twitter_ebooks 2.0.3 → 2.0.4
Sign up to get free protection for your applications and to get access to all the features.
- data/Gemfile.lock +12 -12
- data/bin/ebooks +21 -6
- data/data/ANC-all-count.txt +297241 -0
- data/data/stopwords.txt +204 -0
- data/data/wordfreq.json +1 -0
- data/lib/twitter_ebooks/bot.rb +25 -7
- data/lib/twitter_ebooks/markov.rb +55 -63
- data/lib/twitter_ebooks/model.rb +57 -74
- data/lib/twitter_ebooks/nlp.rb +90 -55
- data/lib/twitter_ebooks/version.rb +1 -1
- data/script/process_anc_data.rb +19 -0
- data/skeleton/Procfile +1 -1
- data/skeleton/bots.rb +0 -6
- data/skeleton/corpus/README.md +1 -1
- data/skeleton/run.rb +9 -0
- data/test/keywords.rb +18 -0
- data/twitter_ebooks.gemspec +3 -5
- metadata +13 -40
- data/skeleton/model/README.md +0 -1
data/Gemfile.lock
CHANGED
|
@@ -1,15 +1,15 @@
|
|
|
1
1
|
PATH
|
|
2
2
|
remote: .
|
|
3
3
|
specs:
|
|
4
|
-
twitter_ebooks (2.0.
|
|
4
|
+
twitter_ebooks (2.0.3)
|
|
5
|
+
bloomfilter-rb
|
|
5
6
|
engtagger
|
|
7
|
+
fast-stemmer
|
|
6
8
|
gingerice
|
|
9
|
+
highscore
|
|
7
10
|
htmlentities
|
|
8
|
-
|
|
9
|
-
ruby-stemmer
|
|
11
|
+
minitest
|
|
10
12
|
rufus-scheduler
|
|
11
|
-
tactful_tokenizer
|
|
12
|
-
tokenizer
|
|
13
13
|
tweetstream
|
|
14
14
|
twitter
|
|
15
15
|
|
|
@@ -19,6 +19,8 @@ GEM
|
|
|
19
19
|
addressable (2.3.5)
|
|
20
20
|
atomic (1.1.14)
|
|
21
21
|
awesome_print (1.2.0)
|
|
22
|
+
bloomfilter-rb (2.1.1)
|
|
23
|
+
redis
|
|
22
24
|
cookiejar (0.3.0)
|
|
23
25
|
daemons (1.1.9)
|
|
24
26
|
em-http-request (1.0.3)
|
|
@@ -37,25 +39,23 @@ GEM
|
|
|
37
39
|
eventmachine (1.0.3)
|
|
38
40
|
faraday (0.8.8)
|
|
39
41
|
multipart-post (~> 1.2.0)
|
|
42
|
+
fast-stemmer (1.0.2)
|
|
40
43
|
gingerice (1.2.1)
|
|
41
44
|
addressable
|
|
42
45
|
awesome_print
|
|
46
|
+
highscore (1.1.0)
|
|
47
|
+
whatlanguage (>= 1.0.0)
|
|
43
48
|
htmlentities (4.3.1)
|
|
44
49
|
http_parser.rb (0.5.3)
|
|
45
|
-
linguistics (2.0.2)
|
|
46
|
-
loggability (~> 0.5)
|
|
47
|
-
loggability (0.8.1)
|
|
48
50
|
minitest (5.0.8)
|
|
49
51
|
multi_json (1.8.2)
|
|
50
52
|
multipart-post (1.2.0)
|
|
51
|
-
|
|
53
|
+
redis (3.0.5)
|
|
52
54
|
rufus-scheduler (3.0.2)
|
|
53
55
|
tzinfo
|
|
54
56
|
simple_oauth (0.2.0)
|
|
55
|
-
tactful_tokenizer (0.0.2)
|
|
56
57
|
thread_safe (0.1.3)
|
|
57
58
|
atomic
|
|
58
|
-
tokenizer (0.1.1)
|
|
59
59
|
tweetstream (2.5.0)
|
|
60
60
|
daemons (~> 1.1)
|
|
61
61
|
em-http-request (~> 1.0.2)
|
|
@@ -68,11 +68,11 @@ GEM
|
|
|
68
68
|
simple_oauth (~> 0.2)
|
|
69
69
|
tzinfo (1.1.0)
|
|
70
70
|
thread_safe (~> 0.1)
|
|
71
|
+
whatlanguage (1.0.5)
|
|
71
72
|
yajl-ruby (1.1.0)
|
|
72
73
|
|
|
73
74
|
PLATFORMS
|
|
74
75
|
ruby
|
|
75
76
|
|
|
76
77
|
DEPENDENCIES
|
|
77
|
-
minitest
|
|
78
78
|
twitter_ebooks!
|
data/bin/ebooks
CHANGED
|
@@ -44,11 +44,7 @@ module Ebooks
|
|
|
44
44
|
end
|
|
45
45
|
|
|
46
46
|
def self.gen(model_path, input)
|
|
47
|
-
|
|
48
|
-
model = nil;
|
|
49
|
-
puts Benchmark.measure {
|
|
50
|
-
model = Model.load(model_path)
|
|
51
|
-
}
|
|
47
|
+
model = Model.load(model_path)
|
|
52
48
|
if input && !input.empty?
|
|
53
49
|
puts "@cmd " + model.markov_response(input, 135)
|
|
54
50
|
else
|
|
@@ -56,16 +52,33 @@ module Ebooks
|
|
|
56
52
|
end
|
|
57
53
|
end
|
|
58
54
|
|
|
55
|
+
def self.score(model_path, input)
|
|
56
|
+
model = Model.load(model_path)
|
|
57
|
+
model.score_interest(input)
|
|
58
|
+
end
|
|
59
|
+
|
|
59
60
|
def self.archive(username, outpath)
|
|
60
61
|
Archiver.new(username, outpath).fetch_tweets
|
|
61
62
|
end
|
|
62
63
|
|
|
64
|
+
def self.tweet(modelpath, username)
|
|
65
|
+
load File.join(APP_PATH, 'bots.rb')
|
|
66
|
+
model = Model.load(modelpath)
|
|
67
|
+
statement = model.markov_statement
|
|
68
|
+
log "@#{username}: #{statement}"
|
|
69
|
+
bot = Bot.get(username)
|
|
70
|
+
bot.configure
|
|
71
|
+
bot.tweet(statement)
|
|
72
|
+
end
|
|
73
|
+
|
|
63
74
|
def self.command(args)
|
|
64
75
|
usage = """Usage:
|
|
65
76
|
ebooks new <reponame>
|
|
66
77
|
ebooks consume <corpus_path> [...]
|
|
67
|
-
ebooks gen <
|
|
78
|
+
ebooks gen <model_path> [input]
|
|
79
|
+
ebooks score <model_path> <input>
|
|
68
80
|
ebooks archive <@user> <outpath>
|
|
81
|
+
ebooks tweet <model_path> <@bot>
|
|
69
82
|
"""
|
|
70
83
|
|
|
71
84
|
if args.length == 0
|
|
@@ -77,7 +90,9 @@ module Ebooks
|
|
|
77
90
|
when "new" then new(args[1])
|
|
78
91
|
when "consume" then consume(args[1..-1])
|
|
79
92
|
when "gen" then gen(args[1], args[2..-1].join(' '))
|
|
93
|
+
when "score" then score(args[1], args[2..-1].join(' '))
|
|
80
94
|
when "archive" then archive(args[1], args[2])
|
|
95
|
+
when "tweet" then tweet(args[1], args[2])
|
|
81
96
|
end
|
|
82
97
|
end
|
|
83
98
|
end
|