mediawiki-title - npm Package Compare versions

Comparing version 0.1.2 to 0.2.0

lib/index.js

 "use strict";
 
-var P          = require('bluebird');
-var preq       = require('preq');
 var sanitizeIP = require('./ip');
@@ -25,9 +23,24 @@ var utils      = require('./utils');
 /**
+ * Information about a wikimedia site required to make correct
+ * normalization.
+ *
+ * @typedef SiteInfo
+ * @type Object
+ * @property {string} lang Site language code.
+ * @property {string} legaltitlechars A perl-like regex for characters
+ *                                    allowed in the page title.
+ * @property {Object} namespaces Site namespaces info in the same format as
+ *                               returned by PHP api.
+ * @property {Object} namespacealiases Site namespace aliases in the same format
+ *                                     as returned by PHP api.
+ */
+
+/**
  * Creates an instance of title normalizer.
  *
  * @param {Object} options the normalizer options
- * @param {Function(string):string} options.apiURI a function
+ * @param {Function(string):Promise(SiteInfo)} options.getSiteInfo a function
  *                 that takes a domain string and returns back
- *                 an API URI that needs to be contacted to get
- *                 the site information used for normalization.
+ *                 the information about the site. The info is cached
+ *                 in-memory, so there's no need for external caching.
  * @constructor
@@ -37,3 +50,3 @@ * @public
 function Normalizer(options) {
-    if (!options || typeof options.apiURI !== 'function') {
+    if (!options || typeof options.getSiteInfo !== 'function') {
         throw new TypeError('Invalid options for Normalizer constructor');
@@ -67,23 +80,15 @@ }
     if (self._siteInfo[domain]) {
-        return P.resolve(self._siteInfo[domain]);
+        return self._siteInfo[domain];
     } else {
-        return preq.post({
-            uri: self._options.apiURI(domain),
-            body: {
-                action: 'query',
-                meta: 'siteinfo',
-                siprop: 'general|namespaces|namespacealiases',
-                format: 'json'
-            }
-        })
-        .then(function(res) {
-            res = res.body.query;
-            self._siteInfo[domain] = {
-                lang: res.general.lang,
-                invalidTitleRegex: self._createInvalidTitleRegex(res.general.legaltitlechars),
-                namespaces: res.namespaces,
-                namespacealiases: res.namespacealiases
+        self._siteInfo[domain] = self._options.getSiteInfo(domain)
+        .then(function(info) {
+            return {
+                lang: info.lang,
+                invalidTitleRegex: self._createInvalidTitleRegex(info.legaltitlechars),
+                namespaces: info.namespaces,
+                namespacealiases: info.namespacealiases
             };
-            return self._siteInfo[domain];
+
         });
+        return self._siteInfo[domain];
     }
@@ -230,3 +235,4 @@ };
     var self = this;
-    return P.try(function() {
+    return self._getSiteInfo(domain)
+    .then(function(siteInfo) {
         title = title.replace(/ /g, '_')
@@ -247,50 +253,48 @@ // Strip Unicode bidi override characters.
         self._checkEmptyTitle(title);
-        return self._getSiteInfo(domain)
-        .then(function(siteInfo) {
-            var result = self._splitNamespace(title, siteInfo);
-            if (result.namespace === NS_TALK) {
-                self._checkTalkNamespace(result.title, siteInfo);
-            }
-            var fragmentIndex = result.title.indexOf('#');
-            if (fragmentIndex >= 0) {
-                var fragment = result.title.substr(fragmentIndex);
-                result.fragment = fragment.substr(1).replace(/_/g, ' ');
-                result.title = result.title
-                .substring(result.title, result.title.length - fragment.length)
-                .replace('/_*$/', '');
-            }
 
-            self._checkLegalTitleCharacters(result.title, siteInfo);
-            self._checkRelativeTitle(result.title);
-            // Magic tilde sequences? Nu-uh!
-            if (result.title.indexOf('~~~') !== -1) {
-                throw new utils.TitleError({
-                    type: 'title-invalid-magic-tilde',
-                    title: title
-                });
-            }
-            self._checkMaxLength(result.title, result.namespace);
+        var result = self._splitNamespace(title, siteInfo);
+        if (result.namespace === NS_TALK) {
+            self._checkTalkNamespace(result.title, siteInfo);
+        }
+        var fragmentIndex = result.title.indexOf('#');
+        if (fragmentIndex >= 0) {
+            var fragment = result.title.substr(fragmentIndex);
+            result.fragment = fragment.substr(1).replace(/_/g, ' ');
+            result.title = result.title
+            .substring(result.title, result.title.length - fragment.length)
+            .replace('/_*$/', '');
+        }
 
-            result = self._capitalizeTitle(result, siteInfo);
+        self._checkLegalTitleCharacters(result.title, siteInfo);
+        self._checkRelativeTitle(result.title);
+        // Magic tilde sequences? Nu-uh!
+        if (result.title.indexOf('~~~') !== -1) {
+            throw new utils.TitleError({
+                type: 'title-invalid-magic-tilde',
+                title: title
+            });
+        }
+        self._checkMaxLength(result.title, result.namespace);
 
-            if (result.namespace !== NS_MAIN) {
-                self._checkEmptyTitle(result.title);
-            }
+        result = self._capitalizeTitle(result, siteInfo);
 
-            if (result.namespace === NS_USER || result.namespace === NS_USER_TALK) {
-                result.title = sanitizeIP(result.title);
-            }
+        if (result.namespace !== NS_MAIN) {
+            self._checkEmptyTitle(result.title);
+        }
 
-            var normalized = result.title;
+        if (result.namespace === NS_USER || result.namespace === NS_USER_TALK) {
+            result.title = sanitizeIP(result.title);
+        }
 
-            if (result.namespace !== NS_MAIN) {
-                normalized = siteInfo.namespaces[result.namespace]['*'] + ':' + normalized;
-            }
+        var normalized = result.title;
 
-            if (result.fragment) {
-                normalized = normalized + '#' + result.fragment;
-            }
+        if (result.namespace !== NS_MAIN) {
+            normalized = siteInfo.namespaces[result.namespace]['*'] + ':' + normalized;
+        }
 
-            return normalized;
-        });
+        if (result.fragment) {
+            normalized = normalized + '#' + result.fragment;
+        }
+
+        return normalized;
     });
@@ -297,0 +301,0 @@ };

package.json

 {
   "name": "mediawiki-title",
-  "version": "0.1.2",
+  "version": "0.2.0",
   "description": "Title normalization library for mediawiki",
@@ -26,6 +26,2 @@ "main": "lib/index.js",
   },
-  "dependencies": {
-    "bluebird": "^3.1.1",
-    "preq": "^0.4.8"
-  },
   "devDependencies": {
@@ -37,4 +33,5 @@ "coveralls": "^2.11.6",
     "mocha-jshint": "^2.2.6",
-    "mocha-lcov-reporter": "^1.0.0"
+    "mocha-lcov-reporter": "^1.0.0",
+    "preq": "^0.4.8"
   }
 }

README.md

@@ -1,2 +0,2 @@
-# mediawiki-title [![npm version](https://badge.fury.io/js/mediawiki-title.svg)](https://badge.fury.io/js/mediawiki-title) [![Build Status](https://travis-ci.org/wikimedia/mediawiki-title.svg?branch=master)](https://travis-ci.org/wikimedia/mediawiki-title) [![Coverage Status](https://coveralls.io/repos/github/wikimedia/analytics-mediawiki-storage/badge.svg?branch=master)](https://coveralls.io/github/wikimedia/analytics-mediawiki-storage?branch=master) [![Dependencies](https://david-dm.org/wikimedia/mediawiki-title.svg?branch=master)](https://david-dm.org/wikimedia/mediawiki-title?branch=master)
+# mediawiki-title [![npm version](https://badge.fury.io/js/mediawiki-title.svg)](https://badge.fury.io/js/mediawiki-title) [![Build Status](https://travis-ci.org/wikimedia/mediawiki-title.svg?branch=master)](https://travis-ci.org/wikimedia/mediawiki-title) [![Coverage Status](https://coveralls.io/repos/github/wikimedia/mediawiki-title/badge.svg?branch=master)](https://coveralls.io/github/wikimedia/mediawiki-title?branch=master) [![Dependencies](https://david-dm.org/wikimedia/mediawiki-title.svg?branch=master)](https://david-dm.org/wikimedia/mediawiki-title?branch=master)
 
@@ -7,4 +7,4 @@ Mediawiki title normalizetion, that conforms to the normalization rules used in [MediaWiki Core](https://www.mediawiki.org/wiki/API:Query#Title_normalization).
 
-<a name="API"></a>
-## API
+<a name="Normalizer"></a>
+## Normalizer
 
@@ -19,6 +19,7 @@ * [Normalizer](#Normalizer)
 
+
 | Param | Type | Description |
 | --- | --- | --- |
 | options | <code>Object</code> | the normalizer options |
-| options.apiURI | <code>function</code> | a function                 that takes a domain string and returns back                 an API URI that needs to be contacted to get                 the site information used for normalization. |
+| options.getSiteInfo | <code>function</code> | a function                 that takes a domain string and returns back                 the information about the site. The info is cached                 in-memory, so there's no need for external caching. |
 
@@ -38,2 +39,16 @@ <a name="Normalizer+normalize"></a>
 
+<a name="SiteInfo"></a>
+## SiteInfo : <code>Object</code>
+Information about a wikimedia site required to make correct
+normalization.
+  
+**Properties**
+
+| Name | Type | Description |
+| --- | --- | --- |
+| lang | <code>string</code> | Site language code. |
+| legaltitlechars | <code>string</code> | A perl-like regex for characters allowed in the page title. |
+| namespaces | <code>Object</code> | Site namespaces info in the same format as returned by PHP api. |
+| namespacealiases | <code>Object</code> | Site namespace aliases in the same format as returned by PHP api. |
+
 ## Usage
@@ -47,3 +62,16 @@
 var normalizer = new Normalizer({
-  apiURI: function(domain) { return 'https://' + domain + '/w/api.php'; }
+  getSiteInfo: function(domain) { 
+    return {
+      lang: 'en',
+      legaltitlechars: " %!\"$&'()*,\\-.\\/0-9:;=?@A-Z\\\\^_`a-z~\\x80-\\xFF+",
+      namespaces: {
+      	"0": {
+		  id: 0,
+          case: "first-letter",
+          content: "",
+          "*": ""
+        },
+      }
+    }; 
+  }
 });
@@ -50,0 +78,0 @@

test/index.js

@@ -6,2 +6,4 @@ 'use strict';
 var utils = require('../lib/utils');
+var preq = require('preq');
+var P = require('bluebird');
 
@@ -14,4 +16,20 @@ // Run jshint as part of normal testing
 var normalizer = new Normalizer({
-    apiURI: function (domain) {
-        return 'https://' + domain + '/w/api.php';
+    getSiteInfo: function (domain) {
+        return preq.post({
+            uri: 'https://' + domain + '/w/api.php',
+            body: {
+                action: 'query',
+                meta: 'siteinfo',
+                siprop: 'general|namespaces|namespacealiases',
+                format: 'json'
+            }
+        })
+        .then(function(res) {
+            return {
+                lang: res.body.query.general.lang,
+                legaltitlechars: res.body.query.general.legaltitlechars,
+                namespaces: res.body.query.namespaces,
+                namespacealiases: res.body.query.namespacealiases
+            }
+        });
     }
@@ -285,2 +303,29 @@ });
     });
+
+    it('Should fetch domains', function() {
+        return preq.get({
+            uri: 'https://en.wikipedia.org/w/api.php?action=sitematrix&format=json'
+        })
+        .then(function(res) {
+            return Object.keys(res.body.sitematrix)
+            .filter(function(idx) {
+                return idx !== 'count' && idx !== 'specials';
+            })
+            .map(function (idx) {
+                return res.body.sitematrix[idx].site[0].url.replace(/^https?:\/\//, '');
+            });
+        })
+        .then(function (domains) {
+            describe('Various domains', function() {
+                domains.forEach(function (domain) {
+                    it('Should work for ' + domain, function() {
+                        return normalizer.normalize('1', domain)
+                        .then(function (res) {
+                            assert.deepEqual(res, '1');
+                        });
+                    });
+                });
+            });
+        });
+    });
 });

New alerts

License Policy Violation

License

This package is not allowed per your license policy. Review the package's license to ensure compliance.

Found 1 instance in 1 package

Fixed alerts

License Policy Violation

License

This package is not allowed per your license policy. Review the package's license to ensure compliance.

Found 1 instance in 1 package

Improved metrics

Total package byte prevSize

35260

increased by7.39%

Dependency count

0

decreased by-100%

Lines of code

793

increased by6.02%

Number of lines in readme file

89

increased by45.9%

Worsened metrics

Dev dependency count

7

increased by16.67%

Dependency changes

• - Removedbluebird@^3.1.1

• - Removedpreq@^0.4.8

• - Removedajv@6.12.6(transitive)

• - Removedasn1@0.2.6(transitive)

• - Removedassert-plus@1.0.0(transitive)

• - Removedasynckit@0.4.0(transitive)

• - Removedaws-sign2@0.7.0(transitive)

• - Removedaws4@1.13.2(transitive)

• - Removedbcrypt-pbkdf@1.0.2(transitive)

• - Removedbluebird@3.7.2(transitive)

• - Removedcaseless@0.12.0(transitive)

• - Removedcombined-stream@1.0.8(transitive)

• - Removedcore-util-is@1.0.2(transitive)

• - Removeddashdash@1.14.1(transitive)

• - Removeddelayed-stream@1.0.0(transitive)

• - Removedecc-jsbn@0.1.2(transitive)

• - Removedextend@3.0.2(transitive)

• - Removedextsprintf@1.3.0(transitive)

• - Removedfast-deep-equal@3.1.3(transitive)

• - Removedfast-json-stable-stringify@2.1.0(transitive)

• - Removedforever-agent@0.6.1(transitive)

• - Removedform-data@2.3.3(transitive)

• - Removedgetpass@0.1.7(transitive)

• - Removedhar-schema@2.0.0(transitive)

• - Removedhar-validator@5.1.5(transitive)

• - Removedhttp-signature@1.2.0(transitive)

• - Removedis-typedarray@1.0.0(transitive)

• - Removedisstream@0.1.2(transitive)

• - Removedjsbn@0.1.1(transitive)

• - Removedjson-schema@0.4.0(transitive)

• - Removedjson-schema-traverse@0.4.1(transitive)

• - Removedjson-stringify-safe@5.0.1(transitive)

• - Removedjsprim@1.4.2(transitive)

• - Removedmime-db@1.52.0(transitive)

• - Removedmime-types@2.1.35(transitive)

• - Removedoauth-sign@0.9.0(transitive)

• - Removedperformance-now@2.1.0(transitive)

• - Removedpreq@0.4.12(transitive)

• - Removedpsl@1.13.0(transitive)

• - Removedpunycode@2.3.1(transitive)

• - Removedqs@6.5.3(transitive)

• - Removedrequest@2.88.2(transitive)

• - Removedsafe-buffer@5.2.1(transitive)

• - Removedsafer-buffer@2.1.2(transitive)

• - Removedsshpk@1.18.0(transitive)

• - Removedtough-cookie@2.5.0(transitive)

• - Removedtunnel-agent@0.6.0(transitive)

• - Removedtweetnacl@0.14.5(transitive)

• - Removeduri-js@4.4.1(transitive)

• - Removeduuid@3.4.0(transitive)

• - Removedverror@1.10.0(transitive)