spn2 0.1.2 → 0.2.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: e968c75da93882e48ac210e17bb599497971ee1f065035758279529b23c53f1b
-  data.tar.gz: 9fa1b6f6125d9347d2418254b8e8838572fd8a4efcf0c6bbabc258f3784c097b
+  metadata.gz: d73702950316690d62ce910059788093af0ba31b094c1208eb7c7ef72cbaa06e
+  data.tar.gz: 87ba2a61a9f76d86d46043da74418b2a67f6ac584dac2c3b50b2aa7baa8c7aad
 SHA512:
-  metadata.gz: bacfeda95f8a40e132496cb69078e8229767de20c99615db7da4bb1d5f35a9a4c7e3e688a1fe3d31d6dd9e5f026d0d6fc8e328aae90915de55e6810db675732f
-  data.tar.gz: e62fe104f074cb9ab5e4a85b89050e8dc0f198613dd9ff62fd56d3c645428e080699e14d57b8b6a655b68ed3be00704ff67ba5e9a58508083fe0f961cdc1ac55
+  metadata.gz: 0ad8a32dc48bd5dbaf5b24428f6a3522c4af9ce3a5a2297df39fb4dd9b36a627565a844cf841ca50e3c3e0200b569c88d73bd27717468ed8b8e2f5444114d316
+  data.tar.gz: b08014dddc62d499a30e5bc8ba1ce20c96a5f01ad8d0d465759897904f297339a9b79ecff1fdd56561c194072e24ba83542b5d0e39f6d6ab8e57a3b55bae8f6d

data/CHANGELOG.md

@@ -11,3 +11,7 @@
 
 - Add user_status
 - Add status calls for multiple job_ids and outlinks
+
+## [0.2.0] - 2022-07-03
+
+- Breaking change: Single method 'status' with kwarg :job_ids now used for status of job(s)

data/README.md

@@ -59,7 +59,7 @@ The key "status_ext" contains an explanatory message - see the API [specificatio
 
 Use the job_id.
 ```rb
-> Spn2.status_job_id(job_id: 'spn2-9c17e047f58f9220a7008d4f18152fee4d111d14')
+> Spn2.status(job_ids: 'spn2-9c17e047f58f9220a7008d4f18152fee4d111d14')
 
 => {"counters"=>{"outlinks"=>1, "embeds"=>2}, "job_id"=>"spn2-9c17e047f58f9220a7008d4f18152fee4d111d14",
     "original_url"=>"http://example.com/", "resources"=>["http://example.com/", "http://example.com/favicon.ico"],
@@ -76,16 +76,16 @@ $ echo "http://example.com/"|tr -d "\n"|shasum
 9c17e047f58f9220a7008d4f18152fee4d111d14  -
 ```
 
-The status of a comma-separated list of job_id's can be obtained with:
+The status of an array of job_id's can be obtained with:
 ```rb
-> Spn2.status_job_ids(job_ids: 'spn2-9c17e047f58f9220a7008d4f18152fee4d111d14,spn2-...')
+> Spn2.status(job_ids: ['spn2-9c17e047f58f9220a7008d4f18152fee4d111d14', 'spn2-...'])
 
 => [.. # an array of status hashes
 ```
 
 Finally, the status of any outlinks captured by using the save option `capture_outlinks: 1` is available by supplying the parent job_id to:
 ```rb
-> Spn2.status_job_id_outlinks(job_id: 'spn2-cce034d987e1d72d8cbf1770bcf99024fe20dddf')
+> Spn2.status(job_ids: 'spn2-cce034d987e1d72d8cbf1770bcf99024fe20dddf', outlinks: true)
 
 => [.. # an array of outlink job status hashes
 ```

data/lib/spn2/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module Spn2
-  VERSION = '0.1.2'
+  VERSION = '0.2.0'
 end

data/lib/spn2.rb

@@ -5,25 +5,12 @@ require 'json'
 require 'nokogiri'
 
 require_relative 'curlable'
+require_relative 'spn2_errors'
 
 # Design decison to not use a class as only 'state' is in 2 env vars
 module Spn2
   extend Curlable
 
-  BAD_AUTH_MSG = 'You need to be logged in to use Save Page Now.'
-  ERROR_CODES = [502].freeze
-
-  class Spn2Error < StandardError; end
-  class Spn2ErrorBadAuth < Spn2Error; end
-  class Spn2ErrorFailedCapture < Spn2Error; end
-  class Spn2ErrorInvalidOption < Spn2Error; end
-  class Spn2ErrorMissingKeys < Spn2Error; end
-  class Spn2ErrorNoOutlinks < Spn2Error; end
-  class Spn2ErrorTooManyRequests < Spn2Error; end
-  class Spn2ErrorUnknownResponse < Spn2Error; end
-  class Spn2ErrorUnknownResponseCode < Spn2Error; end
-  ERROR_CODES.each { |i| Spn2.const_set("Spn2Error#{i}", Class.new(Spn2Error)) }
-
   ESSENTIAL_STATUS_KEYS = %w[job_id resources status].freeze
   JOB_ID_REGEXP = /^(spn2-([a-f]|\d){40})$/
   WEB_ARCHIVE = 'https://web.archive.org'
@@ -65,33 +52,30 @@ module Spn2
     end
     alias capture save
 
-    def status_job_id(job_id:)
-      json = json(auth_post(url: "#{WEB_ARCHIVE}/save/status", params: { job_id: job_id }))
-      raise Spn2ErrorBadAuth, json.inspect if json['message']&.== BAD_AUTH_MSG
-
-      raise Spn2ErrorMissingKeys, json.inspect unless (ESSENTIAL_STATUS_KEYS - json.keys).empty?
+    def status(job_ids:, outlinks: false)
+      params = status_params(job_ids: job_ids, outlinks: outlinks)
+      json = json(auth_post(url: "#{WEB_ARCHIVE}/save/status", params: params))
+      return json if json.is_a? Array # must be valid response
 
+      handle_status_errors(job_ids: job_ids, json: json, outlinks: outlinks)
       json
     end
-    alias status status_job_id
 
-    def status_job_ids(job_ids:)
-      json = json(auth_post(url: "#{WEB_ARCHIVE}/save/status", params: { job_ids: job_ids }))
-      raise Spn2Error, json.inspect unless json.is_a? Array
-
-      json
-    end
-    alias statuses status_job_ids
+    private
 
-    def status_job_id_outlinks(job_id:)
-      json = json(auth_post(url: "#{WEB_ARCHIVE}/save/status", params: { job_id_outlinks: job_id }))
-      raise Spn2ErrorNoOutlinks, json.inspect unless json.is_a? Array
+    def status_params(job_ids:, outlinks:)
+      return { job_ids: job_ids.join(',') } if job_ids.is_a?(Array)
+      return { job_id_outlinks: job_ids } if outlinks
 
-      json
+      { job_id: job_ids } # single job_id
     end
-    alias status_outlinks status_job_id_outlinks
 
-    private
+    def handle_status_errors(job_ids:, json:, outlinks:)
+      raise Spn2ErrorBadAuth, json.inspect if json['message']&.== BAD_AUTH_MSG
+      raise Spn2ErrorNoOutlinks, json.inspect if outlinks
+      raise Spn2ErrorMissingKeys, json.inspect unless (ESSENTIAL_STATUS_KEYS - json.keys).empty?
+      raise Spn2Error, json.inspect if job_ids.is_a?(Array)
+    end
 
     def auth_get(url:)
       get(url: url, headers: accept_header.merge(auth_header))
@@ -115,25 +99,34 @@ module Spn2
 
     def json(html_string)
       JSON.parse(doc = doc(html_string))
-    rescue JSON::ParserError # an html response & therefore an error
+    rescue JSON::ParserError # an html response
       parse_error_code_from_page_title(doc.title) if doc.title
-      parse_error_from_page_body(html_string) # if no title parse body
+      parse_error_from_page_body(html_string)
     end
 
     def parse_error_code_from_page_title(title_string)
-      code = title_string.to_i
-      raise Spn2.const_get("Spn2Error#{code}") if ERROR_CODES.include? code
-
-      raise Spn2ErrorUnknownResponseCode
+      raise_code_response_error_if_code_in_string(title_string)
+      raise Spn2ErrorUnknownResponseCode, title_string # code found but doesn't match any known error classes
     end
 
     def parse_error_from_page_body(html_string)
-      h1 = doc(html_string).xpath('//h1')
-      raise Spn2ErrorTooManyRequests if !h1.empty? && h1.text == 'Too Many Requests'
+      h1_tag_text = h1_tag_text(html_string)
+      raise_code_response_error_if_code_in_string h1_tag_text
+      raise Spn2ErrorTooManyRequests if h1_tag_text == TOO_MANY_REQUESTS
 
       raise Spn2ErrorUnknownResponse, html_string # fall through
     end
 
+    def h1_tag_text(html_string)
+      doc(html_string).xpath('//h1')&.text || ''
+    end
+
+    def raise_code_response_error_if_code_in_string(string)
+      return unless ERROR_CODES.include? code = string.to_i
+
+      raise Spn2.const_get("Spn2Error#{code}")
+    end
+
     def options_valid?(opts)
       opts.keys.all? { |k| (BINARY_OPTS + OTHER_OPTS).include? k.to_s }
     end

data/lib/spn2_errors.rb

@@ -0,0 +1,20 @@
+# frozen_string_literal: true
+
+# namespace
+module Spn2
+  BAD_AUTH_MSG = 'You need to be logged in to use Save Page Now.'
+  ERROR_CODES = [400, 502].freeze
+  TOO_MANY_REQUESTS = 'Too Many Requests'
+
+  class Spn2Error < StandardError; end
+  class Spn2ErrorBadAuth < Spn2Error; end
+  class Spn2ErrorBadParams < Spn2Error; end
+  class Spn2ErrorFailedCapture < Spn2Error; end
+  class Spn2ErrorInvalidOption < Spn2Error; end
+  class Spn2ErrorMissingKeys < Spn2Error; end
+  class Spn2ErrorNoOutlinks < Spn2Error; end
+  class Spn2ErrorTooManyRequests < Spn2Error; end
+  class Spn2ErrorUnknownResponse < Spn2Error; end
+  class Spn2ErrorUnknownResponseCode < Spn2Error; end
+  ERROR_CODES.each { |i| const_set("Spn2Error#{i}", Class.new(Spn2Error)) }
+end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: spn2
 version: !ruby/object:Gem::Version
-  version: 0.1.2
+  version: 0.2.0
 platform: ruby
 authors:
 - MatzFan
 autorequire: 
 bindir: exe
 cert_chain: []
-date: 2022-07-02 00:00:00.000000000 Z
+date: 2022-07-03 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: curb
@@ -166,6 +166,7 @@ files:
 - lib/curlable.rb
 - lib/spn2.rb
 - lib/spn2/version.rb
+- lib/spn2_errors.rb
 - sig/spn2.rbs
 homepage: https://gitlab.com/matzfan/spn2
 licenses: