Sha256: 053bfcd77a5e52f1362c0beb2ab51990c44590d042df30bc81d2ef22f5c9de85
Contents?: true
Size: 754 Bytes
Versions: 26
Compression:
Stored size: 754 Bytes
Contents
#!/usr/bin/env ruby
scrappers = []
results = []
class QuotedUrlScrapper
REGEX = /"\w+:\/\/[^"]+"/
def initialize(text)
@text = text
end
def get_urls
@text.scan(REGEX).map do |url|
url.chop!
url.reverse.chop.reverse
end
end
end
class TextUrlScrapper
REGEX = /\w+:\/\/[^\s]+/
def initialize(text)
@text = text
end
def get_urls
@text.scan(REGEX).map do |url|
url.gsub!(/".*/, '')
url.gsub!(/'.*/, '')
end
end
end
scrappers.push(QuotedUrlScrapper)
scrappers.push(TextUrlScrapper)
STDIN.each_line do |line|
scrappers.each do |scrapper|
parser = scrapper.new(line)
parser.get_urls.each do |url|
results << url
end
end
end
results.uniq.each do |url|
puts url
end
Version data entries
26 entries across 26 versions & 1 rubygems