anemone 0.6.1 → 0.7.0
Sign up to get free protection for your applications and to get access to all the features.
- data/CHANGELOG.rdoc +18 -1
- data/README.rdoc +2 -1
- data/Rakefile +9 -10
- data/VERSION +1 -1
- data/lib/anemone/core.rb +1 -1
- data/lib/anemone/http.rb +1 -1
- data/lib/anemone/page.rb +18 -3
- data/lib/anemone/storage.rb +10 -0
- data/lib/anemone/storage/kyoto_cabinet.rb +72 -0
- data/lib/anemone/storage/sqlite3.rb +90 -0
- data/lib/anemone/storage/tokyo_cabinet.rb +1 -0
- data/spec/core_spec.rb +22 -1
- data/spec/fakeweb_helper.rb +6 -8
- data/spec/http_spec.rb +1 -7
- data/spec/page_spec.rb +97 -13
- data/spec/page_store_spec.rb +19 -1
- data/spec/spec_helper.rb +2 -0
- data/spec/storage_spec.rb +56 -1
- metadata +164 -85
data/CHANGELOG.rdoc
CHANGED
@@ -1,6 +1,23 @@
|
|
1
|
+
== 0.7.0 / 2012-01-19
|
2
|
+
|
3
|
+
* Major enhancements
|
4
|
+
|
5
|
+
* Added support for SQLite3 and Kyoto Cabinet storage
|
6
|
+
|
7
|
+
* Minor enhancements
|
8
|
+
|
9
|
+
* Added Page#base to use base HTML element
|
10
|
+
* Use bundler for development dependencies
|
11
|
+
|
12
|
+
* Bug fixes
|
13
|
+
|
14
|
+
* Encode characters in URLs
|
15
|
+
* Fix specs to run under rake
|
16
|
+
* Fix handling of redirect_to in storage adapters
|
17
|
+
|
1
18
|
== 0.6.1 / 2011-02-24
|
2
19
|
|
3
|
-
*Bug fixes
|
20
|
+
* Bug fixes
|
4
21
|
|
5
22
|
* Fix a bug preventing SSL connections from working
|
6
23
|
|
data/README.rdoc
CHANGED
@@ -16,7 +16,7 @@ See http://anemone.rubyforge.org for more information.
|
|
16
16
|
* Records response time for each page
|
17
17
|
* CLI program can list all pages in a domain, calculate page depths, and more
|
18
18
|
* Obey robots.txt
|
19
|
-
* In-memory or persistent storage of pages during crawl, using TokyoCabinet, MongoDB, or Redis
|
19
|
+
* In-memory or persistent storage of pages during crawl, using TokyoCabinet, SQLite3, MongoDB, or Redis
|
20
20
|
|
21
21
|
== Examples
|
22
22
|
See the scripts under the <tt>lib/anemone/cli</tt> directory for examples of several useful Anemone tasks.
|
@@ -32,5 +32,6 @@ To test and develop this gem, additional requirements are:
|
|
32
32
|
* tokyocabinet
|
33
33
|
* mongo
|
34
34
|
* redis
|
35
|
+
* sqlite3
|
35
36
|
|
36
37
|
You will need to have {Tokyo Cabinet}[http://fallabs.com/tokyocabinet/], {MongoDB}[http://www.mongodb.org/], and {Redis}[http://code.google.com/p/redis/] installed on your system and running.
|
data/Rakefile
CHANGED
@@ -1,26 +1,25 @@
|
|
1
1
|
require 'rubygems'
|
2
2
|
require 'rake'
|
3
|
+
require 'rspec/core/rake_task'
|
3
4
|
|
4
|
-
|
5
|
-
|
6
|
-
spec.
|
7
|
-
spec.spec_files = FileList['spec/**/*_spec.rb']
|
5
|
+
desc "Run all specs"
|
6
|
+
RSpec::Core::RakeTask.new(:rspec) do |spec|
|
7
|
+
spec.pattern = 'spec/**/*_spec.rb'
|
8
8
|
end
|
9
9
|
|
10
|
-
|
11
|
-
spec.libs << 'lib' << 'spec'
|
10
|
+
RSpec::Core::RakeTask.new(:rcov) do |spec|
|
12
11
|
spec.pattern = 'spec/**/*_spec.rb'
|
13
12
|
spec.rcov = true
|
14
13
|
end
|
15
14
|
|
16
|
-
task :default => :
|
15
|
+
task :default => :rspec
|
17
16
|
|
18
|
-
require '
|
19
|
-
|
17
|
+
require 'rdoc/task'
|
18
|
+
RDoc::Task.new do |rdoc|
|
20
19
|
version = File.exist?('VERSION') ? File.read('VERSION') : ""
|
21
20
|
|
22
21
|
rdoc.rdoc_dir = 'rdoc'
|
23
22
|
rdoc.title = "anemone #{version}"
|
24
23
|
rdoc.rdoc_files.include('README*')
|
25
24
|
rdoc.rdoc_files.include('lib/**/*.rb')
|
26
|
-
end
|
25
|
+
end
|
data/VERSION
CHANGED
@@ -1 +1 @@
|
|
1
|
-
0.
|
1
|
+
0.7.0
|
data/lib/anemone/core.rb
CHANGED
data/lib/anemone/http.rb
CHANGED
@@ -112,7 +112,7 @@ module Anemone
|
|
112
112
|
|
113
113
|
response, response_time = get_response(loc, referer)
|
114
114
|
code = Integer(response.code)
|
115
|
-
redirect_to = response.is_a?(Net::HTTPRedirection) ?
|
115
|
+
redirect_to = response.is_a?(Net::HTTPRedirection) ? URI(response['location']).normalize : nil
|
116
116
|
yield response, code, loc, redirect_to, response_time
|
117
117
|
limit -= 1
|
118
118
|
end while (loc = redirect_to) && allowed?(redirect_to, url) && limit > 0
|
data/lib/anemone/page.rb
CHANGED
@@ -62,7 +62,7 @@ module Anemone
|
|
62
62
|
doc.search("//a[@href]").each do |a|
|
63
63
|
u = a['href']
|
64
64
|
next if u.nil? or u.empty?
|
65
|
-
abs = to_absolute(URI(u)) rescue next
|
65
|
+
abs = to_absolute(URI(URI.escape(u))) rescue next
|
66
66
|
@links << abs if in_domain?(abs)
|
67
67
|
end
|
68
68
|
@links.uniq!
|
@@ -131,6 +131,21 @@ module Anemone
|
|
131
131
|
404 == @code
|
132
132
|
end
|
133
133
|
|
134
|
+
#
|
135
|
+
# Base URI from the HTML doc head element
|
136
|
+
# http://www.w3.org/TR/html4/struct/links.html#edef-BASE
|
137
|
+
#
|
138
|
+
def base
|
139
|
+
@base = if doc
|
140
|
+
href = doc.search('//head/base/@href')
|
141
|
+
URI(href.to_s) unless href.nil? rescue nil
|
142
|
+
end unless @base
|
143
|
+
|
144
|
+
return nil if @base && @base.to_s().empty?
|
145
|
+
@base
|
146
|
+
end
|
147
|
+
|
148
|
+
|
134
149
|
#
|
135
150
|
# Converts relative URL *link* into an absolute URL based on the
|
136
151
|
# location of the page
|
@@ -142,7 +157,7 @@ module Anemone
|
|
142
157
|
link = URI.encode(URI.decode(link.to_s.gsub(/#[a-zA-Z0-9_-]*$/,'')))
|
143
158
|
|
144
159
|
relative = URI(link)
|
145
|
-
absolute = @url.merge(relative)
|
160
|
+
absolute = base ? base.merge(relative) : @url.merge(relative)
|
146
161
|
|
147
162
|
absolute.path = '/' if absolute.path.empty?
|
148
163
|
|
@@ -190,7 +205,7 @@ module Anemone
|
|
190
205
|
'@visited' => hash['visited'],
|
191
206
|
'@depth' => hash['depth'].to_i,
|
192
207
|
'@referer' => hash['referer'],
|
193
|
-
'@redirect_to' => URI(hash['redirect_to']),
|
208
|
+
'@redirect_to' => (!!hash['redirect_to'] && !hash['redirect_to'].empty?) ? URI(hash['redirect_to']) : nil,
|
194
209
|
'@response_time' => hash['response_time'].to_i,
|
195
210
|
'@fetched' => hash['fetched']
|
196
211
|
}.each do |var, value|
|
data/lib/anemone/storage.rb
CHANGED
@@ -18,6 +18,11 @@ module Anemone
|
|
18
18
|
self::TokyoCabinet.new(file)
|
19
19
|
end
|
20
20
|
|
21
|
+
def self.KyotoCabinet(file = 'anemone.tch')
|
22
|
+
require 'anemone/storage/kyoto_cabinet'
|
23
|
+
self::KyotoCabinet.new(file)
|
24
|
+
end
|
25
|
+
|
21
26
|
def self.MongoDB(mongo_db = nil, collection_name = 'pages')
|
22
27
|
require 'anemone/storage/mongodb'
|
23
28
|
mongo_db ||= Mongo::Connection.new.db('anemone')
|
@@ -29,6 +34,11 @@ module Anemone
|
|
29
34
|
require 'anemone/storage/redis'
|
30
35
|
self::Redis.new(opts)
|
31
36
|
end
|
37
|
+
|
38
|
+
def self.SQLite3(file = 'anemone.db')
|
39
|
+
require 'anemone/storage/sqlite3'
|
40
|
+
self::SQLite3.new(file)
|
41
|
+
end
|
32
42
|
|
33
43
|
end
|
34
44
|
end
|
@@ -0,0 +1,72 @@
|
|
1
|
+
begin
|
2
|
+
require 'kyotocabinet'
|
3
|
+
rescue LoadError
|
4
|
+
puts $!
|
5
|
+
puts "You need the kyotocabinet-ruby gem to use Anemone::Storage::KyotoCabinet"
|
6
|
+
exit
|
7
|
+
end
|
8
|
+
|
9
|
+
require 'forwardable'
|
10
|
+
|
11
|
+
module Anemone
|
12
|
+
module Storage
|
13
|
+
class KyotoCabinet
|
14
|
+
extend Forwardable
|
15
|
+
|
16
|
+
def_delegators :@db, :close, :size, :each
|
17
|
+
|
18
|
+
def initialize(file)
|
19
|
+
raise "KyotoCabinet filename must have .kch extension" if File.extname(file) != '.kch'
|
20
|
+
@db = ::KyotoCabinet::DB::new
|
21
|
+
@db.open(file, ::KyotoCabinet::DB::OWRITER | ::KyotoCabinet::DB::OCREATE)
|
22
|
+
@db.clear
|
23
|
+
end
|
24
|
+
|
25
|
+
def [](key)
|
26
|
+
if value = @db[key]
|
27
|
+
load_value(value)
|
28
|
+
end
|
29
|
+
end
|
30
|
+
|
31
|
+
def []=(key, value)
|
32
|
+
@db[key] = [Marshal.dump(value)].pack("m")
|
33
|
+
end
|
34
|
+
|
35
|
+
def each
|
36
|
+
@db.each do |k, v|
|
37
|
+
yield(k, load_value(v))
|
38
|
+
end
|
39
|
+
end
|
40
|
+
|
41
|
+
def has_key?(key)
|
42
|
+
# Kyoto Cabinet doesn't have a way to query whether a key exists, so hack it
|
43
|
+
keys = @db.match_prefix(key)
|
44
|
+
!!keys && keys.include?(key)
|
45
|
+
end
|
46
|
+
|
47
|
+
def keys
|
48
|
+
acc = []
|
49
|
+
@db.each_key { |key| acc << key.first }
|
50
|
+
acc
|
51
|
+
end
|
52
|
+
|
53
|
+
def delete(key)
|
54
|
+
value = self[key]
|
55
|
+
@db.delete(key)
|
56
|
+
value
|
57
|
+
end
|
58
|
+
|
59
|
+
def merge!(hash)
|
60
|
+
hash.each { |key, value| self[key] = value }
|
61
|
+
self
|
62
|
+
end
|
63
|
+
|
64
|
+
private
|
65
|
+
|
66
|
+
def load_value(value)
|
67
|
+
Marshal.load(value.unpack("m")[0])
|
68
|
+
end
|
69
|
+
|
70
|
+
end
|
71
|
+
end
|
72
|
+
end
|
@@ -0,0 +1,90 @@
|
|
1
|
+
begin
|
2
|
+
require 'sqlite3'
|
3
|
+
rescue LoadError
|
4
|
+
puts "You need the sqlite3 gem to use Anemone::Storage::SQLite3"
|
5
|
+
exit
|
6
|
+
end
|
7
|
+
|
8
|
+
module Anemone
|
9
|
+
module Storage
|
10
|
+
class SQLite3
|
11
|
+
|
12
|
+
def initialize(file)
|
13
|
+
@db = ::SQLite3::Database.new(file)
|
14
|
+
create_schema
|
15
|
+
end
|
16
|
+
|
17
|
+
def [](url)
|
18
|
+
value = @db.get_first_value('SELECT data FROM anemone_storage WHERE key = ?', url.to_s)
|
19
|
+
if value
|
20
|
+
Marshal.load(value)
|
21
|
+
end
|
22
|
+
end
|
23
|
+
|
24
|
+
def []=(url, value)
|
25
|
+
data = Marshal.dump(value)
|
26
|
+
if has_key?(url)
|
27
|
+
@db.execute('UPDATE anemone_storage SET data = ? WHERE key = ?', data, url.to_s)
|
28
|
+
else
|
29
|
+
@db.execute('INSERT INTO anemone_storage (data, key) VALUES(?, ?)', data, url.to_s)
|
30
|
+
end
|
31
|
+
end
|
32
|
+
|
33
|
+
def delete(url)
|
34
|
+
page = self[url]
|
35
|
+
@db.execute('DELETE FROM anemone_storage WHERE key = ?', url.to_s)
|
36
|
+
page
|
37
|
+
end
|
38
|
+
|
39
|
+
def each
|
40
|
+
@db.execute("SELECT key, data FROM anemone_storage ORDER BY id") do |row|
|
41
|
+
value = Marshal.load(row[1])
|
42
|
+
yield row[0], value
|
43
|
+
end
|
44
|
+
end
|
45
|
+
|
46
|
+
def merge!(hash)
|
47
|
+
hash.each { |key, value| self[key] = value }
|
48
|
+
self
|
49
|
+
end
|
50
|
+
|
51
|
+
def size
|
52
|
+
@db.get_first_value('SELECT COUNT(id) FROM anemone_storage')
|
53
|
+
end
|
54
|
+
|
55
|
+
def keys
|
56
|
+
@db.execute("SELECT key FROM anemone_storage ORDER BY id").map{|t| t[0]}
|
57
|
+
end
|
58
|
+
|
59
|
+
def has_key?(url)
|
60
|
+
!!@db.get_first_value('SELECT id FROM anemone_storage WHERE key = ?', url.to_s)
|
61
|
+
end
|
62
|
+
|
63
|
+
def close
|
64
|
+
@db.close
|
65
|
+
end
|
66
|
+
|
67
|
+
private
|
68
|
+
|
69
|
+
def create_schema
|
70
|
+
@db.execute_batch <<SQL
|
71
|
+
create table if not exists anemone_storage (
|
72
|
+
id INTEGER PRIMARY KEY ASC,
|
73
|
+
key TEXT,
|
74
|
+
data BLOB
|
75
|
+
);
|
76
|
+
create index if not exists anemone_key_idx on anemone_storage (key);
|
77
|
+
SQL
|
78
|
+
end
|
79
|
+
|
80
|
+
def load_page(hash)
|
81
|
+
BINARY_FIELDS.each do |field|
|
82
|
+
hash[field] = hash[field].to_s
|
83
|
+
end
|
84
|
+
Page.from_hash(hash)
|
85
|
+
end
|
86
|
+
|
87
|
+
end
|
88
|
+
end
|
89
|
+
end
|
90
|
+
|
data/spec/core_spec.rb
CHANGED
@@ -1,6 +1,6 @@
|
|
1
1
|
$:.unshift(File.dirname(__FILE__))
|
2
2
|
require 'spec_helper'
|
3
|
-
%w[pstore tokyo_cabinet].each { |file| require "anemone/storage/#{file}.rb" }
|
3
|
+
%w[pstore tokyo_cabinet sqlite3].each { |file| require "anemone/storage/#{file}.rb" }
|
4
4
|
|
5
5
|
module Anemone
|
6
6
|
describe Core do
|
@@ -277,6 +277,27 @@ module Anemone
|
|
277
277
|
end
|
278
278
|
end
|
279
279
|
|
280
|
+
describe Storage::SQLite3 do
|
281
|
+
it_should_behave_like "crawl"
|
282
|
+
|
283
|
+
before(:all) do
|
284
|
+
@test_file = 'test.db'
|
285
|
+
end
|
286
|
+
|
287
|
+
before(:each) do
|
288
|
+
File.delete(@test_file) if File.exists?(@test_file)
|
289
|
+
@opts = {:storage => @store = Storage.SQLite3(@test_file)}
|
290
|
+
end
|
291
|
+
|
292
|
+
after(:each) do
|
293
|
+
@store.close
|
294
|
+
end
|
295
|
+
|
296
|
+
after(:each) do
|
297
|
+
File.delete(@test_file) if File.exists?(@test_file)
|
298
|
+
end
|
299
|
+
end
|
300
|
+
|
280
301
|
describe "options" do
|
281
302
|
it "should accept options for the crawl" do
|
282
303
|
core = Anemone.crawl(SPEC_DOMAIN, :verbose => false,
|
data/spec/fakeweb_helper.rb
CHANGED
@@ -1,10 +1,3 @@
|
|
1
|
-
begin
|
2
|
-
require 'fakeweb'
|
3
|
-
rescue LoadError
|
4
|
-
warn "You need the 'fakeweb' gem installed to test Anemone"
|
5
|
-
exit
|
6
|
-
end
|
7
|
-
|
8
1
|
FakeWeb.allow_net_connect = false
|
9
2
|
|
10
3
|
module Anemone
|
@@ -22,6 +15,7 @@ module Anemone
|
|
22
15
|
@hrefs = [options[:hrefs]].flatten if options.has_key?(:hrefs)
|
23
16
|
@redirect = options[:redirect] if options.has_key?(:redirect)
|
24
17
|
@auth = options[:auth] if options.has_key?(:auth)
|
18
|
+
@base = options[:base] if options.has_key?(:base)
|
25
19
|
@content_type = options[:content_type] || "text/html"
|
26
20
|
@body = options[:body]
|
27
21
|
|
@@ -40,7 +34,11 @@ module Anemone
|
|
40
34
|
private
|
41
35
|
|
42
36
|
def create_body
|
43
|
-
@
|
37
|
+
if @base
|
38
|
+
@body = "<html><head><base href=\"#{@base}\"></head><body>"
|
39
|
+
else
|
40
|
+
@body = "<html><body>"
|
41
|
+
end
|
44
42
|
@links.each{|l| @body += "<a href=\"#{SPEC_DOMAIN}#{l}\"></a>"} if @links
|
45
43
|
@hrefs.each{|h| @body += "<a href=\"#{h}\"></a>"} if @hrefs
|
46
44
|
@body += "</body></html>"
|
data/spec/http_spec.rb
CHANGED
@@ -1,4 +1,3 @@
|
|
1
|
-
$:.unshift(File.dirname(__FILE__))
|
2
1
|
require 'spec_helper'
|
3
2
|
|
4
3
|
module Anemone
|
@@ -10,12 +9,7 @@ module Anemone
|
|
10
9
|
end
|
11
10
|
|
12
11
|
it "should still return a Page if an exception occurs during the HTTP connection" do
|
13
|
-
|
14
|
-
def refresh_connection
|
15
|
-
raise "test exception"
|
16
|
-
end
|
17
|
-
end
|
18
|
-
|
12
|
+
HTTP.stub!(:refresh_connection).and_raise(StandardError)
|
19
13
|
http = Anemone::HTTP.new
|
20
14
|
http.fetch_page(SPEC_DOMAIN).should be_an_instance_of(Page)
|
21
15
|
end
|
data/spec/page_spec.rb
CHANGED
@@ -74,19 +74,103 @@ module Anemone
|
|
74
74
|
@page.cookies.should == []
|
75
75
|
end
|
76
76
|
|
77
|
-
|
78
|
-
|
79
|
-
|
80
|
-
|
81
|
-
|
82
|
-
|
83
|
-
|
84
|
-
|
85
|
-
|
86
|
-
|
87
|
-
|
88
|
-
|
89
|
-
|
77
|
+
describe "#to_hash" do
|
78
|
+
it "converts the page to a hash" do
|
79
|
+
hash = @page.to_hash
|
80
|
+
hash['url'].should == @page.url.to_s
|
81
|
+
hash['referer'].should == @page.referer.to_s
|
82
|
+
hash['links'].should == @page.links.map(&:to_s)
|
83
|
+
end
|
84
|
+
|
85
|
+
context "when redirect_to is nil" do
|
86
|
+
it "sets 'redirect_to' to nil in the hash" do
|
87
|
+
@page.redirect_to.should be_nil
|
88
|
+
@page.to_hash[:redirect_to].should be_nil
|
89
|
+
end
|
90
|
+
end
|
91
|
+
|
92
|
+
context "when redirect_to is a non-nil URI" do
|
93
|
+
it "sets 'redirect_to' to the URI string" do
|
94
|
+
new_page = Page.new(URI(SPEC_DOMAIN), {:redirect_to => URI(SPEC_DOMAIN + '1')})
|
95
|
+
new_page.redirect_to.to_s.should == SPEC_DOMAIN + '1'
|
96
|
+
new_page.to_hash['redirect_to'].should == SPEC_DOMAIN + '1'
|
97
|
+
end
|
98
|
+
end
|
99
|
+
end
|
100
|
+
|
101
|
+
describe "#from_hash" do
|
102
|
+
it "converts from a hash to a Page" do
|
103
|
+
page = @page.dup
|
104
|
+
page.depth = 1
|
105
|
+
converted = Page.from_hash(page.to_hash)
|
106
|
+
converted.links.should == page.links
|
107
|
+
converted.depth.should == page.depth
|
108
|
+
end
|
109
|
+
|
110
|
+
it 'handles a from_hash with a nil redirect_to' do
|
111
|
+
page_hash = @page.to_hash
|
112
|
+
page_hash['redirect_to'] = nil
|
113
|
+
lambda{Page.from_hash(page_hash)}.should_not raise_error(URI::InvalidURIError)
|
114
|
+
Page.from_hash(page_hash).redirect_to.should be_nil
|
115
|
+
end
|
116
|
+
end
|
117
|
+
|
118
|
+
describe "#redirect_to" do
|
119
|
+
context "when the page was a redirect" do
|
120
|
+
it "returns a URI of the page it redirects to" do
|
121
|
+
new_page = Page.new(URI(SPEC_DOMAIN), {:redirect_to => URI(SPEC_DOMAIN + '1')})
|
122
|
+
redirect = new_page.redirect_to
|
123
|
+
redirect.should be_a(URI)
|
124
|
+
redirect.to_s.should == SPEC_DOMAIN + '1'
|
125
|
+
end
|
126
|
+
end
|
127
|
+
end
|
128
|
+
|
129
|
+
it "should detect, store and expose the base url for the page head" do
|
130
|
+
base = "#{SPEC_DOMAIN}path/to/base_url/"
|
131
|
+
page = @http.fetch_page(FakePage.new('body_test', {:base => base}).url)
|
132
|
+
page.base.should == URI(base)
|
133
|
+
@page.base.should be_nil
|
134
|
+
end
|
135
|
+
|
136
|
+
it "should have a method to convert a relative url to an absolute one" do
|
137
|
+
@page.should respond_to(:to_absolute)
|
138
|
+
|
139
|
+
# Identity
|
140
|
+
@page.to_absolute(@page.url).should == @page.url
|
141
|
+
@page.to_absolute("").should == @page.url
|
142
|
+
|
143
|
+
# Root-ness
|
144
|
+
@page.to_absolute("/").should == URI("#{SPEC_DOMAIN}")
|
145
|
+
|
146
|
+
# Relativeness
|
147
|
+
relative_path = "a/relative/path"
|
148
|
+
@page.to_absolute(relative_path).should == URI("#{SPEC_DOMAIN}#{relative_path}")
|
149
|
+
|
150
|
+
deep_page = @http.fetch_page(FakePage.new('home/deep', :links => '1').url)
|
151
|
+
upward_relative_path = "../a/relative/path"
|
152
|
+
deep_page.to_absolute(upward_relative_path).should == URI("#{SPEC_DOMAIN}#{relative_path}")
|
153
|
+
|
154
|
+
# The base URL case
|
155
|
+
base_path = "path/to/base_url/"
|
156
|
+
base = "#{SPEC_DOMAIN}#{base_path}"
|
157
|
+
page = @http.fetch_page(FakePage.new('home', {:base => base}).url)
|
158
|
+
|
159
|
+
# Identity
|
160
|
+
page.to_absolute(page.url).should == page.url
|
161
|
+
# It should revert to the base url
|
162
|
+
page.to_absolute("").should_not == page.url
|
163
|
+
|
164
|
+
# Root-ness
|
165
|
+
page.to_absolute("/").should == URI("#{SPEC_DOMAIN}")
|
166
|
+
|
167
|
+
# Relativeness
|
168
|
+
relative_path = "a/relative/path"
|
169
|
+
page.to_absolute(relative_path).should == URI("#{base}#{relative_path}")
|
170
|
+
|
171
|
+
upward_relative_path = "../a/relative/path"
|
172
|
+
upward_base = "#{SPEC_DOMAIN}path/to/"
|
173
|
+
page.to_absolute(upward_relative_path).should == URI("#{upward_base}#{relative_path}")
|
90
174
|
end
|
91
175
|
|
92
176
|
end
|
data/spec/page_store_spec.rb
CHANGED
@@ -1,6 +1,6 @@
|
|
1
1
|
$:.unshift(File.dirname(__FILE__))
|
2
2
|
require 'spec_helper'
|
3
|
-
%w[pstore tokyo_cabinet mongodb redis].each { |file| require "anemone/storage/#{file}.rb" }
|
3
|
+
%w[pstore tokyo_cabinet sqlite3 mongodb redis].each { |file| require "anemone/storage/#{file}.rb" }
|
4
4
|
|
5
5
|
module Anemone
|
6
6
|
describe PageStore do
|
@@ -125,6 +125,24 @@ module Anemone
|
|
125
125
|
end
|
126
126
|
end
|
127
127
|
|
128
|
+
describe Storage::SQLite3 do
|
129
|
+
it_should_behave_like "page storage"
|
130
|
+
|
131
|
+
before(:each) do
|
132
|
+
@test_file = 'test.db'
|
133
|
+
File.delete(@test_file) if File.exists?(@test_file)
|
134
|
+
@opts = {:storage => @store = Storage.SQLite3(@test_file)}
|
135
|
+
end
|
136
|
+
|
137
|
+
after(:each) do
|
138
|
+
@store.close
|
139
|
+
end
|
140
|
+
|
141
|
+
after(:each) do
|
142
|
+
File.delete(@test_file) if File.exists?(@test_file)
|
143
|
+
end
|
144
|
+
end
|
145
|
+
|
128
146
|
describe Storage::MongoDB do
|
129
147
|
it_should_behave_like "page storage"
|
130
148
|
|
data/spec/spec_helper.rb
CHANGED
data/spec/storage_spec.rb
CHANGED
@@ -1,7 +1,7 @@
|
|
1
1
|
$:.unshift(File.dirname(__FILE__))
|
2
2
|
require 'spec_helper'
|
3
3
|
|
4
|
-
%w[pstore tokyo_cabinet mongodb redis].each { |file| require "anemone/storage/#{file}.rb" }
|
4
|
+
%w[pstore tokyo_cabinet kyoto_cabinet sqlite3 mongodb redis].each { |file| require "anemone/storage/#{file}.rb" }
|
5
5
|
|
6
6
|
module Anemone
|
7
7
|
describe Storage do
|
@@ -25,6 +25,14 @@ module Anemone
|
|
25
25
|
store.close
|
26
26
|
end
|
27
27
|
|
28
|
+
it "should have a class method to produce a SQLite3" do
|
29
|
+
test_file = 'test.db'
|
30
|
+
Anemone::Storage.should respond_to(:SQLite3)
|
31
|
+
store = Anemone::Storage.SQLite3(test_file)
|
32
|
+
store.should be_an_instance_of(Anemone::Storage::SQLite3)
|
33
|
+
store.close
|
34
|
+
end
|
35
|
+
|
28
36
|
it "should have a class method to produce a MongoDB" do
|
29
37
|
Anemone::Storage.should respond_to(:MongoDB)
|
30
38
|
store = Anemone::Storage.MongoDB
|
@@ -105,6 +113,12 @@ module Anemone
|
|
105
113
|
|
106
114
|
merged.should === @store
|
107
115
|
end
|
116
|
+
|
117
|
+
it "should correctly deserialize nil redirect_to when loading" do
|
118
|
+
@page.redirect_to.should be_nil
|
119
|
+
@store[@url] = @page
|
120
|
+
@store[@url].redirect_to.should be_nil
|
121
|
+
end
|
108
122
|
end
|
109
123
|
|
110
124
|
describe PStore do
|
@@ -143,6 +157,47 @@ module Anemone
|
|
143
157
|
end
|
144
158
|
end
|
145
159
|
|
160
|
+
describe KyotoCabinet do
|
161
|
+
it_should_behave_like "storage engine"
|
162
|
+
|
163
|
+
before(:each) do
|
164
|
+
@test_file = 'test.kch'
|
165
|
+
File.delete @test_file rescue nil
|
166
|
+
@store = Anemone::Storage.KyotoCabinet(@test_file)
|
167
|
+
end
|
168
|
+
|
169
|
+
after(:each) do
|
170
|
+
@store.close
|
171
|
+
end
|
172
|
+
|
173
|
+
after(:all) do
|
174
|
+
File.delete @test_file rescue nil
|
175
|
+
end
|
176
|
+
|
177
|
+
it "should raise an error if supplied with a file extension other than .kch" do
|
178
|
+
lambda { Anemone::Storage.KyotoCabinet('test.tmp') }.should raise_error(RuntimeError)
|
179
|
+
end
|
180
|
+
end
|
181
|
+
|
182
|
+
describe SQLite3 do
|
183
|
+
it_should_behave_like "storage engine"
|
184
|
+
|
185
|
+
before(:each) do
|
186
|
+
@test_file = 'test.db'
|
187
|
+
File.delete @test_file rescue nil
|
188
|
+
@store = Anemone::Storage.SQLite3(@test_file)
|
189
|
+
end
|
190
|
+
|
191
|
+
after(:each) do
|
192
|
+
@store.close
|
193
|
+
end
|
194
|
+
|
195
|
+
after(:all) do
|
196
|
+
File.delete @test_file rescue nil
|
197
|
+
end
|
198
|
+
|
199
|
+
end
|
200
|
+
|
146
201
|
describe Storage::MongoDB do
|
147
202
|
it_should_behave_like "storage engine"
|
148
203
|
|
metadata
CHANGED
@@ -1,139 +1,218 @@
|
|
1
|
-
--- !ruby/object:Gem::Specification
|
1
|
+
--- !ruby/object:Gem::Specification
|
2
2
|
name: anemone
|
3
|
-
version: !ruby/object:Gem::Version
|
4
|
-
|
5
|
-
|
6
|
-
- 0
|
7
|
-
- 6
|
8
|
-
- 1
|
9
|
-
version: 0.6.1
|
3
|
+
version: !ruby/object:Gem::Version
|
4
|
+
version: 0.7.0
|
5
|
+
prerelease:
|
10
6
|
platform: ruby
|
11
|
-
authors:
|
7
|
+
authors:
|
12
8
|
- Chris Kite
|
13
9
|
autorequire:
|
14
10
|
bindir: bin
|
15
11
|
cert_chain: []
|
16
|
-
|
17
|
-
|
18
|
-
|
19
|
-
dependencies:
|
20
|
-
- !ruby/object:Gem::Dependency
|
12
|
+
date: 2012-01-20 00:00:00.000000000Z
|
13
|
+
dependencies:
|
14
|
+
- !ruby/object:Gem::Dependency
|
21
15
|
name: nokogiri
|
22
|
-
|
23
|
-
requirement: &id001 !ruby/object:Gem::Requirement
|
16
|
+
requirement: &19111780 !ruby/object:Gem::Requirement
|
24
17
|
none: false
|
25
|
-
requirements:
|
26
|
-
- -
|
27
|
-
- !ruby/object:Gem::Version
|
28
|
-
segments:
|
29
|
-
- 1
|
30
|
-
- 3
|
31
|
-
- 0
|
18
|
+
requirements:
|
19
|
+
- - ! '>='
|
20
|
+
- !ruby/object:Gem::Version
|
32
21
|
version: 1.3.0
|
33
22
|
type: :runtime
|
34
|
-
version_requirements: *id001
|
35
|
-
- !ruby/object:Gem::Dependency
|
36
|
-
name: robots
|
37
23
|
prerelease: false
|
38
|
-
|
24
|
+
version_requirements: *19111780
|
25
|
+
- !ruby/object:Gem::Dependency
|
26
|
+
name: robots
|
27
|
+
requirement: &19111300 !ruby/object:Gem::Requirement
|
39
28
|
none: false
|
40
|
-
requirements:
|
41
|
-
- -
|
42
|
-
- !ruby/object:Gem::Version
|
43
|
-
segments:
|
44
|
-
- 0
|
45
|
-
- 7
|
46
|
-
- 2
|
29
|
+
requirements:
|
30
|
+
- - ! '>='
|
31
|
+
- !ruby/object:Gem::Version
|
47
32
|
version: 0.7.2
|
48
33
|
type: :runtime
|
49
|
-
|
34
|
+
prerelease: false
|
35
|
+
version_requirements: *19111300
|
36
|
+
- !ruby/object:Gem::Dependency
|
37
|
+
name: rake
|
38
|
+
requirement: &19141340 !ruby/object:Gem::Requirement
|
39
|
+
none: false
|
40
|
+
requirements:
|
41
|
+
- - ! '>='
|
42
|
+
- !ruby/object:Gem::Version
|
43
|
+
version: 0.8.7
|
44
|
+
type: :development
|
45
|
+
prerelease: false
|
46
|
+
version_requirements: *19141340
|
47
|
+
- !ruby/object:Gem::Dependency
|
48
|
+
name: rspec
|
49
|
+
requirement: &19140880 !ruby/object:Gem::Requirement
|
50
|
+
none: false
|
51
|
+
requirements:
|
52
|
+
- - ! '>='
|
53
|
+
- !ruby/object:Gem::Version
|
54
|
+
version: 2.6.0
|
55
|
+
type: :development
|
56
|
+
prerelease: false
|
57
|
+
version_requirements: *19140880
|
58
|
+
- !ruby/object:Gem::Dependency
|
59
|
+
name: fakeweb
|
60
|
+
requirement: &19140420 !ruby/object:Gem::Requirement
|
61
|
+
none: false
|
62
|
+
requirements:
|
63
|
+
- - ! '>='
|
64
|
+
- !ruby/object:Gem::Version
|
65
|
+
version: 1.3.0
|
66
|
+
type: :development
|
67
|
+
prerelease: false
|
68
|
+
version_requirements: *19140420
|
69
|
+
- !ruby/object:Gem::Dependency
|
70
|
+
name: redis
|
71
|
+
requirement: &19139960 !ruby/object:Gem::Requirement
|
72
|
+
none: false
|
73
|
+
requirements:
|
74
|
+
- - ! '>='
|
75
|
+
- !ruby/object:Gem::Version
|
76
|
+
version: 2.2.0
|
77
|
+
type: :development
|
78
|
+
prerelease: false
|
79
|
+
version_requirements: *19139960
|
80
|
+
- !ruby/object:Gem::Dependency
|
81
|
+
name: mongo
|
82
|
+
requirement: &19139500 !ruby/object:Gem::Requirement
|
83
|
+
none: false
|
84
|
+
requirements:
|
85
|
+
- - ! '>='
|
86
|
+
- !ruby/object:Gem::Version
|
87
|
+
version: 1.3.1
|
88
|
+
type: :development
|
89
|
+
prerelease: false
|
90
|
+
version_requirements: *19139500
|
91
|
+
- !ruby/object:Gem::Dependency
|
92
|
+
name: bson_ext
|
93
|
+
requirement: &19139040 !ruby/object:Gem::Requirement
|
94
|
+
none: false
|
95
|
+
requirements:
|
96
|
+
- - ! '>='
|
97
|
+
- !ruby/object:Gem::Version
|
98
|
+
version: 1.3.1
|
99
|
+
type: :development
|
100
|
+
prerelease: false
|
101
|
+
version_requirements: *19139040
|
102
|
+
- !ruby/object:Gem::Dependency
|
103
|
+
name: tokyocabinet
|
104
|
+
requirement: &19138580 !ruby/object:Gem::Requirement
|
105
|
+
none: false
|
106
|
+
requirements:
|
107
|
+
- - ! '>='
|
108
|
+
- !ruby/object:Gem::Version
|
109
|
+
version: '1.29'
|
110
|
+
type: :development
|
111
|
+
prerelease: false
|
112
|
+
version_requirements: *19138580
|
113
|
+
- !ruby/object:Gem::Dependency
|
114
|
+
name: kyotocabinet-ruby
|
115
|
+
requirement: &19138120 !ruby/object:Gem::Requirement
|
116
|
+
none: false
|
117
|
+
requirements:
|
118
|
+
- - ! '>='
|
119
|
+
- !ruby/object:Gem::Version
|
120
|
+
version: 1.27.1
|
121
|
+
type: :development
|
122
|
+
prerelease: false
|
123
|
+
version_requirements: *19138120
|
124
|
+
- !ruby/object:Gem::Dependency
|
125
|
+
name: sqlite3
|
126
|
+
requirement: &19137660 !ruby/object:Gem::Requirement
|
127
|
+
none: false
|
128
|
+
requirements:
|
129
|
+
- - ! '>='
|
130
|
+
- !ruby/object:Gem::Version
|
131
|
+
version: 1.3.4
|
132
|
+
type: :development
|
133
|
+
prerelease: false
|
134
|
+
version_requirements: *19137660
|
50
135
|
description:
|
51
136
|
email:
|
52
|
-
executables:
|
137
|
+
executables:
|
53
138
|
- anemone
|
54
139
|
extensions: []
|
55
|
-
|
56
|
-
extra_rdoc_files:
|
140
|
+
extra_rdoc_files:
|
57
141
|
- README.rdoc
|
58
|
-
files:
|
142
|
+
files:
|
59
143
|
- VERSION
|
60
144
|
- LICENSE.txt
|
61
145
|
- CHANGELOG.rdoc
|
62
146
|
- README.rdoc
|
63
147
|
- Rakefile
|
64
|
-
- lib/anemone.rb
|
65
|
-
- lib/anemone/cookie_store.rb
|
66
|
-
- lib/anemone/storage.rb
|
67
|
-
- lib/anemone/core.rb
|
68
|
-
- lib/anemone/cli.rb
|
69
|
-
- lib/anemone/exceptions.rb
|
70
|
-
- lib/anemone/tentacle.rb
|
71
|
-
- lib/anemone/storage/tokyo_cabinet.rb
|
72
|
-
- lib/anemone/storage/base.rb
|
73
|
-
- lib/anemone/storage/exceptions.rb
|
74
148
|
- lib/anemone/storage/pstore.rb
|
75
149
|
- lib/anemone/storage/mongodb.rb
|
150
|
+
- lib/anemone/storage/tokyo_cabinet.rb
|
151
|
+
- lib/anemone/storage/exceptions.rb
|
76
152
|
- lib/anemone/storage/redis.rb
|
77
|
-
- lib/anemone/
|
153
|
+
- lib/anemone/storage/sqlite3.rb
|
154
|
+
- lib/anemone/storage/base.rb
|
155
|
+
- lib/anemone/storage/kyoto_cabinet.rb
|
78
156
|
- lib/anemone/page_store.rb
|
79
|
-
- lib/anemone/
|
80
|
-
- lib/anemone/
|
81
|
-
- lib/anemone/
|
157
|
+
- lib/anemone/storage.rb
|
158
|
+
- lib/anemone/tentacle.rb
|
159
|
+
- lib/anemone/http.rb
|
160
|
+
- lib/anemone/cli.rb
|
161
|
+
- lib/anemone/page.rb
|
162
|
+
- lib/anemone/exceptions.rb
|
163
|
+
- lib/anemone/core.rb
|
82
164
|
- lib/anemone/cli/url_list.rb
|
83
165
|
- lib/anemone/cli/serialize.rb
|
84
|
-
- lib/anemone/
|
85
|
-
-
|
86
|
-
-
|
87
|
-
-
|
166
|
+
- lib/anemone/cli/count.rb
|
167
|
+
- lib/anemone/cli/cron.rb
|
168
|
+
- lib/anemone/cli/pagedepth.rb
|
169
|
+
- lib/anemone/cookie_store.rb
|
170
|
+
- lib/anemone.rb
|
88
171
|
- spec/fakeweb_helper.rb
|
89
172
|
- spec/page_spec.rb
|
90
|
-
- spec/cookie_store_spec.rb
|
91
173
|
- spec/anemone_spec.rb
|
92
|
-
- spec/
|
174
|
+
- spec/core_spec.rb
|
93
175
|
- spec/storage_spec.rb
|
176
|
+
- spec/page_store_spec.rb
|
177
|
+
- spec/cookie_store_spec.rb
|
178
|
+
- spec/http_spec.rb
|
179
|
+
- spec/spec_helper.rb
|
94
180
|
- bin/anemone
|
95
|
-
has_rdoc: true
|
96
181
|
homepage: http://anemone.rubyforge.org
|
97
182
|
licenses: []
|
98
|
-
|
99
183
|
post_install_message:
|
100
|
-
rdoc_options:
|
184
|
+
rdoc_options:
|
101
185
|
- -m
|
102
186
|
- README.rdoc
|
103
187
|
- -t
|
104
188
|
- Anemone
|
105
|
-
require_paths:
|
189
|
+
require_paths:
|
106
190
|
- lib
|
107
|
-
required_ruby_version: !ruby/object:Gem::Requirement
|
191
|
+
required_ruby_version: !ruby/object:Gem::Requirement
|
108
192
|
none: false
|
109
|
-
requirements:
|
110
|
-
- -
|
111
|
-
- !ruby/object:Gem::Version
|
112
|
-
|
113
|
-
|
114
|
-
version: "0"
|
115
|
-
required_rubygems_version: !ruby/object:Gem::Requirement
|
193
|
+
requirements:
|
194
|
+
- - ! '>='
|
195
|
+
- !ruby/object:Gem::Version
|
196
|
+
version: '0'
|
197
|
+
required_rubygems_version: !ruby/object:Gem::Requirement
|
116
198
|
none: false
|
117
|
-
requirements:
|
118
|
-
- -
|
119
|
-
- !ruby/object:Gem::Version
|
120
|
-
|
121
|
-
- 0
|
122
|
-
version: "0"
|
199
|
+
requirements:
|
200
|
+
- - ! '>='
|
201
|
+
- !ruby/object:Gem::Version
|
202
|
+
version: '0'
|
123
203
|
requirements: []
|
124
|
-
|
125
204
|
rubyforge_project: anemone
|
126
|
-
rubygems_version: 1.
|
205
|
+
rubygems_version: 1.8.15
|
127
206
|
signing_key:
|
128
207
|
specification_version: 3
|
129
208
|
summary: Anemone web-spider framework
|
130
|
-
test_files:
|
131
|
-
- spec/http_spec.rb
|
132
|
-
- spec/page_store_spec.rb
|
133
|
-
- spec/core_spec.rb
|
209
|
+
test_files:
|
134
210
|
- spec/fakeweb_helper.rb
|
135
211
|
- spec/page_spec.rb
|
136
|
-
- spec/cookie_store_spec.rb
|
137
212
|
- spec/anemone_spec.rb
|
138
|
-
- spec/
|
213
|
+
- spec/core_spec.rb
|
139
214
|
- spec/storage_spec.rb
|
215
|
+
- spec/page_store_spec.rb
|
216
|
+
- spec/cookie_store_spec.rb
|
217
|
+
- spec/http_spec.rb
|
218
|
+
- spec/spec_helper.rb
|