Class: HTMLProofer::Cache
- Inherits:
-
Object
- Object
- HTMLProofer::Cache
- Includes:
- Utils
- Defined in:
- lib/html-proofer/cache.rb
Constant Summary collapse
- DEFAULT_STORAGE_DIR =
File.join('tmp', '.htmlproofer')
- DEFAULT_CACHE_FILE_NAME =
'cache.log'
- URI_REGEXP =
URI::DEFAULT_PARSER.make_regexp
Instance Attribute Summary collapse
-
#cache_file ⇒ Object
readonly
Returns the value of attribute cache_file.
-
#cache_log ⇒ Object
readonly
Returns the value of attribute cache_log.
-
#exists ⇒ Object
readonly
Returns the value of attribute exists.
-
#storage_dir ⇒ Object
readonly
Returns the value of attribute storage_dir.
Instance Method Summary collapse
- #add(url, filenames, status, msg = '') ⇒ Object
- #clean_url(url) ⇒ Object
- #detect_url_changes(found, type) ⇒ Object
-
#initialize(logger, options) ⇒ Cache
constructor
A new instance of Cache.
- #load? ⇒ Boolean
- #parsed_timeframe(timeframe) ⇒ Object
- #retrieve_urls(urls, type) ⇒ Object
- #setup_cache!(options) ⇒ Object
- #size ⇒ Object
-
#unescape_url(url) ⇒ Object
FIXME: it seems that Typhoeus actually acts on escaped URLs, but there’s no way to get at that information, and the cache stores unescaped URLs.
- #urls ⇒ Object
- #within_timeframe?(time) ⇒ Boolean
-
#write ⇒ Object
TODO: Garbage performance–both the external and internal caches need access to this file.
Methods included from Utils
#create_nokogiri, #pluralize, #swap
Constructor Details
#initialize(logger, options) ⇒ Cache
Returns a new instance of Cache.
19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 |
# File 'lib/html-proofer/cache.rb', line 19 def initialize(logger, ) @logger = logger @cache_log = {} @cache_datetime = DateTime.now @cache_time = @cache_datetime.to_time if .nil? || .empty? define_singleton_method('use_cache?') { false } else define_singleton_method('use_cache?') { true } setup_cache!() @parsed_timeframe = parsed_timeframe([:timeframe]) end end |
Instance Attribute Details
#cache_file ⇒ Object (readonly)
Returns the value of attribute cache_file.
17 18 19 |
# File 'lib/html-proofer/cache.rb', line 17 def cache_file @cache_file end |
#cache_log ⇒ Object (readonly)
Returns the value of attribute cache_log.
17 18 19 |
# File 'lib/html-proofer/cache.rb', line 17 def cache_log @cache_log end |
#exists ⇒ Object (readonly)
Returns the value of attribute exists.
17 18 19 |
# File 'lib/html-proofer/cache.rb', line 17 def exists @exists end |
#storage_dir ⇒ Object (readonly)
Returns the value of attribute storage_dir.
17 18 19 |
# File 'lib/html-proofer/cache.rb', line 17 def storage_dir @storage_dir end |
Instance Method Details
#add(url, filenames, status, msg = '') ⇒ Object
66 67 68 69 70 71 72 73 74 75 76 77 |
# File 'lib/html-proofer/cache.rb', line 66 def add(url, filenames, status, msg = '') return unless use_cache? data = { time: @cache_time, filenames: filenames, status: status, message: msg } @cache_log[clean_url(url)] = data end |
#clean_url(url) ⇒ Object
155 156 157 |
# File 'lib/html-proofer/cache.rb', line 155 def clean_url(url) unescape_url(url) end |
#detect_url_changes(found, type) ⇒ Object
79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 |
# File 'lib/html-proofer/cache.rb', line 79 def detect_url_changes(found, type) found_urls = found.keys.map { |url| clean_url(url) } # if there were no urls, bail return {} if found_urls.empty? existing_urls = @cache_log.keys.map { |url| clean_url(url) } # prepare to add new URLs detected additions = found.reject do |url, _| url = clean_url(url) if existing_urls.include?(url) true else @logger.log :debug, "Adding #{url} to cache check" false end end new_link_count = additions.length new_link_text = pluralize(new_link_count, 'link', 'links') @logger.log :info, "Adding #{new_link_text} to the cache..." # remove from cache URLs that no longer exist deletions = 0 @cache_log.delete_if do |url, _| url = clean_url(url) if found_urls.include?(url) false elsif url_matches_type?(url, type) @logger.log :debug, "Removing #{url} from cache check" deletions += 1 true end end del_link_text = pluralize(deletions, 'link', 'links') @logger.log :info, "Removing #{del_link_text} from the cache..." additions end |
#load? ⇒ Boolean
129 130 131 |
# File 'lib/html-proofer/cache.rb', line 129 def load? @load.nil? end |
#parsed_timeframe(timeframe) ⇒ Object
49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 |
# File 'lib/html-proofer/cache.rb', line 49 def parsed_timeframe(timeframe) time, date = timeframe.match(/(\d+)(\D)/).captures time = time.to_i case date when 'M' time_ago(time, :months) when 'w' time_ago(time, :weeks) when 'd' time_ago(time, :days) when 'h' time_ago(time, :hours) else raise ArgumentError, "#{date} is not a valid timeframe!" end end |
#retrieve_urls(urls, type) ⇒ Object
133 134 135 136 137 138 139 140 141 142 143 144 |
# File 'lib/html-proofer/cache.rb', line 133 def retrieve_urls(urls, type) urls_to_check = detect_url_changes(urls, type) @cache_log.each_pair do |url, cache| next if within_timeframe?(cache['time']) && cache['message'].empty? # these were successes to skip if url_matches_type?(url, type) urls_to_check[url] = cache['filenames'] # recheck expired links end end urls_to_check end |
#setup_cache!(options) ⇒ Object
159 160 161 162 163 164 165 166 167 168 169 170 171 172 |
# File 'lib/html-proofer/cache.rb', line 159 def setup_cache!() @storage_dir = [:storage_dir] || DEFAULT_STORAGE_DIR FileUtils.mkdir_p(storage_dir) unless Dir.exist?(storage_dir) cache_file_name = [:cache_file] || DEFAULT_CACHE_FILE_NAME @cache_file = File.join(storage_dir, cache_file_name) return unless File.exist?(@cache_file) contents = File.read(@cache_file) @cache_log = contents.empty? ? {} : JSON.parse(contents) end |
#size ⇒ Object
45 46 47 |
# File 'lib/html-proofer/cache.rb', line 45 def size @cache_log.length end |
#unescape_url(url) ⇒ Object
FIXME: it seems that Typhoeus actually acts on escaped URLs, but there’s no way to get at that information, and the cache stores unescaped URLs. Because of this, some links, such as github.com/search/issues?q=is:open+is:issue+fig are not matched as github.com/search/issues?q=is%3Aopen+is%3Aissue+fig
151 152 153 |
# File 'lib/html-proofer/cache.rb', line 151 def unescape_url(url) Addressable::URI.unescape(url) end |
#urls ⇒ Object
41 42 43 |
# File 'lib/html-proofer/cache.rb', line 41 def urls @cache_log['urls'] || [] end |
#within_timeframe?(time) ⇒ Boolean
35 36 37 38 39 |
# File 'lib/html-proofer/cache.rb', line 35 def within_timeframe?(time) return false if time.nil? (@parsed_timeframe..@cache_time).cover?(Time.parse(time)) end |
#write ⇒ Object
TODO: Garbage performance–both the external and internal caches need access to this file. Write a proper versioned schema in the future
125 126 127 |
# File 'lib/html-proofer/cache.rb', line 125 def write File.write(cache_file, @cache_log.to_json) end |