Search a mediawiki

Question

Is there any api or other way, with Javascript, to search any mediawiki and print the found page (and if not found print that).

I'd prefer something like this:

function searchWiki(wikipage, search) {
    //the function here
    document.write(foundPage);
}

//run it

searchWiki('https://en.wikipedia.org', 'banana');

//it would print 'https://en.wikipedia.org/wiki/Banana'

Answer 1

Here is my implementation of such a function. It uses the MediaWiki API through JSONP and is quite flexible. I guess a jQuery solution is fine. I created a small fiddle.

searchWiki(site, search, [callback], [options])

function searchWiki(site, search, callback, opts) {
    if(typeof callback == 'object') {
        opts = callback;
        callback = null;
    } else {
        opts = opts || {};
    }
    // Build the required URLs
    var siteUrl = (opts.ssl ? 'https' : 'http') + '://' + site;
    var apiUrl = siteUrl + (opts.apiBase || '/w/') + 'api.php';
    var queryUrl = apiUrl + '?action=query&list=search&srsearch=' + encodeURIComponent(search) + '&srlimit=' + (opts.maxResults || 1) + '&format=json';
    // Issue the JSONP request
    $.ajax(queryUrl + '&callback=?', {
        dataType: 'jsonp',
        // This prevents warnings about the unrecognized parameter "_"
        cache: true,
        success: function(data) {
            // Get all returned pages
            var titles = [], links = [];
            for(var i = 0; i < data.query.search.length; i++) {
                var title = data.query.search[i].title,
                    link = siteUrl + (opts.wikiBase || '/wiki/') + encodeURIComponent(title);
                titles.push(title);
                links.push(link);
            }
            if(!opts.maxResults) {
                // Single result requested
                if(data.query.search.length == 0) {
                    titles = links = null;
                } else {
                    titles = titles[0];
                    links = links[0];
                }
            }
            // Call the callback
            (callback || opts.success || function(){})(titles, links);
        }
    });
}

Example 1: Single wikipedia search

searchWiki('en.wikipedia.org', 'banana fruit', {
    ssl: true,
    success: function(title, link) {
        // link is now "https://en.wikipedia.org/wiki/Banana"
        if(title === null) {
            $('#search-msg').text('Not found');
        } else {
            var anchor = $('<a>').text(title).attr('href', link);
            $('#search-msg').append(anchor);
        }
    }
});

This example shows a link to a wikipedia page with the associated title.

Example 2: Multiple results

searchWiki('www.mediawiki.org', 'Release notes', {
    ssl: true,
    maxResults: 5,
    success: function(titles, links) {
        for(var i = 0; i < titles.length; i++) {
            alert('MediaWiki ' + titles[i] + ' at ' + links[i]);
        }
    }
});

This example displays up to five links to MediaWiki pages which match the query "Release notes".

Options:

• ssl: Use HTTPS instead of HTTP
• maxResults: Return multiple (up to n) results
• apiBase: API directory on target site (defaults to /w/)
• wikiBase: Wiki directory on target site (defaults to /wiki/)
• success: Function to invoke after retrieving the result list

You can either pass the callback as a function argument (before the options) or as the success option.

---

Update: Here is the pure JS solution (no jQuery required). And there is another fiddle, this time without jQuery.

function searchWiki(site, search, callback, opts) {
    if(typeof callback == 'object') {
        opts = callback;
        callback = null;
    } else {
        opts = opts || {};
    }
    // Build the required URLs
    var siteUrl = (opts.ssl ? 'https' : 'http') + '://' + site;
    var apiUrl = siteUrl + (opts.apiBase || '/w/') + 'api.php';
    var queryUrl = apiUrl + '?action=query&list=search&srsearch=' + encodeURIComponent(search) + '&srlimit=' + (opts.maxResults || 1) + '&format=json';
    var fnName = '_cb_' + Math.floor(Math.random() * 4294967296);
    window[fnName] = function(data) {
        // Clear references to this function
        window[fnName] = null;
        // Get all returned pages
        var titles = [], links = [];
        for(var i = 0; i < data.query.search.length; i++) {
            var title = data.query.search[i].title,
                link = siteUrl + (opts.wikiBase || '/wiki/') + encodeURIComponent(title);
            titles.push(title);
            links.push(link);
        }
        if(!opts.maxResults) {
            // Single result requested
            if(data.query.search.length == 0) {
                titles = links = null;
            } else {
                titles = titles[0];
                links = links[0];
            }
        }
        // Call the callback
        (callback || opts.success || function(){})(titles, links);
    }
    // Issue the JSONP request
    var scriptTag = document.createElement('script');
    scriptTag.setAttribute('src', queryUrl + '&callback=' + fnName);
    document.head.appendChild(scriptTag);
}

---

Update 2: Finally a solution for node.js. The API is still the same, but it provides some additional options:

• error: An error callback (this was impossible in browser-based JS)
• userAgent: A custom user agent string as suggested in the docs
• port: Target port (defaults to 80/443)
• encoding: Response encoding (default to utf8)

I did not test this much, but the examples (see above) should still work.

var http = require('http'),
    https = require('https');

function searchWiki(site, search, callback, opts) {
    if(typeof callback == 'object') {
        opts = callback;
        callback = null;
    } else {
        opts = opts || {};
    }
    // Build the required paths
    var apiPath = (opts.apiBase || '/w/') + 'api.php';
    var queryPath = apiPath + '?action=query&list=search&srsearch=' + encodeURIComponent(search) + '&srlimit=' + (opts.maxResults || 1) + '&format=json';
    // Request options
    var httpOpts = {
        hostname: site,
        port: (opts.port ? opts.port : (opts.ssl ? 443 : 80)),
        method: 'GET',
        path: queryPath,
        agent: false
    };
    // Custom user agent
    if(opts.userAgent) {
        httpOpts.headers = {
            'User-Agent': opts.userAgent
        };
    }
    // Make the request
    var req = (opts.ssl ? https : http).request(httpOpts, function(res) {
        var msgBody = '';
        res.setEncoding(opts.encoding || 'utf8');

        res.on('data', function(chunk) {
            msgBody += chunk;
        });

        res.on('end', function() {
            // Parse response as JSON
            var data;
            try {
                data = JSON.parse(msgBody);
            } catch(err) {
                (opts.error || function(){})(err);
                return;
            }
            // Get all returned pages
            var siteUrl = (opts.ssl ? 'https' : 'http') + '://' + site;
            var titles = [], links = [];
            for(var i = 0; i < data.query.search.length; i++) {
                var title = data.query.search[i].title,
                    link = siteUrl + (opts.wikiBase || '/wiki/') + encodeURIComponent(title);
                titles.push(title);
                links.push(link);
            }
            if(!opts.maxResults) {
                // Single result requested
                if(data.query.search.length == 0) {
                    titles = links = null;
                } else {
                    titles = titles[0];
                    links = links[0];
                }
            }
            // Call the callback
            (callback || opts.success || function(){})(titles, links);
        });
    });
    req.on('error', function(err) {
        (opts.error || function(){})(err);
    });
    req.end();
}