    protected static function isValidSmiles($smiles) {
+        return preg_match('/^[a-zA-Z0-9@%=\\\\\/\[\]\.\+\-\_\*\(\)]+$/', $smiles) ? true : false;
+    }
+
+    /**
+     * @param $name
+     * @return string
+     */
+    protected function getIfCachedAndValid($name) {
+        if ($this->isCacheEnabled()) {
+            $cachedSmiles = $this->db->get('smiles', 'name', $name);
+            if ($cachedSmiles !== null && self::isValidSmiles($cachedSmiles)) {
+                return $cachedSmiles;
+            }
+        }
+        return null;
+    }
+
+    protected function cache($name, $smiles) {
+        if ($this->isCacheEnabled()) {
+            $this->db->insert(['name' => $name, 'smiles' => $smiles]); // Cache name for future.
+        }
+    }
+
+    /**
+     * Falls back to another SMILES conversion service.
+     *
+     * @param string $name  the compound name to convert
+     * @return null|string  the converted compound name or null if not found
+     */
+    protected function fallback($name) {
+        if ($this->fallback === null) {
+            return null;
+        }
+        return $this->fallback->nameToSmiles($name);
+    }
+
+    /**
+     * Converts a compound name to SMILES notation.
+     *
+     * @param string $name  the name of the compound
+     * @return null|string  the SMILES notation for the named compound or null if not found
+     */
+    public abstract function nameToSmiles($name);
+}

src/WikipediaSmilesConverter.php

@@ -0,0 +1,70 @@
+<?php
+
+namespace Avogadrio;
+
+use Condense\Database;
+use \GuzzleHttp\Client;
+use pQuery;
+
+/**
+ * Provides a conversion service between compound names and SMILES strings that scrapes SMILES strings from Wikipedia.
+ *
+ * @author Saul Johnson <saul.a.johnson@gmail.com>
+ * @since 27/08/2017
+ * @package Avogadrio
+ */
+class WikipediaSmilesConverter extends SmilesConverter
+{
+    /**
+     * Initializes a new instance of a conversion service between compound names and SMILES strings.
+     *
+     * @param Database $db              the database in which to cache lookup results
+     * @param SmilesConverter $fallback the conversion service to fall back to in case of error
+     */
+    public function __construct($db = null, $fallback = null) {
+        parent::__construct($db, $fallback);
+    }
+
+    /**
+     * @inheritdoc
+     */
+    public function nameToSmiles($name)
+    {
+        // Sanitize name for URLs.
+        $encoded = str_replace(' ', '_', $name);
+
+        // Check if we've got the name cached already.
+        $cached = $this->getIfCachedAndValid($encoded);
+        if ($cached !== null) {
+            return $cached;
+        }
+
+        // Get Wikipedia page for compound.
+        $client = new Client();
+        $response = $client->request('GET', "https://en.wikipedia.org/wiki/$encoded");
+
+        // If there is a page.
+        if ($response->getStatusCode() === 200) {
+
+            // Parse page into DOM.
+            $page = $response->getBody()->getContents();
+            $dom = pQuery::parseStr($page);
+
+            // Get all hyperlinks.
+            $links = $dom->select('a');
+
+            // Look for the SMILES hyperlink.
+            foreach ($links as $link) {
+                if (strstr($link->text(), 'SMILES')) {
+                    $smiles = trim($link->parent->getNextSibling()->text()); // Get actual SMILES string.
+                    if (self::isValidSmiles($smiles)) {
+                        $this->cache($encoded, $smiles);
+                        return $smiles;
+                    }
+                }
+            }
+        }
+
+        return $this->fallback($name); // Fall back.
+    }
+}

web/index.php

@@ -9,6 +9,7 @@ use Symfony\Component\HttpKernel\HttpKernelInterface;
 
 use Condense\Database;
 use Avogadrio\CactusSmilesConverter;
+use Avogadrio\WikipediaSmilesConverter;
 use Avogadrio\MoleculeRenderer;
 
 $app = new Silex\Application();
@@ -20,7 +21,7 @@ $app = new Silex\Application();
 $config = Spyc::YAMLLoad(__DIR__ . '/../config/config.yaml');
 
 // Services.
-$smilesConverter = new CactusSmilesConverter(new Database('names', __DIR__ . '/../db'));
+$smilesConverter = new CactusSmilesConverter(new Database('names', __DIR__ . '/../db'), new WikipediaSmilesConverter());
 $moleculeRenderer = new MoleculeRenderer($config['sourire_service']);
 
 $moleculeRenderer->setRenderChiralLabels(false); // Disable chiral labels.