goog 0.0.2 → 0.0.3

data/bin/goog

@@ -6,6 +6,7 @@ require 'yaml'
 require 'uri'
 
 if ARGV[0] =~ /^-h|--help/
+  require 'goog'
   puts <<END
 Usage: goog [-n pages] [query]
 
@@ -13,7 +14,7 @@ Synopsis:
     -n [pages]                      Return n pages of results
     -h, --help                      Show this message
 
-goog 0.0.2 
+goog #{Goog::VERSION}
 http://github.com/danchoi/goog
 Author: Daniel Choi <dhchoi@gmail.com>
 END
@@ -26,20 +27,21 @@ query = ARGV.join(' ')
 unless query
   abort "Please provide a search query"
 end
-CACHE = "#{ENV['HOME']}/.goog.cache.yml"
+
 unless `which tidy` =~ /tidy/
   abort "No tidy found. Please install tidy."
 end
 if RUBY_VERSION !~ /^1.9/
   abort "Requires Ruby 1.9"
 end
-query = CGI.escape query
-
+query = "/search?q=#{CGI.escape query}"
 (1..pages).each do |page| 
-  q = page == 1 ? query : "#{query}&start=#{page * 10}"
-  curl = "curl -s -A Mozilla http://www.google.com/search?q=#{q} | tidy --wrap 0 -indent -ashtml --merge-divs yes 2>/dev/null"
+  if query.nil?
+    exit
+  end
+  curl = "curl -s -A Mozilla 'http://www.google.com#{query}' | tidy --wrap 0 -indent -ashtml --merge-divs yes 2>/dev/null"
   resp = %x{#{curl}}
-  doc = Nokogiri::HTML resp
+  doc = Nokogiri::HTML resp, nil, 'iso-8859-1'
   doc.search('ol li.g').each_with_index {|li, index|
     next unless li.at('h3 a')
     link = li.at('h3 a')['href'].sub(/^\/url\?q=/, '')
@@ -52,8 +54,6 @@ query = CGI.escape query
                 description.search('span').remove 
                 excerpt = begin 
                   s = description.inner_text.strip
-                  s.force_encoding('iso-8859-1')
-                  s.encode!('utf-8', undef: :replace, invalid: :replace)
                   s.gsub(/\s{2,}/, ' ')
                 rescue
                   puts "ERROR"
@@ -70,6 +70,10 @@ query = CGI.escape query
     puts res.compact
     puts
   }
+  # find next page link
+  # <a href="/search?q=why+the+lucky+stiff&amp;hl=en&amp;ie=UTF-8&amp;prmd=ivns&amp;ei=K6akT9bRBeaM6QHjifmwBA&amp;start=10&amp;sa=N" style="text-align:left"><span style="display:block;margin-left:53px">Next</span></a>
+  next_page_href = (nextspan = doc.at("//span[contains(child::text(),'Next')]")) && nextspan.parent[:href]
+  query = next_page_href
 end
 
 

data/lib/goog.rb

@@ -0,0 +1,3 @@
+module Goog
+  VERSION = '0.0.3'
+end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: goog
 version: !ruby/object:Gem::Version
-  version: 0.0.2
+  version: 0.0.3
   prerelease: 
 platform: ruby
 authors:
@@ -36,6 +36,7 @@ extensions: []
 extra_rdoc_files: []
 files:
 - bin/goog
+- lib/goog.rb
 homepage: https://github.com/danchoi/goog
 licenses: []
 post_install_message: