Récupérer le contenu des balises meta d'un page HTML distante

tags = get_meta_tags ('http://[adresse du site où sont les métas.]/PWS.html');

function file_get_contents (url, flags, context, offset, maxLen) {
    // Read the entire file into a string
    //
    // version: 906.111
    // discuss at: http://phpjs.org/functions/file_get_contents
    // +   original by: Legaev Andrey
    // +      input by: Jani Hartikainen
    // +   improved by: Kevin van Zonneveld (http://kevin.vanzonneveld.net)
    // +   improved by: Brett Zamir (http://brett-zamir.me)
    // +   input by: Raphael (Ao) RUDLER
    // +   bugfixed by: Brett Zamir (http://brett-zamir.me)
    // %        note 1: This function uses XmlHttpRequest and cannot retrieve resource from different domain without modifications.
    // %        note 2: Synchronous by default (as in PHP) so may lock up browser. Can
    // %        note 2: get async by setting a custom "phpjs.async" property to true and "notification" for an
    // %        note 2: optional callback (both as context params, with responseText, and other JS-specific
    // %        note 2: request properties available via 'this'). Note that file_get_contents() will not return the text
    // %        note 2: in such a case (use this.responseText within the callback). Or, consider using
    // %        note 2: jQuery's: $('#divId').load('http://url') instead.
    // %        note 3: The context argument is only implemented for http, and only partially (see below for
    // %        note 3: "Presently unimplemented HTTP context options"); also the arguments passed to
    // %        note 3: notification are incomplete
    // *     example 1: file_get_contents('http://kevin.vanzonneveld.net/pj_test_supportfile_1.htm');
    // *     returns 1: '123'
    // Note: could also be made to optionally add to global $http_response_header as per http://php.net/manual/en/reserved.variables.httpresponseheader.php

    var tmp, headers = [], newTmp = [], k=0, i=0, href = '', pathPos = -1, flagNames = 0, content = null, http_stream = false;
    var func = function (value) { return value.substring(1) !== ''; };

    // BEGIN REDUNDANT
    this.php_js = this.php_js || {};
    this.php_js.ini = this.php_js.ini || {};
    // END REDUNDANT

    var ini = this.php_js.ini;
    context = context || this.php_js.default_streams_context || null;

    if (!flags) {flags = 0;}
    var OPTS = {
        FILE_USE_INCLUDE_PATH : 1,
        FILE_TEXT : 32,
        FILE_BINARY : 64
    };
    if (typeof flags === 'number') { // Allow for a single string or an array of string flags
        flagNames = flags;
    }
    else {
        flags = [].concat(flags);
        for (i=0; i < flags.length; i++) {
            if (OPTS[flags[i]]) {
                flagNames = flagNames | OPTS[flags[i]];
            }
        }
    }

    if (flagNames & OPTS.FILE_BINARY && (flagNames & OPTS.FILE_TEXT)) { // These flags shouldn't be together
        throw 'You cannot pass both FILE_BINARY and FILE_TEXT to file_get_contents()';
    }

    if ((flagNames & OPTS.FILE_USE_INCLUDE_PATH) && ini.include_path &&
            ini.include_path.local_value) {
        var slash = ini.include_path.local_value.indexOf('/') !== -1 ? '/' : '\\';
        url = ini.include_path.local_value+slash+url;
    }
    else if (!/^(https?|file):/.test(url)) { // Allow references within or below the same directory (should fix to allow other relative references or root reference; could make dependent on parse_url())
        href = this.window.location.href;
        pathPos = url.indexOf('/') === 0 ? href.indexOf('/', 8)-1 : href.lastIndexOf('/');
        url = href.slice(0, pathPos+1)+url;
    }

    if (context) {
        var http_options = context.stream_options && context.stream_options.http;
        http_stream = !!http_options;
    }

    if (!context || http_stream) {
        var req = this.window.ActiveXObject ? new ActiveXObject('Microsoft.XMLHTTP') : new XMLHttpRequest();
        if (!req) {throw new Error('XMLHttpRequest not supported');}

        var method = http_stream ? http_options.method : 'GET';
        var async = !!(context && context.stream_params && context.stream_params['phpjs.async']);
        
        if (ini['phpjs.ajaxBypassCache'] && ini['phpjs.ajaxBypassCache'].local_value) {
            url += (url.match(/\?/) == null ? "?" : "&") + (new Date()).getTime(); // Give optional means of forcing bypass of cache
        }
        
        req.open(method, url, async);
        if (async) {
            var notification = context.stream_params.notification;
            if (typeof notification === 'function') {
                // Fix: make work with req.addEventListener if available: https://developer.mozilla.org/En/Using_XMLHttpRequest
                if (0 && req.addEventListener) { // Unimplemented so don't allow to get here
                    /*
                    req.addEventListener('progress', updateProgress, false);
                    req.addEventListener('load', transferComplete, false);
                    req.addEventListener('error', transferFailed, false);
                    req.addEventListener('abort', transferCanceled, false);
                    */
                }
                else {
                    req.onreadystatechange = function (aEvt) { // aEvt has stopPropagation(), preventDefault(); see https://developer.mozilla.org/en/NsIDOMEvent

    // Other XMLHttpRequest properties: multipart, responseXML, status, statusText, upload, withCredentials
    /*
    PHP Constants:
    STREAM_NOTIFY_RESOLVE   1       A remote address required for this stream has been resolved, or the resolution failed. See severity  for an indication of which happened.
    STREAM_NOTIFY_CONNECT   2     A connection with an external resource has been established.
    STREAM_NOTIFY_AUTH_REQUIRED 3     Additional authorization is required to access the specified resource. Typical issued with severity level of STREAM_NOTIFY_SEVERITY_ERR.
    STREAM_NOTIFY_MIME_TYPE_IS  4     The mime-type of resource has been identified, refer to message for a description of the discovered type.
    STREAM_NOTIFY_FILE_SIZE_IS  5     The size of the resource has been discovered.
    STREAM_NOTIFY_REDIRECTED    6     The external resource has redirected the stream to an alternate location. Refer to message .
    STREAM_NOTIFY_PROGRESS  7     Indicates current progress of the stream transfer in bytes_transferred and possibly bytes_max as well.
    STREAM_NOTIFY_COMPLETED 8     There is no more data available on the stream.
    STREAM_NOTIFY_FAILURE   9     A generic error occurred on the stream, consult message and message_code for details.
    STREAM_NOTIFY_AUTH_RESULT   10     Authorization has been completed (with or without success).

    STREAM_NOTIFY_SEVERITY_INFO 0     Normal, non-error related, notification.
    STREAM_NOTIFY_SEVERITY_WARN 1     Non critical error condition. Processing may continue.
    STREAM_NOTIFY_SEVERITY_ERR  2     A critical error occurred. Processing cannot continue.
    */
                        var objContext = {
                            responseText : req.responseText,
                            responseXML : req.responseXML,
                            status : req.status,
                            statusText : req.statusText,
                            readyState : req.readyState,
                            evt : aEvt
                        }; // properties are not available in PHP, but offered on notification via 'this' for convenience

                        // notification args: notification_code, severity, message, message_code, bytes_transferred, bytes_max (all int's except string 'message')
                        // Need to add message, etc.
                        var bytes_transferred;
                        switch (req.readyState) {
                            case 0: //     UNINITIALIZED     open() has not been called yet.
                                notification.call(objContext, 0, 0, '', 0, 0, 0);
                                break;
                            case 1: //     LOADING     send() has not been called yet.
                                notification.call(objContext, 0, 0, '', 0, 0, 0);
                                break;
                            case 2: //     LOADED     send() has been called, and headers and status are available.
                                notification.call(objContext, 0, 0, '', 0, 0, 0);
                                break;
                            case 3: //     INTERACTIVE     Downloading; responseText holds partial data.
                                bytes_transferred = req.responseText.length*2; // One character is two bytes
                                notification.call(objContext, 7, 0, '', 0, bytes_transferred, 0);
                                break;
                            case 4: //     COMPLETED     The operation is complete.
                                if (req.status >= 200 && req.status < 400) {
                                    bytes_transferred = req.responseText.length*2; // One character is two bytes
                                    notification.call(objContext, 8, 0, '', req.status, bytes_transferred, 0);
                                }
                                else if (req.status === 403) { // Fix: These two are finished except for message
                                    notification.call(objContext, 10, 2, '', req.status, 0, 0);
                                }
                                else { // Errors
                                    notification.call(objContext, 9, 2, '', req.status, 0, 0);
                                }
                                break;
                            default:
                                throw 'Unrecognized ready state for file_get_contents()';
                        }
                    }
                }
            }
        }

        if (http_stream) {
            var sendHeaders = http_options.header && http_options.header.split(/\r?\n/);
            var userAgentSent = false;
            for (i=0; i < sendHeaders.length; i++) {
                var sendHeader = sendHeaders[i];
                var breakPos = sendHeader.search(/:\s*/);
                var sendHeaderName = sendHeader.substring(0, breakPos);
                req.setRequestHeader(sendHeaderName, sendHeader.substring(breakPos+1));
                if (sendHeaderName === 'User-Agent') {
                    userAgentSent = true;
                }
            }
            if (!userAgentSent) {
                var user_agent = http_options.user_agent ||
                                                                    (ini.user_agent && ini.user_agent.local_value);
                if (user_agent) {
                    req.setRequestHeader('User-Agent', user_agent);
                }
            }
            content = http_options.content || null;
            /*
            // Presently unimplemented HTTP context options
            var request_fulluri = http_options.request_fulluri || false; // When set to TRUE, the entire URI will be used when constructing the request. (i.e. GET http://www.example.com/path/to/file.html HTTP/1.0). While this is a non-standard request format, some proxy servers require it.
            var max_redirects = http_options.max_redirects || 20; // The max number of redirects to follow. Value 1 or less means that no redirects are followed.
            var protocol_version = http_options.protocol_version || 1.0; // HTTP protocol version
            var timeout = http_options.timeout || (ini.default_socket_timeout && ini.default_socket_timeout.local_value); // Read timeout in seconds, specified by a float
            var ignore_errors = http_options.ignore_errors || false; // Fetch the content even on failure status codes.
            */
        }

        if (flagNames & OPTS.FILE_TEXT) { // Overrides how encoding is treated (regardless of what is returned from the server)
            var content_type = 'text/html';
            if (http_options && http_options['phpjs.override']) { // Fix: Could allow for non-HTTP as well
                content_type = http_options['phpjs.override']; // We use this, e.g., in gettext-related functions if character set
                                                                                                        //   overridden earlier by bind_textdomain_codeset()
            }
            else {
                var encoding = (ini['unicode.stream_encoding'] && ini['unicode.stream_encoding'].local_value) || 'UTF-8';
                if (http_options && http_options.header && (/^content-type:/im).test(http_options.header)) { // We'll assume a content-type expects its own specified encoding if present
                    content_type = http_options.header.match(/^content-type:\s*(.*)$/im)[1]; // We let any header encoding stand
                }
                if (!(/;\s*charset=/).test(content_type)) { // If no encoding
                    content_type += '; charset='+encoding;
                }
            }
            req.overrideMimeType(content_type);
        }
        // Default is FILE_BINARY, but for binary, we apparently deviate from PHP in requiring the flag, since many if not
        //     most people will also want a way to have it be auto-converted into native JavaScript text instead
        else if (flagNames & OPTS.FILE_BINARY) { // Trick at https://developer.mozilla.org/En/Using_XMLHttpRequest to get binary
            req.overrideMimeType('text/plain; charset=x-user-defined');
            // Getting an individual byte then requires:
            // responseText.charCodeAt(x) & 0xFF; // throw away high-order byte (f7) where x is 0 to responseText.length-1 (see notes in our substr())
        }

        if (http_options && http_options['phpjs.sendAsBinary']) { // For content sent in a POST or PUT request (use with file_put_contents()?)
            req.sendAsBinary(content); // In Firefox, only available FF3+
        }
        else {
            req.send(content);
        }

        tmp = req.getAllResponseHeaders();
        if (tmp) {
            tmp = tmp.split('\n');
            for (k=0; k < tmp.length; k++) {
                if (func(tmp[k])) {
                    newTmp.push(tmp[k]);
                }
            }
            tmp = newTmp;
            for (i=0; i < tmp.length; i++) {
                headers[i] = tmp[i];
            }
            this.$http_response_header = headers; // see http://php.net/manual/en/reserved.variables.httpresponseheader.php
        }

        if (offset || maxLen) {
            if (maxLen) {
                return req.responseText.substr(offset || 0, maxLen);
            }
            return req.responseText.substr(offset);
        }
        return req.responseText;
    }
    return false;
}



function get_meta_tags (file) {
    // Extracts all meta tag content attributes from a file and returns an array
    //
    // version: 905.3122
    // discuss at: http://phpjs.org/functions/get_meta_tags
    // +   original by: Brett Zamir (http://brett-zamir.me)
    // %        note 1: This function uses XmlHttpRequest and cannot retrieve resource from different domain.
    // %        note 1: Synchronous so may lock up browser, mainly here for study purposes.
    // -    depends on: file_get_contents
    // *     example 1: get_meta_tags('http://kevin.vanzonneveld.net/pj_test_supportfile_2.htm');
    // *     returns 1: {description: 'a php manual', author: 'name', keywords: 'php documentation', 'geo_position': '49.33;-86.59'}
    var fulltxt = '';

    if (false) {
        // Use this for testing instead of the line above:
        fulltxt = '<meta name="author" content="name">'+
        '<meta name="keywords" content="php documentation">'+
        '<meta name="DESCRIPTION" content="a php manual">'+
        '<meta name="geo.position" content="49.33;-86.59">'+
        '</head>';
    } else {
        fulltxt = this.file_get_contents(file).match(/^[\s\S]*<\/head>/i); // We have to disallow some character, so we choose a Unicode non-character
    }

    var patt = /<meta[^>]*?>/gim;
    var patt1 = /<meta\s+.*?name\s*=\s*(['"]?)(.*?)\1\s+.*?content\s*=\s*(['"]?)(.*?)\3/gim;
    var patt2 = /<meta\s+.*?content\s*=\s*(['"?])(.*?)\1\s+.*?name\s*=\s*(['"]?)(.*?)\3/gim;
    var txt, match, name, arr={};

    while ((txt = patt.exec(fulltxt)) !== null) {
        while ((match = patt1.exec(txt)) !== null) {
            name = match[2].replace(/\W/g, '_').toLowerCase();
            arr[name] = match[4];
        }
        while ((match = patt2.exec(txt)) !== null) {
            name = match[4].replace(/\W/g, '_').toLowerCase();
            arr[name] = match[2];
        }
    }
    return arr;
}