Initial work

This commit is contained in:
erwanlr
2013-02-20 17:34:17 +01:00
parent a01e1ab29f
commit 8b9aec468a
11 changed files with 26 additions and 78 deletions

View File

@@ -48,7 +48,7 @@ class Browser
@hydra = Typhoeus::Hydra.new(
max_concurrency: @max_threads,
timeout: @request_timeout
#connecttimeout: @request_timeout
)
# TODO : add an option for the cache dir instead of using a constant
@@ -56,8 +56,7 @@ class Browser
@cache.clean
# might be in CacheFileStore
setup_cache_handlers
#Typhoeus::Config.cache = @cache
end
private_class_method :new
@@ -146,24 +145,6 @@ class Browser
end
end
def setup_cache_handlers
@hydra.cache_setter do |request|
@cache.write_entry(
Browser.generate_cache_key_from_request(request),
request.response,
request.cache_timeout
)
end
@hydra.cache_getter do |request|
@cache.read_entry(
Browser.generate_cache_key_from_request(request)
) rescue nil
end
end
private :setup_cache_handlers
def get(url, params = {})
run_request(
forge_request(url, params.merge(method: :get))
@@ -177,10 +158,10 @@ class Browser
end
def get_and_follow_location(url, params = {})
params[:max_redirects] ||= 2
params[:maxredirs] ||= 2
run_request(
forge_request(url, params.merge(method: :get, follow_location: true))
forge_request(url, params.merge(method: :get, followlocation: true))
)
end
@@ -208,12 +189,13 @@ class Browser
end
end
unless params.has_key?(:disable_ssl_host_verification)
params = params.merge(:disable_ssl_host_verification => true)
# TODO : check if it's the default value into ethon. If so, removed the lines from here
unless params.has_key?(:ssl_verifyhost)
params = params.merge(ssl_verifyhost: 0)
end
unless params.has_key?(:disable_ssl_peer_verification)
params = params.merge(:disable_ssl_peer_verification => true)
unless params.has_key?(:ssl_verifypeer)
params = params.merge(ssl_verifypeer: false)
end
if !params.has_key?(:headers)
@@ -223,9 +205,9 @@ class Browser
end
# Used to enable the cache system if :cache_timeout > 0
unless params.has_key?(:cache_timeout)
params = params.merge(:cache_timeout => @cache_timeout)
end
#unless params.has_key?(:cache_ttl)
# params = params.merge(cache_ttl: @cache_timeout)
#end
params
end
@@ -247,17 +229,4 @@ class Browser
end
end
end
# The Typhoeus::Request.cache_key only hash the url :/
# this one will include the params
# TODO : include also the method (:get, :post, :any)
def self.generate_cache_key_from_request(request)
cache_key = request.cache_key
if request.params
cache_key = Digest::SHA1.hexdigest("#{cache_key}-#{request.params.hash}")
end
cache_key
end
end

View File

@@ -51,7 +51,7 @@ class CacheFileStore
end
end
def read_entry(key)
def get(key)
entry_file_path = get_entry_file_path(key)
if File.exists?(entry_file_path)
@@ -59,7 +59,7 @@ class CacheFileStore
end
end
def write_entry(key, data_to_store, cache_timeout)
def set(key, data_to_store, cache_timeout)
if cache_timeout > 0
File.open(get_entry_file_path(key), 'w') do |f|
f.write(@serializer.dump(data_to_store))

View File

@@ -33,7 +33,6 @@ begin
require 'rbconfig'
require 'pp'
# Third party libs
gem 'typhoeus', '=0.4.2'
require 'typhoeus'
require 'json'
require 'nokogiri'

View File

@@ -60,7 +60,7 @@ module WpUsernames
end
def get_nickname_from_url(url)
resp = Browser.instance.get(url, { follow_location: true, max_redirects: 2 })
resp = Browser.instance.get_and_follow_location(url)
nickname = nil
if resp.code == 200
nickname = extract_nickname_from_body(resp.body)

View File

@@ -55,7 +55,7 @@ class WpEnumerator
targets.each do |target|
url = target.get_full_url
request = enum_browser.forge_request(url, { cache_timeout: 0, follow_location: true })
request = enum_browser.forge_request(url, { cache_ttl: 0, followlocation: true })
request_count += 1
request.on_complete do |response|

View File

@@ -74,7 +74,7 @@ class WpTheme < WpItem
# Discover the wordpress theme name by parsing the css link rel
def self.find_from_css_link(target_uri)
response = Browser.instance.get(target_uri.to_s, { follow_location: true, max_redirects: 2 })
response = Browser.instance.get_and_follow_location(target_uri.to_s)
# https + domain is optional because of relative links
matches = %r{(?:https?://[^"']+)?/([^/]+)/themes/([^"']+)/style.css}i.match(response.body)

View File

@@ -63,7 +63,7 @@ class CheckerPlugin < Plugin
number_of_urls = urls.size
urls.each do |url|
request = browser.forge_request(url, { cache_timeout: 0, follow_location: true })
request = browser.forge_request(url, { cache_ttl: 0, followlocation: true })
request_count += 1
request.on_complete do |response|