porras-imdb 0.0.6 → 0.0.7
Sign up to get free protection for your applications and to get access to all the features.
- data/README +11 -11
- data/lib/imdb/imdb_movie.rb +4 -0
- data/lib/imdb/imdb_search.rb +23 -9
- data/spec/imdb_movie_spec.rb +4 -0
- data/spec/imdb_search_spec.rb +35 -1
- metadata +1 -1
data/README
CHANGED
@@ -1,6 +1,4 @@
|
|
1
1
|
|
2
|
-
ImdbMovie
|
3
|
-
|
4
2
|
ImdbMovie Indiana Jones and the Last Crusade
|
5
3
|
- should query IMDB url
|
6
4
|
- should get director
|
@@ -8,6 +6,7 @@ ImdbMovie Indiana Jones and the Last Crusade
|
|
8
6
|
- should get the rating
|
9
7
|
- should get cast members
|
10
8
|
- should get the writers
|
9
|
+
- should get the year
|
11
10
|
- should get the release date
|
12
11
|
- should get the genres
|
13
12
|
- should get the plot
|
@@ -40,24 +39,25 @@ ImdbMovie Han robado una estrella
|
|
40
39
|
- should get the company
|
41
40
|
- should not get any photos
|
42
41
|
|
43
|
-
ImdbSearch
|
44
|
-
|
45
|
-
ImdbSearch Indiana Jones
|
42
|
+
ImdbSearch search that returns multiple movies
|
46
43
|
- should query IMDB url
|
44
|
+
- should not allow to change the query
|
47
45
|
|
48
|
-
ImdbSearch
|
46
|
+
ImdbSearch search that returns multiple movies movies
|
49
47
|
- should be a collection of ImdbMovie instances
|
50
48
|
- should include 'Indiana Jones and the Last Crusade'
|
51
49
|
- should have titles
|
52
50
|
- should not have titles with HTML tags
|
51
|
+
- should not have duplicate movies
|
53
52
|
|
54
|
-
ImdbSearch
|
53
|
+
ImdbSearch search that redirects to the lone matching movie movies
|
54
|
+
- should be a collection containing a single ImdbMovie instance
|
55
|
+
- should have the correct ID
|
56
|
+
- should have the correct title
|
55
57
|
|
56
58
|
ImdbSearch searches with potential encoding issues movies
|
57
59
|
- should include 'Misión en Marbella'
|
58
60
|
|
59
|
-
String
|
60
|
-
|
61
61
|
String unescape_html
|
62
62
|
- should convert & to &
|
63
63
|
- should convert ó to ó
|
@@ -65,6 +65,6 @@ String unescape_html
|
|
65
65
|
String strip_tags
|
66
66
|
- should strip HTML tags
|
67
67
|
|
68
|
-
Finished in
|
68
|
+
Finished in 0.829689 seconds
|
69
69
|
|
70
|
-
|
70
|
+
48 examples, 0 failures
|
data/lib/imdb/imdb_movie.rb
CHANGED
@@ -27,6 +27,10 @@ class ImdbMovie
|
|
27
27
|
def writers
|
28
28
|
document.search("h5[text()^='Writers'] ~ a").map { |link| link.innerHTML.strip.unescape_html }.reject { |w| w == 'more' }.uniq rescue []
|
29
29
|
end
|
30
|
+
|
31
|
+
def year
|
32
|
+
document.search('a[@href^="/Sections/Years/"]').innerHTML.to_i
|
33
|
+
end
|
30
34
|
|
31
35
|
def release_date
|
32
36
|
date = document.search("//h5[text()^='Release Date']/..").innerHTML[/^\d{1,2} \w+ \d{4}/]
|
data/lib/imdb/imdb_search.rb
CHANGED
@@ -1,23 +1,37 @@
|
|
1
1
|
class ImdbSearch
|
2
2
|
|
3
|
-
attr_accessor :query
|
4
|
-
|
5
3
|
def initialize(query)
|
6
|
-
|
4
|
+
@query = query
|
7
5
|
end
|
8
6
|
|
9
7
|
def movies
|
10
|
-
@movies ||=
|
11
|
-
element.innerHTML.strip_tags.empty?
|
12
|
-
end.map do |element|
|
13
|
-
ImdbMovie.new(element['href'][/\d+/], element.innerHTML.strip_tags.unescape_html)
|
14
|
-
end
|
8
|
+
@movies ||= (exact_match? ? parse_movie : parse_movies)
|
15
9
|
end
|
16
10
|
|
17
11
|
private
|
18
12
|
|
19
13
|
def document
|
20
|
-
@document ||= Hpricot(open("http://www.imdb.com/find?q=#{CGI::escape(query)};s=tt").read)
|
14
|
+
@document ||= Hpricot(open("http://www.imdb.com/find?q=#{CGI::escape(@query)};s=tt").read)
|
15
|
+
end
|
16
|
+
|
17
|
+
def parse_movies
|
18
|
+
document.search('a[@href^="/title/tt"]').reject do |element|
|
19
|
+
element.innerHTML.strip_tags.empty?
|
20
|
+
end.map do |element|
|
21
|
+
[element['href'][/\d+/], element.innerHTML.strip_tags.unescape_html]
|
22
|
+
end.uniq.map do |values|
|
23
|
+
ImdbMovie.new(*values)
|
24
|
+
end
|
25
|
+
end
|
26
|
+
|
27
|
+
def parse_movie
|
28
|
+
id = document.at("a[@name='poster']")['href'][/\d+$/]
|
29
|
+
title = document.at("h1").innerHTML.split('<span').first.strip.unescape_html
|
30
|
+
[ImdbMovie.new(id, title)]
|
31
|
+
end
|
32
|
+
|
33
|
+
def exact_match?
|
34
|
+
document.at("title[text()='IMDb Search']").nil?
|
21
35
|
end
|
22
36
|
|
23
37
|
end
|
data/spec/imdb_movie_spec.rb
CHANGED
@@ -40,6 +40,10 @@ describe ImdbMovie do
|
|
40
40
|
@imdb_movie.writers.should include('Philip Kaufman')
|
41
41
|
@imdb_movie.writers.should_not include('more')
|
42
42
|
end
|
43
|
+
|
44
|
+
it "should get the year" do
|
45
|
+
@imdb_movie.year.should == 1989
|
46
|
+
end
|
43
47
|
|
44
48
|
it "should get the release date" do
|
45
49
|
@imdb_movie.release_date.should be_an_instance_of(Date)
|
data/spec/imdb_search_spec.rb
CHANGED
@@ -2,7 +2,7 @@ require File.dirname(__FILE__) + '/spec_helper'
|
|
2
2
|
|
3
3
|
describe ImdbSearch do
|
4
4
|
|
5
|
-
describe '
|
5
|
+
describe 'search that returns multiple movies' do
|
6
6
|
|
7
7
|
before(:each) do
|
8
8
|
@imdb_search = ImdbSearch.new('indiana jones')
|
@@ -13,6 +13,10 @@ describe ImdbSearch do
|
|
13
13
|
@imdb_search.should_receive(:open).with("http://www.imdb.com/find?q=indiana+jones;s=tt").and_return(open("#{$samples_dir}/sample_search.html"))
|
14
14
|
@imdb_search.send(:document)
|
15
15
|
end
|
16
|
+
|
17
|
+
it "should not allow to change the query" do
|
18
|
+
lambda { @imdb_search.query = 'wadus' }.should raise_error(NoMethodError)
|
19
|
+
end
|
16
20
|
|
17
21
|
describe "movies" do
|
18
22
|
|
@@ -39,7 +43,37 @@ describe ImdbSearch do
|
|
39
43
|
movie.title.should_not match(/<.+>/)
|
40
44
|
end
|
41
45
|
end
|
46
|
+
|
47
|
+
it "should not have duplicate movies" do
|
48
|
+
all_movie_ids = @imdb_search.movies.collect {|m| m.id}
|
49
|
+
unique_movie_ids = all_movie_ids.uniq
|
50
|
+
all_movie_ids.should == unique_movie_ids
|
51
|
+
end
|
52
|
+
end
|
42
53
|
|
54
|
+
end
|
55
|
+
|
56
|
+
describe 'search that redirects to the lone matching movie' do
|
57
|
+
|
58
|
+
before(:each) do
|
59
|
+
@imdb_search = ImdbSearch.new('some extremely specific search for indiana jones')
|
60
|
+
@imdb_search.stub!(:open).and_return(open("#{$samples_dir}/sample_movie.html"))
|
61
|
+
end
|
62
|
+
|
63
|
+
describe "movies" do
|
64
|
+
|
65
|
+
it "should be a collection containing a single ImdbMovie instance" do
|
66
|
+
@imdb_search.movies.size.should == 1
|
67
|
+
@imdb_search.movies.first.should be_an_instance_of(ImdbMovie)
|
68
|
+
end
|
69
|
+
|
70
|
+
it "should have the correct ID" do
|
71
|
+
@imdb_search.movies.first.id.should == '0097576'
|
72
|
+
end
|
73
|
+
|
74
|
+
it "should have the correct title" do
|
75
|
+
@imdb_search.movies.first.title.should == 'Indiana Jones and the Last Crusade'
|
76
|
+
end
|
43
77
|
end
|
44
78
|
|
45
79
|
end
|