apollo-crawler 0.1.5 → 0.1.6
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +8 -8
- data/bin/apollo-crawler +12 -410
- data/lib/apollo_crawler.rb +31 -20
- data/lib/apollo_crawler/{cache.rb → cache/cache_base.rb} +37 -34
- data/lib/apollo_crawler/cache/factory.rb +35 -0
- data/lib/apollo_crawler/{caches → cache}/filesystem_cache.rb +37 -34
- data/lib/apollo_crawler/cache/memcached_cache.rb +51 -0
- data/lib/apollo_crawler/{caches → cache}/memory_cache.rb +46 -43
- data/lib/apollo_crawler/{caches → cache}/null_cache.rb +33 -30
- data/lib/apollo_crawler/config.rb +53 -0
- data/lib/apollo_crawler/{crawler.rb → crawler/crawler_base.rb} +157 -155
- data/lib/apollo_crawler/{crawler_template.rb → crawler/crawler_template.rb} +24 -24
- data/lib/apollo_crawler/{crawlers → crawler}/google_com/google.rb +40 -40
- data/lib/apollo_crawler/{crawlers → crawler}/slashdot_org/slashdot.rb +40 -40
- data/lib/apollo_crawler/{crawlers → crawler}/stackoverflow_com/stackoverflow.rb +44 -44
- data/lib/apollo_crawler/{crawlers → crawler}/xkcd_com/xkcd.rb +35 -35
- data/lib/apollo_crawler/{crawlers → crawler}/ycombinator_com/hacker_news.rb +44 -44
- data/lib/apollo_crawler/fetcher/fetcher_base.rb +6 -0
- data/lib/apollo_crawler/fetcher/simple_fetcher.rb +8 -0
- data/lib/apollo_crawler/formatter/formatter_base.rb +6 -0
- data/lib/apollo_crawler/{formatters → formatter}/formatter_json.rb +17 -17
- data/lib/apollo_crawler/{formatters → formatter}/formatter_plain.rb +17 -17
- data/lib/apollo_crawler/{formatters → formatter}/formatter_table.rb +35 -35
- data/lib/apollo_crawler/lib.rb +28 -0
- data/lib/apollo_crawler/program.rb +406 -0
- data/lib/apollo_crawler/store/store_base.rb +6 -0
- data/lib/apollo_crawler/version.rb +2 -2
- metadata +52 -17
- data/lib/apollo_crawler/caches/factory.rb +0 -30
- data/lib/apollo_crawler/formatter.rb +0 -6
@@ -1,17 +1,17 @@
|
|
1
|
-
require 'json'
|
2
|
-
|
3
|
-
require File.join(File.dirname(__FILE__), '
|
4
|
-
|
5
|
-
module Apollo
|
6
|
-
module
|
7
|
-
class Json <
|
8
|
-
def format(obj)
|
9
|
-
return Json.format(obj)
|
10
|
-
end
|
11
|
-
|
12
|
-
def self.format(obj)
|
13
|
-
return JSON.pretty_generate(obj)
|
14
|
-
end
|
15
|
-
end
|
16
|
-
end #
|
17
|
-
end # Apollo
|
1
|
+
require 'json'
|
2
|
+
|
3
|
+
require File.join(File.dirname(__FILE__), 'formatter_base')
|
4
|
+
|
5
|
+
module Apollo
|
6
|
+
module Formatter
|
7
|
+
class Json < FormatterBase
|
8
|
+
def format(obj)
|
9
|
+
return Json.format(obj)
|
10
|
+
end
|
11
|
+
|
12
|
+
def self.format(obj)
|
13
|
+
return JSON.pretty_generate(obj)
|
14
|
+
end
|
15
|
+
end
|
16
|
+
end # Formatter
|
17
|
+
end # Apollo
|
@@ -1,17 +1,17 @@
|
|
1
|
-
require 'awesome_print'
|
2
|
-
|
3
|
-
require File.join(File.dirname(__FILE__), '
|
4
|
-
|
5
|
-
module Apollo
|
6
|
-
module
|
7
|
-
class Plain <
|
8
|
-
def format(obj)
|
9
|
-
return Plain.format(obj)
|
10
|
-
end
|
11
|
-
|
12
|
-
def self.format(obj)
|
13
|
-
return obj.inspect
|
14
|
-
end
|
15
|
-
end
|
16
|
-
end #
|
17
|
-
end # Apollo
|
1
|
+
require 'awesome_print'
|
2
|
+
|
3
|
+
require File.join(File.dirname(__FILE__), 'formatter_base')
|
4
|
+
|
5
|
+
module Apollo
|
6
|
+
module Formatter
|
7
|
+
class Plain < FormatterBase
|
8
|
+
def format(obj)
|
9
|
+
return Plain.format(obj)
|
10
|
+
end
|
11
|
+
|
12
|
+
def self.format(obj)
|
13
|
+
return obj.inspect
|
14
|
+
end
|
15
|
+
end
|
16
|
+
end # Formatter
|
17
|
+
end # Apollo
|
@@ -1,35 +1,35 @@
|
|
1
|
-
require 'terminal-table'
|
2
|
-
|
3
|
-
require File.join(File.dirname(__FILE__), '
|
4
|
-
|
5
|
-
module Apollo
|
6
|
-
module
|
7
|
-
class Table <
|
8
|
-
def format(obj)
|
9
|
-
return Table.format(obj)
|
10
|
-
end
|
11
|
-
|
12
|
-
def self.format(obj)
|
13
|
-
headings = []
|
14
|
-
if(obj[:data].length > 0)
|
15
|
-
headings = obj[:data][0].keys
|
16
|
-
end
|
17
|
-
|
18
|
-
rows = []
|
19
|
-
obj[:data].each do |line|
|
20
|
-
next if (line.nil? || line.empty?)
|
21
|
-
|
22
|
-
data = []
|
23
|
-
headings.each do |column|
|
24
|
-
data << line[column]
|
25
|
-
end
|
26
|
-
|
27
|
-
rows << data
|
28
|
-
end
|
29
|
-
|
30
|
-
table = Terminal::Table.new :headings => headings, :rows => rows
|
31
|
-
return table
|
32
|
-
end
|
33
|
-
end
|
34
|
-
end #
|
35
|
-
end # Apollo
|
1
|
+
require 'terminal-table'
|
2
|
+
|
3
|
+
require File.join(File.dirname(__FILE__), 'formatter_base')
|
4
|
+
|
5
|
+
module Apollo
|
6
|
+
module Formatter
|
7
|
+
class Table < FormatterBase
|
8
|
+
def format(obj)
|
9
|
+
return Table.format(obj)
|
10
|
+
end
|
11
|
+
|
12
|
+
def self.format(obj)
|
13
|
+
headings = []
|
14
|
+
if(obj[:data].length > 0)
|
15
|
+
headings = obj[:data][0].keys
|
16
|
+
end
|
17
|
+
|
18
|
+
rows = []
|
19
|
+
obj[:data].each do |line|
|
20
|
+
next if (line.nil? || line.empty?)
|
21
|
+
|
22
|
+
data = []
|
23
|
+
headings.each do |column|
|
24
|
+
data << line[column]
|
25
|
+
end
|
26
|
+
|
27
|
+
rows << data
|
28
|
+
end
|
29
|
+
|
30
|
+
table = Terminal::Table.new :headings => headings, :rows => rows
|
31
|
+
return table
|
32
|
+
end
|
33
|
+
end
|
34
|
+
end # Formatter
|
35
|
+
end # Apollo
|
@@ -0,0 +1,28 @@
|
|
1
|
+
# Main
|
2
|
+
require File.join(File.dirname(__FILE__), 'program')
|
3
|
+
|
4
|
+
# Caches
|
5
|
+
require File.join(File.dirname(__FILE__), 'cache/cache_base')
|
6
|
+
require File.join(File.dirname(__FILE__), 'cache/factory')
|
7
|
+
require File.join(File.dirname(__FILE__), 'cache/filesystem_cache')
|
8
|
+
require File.join(File.dirname(__FILE__), 'cache/memcached_cache')
|
9
|
+
require File.join(File.dirname(__FILE__), 'cache/memory_cache')
|
10
|
+
require File.join(File.dirname(__FILE__), 'cache/null_cache')
|
11
|
+
|
12
|
+
# Crawlers
|
13
|
+
require File.join(File.dirname(__FILE__), 'crawler/crawler_base')
|
14
|
+
require File.join(File.dirname(__FILE__), 'crawler/google_com/google')
|
15
|
+
require File.join(File.dirname(__FILE__), 'crawler/slashdot_org/slashdot')
|
16
|
+
require File.join(File.dirname(__FILE__), 'crawler/stackoverflow_com/stackoverflow')
|
17
|
+
require File.join(File.dirname(__FILE__), 'crawler/xkcd_com/xkcd')
|
18
|
+
require File.join(File.dirname(__FILE__), 'crawler/ycombinator_com/hacker_news')
|
19
|
+
|
20
|
+
# Fetchers
|
21
|
+
require File.join(File.dirname(__FILE__), 'fetcher/fetcher_base')
|
22
|
+
require File.join(File.dirname(__FILE__), 'fetcher/simple_fetcher')
|
23
|
+
|
24
|
+
# Formatters
|
25
|
+
require File.join(File.dirname(__FILE__), 'formatter/formatter_base')
|
26
|
+
require File.join(File.dirname(__FILE__), 'formatter/formatter_json')
|
27
|
+
require File.join(File.dirname(__FILE__), 'formatter/formatter_plain')
|
28
|
+
require File.join(File.dirname(__FILE__), 'formatter/formatter_table')
|
@@ -0,0 +1,406 @@
|
|
1
|
+
require "rubygems"
|
2
|
+
require "bundler/setup"
|
3
|
+
|
4
|
+
require 'json'
|
5
|
+
|
6
|
+
require "thor"
|
7
|
+
|
8
|
+
require "open-uri"
|
9
|
+
require "nokogiri"
|
10
|
+
|
11
|
+
require "pp"
|
12
|
+
require "optparse"
|
13
|
+
|
14
|
+
require 'active_support'
|
15
|
+
require 'active_support/inflector'
|
16
|
+
|
17
|
+
require 'terminal-table'
|
18
|
+
|
19
|
+
require File.join(File.dirname(__FILE__), 'version')
|
20
|
+
|
21
|
+
# require File.join(File.dirname(__FILE__), 'config/crawler')
|
22
|
+
# puts Apollo::CrawlerProgramConfig
|
23
|
+
|
24
|
+
module Apollo
|
25
|
+
class CrawlerProgram
|
26
|
+
require File.join(File.dirname(__FILE__), "config")
|
27
|
+
|
28
|
+
# This hash will hold all of the options
|
29
|
+
# parsed from the command-line by OptionParser.
|
30
|
+
@caches = nil
|
31
|
+
@crawlers = nil
|
32
|
+
@formatter = nil
|
33
|
+
@formatters = nil
|
34
|
+
@options = nil
|
35
|
+
@optparser = nil
|
36
|
+
|
37
|
+
# Initializer - Constructor
|
38
|
+
def initialize
|
39
|
+
@caches = {}
|
40
|
+
@crawlers = {}
|
41
|
+
@formatter = RbConfig::DEFAULT_FORMATTER
|
42
|
+
@formatters = {}
|
43
|
+
end
|
44
|
+
|
45
|
+
# Initialize command-line options
|
46
|
+
def init_options
|
47
|
+
@options = {}
|
48
|
+
@options[:verbose] = false
|
49
|
+
@options[:version] = false
|
50
|
+
@options[:cache_dirs] = [
|
51
|
+
RbConfig::CACHES_DIR
|
52
|
+
]
|
53
|
+
@options[:crawler_dirs] = [
|
54
|
+
RbConfig::CRAWLERS_DIR
|
55
|
+
]
|
56
|
+
@options[:formatter_dirs] = [
|
57
|
+
RbConfig::FORMATTERS_DIR
|
58
|
+
]
|
59
|
+
@options[:generate_crawler] = nil
|
60
|
+
|
61
|
+
@optparser = OptionParser.new do | opts |
|
62
|
+
opts.banner = "Usage: apollo-crawler [OPTIONS] CRAWLER_NAME [START_URL]"
|
63
|
+
|
64
|
+
opts.separator ""
|
65
|
+
opts.separator "Specific options:"
|
66
|
+
|
67
|
+
# This displays the help screen, all programs are
|
68
|
+
# assumed to have this option.
|
69
|
+
opts.on('-h', '--help', 'Display this screen') do
|
70
|
+
@options[:show_help]
|
71
|
+
end
|
72
|
+
|
73
|
+
opts.on('-a', '--all', 'Run all crawlers') do
|
74
|
+
@options[:run_all] = true
|
75
|
+
end
|
76
|
+
|
77
|
+
opts.on('-f', '--format [NAME]', "Formatter used") do |name|
|
78
|
+
@options[:formatter] = name
|
79
|
+
end
|
80
|
+
|
81
|
+
opts.on('-g', '--generate [NAME]', "Generate scaffold for new crawler") do |name|
|
82
|
+
@options[:generate_crawler] = name
|
83
|
+
end
|
84
|
+
|
85
|
+
opts.on('-i', '--include [PATH]', 'Include additional crawler or crawler directory') do |path|
|
86
|
+
@options[:crawler_dirs] << path
|
87
|
+
end
|
88
|
+
|
89
|
+
opts.on('-v', '--verbose', 'Enable verbose output') do
|
90
|
+
@options[:verbose] = true
|
91
|
+
end
|
92
|
+
|
93
|
+
opts.on('-V', '--version', 'Show version info') do
|
94
|
+
@options[:version] = true
|
95
|
+
end
|
96
|
+
|
97
|
+
opts.on('-l', '--list-crawlers', 'List of crawlers') do
|
98
|
+
@options[:list_crawlers] = true
|
99
|
+
end
|
100
|
+
|
101
|
+
opts.on(nil, '--list-formatters', 'List of formatters available') do
|
102
|
+
@options[:list_formatters] = true
|
103
|
+
end
|
104
|
+
end
|
105
|
+
end
|
106
|
+
|
107
|
+
# Parse the options passed to command-line
|
108
|
+
def parse_options(args = ARGV)
|
109
|
+
# Parse the command-line. Remember there are two forms
|
110
|
+
# of the parse method. The 'parse' method simply parses
|
111
|
+
# ARGV, while the 'parse!' method parses ARGV and removes
|
112
|
+
# any options found there, as well as any parameters for
|
113
|
+
# the options. What's left is the list of files to resize.
|
114
|
+
@optparser.parse!(args)
|
115
|
+
end
|
116
|
+
|
117
|
+
# Load global options first
|
118
|
+
# Merge it with local options (if they exists)
|
119
|
+
def load_config_file()
|
120
|
+
config = File.join(File.dirname(__FILE__), "config", "crawler.rb")
|
121
|
+
if(File.exists?(config))
|
122
|
+
if(@options[:verbose])
|
123
|
+
puts "Loading config '#{config}'"
|
124
|
+
end
|
125
|
+
|
126
|
+
# puts "Let's require '#{@options[:verbose]}'"
|
127
|
+
require config
|
128
|
+
else
|
129
|
+
if(@options[:verbose])
|
130
|
+
# TODO: Add support for initial rake task generation
|
131
|
+
# Something like this:
|
132
|
+
# rake config:init # Initializes config files with
|
133
|
+
# their defaults (if not exists already)
|
134
|
+
puts "Default config does not exist, skipping - '#{config}'"
|
135
|
+
end
|
136
|
+
end
|
137
|
+
end
|
138
|
+
|
139
|
+
# Register caches
|
140
|
+
def register_cache(dir)
|
141
|
+
if(@options[:verbose])
|
142
|
+
puts "Registering caches - '#{dir}'"
|
143
|
+
end
|
144
|
+
|
145
|
+
files = File.join(dir, "**", "*.rb")
|
146
|
+
Dir.glob(files).each do |file|
|
147
|
+
require file
|
148
|
+
end
|
149
|
+
|
150
|
+
tmp = Apollo::Cache.constants.select { |c|
|
151
|
+
Class === Apollo::Cache.const_get(c)
|
152
|
+
}
|
153
|
+
|
154
|
+
tmp.each do |x|
|
155
|
+
klass = Object.const_get('Apollo').const_get('Cache').const_get(x)
|
156
|
+
@caches.merge!({ x.downcase.to_s => klass})
|
157
|
+
end
|
158
|
+
|
159
|
+
if(@options[:verbose])
|
160
|
+
@caches.each do |cache, klass|
|
161
|
+
name = klass
|
162
|
+
|
163
|
+
# klass.ancestors.include?(Apollo::Caches::Cache)
|
164
|
+
if name == "Apollo::Caches::Cache"
|
165
|
+
next
|
166
|
+
end
|
167
|
+
|
168
|
+
puts "Registered cache '#{cache}' -> '#{name}'"
|
169
|
+
end
|
170
|
+
end
|
171
|
+
end
|
172
|
+
|
173
|
+
# Register crawlers
|
174
|
+
def register_crawlers(dir)
|
175
|
+
if(@options[:verbose])
|
176
|
+
puts "Registering crawlers - '#{dir}'"
|
177
|
+
end
|
178
|
+
|
179
|
+
files = File.join(dir, "**", "*.rb")
|
180
|
+
Dir.glob(files).each do |file|
|
181
|
+
require file
|
182
|
+
end
|
183
|
+
|
184
|
+
tmp = Apollo::Crawler.constants.select { |c|
|
185
|
+
Class === Apollo::Crawler.const_get(c)
|
186
|
+
}
|
187
|
+
|
188
|
+
tmp.each do |x|
|
189
|
+
klass = Object.const_get('Apollo').const_get('Crawler').const_get(x)
|
190
|
+
@crawlers.merge!({ x.downcase.to_s => klass})
|
191
|
+
end
|
192
|
+
|
193
|
+
if(@options[:verbose])
|
194
|
+
@crawlers.each do |crawler, klass|
|
195
|
+
name = klass.new.class.name
|
196
|
+
|
197
|
+
if name == "Apollo::Crawler::Crawler"
|
198
|
+
next
|
199
|
+
end
|
200
|
+
|
201
|
+
puts "Registered crawler '#{crawler}' -> '#{name}'"
|
202
|
+
end
|
203
|
+
end
|
204
|
+
end
|
205
|
+
|
206
|
+
# Register formatters
|
207
|
+
def register_formatters(dir)
|
208
|
+
if(@options[:verbose])
|
209
|
+
puts "Registering formatters - '#{dir}'"
|
210
|
+
end
|
211
|
+
|
212
|
+
files = File.join(dir, "**", "*.rb")
|
213
|
+
Dir.glob(files).each do |file|
|
214
|
+
require file
|
215
|
+
end
|
216
|
+
|
217
|
+
tmp = Apollo::Formatter.constants.select { |c|
|
218
|
+
Class === Apollo::Formatter.const_get(c)
|
219
|
+
}
|
220
|
+
|
221
|
+
tmp.each do |x|
|
222
|
+
klass = Object.const_get('Apollo').const_get('Formatter').const_get(x)
|
223
|
+
@formatters.merge!({ x.downcase.to_s => klass})
|
224
|
+
end
|
225
|
+
|
226
|
+
if(@options[:verbose])
|
227
|
+
@formatters.each do |formatter, klass|
|
228
|
+
name = klass.new.class.name
|
229
|
+
|
230
|
+
if name == "Apollo::Formatters::Formatter"
|
231
|
+
next
|
232
|
+
end
|
233
|
+
|
234
|
+
puts "Registered formatter '#{formatter}' -> '#{name}'"
|
235
|
+
end
|
236
|
+
end
|
237
|
+
end
|
238
|
+
|
239
|
+
def generate_crawler(name, url = nil, matcher = nil)
|
240
|
+
name = name.titleize.gsub(" ", "")
|
241
|
+
|
242
|
+
if(@options[:verbose])
|
243
|
+
puts "Generating new crawler '#{name}'"
|
244
|
+
end
|
245
|
+
|
246
|
+
template_path = RbConfig::CRAWLER_TEMPLATE_PATH
|
247
|
+
if(File.exists?(template_path) == false)
|
248
|
+
puts "Template file '#{template_path}' does not exists!"
|
249
|
+
return
|
250
|
+
end
|
251
|
+
|
252
|
+
if(@options[:verbose])
|
253
|
+
puts "Using template '#{template_path}'"
|
254
|
+
end
|
255
|
+
|
256
|
+
dest_path = File.join(Dir.pwd, "#{name.underscore}.rb")
|
257
|
+
|
258
|
+
url = url ? url : "http://some-url-here"
|
259
|
+
matcher = matcher ? matcher : "//a"
|
260
|
+
|
261
|
+
placeholders = {
|
262
|
+
"CRAWLER_CLASS_NAME" => name,
|
263
|
+
"CRAWLER_NAME" => name.titleize,
|
264
|
+
"CRAWLER_URL" => url,
|
265
|
+
"CRAWLER_MATCHER" => matcher
|
266
|
+
}
|
267
|
+
|
268
|
+
puts "Generating crawler '#{name.titleize}', class: '#{name}', path: '#{dest_path}'"
|
269
|
+
|
270
|
+
File.open(template_path, 'r') do |tmpl|
|
271
|
+
File.open(dest_path, 'w') do |crawler|
|
272
|
+
while line = tmpl.gets
|
273
|
+
#puts line
|
274
|
+
placeholders.each do |k, v|
|
275
|
+
line.gsub!(k, v)
|
276
|
+
end
|
277
|
+
|
278
|
+
crawler.puts line
|
279
|
+
end
|
280
|
+
end
|
281
|
+
end
|
282
|
+
end
|
283
|
+
|
284
|
+
def run(args = ARGV)
|
285
|
+
puts "#{ARGV.inspect}"
|
286
|
+
|
287
|
+
init_options()
|
288
|
+
|
289
|
+
parse_options(args)
|
290
|
+
|
291
|
+
if(@options[:version])
|
292
|
+
puts Apollo::VERSION
|
293
|
+
return 0
|
294
|
+
end
|
295
|
+
|
296
|
+
if(@options[:show_help])
|
297
|
+
puts @optparser
|
298
|
+
return 0
|
299
|
+
end
|
300
|
+
|
301
|
+
load_config_file()
|
302
|
+
|
303
|
+
if(@options[:generate_crawler])
|
304
|
+
name = @options[:generate_crawler]
|
305
|
+
url = args.length > 0 ? args[0] : nil
|
306
|
+
matcher = args.length > 1 ? args[1] : nil
|
307
|
+
|
308
|
+
self.generate_crawler(name, url, matcher)
|
309
|
+
return 0
|
310
|
+
end
|
311
|
+
|
312
|
+
# Register caches which can be used
|
313
|
+
@options[:cache_dirs].each do |dir|
|
314
|
+
register_cache(dir)
|
315
|
+
end
|
316
|
+
|
317
|
+
# Register sites which can be crawled
|
318
|
+
@options[:crawler_dirs].each do |dir|
|
319
|
+
register_crawlers(dir)
|
320
|
+
end
|
321
|
+
|
322
|
+
# Register sites which can be crawled
|
323
|
+
@options[:formatter_dirs].each do |dir|
|
324
|
+
register_formatters(dir)
|
325
|
+
end
|
326
|
+
|
327
|
+
# Set default formatter here
|
328
|
+
formatter_name = "json"
|
329
|
+
if(@options[:formatter])
|
330
|
+
formatter_name = @options[:formatter]
|
331
|
+
end
|
332
|
+
|
333
|
+
# Look for specified formatter
|
334
|
+
f = @formatters.select { |k, v|
|
335
|
+
k.downcase == formatter_name.downcase
|
336
|
+
}
|
337
|
+
|
338
|
+
if(f)
|
339
|
+
@formatter = f[f.keys[0]]
|
340
|
+
end
|
341
|
+
|
342
|
+
if(@options[:list_formatters])
|
343
|
+
headings = ['name', 'class']
|
344
|
+
rows = @formatters
|
345
|
+
|
346
|
+
table = Terminal::Table.new :headings => headings, :rows => rows
|
347
|
+
|
348
|
+
puts table
|
349
|
+
return 0
|
350
|
+
end
|
351
|
+
|
352
|
+
if(@options[:list_crawlers])
|
353
|
+
headings = ['name', 'class']
|
354
|
+
rows = @crawlers
|
355
|
+
|
356
|
+
table = Terminal::Table.new :headings => headings, :rows => rows
|
357
|
+
|
358
|
+
puts table
|
359
|
+
return 0
|
360
|
+
end
|
361
|
+
|
362
|
+
crawlers = []
|
363
|
+
if(args.length > 0)
|
364
|
+
crawlers << args.shift
|
365
|
+
end
|
366
|
+
|
367
|
+
if(@options[:run_all])
|
368
|
+
crawlers = @crawlers.keys
|
369
|
+
end
|
370
|
+
|
371
|
+
if(crawlers.empty?)
|
372
|
+
puts @optparser
|
373
|
+
return 0
|
374
|
+
end
|
375
|
+
|
376
|
+
crawlers.each do |crawler|
|
377
|
+
p = @crawlers[crawler.downcase]
|
378
|
+
if(p == nil)
|
379
|
+
puts "Invalid crawler name - '#{crawler}'"
|
380
|
+
puts "See program help"
|
381
|
+
return 0
|
382
|
+
end
|
383
|
+
|
384
|
+
if(@options[:verbose])
|
385
|
+
puts "Running '#{crawler}'"
|
386
|
+
end
|
387
|
+
|
388
|
+
res = p.new.etl(args) { | docs |
|
389
|
+
if(docs.nil?)
|
390
|
+
next
|
391
|
+
end
|
392
|
+
|
393
|
+
if(docs.kind_of?(Array) == false)
|
394
|
+
docs = [docs]
|
395
|
+
end
|
396
|
+
|
397
|
+
docs.each do |doc|
|
398
|
+
puts @formatter.format(doc)
|
399
|
+
end
|
400
|
+
}
|
401
|
+
end
|
402
|
+
|
403
|
+
return 0
|
404
|
+
end
|
405
|
+
end
|
406
|
+
end
|