site_checker 0.0.0 → 0.1.1
Sign up to get free protection for your applications and to get access to all the features.
- data/lib/site_checker.rb +224 -0
- metadata +45 -11
data/lib/site_checker.rb
CHANGED
@@ -0,0 +1,224 @@
|
|
1
|
+
require 'nokogiri'
|
2
|
+
require 'open-uri'
|
3
|
+
|
4
|
+
class SiteChecker
|
5
|
+
attr_accessor :problems
|
6
|
+
attr_accessor :ignore_list, :visit_references, :max_recursion_depth
|
7
|
+
|
8
|
+
def initialize()
|
9
|
+
yield self if block_given?
|
10
|
+
@ignore_list ||= []
|
11
|
+
@visit_references ||= false
|
12
|
+
@max_recursion_depth ||= -1
|
13
|
+
end
|
14
|
+
|
15
|
+
def check(url, root)
|
16
|
+
@visits = {}
|
17
|
+
@problems = {}
|
18
|
+
@recursion_depth = 0
|
19
|
+
|
20
|
+
@root = root
|
21
|
+
|
22
|
+
register_visit(:local_page, url)
|
23
|
+
process_local_page(url, nil)
|
24
|
+
end
|
25
|
+
|
26
|
+
def local_pages
|
27
|
+
@visits[:local_page]
|
28
|
+
end
|
29
|
+
|
30
|
+
def remote_pages
|
31
|
+
@visits[:remote_page]
|
32
|
+
end
|
33
|
+
|
34
|
+
def local_images
|
35
|
+
@visits[:local_image]
|
36
|
+
end
|
37
|
+
|
38
|
+
def remote_images
|
39
|
+
@visits[:remote_image]
|
40
|
+
end
|
41
|
+
|
42
|
+
private
|
43
|
+
def process_local_page(url, parent_url)
|
44
|
+
links = collect_links(url, parent_url)
|
45
|
+
|
46
|
+
filter_out_working_anchors!(links)
|
47
|
+
report_and_remove_anchors!(links, parent_url)
|
48
|
+
|
49
|
+
links.each do |link, kind|
|
50
|
+
if kind != :anchor
|
51
|
+
visit(kind, url, link) unless visited?(kind, link)
|
52
|
+
else
|
53
|
+
end
|
54
|
+
end
|
55
|
+
end
|
56
|
+
|
57
|
+
def register_visit(kind, link)
|
58
|
+
@visits[kind] = [] unless @visits.has_key?(kind)
|
59
|
+
@visits[kind] << link
|
60
|
+
end
|
61
|
+
|
62
|
+
def visited?(kind, link)
|
63
|
+
@visits[kind] = [] unless @visits.has_key?(kind)
|
64
|
+
@visits[kind].include?(link)
|
65
|
+
end
|
66
|
+
|
67
|
+
def visit(kind, parent_url, link)
|
68
|
+
register_visit(kind, link)
|
69
|
+
if kind != :local_page
|
70
|
+
open_reference(kind, link, parent_url)
|
71
|
+
else
|
72
|
+
unless stop_recursion?
|
73
|
+
@recursion_depth += 1
|
74
|
+
process_local_page(link, parent_url)
|
75
|
+
@recursion_depth -= 1
|
76
|
+
end
|
77
|
+
end
|
78
|
+
end
|
79
|
+
|
80
|
+
def open_reference(kind, link, parent_url)
|
81
|
+
content = nil
|
82
|
+
begin
|
83
|
+
if kind == :local_page
|
84
|
+
if URI(@root).absolute?
|
85
|
+
content = open(link)
|
86
|
+
else
|
87
|
+
link = add_index_html(link)
|
88
|
+
content = File.open(link).read
|
89
|
+
end
|
90
|
+
elsif kind == :local_image
|
91
|
+
if URI(@root).absolute?
|
92
|
+
open(link)
|
93
|
+
else
|
94
|
+
File.open(link)
|
95
|
+
end
|
96
|
+
elsif @visit_references
|
97
|
+
open(link)
|
98
|
+
end
|
99
|
+
rescue OpenURI::HTTPError => e
|
100
|
+
new_problem(strip_root(parent_url), "#{strip_root(link)} (#{e.message.strip})")
|
101
|
+
rescue Errno::ENOENT => e
|
102
|
+
link = remove_index_html(link) if kind == :local_page
|
103
|
+
new_problem(strip_root(parent_url), "#{strip_root(link)} (404 Not Found)")
|
104
|
+
rescue => e
|
105
|
+
new_problem(strip_root(parent_url), "#{strip_root(link)} (#{e.message.strip})")
|
106
|
+
end
|
107
|
+
content
|
108
|
+
end
|
109
|
+
|
110
|
+
def filter_out_working_anchors!(links)
|
111
|
+
links.delete_if{ |link, kind| (kind == :local_page && has_anchor?(links, link)) }
|
112
|
+
end
|
113
|
+
|
114
|
+
def report_and_remove_anchors!(links, parent_url)
|
115
|
+
anchors = links.select {|link, kind| link.match(/^.+#.+$/) && kind == :local_page}
|
116
|
+
anchors.each do |anchor, kind|
|
117
|
+
new_problem(strip_root(parent_url), "#{strip_root(anchor)} (404 Not Found)")
|
118
|
+
links.delete(anchor)
|
119
|
+
end
|
120
|
+
end
|
121
|
+
|
122
|
+
def has_anchor?(links, link)
|
123
|
+
anchor = link.gsub(/^.+#/, "")
|
124
|
+
links.has_key?(anchor) && links[anchor] == :anchor
|
125
|
+
end
|
126
|
+
|
127
|
+
|
128
|
+
def absolute_reference?(link)
|
129
|
+
link.start_with?(@root)
|
130
|
+
end
|
131
|
+
|
132
|
+
def relative_reference?(link)
|
133
|
+
link =~ /^\/.+/
|
134
|
+
end
|
135
|
+
|
136
|
+
def collect_links(url, parent_url)
|
137
|
+
links = {}
|
138
|
+
content = open_reference(:local_page, url, parent_url)
|
139
|
+
if content
|
140
|
+
doc = Nokogiri(content)
|
141
|
+
doc.xpath("//img").reject {|img| ignored?(img['src'])}.each do |img|
|
142
|
+
link_kind = detect_link_and_kind(img['src'], url, :remote_image, :local_image)
|
143
|
+
links.merge!(link_kind) unless link_kind.empty?
|
144
|
+
end
|
145
|
+
doc.xpath("//a").reject {|a| ignored?(a['href'])}.each do |a|
|
146
|
+
link_kind = detect_link_and_kind(a['href'], url, :remote_page, :local_page)
|
147
|
+
links.merge!(link_kind) unless link_kind.empty?
|
148
|
+
end
|
149
|
+
|
150
|
+
doc.xpath("//a").reject {|a| !a['id']}.each do |a|
|
151
|
+
links.merge!({a['id'] => :anchor})
|
152
|
+
end
|
153
|
+
end
|
154
|
+
links
|
155
|
+
end
|
156
|
+
|
157
|
+
def detect_link_and_kind(reference, url, external_kind, local_kind)
|
158
|
+
link_kind = {}
|
159
|
+
link = URI(strip_trailing_slash(reference))
|
160
|
+
if link.to_s.start_with?(@root)
|
161
|
+
new_problem(url, "#{link} (absolute path)")
|
162
|
+
else
|
163
|
+
if URI(reference).absolute?
|
164
|
+
link_kind[link.to_s] = external_kind
|
165
|
+
else
|
166
|
+
link_kind[create_absolute_reference(link.to_s)] = local_kind
|
167
|
+
end
|
168
|
+
end
|
169
|
+
link_kind
|
170
|
+
end
|
171
|
+
|
172
|
+
def strip_trailing_slash(link)
|
173
|
+
link.gsub(/\/$/, "")
|
174
|
+
end
|
175
|
+
|
176
|
+
def strip_root(link)
|
177
|
+
if link
|
178
|
+
link.gsub(/^#{@root}[\/]?/, "")
|
179
|
+
else
|
180
|
+
""
|
181
|
+
end
|
182
|
+
end
|
183
|
+
|
184
|
+
def add_index_html(path)
|
185
|
+
path.end_with?(".html") ? path : File.join(path, "index.html")
|
186
|
+
end
|
187
|
+
|
188
|
+
def remove_index_html(path)
|
189
|
+
path.gsub(/\/index.html$/, "")
|
190
|
+
end
|
191
|
+
|
192
|
+
def create_absolute_reference(link)
|
193
|
+
root = URI(@root)
|
194
|
+
if root.absolute?
|
195
|
+
root.merge(link).to_s.gsub(/\/$/, "")
|
196
|
+
else
|
197
|
+
File.join(root.path, link)
|
198
|
+
end
|
199
|
+
end
|
200
|
+
|
201
|
+
def new_problem(url, message)
|
202
|
+
url = @root if url.empty?
|
203
|
+
@problems[url] = [] unless problems.has_key?(url)
|
204
|
+
@problems[url] << message
|
205
|
+
end
|
206
|
+
|
207
|
+
def ignored?(link)
|
208
|
+
if link
|
209
|
+
@ignore_list.include? link
|
210
|
+
else
|
211
|
+
true
|
212
|
+
end
|
213
|
+
end
|
214
|
+
|
215
|
+
def stop_recursion?
|
216
|
+
if @max_recursion_depth == -1
|
217
|
+
false
|
218
|
+
elsif @max_recursion_depth > @recursion_depth
|
219
|
+
false
|
220
|
+
else
|
221
|
+
true
|
222
|
+
end
|
223
|
+
end
|
224
|
+
end
|
metadata
CHANGED
@@ -1,12 +1,13 @@
|
|
1
1
|
--- !ruby/object:Gem::Specification
|
2
2
|
name: site_checker
|
3
3
|
version: !ruby/object:Gem::Version
|
4
|
-
|
4
|
+
hash: 25
|
5
|
+
prerelease:
|
5
6
|
segments:
|
6
7
|
- 0
|
7
|
-
-
|
8
|
-
-
|
9
|
-
version: 0.
|
8
|
+
- 1
|
9
|
+
- 1
|
10
|
+
version: 0.1.1
|
10
11
|
platform: ruby
|
11
12
|
authors:
|
12
13
|
- Zsolt Fabok
|
@@ -14,10 +15,40 @@ autorequire:
|
|
14
15
|
bindir: bin
|
15
16
|
cert_chain: []
|
16
17
|
|
17
|
-
date: 2012-07-
|
18
|
-
|
19
|
-
|
20
|
-
|
18
|
+
date: 2012-07-18 00:00:00 Z
|
19
|
+
dependencies:
|
20
|
+
- !ruby/object:Gem::Dependency
|
21
|
+
name: rspec
|
22
|
+
prerelease: false
|
23
|
+
requirement: &id001 !ruby/object:Gem::Requirement
|
24
|
+
none: false
|
25
|
+
requirements:
|
26
|
+
- - ">="
|
27
|
+
- !ruby/object:Gem::Version
|
28
|
+
hash: 35
|
29
|
+
segments:
|
30
|
+
- 2
|
31
|
+
- 11
|
32
|
+
- 0
|
33
|
+
version: 2.11.0
|
34
|
+
type: :development
|
35
|
+
version_requirements: *id001
|
36
|
+
- !ruby/object:Gem::Dependency
|
37
|
+
name: nokogiri
|
38
|
+
prerelease: false
|
39
|
+
requirement: &id002 !ruby/object:Gem::Requirement
|
40
|
+
none: false
|
41
|
+
requirements:
|
42
|
+
- - ">="
|
43
|
+
- !ruby/object:Gem::Version
|
44
|
+
hash: 9
|
45
|
+
segments:
|
46
|
+
- 1
|
47
|
+
- 5
|
48
|
+
- 5
|
49
|
+
version: 1.5.5
|
50
|
+
type: :runtime
|
51
|
+
version_requirements: *id002
|
21
52
|
description: A simple tool for checking references on your website
|
22
53
|
email: me@zsoltfabok.com
|
23
54
|
executables: []
|
@@ -28,7 +59,6 @@ extra_rdoc_files: []
|
|
28
59
|
|
29
60
|
files:
|
30
61
|
- lib/site_checker.rb
|
31
|
-
has_rdoc: true
|
32
62
|
homepage: https://github.com/ZsoltFabok/site_checker
|
33
63
|
licenses: []
|
34
64
|
|
@@ -38,25 +68,29 @@ rdoc_options: []
|
|
38
68
|
require_paths:
|
39
69
|
- lib
|
40
70
|
required_ruby_version: !ruby/object:Gem::Requirement
|
71
|
+
none: false
|
41
72
|
requirements:
|
42
73
|
- - ">="
|
43
74
|
- !ruby/object:Gem::Version
|
75
|
+
hash: 3
|
44
76
|
segments:
|
45
77
|
- 0
|
46
78
|
version: "0"
|
47
79
|
required_rubygems_version: !ruby/object:Gem::Requirement
|
80
|
+
none: false
|
48
81
|
requirements:
|
49
82
|
- - ">="
|
50
83
|
- !ruby/object:Gem::Version
|
84
|
+
hash: 3
|
51
85
|
segments:
|
52
86
|
- 0
|
53
87
|
version: "0"
|
54
88
|
requirements: []
|
55
89
|
|
56
90
|
rubyforge_project:
|
57
|
-
rubygems_version: 1.
|
91
|
+
rubygems_version: 1.8.15
|
58
92
|
signing_key:
|
59
93
|
specification_version: 3
|
60
|
-
summary: site_checker-0.
|
94
|
+
summary: site_checker-0.1.1
|
61
95
|
test_files: []
|
62
96
|
|