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