default_tests.int_link_info - Code Metrics - Inspection of "#18 Bug Fix & QA Fix" - dylangrech92/seotoolbox - Measure and Improve Code Quality continuously with Scrutinizer

Completed

Push — develop ( f2361e...8024f0 )

by Dylan

created 2016-09-29 11:48 UTC

default_tests.int_link_info D

↳ Parent: js/default_tests.js

Complexity

Conditions	9
Paths	48

Size

Total Lines

Duplication

Lines	0
Ratio	0 %

Importance

Changes

Metric	Value
cc	9
c	0
b	0
f	0
nc	48
nop	5
dl	0
loc	36
rs	4.909

1 Function

Rating	Name	Duplication	Size	Complexity
A		0	4	4

const default_tests = {

    // [name, title, headers, type]
    tests: [
        ['error_pages', 'ERROR PAGES', ['URL'], 'success'],
        ['h1_info', 'H1 INFO', ['URL', 'Count', 'Text', 'Status']],
        ['h2_info', 'H2 INFO', ['URL', 'Count', 'Text', 'Status']],
        ['word_count', 'WORD COUNT', ['URL', 'Word Count', 'Article Word Count']],
        ['int_link_info', 'INTERNAL LINKS',
            ['URL', 'Article Links', 'Article Link Count', 'Article Density', 'Total Link Count', 'Total Density', 'Status'],
            'info'],
        ['ext_link_info', 'EXTERNAL LINKS', ['URL', 'External Link Count', 'External Links'], 'success'],
        ['img_info', 'IMAGES', ['URL', 'Count', 'Missing Alt Tag', 'Missing Title Tag', 'Fields Missing Images', 'Status'], 'success'],
        ['title_info', 'META TITLE', ['URL', 'Meta Title', 'Length', 'Status']],
        ['description_info', 'META DESCRIPTION', ['URL', 'Meta Description', 'Length', 'Status']],
        ['canonical_info', 'CANONICAL', ['URL', 'Status'], 'success'],
        ['noindex_pages', 'NO-INDEX PAGES', ['URL'], 'success'],
        ['urls_test', 'URL STRUCTURE', ['URL', 'Status'], 'success'],
        ['duplicate_meta_tags', 'DUPLICATE META TAGS', ['URL', 'Status'], 'success'],
        ['href_langs', 'LANG TAGS', ['URL', 'Tags'], 'info'],
        ['orphan_pages', 'ORPHAN PAGES', ['URL'], 'info']
    ],

    /**
     * Test the h1s on the page provided
     *
     * @param {string} url
     * @param {jQuery} html
     * @returns {undefined}
     */
    h1_info: function(url, html){
        var h1      = html.find( 'h1' ),
            link    = crawler_painter.create_link(url, url),
            joined  = [],
            status;

        h1.each(function(){ joined.push(this.innerHTML); });

        if(h1.length != 1)
            status = crawler_painter.create_status('error', (h1.length < 1) ? 'Missing H1' : 'Multiple H1 tags');
        else status = crawler_painter.create_status('success', 'OK!');

        return crawler_painter.add_row(this.name, [link, h1.length, joined.join(', '), status]);
    },

    /**
     * Test the h2s on the page provided
     *
     * @param {string} url
     * @param {jQuery} html
     * @returns {undefined}
     */
    h2_info: function(url, html){
        var h2      = html.find( 'h2' ),
            link    = crawler_painter.create_link(url, url),
            joined  = [], status;

        h2.each(function(){ joined.push(this.innerHTML); });

        if(h2.length < 1) status = crawler_painter.create_status('warning', 'Missing H2');
        else status = crawler_painter.create_status('success', 'OK!');

        return crawler_painter.add_row(this.name, [link, h2.length, joined.join(', '), status]);
    },

    /**
     * Check the word count for the passed page
     *
     * @param {string} url
     * @param {jQuery} html
     * @param {string} headers
     * @param {Array} field_data
     * @param {Array} phrases
     * @returns {undefined}
     */
    word_count: function(url, html, headers, field_data, phrases){
        var link        = crawler_painter.create_link(url, url),
            word_count  = crawler.get_word_count(phrases),
            art_count   = crawler.get_word_count(field_data[3]);

        return crawler_painter.add_row(this.name, [link, word_count, art_count]);
    },

    /**
     * Test the internal links found on the page provided
     *
     * @param {string} url
     * @param {jQuery} html
     * @param {string} headers
     * @param {Array} field_data
     * @param {Array} phrases
     * @returns {undefined}
     */
    int_link_info: function(url, html, headers, field_data, phrases){
        var link = crawler_painter.create_link(url, url),
            art_links = [], links = [];

        // Article links
        for( var field in field_data[2] ) {
            $.each($(field_data[2][field]).find('a'), function () {
                var href = $(this).attr('href');
                if(href && !crawler.is_external(href) && !crawler.is_anchor(href, url)) art_links.push(href);
            });
        }

        // Full page links
        $.each(html.find('a'), function () {
            var href = $(this).attr('href');
            if(href && !crawler.is_external(href) && !crawler.is_anchor(href, url)) links.push(href);
        });

        var art_word_count  = crawler.get_word_count(field_data[3]),
            art_density     = (art_links.length > 0) ? art_word_count / art_links.length : false,
            art_dens_text   = (art_density != false) ? art_density.toFixed(2) +' words/link' : 'No internal links',
            word_count      = crawler.get_word_count(phrases),
            density         = (links.length > 0) ? word_count / links.length : false,
            dens_text       = (density != false) ? density.toFixed(2) +' words/link' : 'No internal links',
            status          = crawler_painter.create_status('success', 'OK!');

        if( ( art_density !== false && art_density < 100 ) )
            status = crawler_painter.create_status('warning', 'This page might be considered spammy');

        if(links.length > 0)
            crawler_painter.add_row( this.name, [
                link, art_links.join('<br />'), art_links.length, art_dens_text, links.length, dens_text, status
            ]);

        return undefined;
    },

    /**
     * Test the external links on the page provided
     *
     * @param {string} url
     * @param {jQuery} html
     * @param {string} headers
     * @param {Array} field_data
     * @returns {undefined}
     */
    ext_link_info: function(url, html, headers, field_data){
        var link = crawler_painter.create_link(url, url),
            links = [];

        for( var field in field_data[2] ) {
            $.each($(field_data[2][field]).find('a'), function () {
                var $this = $(this),
                    href = $this.attr('href');
                if(href && crawler.is_external(href)){
                    var type = ( !$this.attr('rel') || $this.attr('rel').toLowerCase().indexOf('nofollow') < 0 )
                        ? 'warning' : 'info';
                    links.push(
                        $('<div class="clearfix"></div>').append([
                            crawler_painter.create_status(type, href),
                            '<p>&nbsp;</p>'
                        ])
                    );
                }
            });
        }

        if(links.length > 0){
            crawler_painter.add_row(this.name, [link, links.length, links]);
        }

        return undefined;
    },

    /**
     * Test the images on the page provided
     *
     * @param {string} url
     * @param {jQuery} html
     * @param {string} headers
     * @param {Array} field_data
     * @returns {undefined}
     */
    img_info: function(url, html, headers, field_data) {
        var link = crawler_painter.create_link(url, url),
            imgs = html.find('img'),
            alt = 0, title = 0, fields = [], status = '';

        // Check alt and title tags
        $.each(imgs, function () {
            var $this = $(this);
            if (!$this.attr('alt') || $this.attr('alt').length < 1) alt += 1;
            if (!$this.attr('title') || $this.attr('title').length < 1) title += 1;
        });

        // Check the fields
        for (var f in field_data[2]) if ($(field_data[2][f]).find('img').length < 1) fields.push(field_data[1][f]);

        // Construct Result
        if (alt > 0)
            status = crawler_painter.create_status('error',
                (alt > 1) ? alt + ' images missing alt tag' : '1 image missing alt tag');
        else if(fields.length > 0)
            status = crawler_painter.create_status('warning',
                (fields.length > 1) ? fields.join(' and ') + ' are missing images' : fields[0] + ' is missing images');
        else if(title > 0)
            status = crawler_painter.create_status('info',
                (title > 1) ? title + ' images missing title tag' : '1 image is missing title tag');
        else
            status = crawler_painter.create_status('success', 'OK!');

        return crawler_painter.add_row(this.name, [link, imgs.length, alt, title, fields.join(', '), status]);
    },

    /**
     * Test the meta title of the page provided
     *
     * @param {string} url
     * @param {jQuery} html
     * @returns {undefined}
     */
    title_info: function(url, html){
        var title   = html.filter( 'title' ),
            link    = crawler_painter.create_link(url, url),
            text    = (title.length == 1) ? title.html() : '',
            status  = default_tests.get_meta_tags_status(title, 'meta title', text, 40, 56);

        if(title.length == 1){
            crawler.set_property('meta_titles', text, url);
        }

        return crawler_painter.add_row(this.name, [link, text, text.length, status]);
    },

    /**
     * Test the meta description for the page provided
     *
     * @param {string} url
     * @param {jQuery} html
     * @returns {undefined}
     */
    description_info: function(url, html){
        var desc    = html.filter( 'meta[name=description]' ),
            link    = crawler_painter.create_link(url, url),
            text    = (desc.length == 1) ? desc.attr('content') : '',
            status  = default_tests.get_meta_tags_status(desc, 'meta description', text, 70, 156);

        if( desc.length == 1 ){
            crawler.set_property('descriptions', text, url);
        }

        return crawler_painter.add_row(this.name, [link, text, text.length, status]);
    },

    /**
     * Test the canonical rules for the page provided
     *
     * @param {string} url
     * @param {jQuery} html
     * @returns {undefined}
     */
    canonical_info: function(url, html){
        var tags = default_tests.get_tags(html, 'link', 'rel', 'canonical');

        if(tags.length != 1) {
            var status = crawler_painter.create_status('error', 'Missing / Multiple canonicals found');
            crawler_painter.add_row(this.name, [crawler_painter.create_link(url, url), status]);
        }else{
            crawler.set_property('canonicals', tags[0].attr('href'), url);
        }

        return undefined;
    },

    /**
     * Check if the page provided has a no-index header
     *
     * @param {string} url
     * @param {jQuery} html
     * @returns {undefined}
     */
    noindex_pages: function(url, html) {
        if(default_tests.get_tags(html, 'meta', 'content', 'noindex').length > 0){
            crawler_painter.add_row(this.name, [crawler_painter.create_link(url, url)]);
            crawler_painter.set_type(this.name, 'error');
        }

        return undefined;
    },

    /**
     * Test the url passed for it's structure
     *
     * @param url
     * @returns {undefined}
     */
    urls_test: function(url){
        var link = crawler_painter.create_link(url, url),
            msg;

        if( url.length > 115 )                  msg = 'URL is too long';
        else if( url.toLowerCase() != url )     msg = 'URL is not in lower case';
        else if( url.replace('_','') !== url )  msg = 'URL contains under scores';
        else return undefined;

        return crawler_painter.add_row(this.name, [link, crawler_painter.create_status('warning', msg)]);
    },

    /**
     * Check for href lang tags in the page provided
     *
     * @param {string} url
     * @param {jQuery} html
     * @returns {undefined}
     */
    href_langs: function(url, html){
        var link    = crawler_painter.create_link(url, url),
            tags    = [];

        $.each( html.filter( 'link' ), function(){
            if( $(this).attr( 'hreflang' ) )
                tags.push( $('<p>').text( $(this).clone().wrap('<p>').parent().html() ).html() );
        });

        if( tags.length > 0 ) crawler_painter.add_row(this.name, [link, tags.join('<br />')] );

        return undefined;
    },

    /**
     * Returns a list of jQuery Objects that are of type {tag},
     * have an attribute {key} an it's value is {value}
     *
     * @param {jQuery} html
     * @param {string} tag
     * @param {string} key
     * @param {string} value
     * @returns {Array}
     */
    get_tags: function(html, tag, key, value){
        var returns = [];

        $.each(html.filter(tag), function(){
            var $this = $(this);
            if( $this.attr(key) && $this.attr(key) == value ){
                returns.push($this);
            }
        });

        return returns;
    },

    /**
     * Goes through an object and tries to find a key that has a value matching the value passed
     *
     * @param {*} object
     * @param {*} search
     * @returns {*}
     */
    get_key_from_object: function(object, search){
        for( var key in object ) if( object[key].indexOf(search) >= 0 ) return key;
        return undefined;
    },

    /**
     * Gets the status box for the meta tag being tested
     * Append to the crawler_painter
     *
     * @param {Array} tags
     * @param {string} tag_name
     * @param {string} text
     * @param {int} min_char
     * @param {int} max_char
     * @returns {jQuery}
     */
    get_meta_tags_status: function(tags, tag_name, text, min_char, max_char){
        if( tags.length > 1 ){
            return crawler_painter.create_status('error', 'Multiple '+tag_name+' tags');
        }else if( tags.length < 1 ){
            return crawler_painter.create_status('error', 'Missing '+tag_name+' tag');
        }else{
            var len = text.length;
            if(len < min_char){
                return crawler_painter.create_status('warning', tag_name+' is too short');
            }else if(len > max_char){
                return crawler_painter.create_status('warning', tag_name+' is too long');
            }else{
                return crawler_painter.create_status('success', 'OK!');
            }
        }
    }
}

// Register the tests
crawler.on('BEFORE_INIT', function(){
    for( var t in default_tests.tests ){
        var test = default_tests.tests[t],
            func = default_tests.hasOwnProperty( test[0] ) ? default_tests[test[0]] : false;

        crawler.regiser_test(test[0], test[1], test[2], func);
        crawler_painter.set_type(test[0], test[3] || 'default');
    }
});

// When a crawl fails add an error page
crawler.on('CRAWL_LOAD_FAILED', function(url){
    crawler_painter.add_row('error_pages', [url]);
    crawler_painter.set_type('error_pages', 'error');
});

// When crawler is done check for orphan pages
crawler.on('ALL_CRAWLS_FINISHED', function(){
    pages_loop:
        for( var i in crawler.tested ){
            var url = crawler.tested[i];

            if( crawler.failed.indexOf(url) >= 0 ){
                continue;
            }

            if( crawler.linked_from.hasOwnProperty(url) ) {
                for (var x in crawler.linked_from[url])
                    if (crawler.linked_from[url][x] != url) continue pages_loop;
            }

            crawler.add_row('orphan_pages', [crawler_painter.create_link(crawler.tested[i], crawler.tested[i])]);
            crawler_painter.set_type('orphan_pages', 'error');
        }

    return true;
});

// When crawler is done check for orphan pages
crawler.on('ALL_CRAWLS_FINISHED', function(){
    var canonicals = crawler.canonicals,
        tests      = {
            'meta_titles'   : 'Urls have same meta title but different canonicals',
            'descriptions'  : 'Urls have same meta description but different canonicals'
        };

    for(var test in tests){
        for(var x in crawler[test]){
            var urls = crawler[test][x];
            if( urls < 2 ) continue;
            var canonical = default_tests.get_key_from_object(canonicals, urls[0]);
            for( var i in urls )
                if( canonical != default_tests.get_key_from_object(canonicals, urls[i]) ) {
                    var status = crawler_painter.create_status('error', tests[test]);
                    crawler_painter.add_row('duplicate_meta_tags', [urls.join(', '), status]);
                    break;
                }
        }
    }

    return undefined;
});


1			const default_tests = {
2
3			// [name, title, headers, type]
4			tests: [
5			['error_pages', 'ERROR PAGES', ['URL'], 'success'],
6			['h1_info', 'H1 INFO', ['URL', 'Count', 'Text', 'Status']],
7			['h2_info', 'H2 INFO', ['URL', 'Count', 'Text', 'Status']],
8			['word_count', 'WORD COUNT', ['URL', 'Word Count', 'Article Word Count']],
9			['int_link_info', 'INTERNAL LINKS',
10			['URL', 'Article Links', 'Article Link Count', 'Article Density', 'Total Link Count', 'Total Density', 'Status'],
11			'info'],
12			['ext_link_info', 'EXTERNAL LINKS', ['URL', 'External Link Count', 'External Links'], 'success'],
13			['img_info', 'IMAGES', ['URL', 'Count', 'Missing Alt Tag', 'Missing Title Tag', 'Fields Missing Images', 'Status'], 'success'],
14			['title_info', 'META TITLE', ['URL', 'Meta Title', 'Length', 'Status']],
15			['description_info', 'META DESCRIPTION', ['URL', 'Meta Description', 'Length', 'Status']],
16			['canonical_info', 'CANONICAL', ['URL', 'Status'], 'success'],
17			['noindex_pages', 'NO-INDEX PAGES', ['URL'], 'success'],
18			['urls_test', 'URL STRUCTURE', ['URL', 'Status'], 'success'],
19			['duplicate_meta_tags', 'DUPLICATE META TAGS', ['URL', 'Status'], 'success'],
20			['href_langs', 'LANG TAGS', ['URL', 'Tags'], 'info'],
21			['orphan_pages', 'ORPHAN PAGES', ['URL'], 'info']
22			],
23
24			/**
25			* Test the h1s on the page provided
26			*
27			* @param {string} url
28			* @param {jQuery} html
29			* @returns {undefined}
30			*/
31			h1_info: function(url, html){
32			var h1 = html.find( 'h1' ),
33			link = crawler_painter.create_link(url, url),
34			joined = [],
35			status;
36
37			h1.each(function(){ joined.push(this.innerHTML); });
38
39			if(h1.length != 1)
40			status = crawler_painter.create_status('error', (h1.length < 1) ? 'Missing H1' : 'Multiple H1 tags');
41			else status = crawler_painter.create_status('success', 'OK!');
42
43			return crawler_painter.add_row(this.name, [link, h1.length, joined.join(', '), status]);
44			},
45
46			/**
47			* Test the h2s on the page provided
48			*
49			* @param {string} url
50			* @param {jQuery} html
51			* @returns {undefined}
52			*/
53			h2_info: function(url, html){
54			var h2 = html.find( 'h2' ),
55			link = crawler_painter.create_link(url, url),
56			joined = [], status;
57
58			h2.each(function(){ joined.push(this.innerHTML); });
59
60			if(h2.length < 1) status = crawler_painter.create_status('warning', 'Missing H2');
61			else status = crawler_painter.create_status('success', 'OK!');
62
63			return crawler_painter.add_row(this.name, [link, h2.length, joined.join(', '), status]);
64			},
65
66			/**
67			* Check the word count for the passed page
68			*
69			* @param {string} url
70			* @param {jQuery} html
71			* @param {string} headers
72			* @param {Array} field_data
73			* @param {Array} phrases
74			* @returns {undefined}
75			*/
76			word_count: function(url, html, headers, field_data, phrases){
77			var link = crawler_painter.create_link(url, url),
78			word_count = crawler.get_word_count(phrases),
79			art_count = crawler.get_word_count(field_data[3]);
80
81			return crawler_painter.add_row(this.name, [link, word_count, art_count]);
82			},
83
84			/**
85			* Test the internal links found on the page provided
86			*
87			* @param {string} url
88			* @param {jQuery} html
89			* @param {string} headers
90			* @param {Array} field_data
91			* @param {Array} phrases
92			* @returns {undefined}
93			*/
94			int_link_info: function(url, html, headers, field_data, phrases){
95			var link = crawler_painter.create_link(url, url),
96			art_links = [], links = [];
97
98			// Article links
99			for( var field in field_data[2] ) {
100			$.each($(field_data[2][field]).find('a'), function () {
101			var href = $(this).attr('href');
102			if(href && !crawler.is_external(href) && !crawler.is_anchor(href, url)) art_links.push(href);
103			});
104			}
105
106			// Full page links
107			$.each(html.find('a'), function () {
108			var href = $(this).attr('href');
109			if(href && !crawler.is_external(href) && !crawler.is_anchor(href, url)) links.push(href);
110			});
111
112			var art_word_count = crawler.get_word_count(field_data[3]),
113			art_density = (art_links.length > 0) ? art_word_count / art_links.length : false,
114			art_dens_text = (art_density != false) ? art_density.toFixed(2) +' words/link' : 'No internal links',
115			word_count = crawler.get_word_count(phrases),
116			density = (links.length > 0) ? word_count / links.length : false,
117			dens_text = (density != false) ? density.toFixed(2) +' words/link' : 'No internal links',
118			status = crawler_painter.create_status('success', 'OK!');
119
120			if( ( art_density !== false && art_density < 100 ) )
121			status = crawler_painter.create_status('warning', 'This page might be considered spammy');
122
123			if(links.length > 0)
124			crawler_painter.add_row( this.name, [
125			link, art_links.join('<br />'), art_links.length, art_dens_text, links.length, dens_text, status
126			]);
127
128			return undefined;
129			},
130
131			/**
132			* Test the external links on the page provided
133			*
134			* @param {string} url
135			* @param {jQuery} html
136			* @param {string} headers
137			* @param {Array} field_data
138			* @returns {undefined}
139			*/
140			ext_link_info: function(url, html, headers, field_data){
141			var link = crawler_painter.create_link(url, url),
142			links = [];
143
144			for( var field in field_data[2] ) {
145			$.each($(field_data[2][field]).find('a'), function () {
146			var $this = $(this),
147			href = $this.attr('href');
148			if(href && crawler.is_external(href)){
149			var type = ( !$this.attr('rel') \|\| $this.attr('rel').toLowerCase().indexOf('nofollow') < 0 )
150			? 'warning' : 'info';
151			links.push(
152			$('<div class="clearfix"></div>').append([
153			crawler_painter.create_status(type, href),
154			'<p> </p>'
155			])
156			);
157			}
158			});
159			}
160
161			if(links.length > 0){
162			crawler_painter.add_row(this.name, [link, links.length, links]);
163			}
164
165			return undefined;
166			},
167
168			/**
169			* Test the images on the page provided
170			*
171			* @param {string} url
172			* @param {jQuery} html
173			* @param {string} headers
174			* @param {Array} field_data
175			* @returns {undefined}
176			*/
177			img_info: function(url, html, headers, field_data) {
178			var link = crawler_painter.create_link(url, url),
179			imgs = html.find('img'),
180			alt = 0, title = 0, fields = [], status = '';
181
182			// Check alt and title tags
183			$.each(imgs, function () {
184			var $this = $(this);
185			if (!$this.attr('alt') \|\| $this.attr('alt').length < 1) alt += 1;
186			if (!$this.attr('title') \|\| $this.attr('title').length < 1) title += 1;
187			});
188
189			// Check the fields
190			for (var f in field_data[2]) if ($(field_data[2][f]).find('img').length < 1) fields.push(field_data[1][f]);
191
192			// Construct Result
193			if (alt > 0)
194			status = crawler_painter.create_status('error',
195			(alt > 1) ? alt + ' images missing alt tag' : '1 image missing alt tag');
196			else if(fields.length > 0)
197			status = crawler_painter.create_status('warning',
198			(fields.length > 1) ? fields.join(' and ') + ' are missing images' : fields[0] + ' is missing images');
199			else if(title > 0)
200			status = crawler_painter.create_status('info',
201			(title > 1) ? title + ' images missing title tag' : '1 image is missing title tag');
202			else
203			status = crawler_painter.create_status('success', 'OK!');
204
205			return crawler_painter.add_row(this.name, [link, imgs.length, alt, title, fields.join(', '), status]);
206			},
207
208			/**
209			* Test the meta title of the page provided
210			*
211			* @param {string} url
212			* @param {jQuery} html
213			* @returns {undefined}
214			*/
215			title_info: function(url, html){
216			var title = html.filter( 'title' ),
217			link = crawler_painter.create_link(url, url),
218			text = (title.length == 1) ? title.html() : '',
219			status = default_tests.get_meta_tags_status(title, 'meta title', text, 40, 56);
220
221			if(title.length == 1){
222			crawler.set_property('meta_titles', text, url);
223			}
224
225			return crawler_painter.add_row(this.name, [link, text, text.length, status]);
226			},
227
228			/**
229			* Test the meta description for the page provided
230			*
231			* @param {string} url
232			* @param {jQuery} html
233			* @returns {undefined}
234			*/
235			description_info: function(url, html){
236			var desc = html.filter( 'meta[name=description]' ),
237			link = crawler_painter.create_link(url, url),
238			text = (desc.length == 1) ? desc.attr('content') : '',
239			status = default_tests.get_meta_tags_status(desc, 'meta description', text, 70, 156);
240
241			if( desc.length == 1 ){
242			crawler.set_property('descriptions', text, url);
243			}
244
245			return crawler_painter.add_row(this.name, [link, text, text.length, status]);
246			},
247
248			/**
249			* Test the canonical rules for the page provided
250			*
251			* @param {string} url
252			* @param {jQuery} html
253			* @returns {undefined}
254			*/
255			canonical_info: function(url, html){
256			var tags = default_tests.get_tags(html, 'link', 'rel', 'canonical');
257
258			if(tags.length != 1) {
259			var status = crawler_painter.create_status('error', 'Missing / Multiple canonicals found');
260			crawler_painter.add_row(this.name, [crawler_painter.create_link(url, url), status]);
261			}else{
262			crawler.set_property('canonicals', tags[0].attr('href'), url);
263			}
264
265			return undefined;
266			},
267
268			/**
269			* Check if the page provided has a no-index header
270			*
271			* @param {string} url
272			* @param {jQuery} html
273			* @returns {undefined}
274			*/
275			noindex_pages: function(url, html) {
276			if(default_tests.get_tags(html, 'meta', 'content', 'noindex').length > 0){
277			crawler_painter.add_row(this.name, [crawler_painter.create_link(url, url)]);
278			crawler_painter.set_type(this.name, 'error');
279			}
280
281			return undefined;
282			},
283
284			/**
285			* Test the url passed for it's structure
286			*
287			* @param url
288			* @returns {undefined}
289			*/
290			urls_test: function(url){
291			var link = crawler_painter.create_link(url, url),
292			msg;
293
294			if( url.length > 115 ) msg = 'URL is too long';
295			else if( url.toLowerCase() != url ) msg = 'URL is not in lower case';
296			else if( url.replace('_','') !== url ) msg = 'URL contains under scores';
297			else return undefined;
298
299			return crawler_painter.add_row(this.name, [link, crawler_painter.create_status('warning', msg)]);
300			},
301
302			/**
303			* Check for href lang tags in the page provided
304			*
305			* @param {string} url
306			* @param {jQuery} html
307			* @returns {undefined}
308			*/
309			href_langs: function(url, html){
310			var link = crawler_painter.create_link(url, url),
311			tags = [];
312
313			$.each( html.filter( 'link' ), function(){
314			if( $(this).attr( 'hreflang' ) )
315			tags.push( $('<p>').text( $(this).clone().wrap('<p>').parent().html() ).html() );
316			});
317
318			if( tags.length > 0 ) crawler_painter.add_row(this.name, [link, tags.join('<br />')] );
319
320			return undefined;
321			},
322
323			/**
324			* Returns a list of jQuery Objects that are of type {tag},
325			* have an attribute {key} an it's value is {value}
326			*
327			* @param {jQuery} html
328			* @param {string} tag
329			* @param {string} key
330			* @param {string} value
331			* @returns {Array}
332			*/
333			get_tags: function(html, tag, key, value){
334			var returns = [];
335
336			$.each(html.filter(tag), function(){
337			var $this = $(this);
338			if( $this.attr(key) && $this.attr(key) == value ){
339			returns.push($this);
340			}
341			});
342
343			return returns;
344			},
345
346			/**
347			* Goes through an object and tries to find a key that has a value matching the value passed
348			*
349			* @param {*} object
350			* @param {*} search
351			* @returns {*}
352			*/
353			get_key_from_object: function(object, search){
354			for( var key in object ) if( object[key].indexOf(search) >= 0 ) return key;
355			return undefined;
356			},
357
358			/**
359			* Gets the status box for the meta tag being tested
360			* Append to the crawler_painter
361			*
362			* @param {Array} tags
363			* @param {string} tag_name
364			* @param {string} text
365			* @param {int} min_char
366			* @param {int} max_char
367			* @returns {jQuery}
368			*/
369			get_meta_tags_status: function(tags, tag_name, text, min_char, max_char){
370			if( tags.length > 1 ){
371			return crawler_painter.create_status('error', 'Multiple '+tag_name+' tags');
372			}else if( tags.length < 1 ){
373			return crawler_painter.create_status('error', 'Missing '+tag_name+' tag');
374			}else{
375			var len = text.length;
376			if(len < min_char){
377			return crawler_painter.create_status('warning', tag_name+' is too short');
378			}else if(len > max_char){
379			return crawler_painter.create_status('warning', tag_name+' is too long');
380			}else{
381			return crawler_painter.create_status('success', 'OK!');
382			}
383			}
384			}
385			}
386
387			// Register the tests
388			crawler.on('BEFORE_INIT', function(){
389			for( var t in default_tests.tests ){
390			var test = default_tests.tests[t],
391			func = default_tests.hasOwnProperty( test[0] ) ? default_tests[test[0]] : false;
392
393			crawler.regiser_test(test[0], test[1], test[2], func);
394			crawler_painter.set_type(test[0], test[3] \|\| 'default');
395			}
396			});
397
398			// When a crawl fails add an error page
399			crawler.on('CRAWL_LOAD_FAILED', function(url){
400			crawler_painter.add_row('error_pages', [url]);
401			crawler_painter.set_type('error_pages', 'error');
402			});
403
404			// When crawler is done check for orphan pages
405			crawler.on('ALL_CRAWLS_FINISHED', function(){
406			pages_loop:
407			for( var i in crawler.tested ){
408			var url = crawler.tested[i];
409
410			if( crawler.failed.indexOf(url) >= 0 ){
411			continue;
412			}
413
414			if( crawler.linked_from.hasOwnProperty(url) ) {
415			for (var x in crawler.linked_from[url])
416			if (crawler.linked_from[url][x] != url) continue pages_loop;
417			}
418
419			crawler.add_row('orphan_pages', [crawler_painter.create_link(crawler.tested[i], crawler.tested[i])]);
420			crawler_painter.set_type('orphan_pages', 'error');
421			}
422
423			return true;
424			});
425
426			// When crawler is done check for orphan pages
427			crawler.on('ALL_CRAWLS_FINISHED', function(){
428			var canonicals = crawler.canonicals,
429			tests = {
430			'meta_titles' : 'Urls have same meta title but different canonicals',
431			'descriptions' : 'Urls have same meta description but different canonicals'
432			};
433
434			for(var test in tests){
435			for(var x in crawler[test]){
436			var urls = crawler[test][x];
437			if( urls < 2 ) continue;
438			var canonical = default_tests.get_key_from_object(canonicals, urls[0]);
439			for( var i in urls )
440			if( canonical != default_tests.get_key_from_object(canonicals, urls[i]) ) {
441			var status = crawler_painter.create_status('error', tests[test]);
442			crawler_painter.add_row('duplicate_meta_tags', [urls.join(', '), status]);
443			break;
444			}
445			}
446			}
447
448			return undefined;
449			});
450

dylangrech92 / seotoolbox

Push — develop ( f2361e...8024f0 )

default_tests.int_link_info D

Complexity

Size

Duplication

Importance

1 Function

Duplication Side-by-Side

Filter issues like