ul-wukong 4.1.0

Sign up to get free protection for your applications and to get access to all the features.
Files changed (261) hide show
  1. checksums.yaml +15 -0
  2. data/.gitignore +60 -0
  3. data/.gitmodules +6 -0
  4. data/.rspec +2 -0
  5. data/.travis.yml +19 -0
  6. data/.yardopts +6 -0
  7. data/CHANGELOG.md +7 -0
  8. data/Gemfile +17 -0
  9. data/Guardfile +12 -0
  10. data/LICENSE.md +95 -0
  11. data/NOTES-travis.md +31 -0
  12. data/README-old.md +422 -0
  13. data/README.md +1308 -0
  14. data/Rakefile +28 -0
  15. data/TODO.md +99 -0
  16. data/bin/cutc +30 -0
  17. data/bin/cuttab +5 -0
  18. data/bin/greptrue +6 -0
  19. data/bin/md5sort +20 -0
  20. data/bin/setcat +11 -0
  21. data/bin/tabchar +5 -0
  22. data/bin/uniq-ord +59 -0
  23. data/bin/uniqc +3 -0
  24. data/bin/wu +34 -0
  25. data/bin/wu-clean-encoding +31 -0
  26. data/bin/wu-date +13 -0
  27. data/bin/wu-datetime +13 -0
  28. data/bin/wu-hist +3 -0
  29. data/bin/wu-lign +186 -0
  30. data/bin/wu-local +4 -0
  31. data/bin/wu-plus +9 -0
  32. data/bin/wu-source +5 -0
  33. data/bin/wu-sum +31 -0
  34. data/diagrams/wu_local.dot +39 -0
  35. data/diagrams/wu_local.dot.png +0 -0
  36. data/examples/Gemfile +38 -0
  37. data/examples/README.md +9 -0
  38. data/examples/basic/string_reverser.rb +23 -0
  39. data/examples/basic/tiny_count.rb +8 -0
  40. data/examples/basic/word_count/accumulator.rb +26 -0
  41. data/examples/basic/word_count/tokenizer.rb +13 -0
  42. data/examples/basic/word_count/word_count.rb +6 -0
  43. data/examples/dataflow/scraper_macro_flow.rb +28 -0
  44. data/examples/deploy_pack/Gemfile +6 -0
  45. data/examples/deploy_pack/README.md +6 -0
  46. data/examples/deploy_pack/a/b/c/.gitkeep +0 -0
  47. data/examples/deploy_pack/app/processors/string_reverser.rb +5 -0
  48. data/examples/deploy_pack/config/environment.rb +1 -0
  49. data/examples/dsl/dataflow/fibonacci_series.rb +101 -0
  50. data/examples/dsl/dataflow/scraper_macro_flow.rb +28 -0
  51. data/examples/dsl/dataflow/simple.rb +12 -0
  52. data/examples/dsl/dataflow/telegram.rb +45 -0
  53. data/examples/dsl/workflow/cherry_pie.dot +97 -0
  54. data/examples/dsl/workflow/cherry_pie.md +104 -0
  55. data/examples/dsl/workflow/cherry_pie.png +0 -0
  56. data/examples/dsl/workflow/cherry_pie.rb +101 -0
  57. data/examples/empty/.gitkeep +0 -0
  58. data/examples/examples_helper.rb +9 -0
  59. data/examples/geo.rb +4 -0
  60. data/examples/geo/geo_grids.numbers +0 -0
  61. data/examples/geo/geolocated.rb +331 -0
  62. data/examples/geo/quadtile.rb +69 -0
  63. data/examples/geo/spec/geolocated_spec.rb +247 -0
  64. data/examples/geo/tile_fetcher.rb +77 -0
  65. data/examples/graph/implied_geolocation/README.md +63 -0
  66. data/examples/graph/minimum_spanning_tree/airfares_graphviz.rb +73 -0
  67. data/examples/improver/tweet_summary.rb +73 -0
  68. data/examples/loadable.rb +2 -0
  69. data/examples/munging/airline_flights/airline_flights.rake +83 -0
  70. data/examples/munging/airline_flights/airplane.rb +0 -0
  71. data/examples/munging/airline_flights/airport_id_unification.rb +129 -0
  72. data/examples/munging/airline_flights/airport_ok_chars.rb +4 -0
  73. data/examples/munging/airline_flights/indexable.rb +75 -0
  74. data/examples/munging/airline_flights/indexable_spec.rb +90 -0
  75. data/examples/munging/airline_flights/reconcile_airports.rb +142 -0
  76. data/examples/munging/airline_flights/tasks.rake +83 -0
  77. data/examples/munging/airline_flights/topcities.rb +167 -0
  78. data/examples/munging/geo/geo_json.rb +54 -0
  79. data/examples/munging/geo/geo_models.rb +69 -0
  80. data/examples/munging/geo/geonames_models.rb +107 -0
  81. data/examples/munging/geo/iso_codes.rb +172 -0
  82. data/examples/munging/geo/reconcile_countries.rb +124 -0
  83. data/examples/munging/geo/tasks.rake +71 -0
  84. data/examples/munging/wikipedia/articles/extract_articles-parsed.rb +79 -0
  85. data/examples/munging/wikipedia/articles/extract_articles-templated.rb +136 -0
  86. data/examples/munging/wikipedia/articles/textualize_articles.rb +54 -0
  87. data/examples/munging/wikipedia/articles/verify_structure.rb +43 -0
  88. data/examples/munging/wikipedia/articles/wp2txt-LICENSE.txt +22 -0
  89. data/examples/munging/wikipedia/articles/wp2txt_article.rb +259 -0
  90. data/examples/munging/wikipedia/articles/wp2txt_utils.rb +452 -0
  91. data/examples/munging/wikipedia/dbpedia/dbpedia_common.rb +5 -0
  92. data/examples/munging/wikipedia/dbpedia/dbpedia_extract_geocoordinates.rb +78 -0
  93. data/examples/munging/wikipedia/dbpedia/extract_links-cruft.rb +66 -0
  94. data/examples/munging/wikipedia/dbpedia/extract_links.rb +260 -0
  95. data/examples/munging/wikipedia/dbpedia/sameas_extractor.rb +20 -0
  96. data/examples/rake_helper.rb +97 -0
  97. data/examples/ruby_project/Gemfile +6 -0
  98. data/examples/ruby_project/README.md +6 -0
  99. data/examples/ruby_project/a/b/c/.gitkeep +0 -0
  100. data/examples/server_logs/geo_ip_mapping/munge_geolite.rb +82 -0
  101. data/examples/server_logs/logline.rb +95 -0
  102. data/examples/server_logs/models.rb +66 -0
  103. data/examples/server_logs/page_counts.pig +48 -0
  104. data/examples/server_logs/server_logs-01-parse-script.rb +13 -0
  105. data/examples/server_logs/server_logs-02-histograms-full.rb +33 -0
  106. data/examples/server_logs/server_logs-02-histograms-mapper.rb +14 -0
  107. data/examples/server_logs/server_logs-03-breadcrumbs-full.rb +71 -0
  108. data/examples/server_logs/server_logs-04-page_page_edges-full.rb +40 -0
  109. data/examples/serverlogs/geo_ip_mapping/munge_geolite.rb +82 -0
  110. data/examples/serverlogs/models/logline.rb +102 -0
  111. data/examples/serverlogs/parser/apache_parser_widget.rb +46 -0
  112. data/examples/serverlogs/visit_paths/common.rb +4 -0
  113. data/examples/serverlogs/visit_paths/page_counts.pig +48 -0
  114. data/examples/serverlogs/visit_paths/serverlogs-01-parse-script.rb +11 -0
  115. data/examples/serverlogs/visit_paths/serverlogs-02-histograms-full.rb +31 -0
  116. data/examples/serverlogs/visit_paths/serverlogs-02-histograms-mapper.rb +12 -0
  117. data/examples/serverlogs/visit_paths/serverlogs-03-breadcrumbs-full.rb +67 -0
  118. data/examples/serverlogs/visit_paths/serverlogs-04-page_page_edges-full.rb +38 -0
  119. data/examples/splitter.rb +94 -0
  120. data/examples/string_reverser.rb +7 -0
  121. data/examples/text/pig_latin/pig_latinizer.rb +35 -0
  122. data/examples/text/pig_latin/pig_latinizer_widget.rb +16 -0
  123. data/examples/text/regional_flavor/README.md +14 -0
  124. data/examples/text/regional_flavor/article_wordbags.pig +39 -0
  125. data/examples/text/regional_flavor/j01-article_wordbags.rb +4 -0
  126. data/examples/text/regional_flavor/simple_pig_script.pig +27 -0
  127. data/examples/twitter.rb +5 -0
  128. data/lib/hanuman.rb +36 -0
  129. data/lib/hanuman/graph.rb +97 -0
  130. data/lib/hanuman/graphvizzer.rb +206 -0
  131. data/lib/hanuman/graphvizzer/gv_models.rb +161 -0
  132. data/lib/hanuman/graphvizzer/gv_presenter.rb +97 -0
  133. data/lib/hanuman/link.rb +35 -0
  134. data/lib/hanuman/registry.rb +46 -0
  135. data/lib/hanuman/stage.rb +128 -0
  136. data/lib/hanuman/tree.rb +67 -0
  137. data/lib/wu/geo.rb +4 -0
  138. data/lib/wu/geo/geo_grids.numbers +0 -0
  139. data/lib/wu/geo/geolocated.rb +331 -0
  140. data/lib/wu/geo/quadtile.rb +69 -0
  141. data/lib/wu/graph/union_find.rb +62 -0
  142. data/lib/wu/model/reconcilable.rb +63 -0
  143. data/lib/wu/munging.rb +71 -0
  144. data/lib/wu/social/models/twitter.rb +31 -0
  145. data/lib/wu/wikipedia/models.rb +20 -0
  146. data/lib/wukong.rb +54 -0
  147. data/lib/wukong/dataflow.rb +43 -0
  148. data/lib/wukong/doc_helpers.rb +14 -0
  149. data/lib/wukong/doc_helpers/dataflow_handler.rb +29 -0
  150. data/lib/wukong/doc_helpers/field_handler.rb +91 -0
  151. data/lib/wukong/doc_helpers/processor_handler.rb +29 -0
  152. data/lib/wukong/driver.rb +214 -0
  153. data/lib/wukong/driver/event_machine_driver.rb +15 -0
  154. data/lib/wukong/driver/wiring.rb +68 -0
  155. data/lib/wukong/local.rb +42 -0
  156. data/lib/wukong/local/runner.rb +96 -0
  157. data/lib/wukong/local/stdio_driver.rb +104 -0
  158. data/lib/wukong/logger.rb +102 -0
  159. data/lib/wukong/model/faker.rb +136 -0
  160. data/lib/wukong/model/flatpack_parser/flat.rb +60 -0
  161. data/lib/wukong/model/flatpack_parser/flatpack.rb +4 -0
  162. data/lib/wukong/model/flatpack_parser/lang.rb +46 -0
  163. data/lib/wukong/model/flatpack_parser/parser.rb +55 -0
  164. data/lib/wukong/model/flatpack_parser/tokens.rb +130 -0
  165. data/lib/wukong/plugin.rb +48 -0
  166. data/lib/wukong/processor.rb +110 -0
  167. data/lib/wukong/rake_helper.rb +6 -0
  168. data/lib/wukong/runner.rb +169 -0
  169. data/lib/wukong/runner/boot_sequence.rb +123 -0
  170. data/lib/wukong/runner/code_loader.rb +52 -0
  171. data/lib/wukong/runner/command_runner.rb +44 -0
  172. data/lib/wukong/runner/deploy_pack_loader.rb +75 -0
  173. data/lib/wukong/runner/help_message.rb +42 -0
  174. data/lib/wukong/source.rb +33 -0
  175. data/lib/wukong/source/source_driver.rb +74 -0
  176. data/lib/wukong/source/source_runner.rb +38 -0
  177. data/lib/wukong/spec_helpers.rb +74 -0
  178. data/lib/wukong/spec_helpers/integration_tests.rb +150 -0
  179. data/lib/wukong/spec_helpers/integration_tests/integration_test_matchers.rb +207 -0
  180. data/lib/wukong/spec_helpers/integration_tests/integration_test_runner.rb +97 -0
  181. data/lib/wukong/spec_helpers/shared_examples.rb +22 -0
  182. data/lib/wukong/spec_helpers/unit_tests.rb +135 -0
  183. data/lib/wukong/spec_helpers/unit_tests/unit_test_driver.rb +132 -0
  184. data/lib/wukong/spec_helpers/unit_tests/unit_test_matchers.rb +169 -0
  185. data/lib/wukong/spec_helpers/unit_tests/unit_test_runner.rb +60 -0
  186. data/lib/wukong/version.rb +3 -0
  187. data/lib/wukong/widget/echo.rb +55 -0
  188. data/lib/wukong/widget/extract.rb +122 -0
  189. data/lib/wukong/widget/filters.rb +452 -0
  190. data/lib/wukong/widget/logger.rb +56 -0
  191. data/lib/wukong/widget/operators.rb +82 -0
  192. data/lib/wukong/widget/reducers.rb +10 -0
  193. data/lib/wukong/widget/reducers/accumulator.rb +73 -0
  194. data/lib/wukong/widget/reducers/bin.rb +368 -0
  195. data/lib/wukong/widget/reducers/count.rb +73 -0
  196. data/lib/wukong/widget/reducers/group.rb +128 -0
  197. data/lib/wukong/widget/reducers/group_concat.rb +98 -0
  198. data/lib/wukong/widget/reducers/improver.rb +71 -0
  199. data/lib/wukong/widget/reducers/join_xml.rb +37 -0
  200. data/lib/wukong/widget/reducers/moments.rb +72 -0
  201. data/lib/wukong/widget/reducers/sort.rb +180 -0
  202. data/lib/wukong/widget/reducers/uniq.rb +91 -0
  203. data/lib/wukong/widget/serializers.rb +317 -0
  204. data/lib/wukong/widget/utils.rb +46 -0
  205. data/lib/wukong/widgets.rb +7 -0
  206. data/spec/examples/dataflow/fibonacci_series_spec.rb +18 -0
  207. data/spec/examples/dataflow/parse_apache_logs_spec.rb +8 -0
  208. data/spec/examples/dataflow/parsing_spec.rb +14 -0
  209. data/spec/examples/dataflow/simple_spec.rb +34 -0
  210. data/spec/examples/dataflow/telegram_spec.rb +43 -0
  211. data/spec/examples/graph/minimum_spanning_tree_spec.rb +34 -0
  212. data/spec/examples/munging/airline_flights/identifiers_spec.rb +16 -0
  213. data/spec/examples/munging/airline_flights_spec.rb +202 -0
  214. data/spec/examples/text/pig_latin_spec.rb +18 -0
  215. data/spec/examples/workflow/cherry_pie_spec.rb +36 -0
  216. data/spec/hanuman/graph_spec.rb +119 -0
  217. data/spec/hanuman/hanuman_spec.rb +10 -0
  218. data/spec/hanuman/registry_spec.rb +123 -0
  219. data/spec/hanuman/stage_spec.rb +81 -0
  220. data/spec/hanuman/tree_spec.rb +119 -0
  221. data/spec/spec.opts +1 -0
  222. data/spec/spec_helper.rb +43 -0
  223. data/spec/support/example_test_helpers.rb +95 -0
  224. data/spec/support/hanuman_test_helpers.rb +92 -0
  225. data/spec/support/integration_helper.rb +38 -0
  226. data/spec/support/model_test_helpers.rb +115 -0
  227. data/spec/support/shared_context_for_graphs.rb +57 -0
  228. data/spec/support/shared_context_for_reducers.rb +37 -0
  229. data/spec/support/shared_examples_for_builders.rb +94 -0
  230. data/spec/support/shared_examples_for_shortcuts.rb +57 -0
  231. data/spec/wu/model/reconcilable_spec.rb +152 -0
  232. data/spec/wukong/dataflow_spec.rb +87 -0
  233. data/spec/wukong/driver_spec.rb +154 -0
  234. data/spec/wukong/local/runner_spec.rb +29 -0
  235. data/spec/wukong/local/stdio_driver_spec.rb +73 -0
  236. data/spec/wukong/local_spec.rb +6 -0
  237. data/spec/wukong/logger_spec.rb +49 -0
  238. data/spec/wukong/model/faker_spec.rb +132 -0
  239. data/spec/wukong/processor_spec.rb +21 -0
  240. data/spec/wukong/runner_spec.rb +132 -0
  241. data/spec/wukong/source_spec.rb +6 -0
  242. data/spec/wukong/widget/extract_spec.rb +101 -0
  243. data/spec/wukong/widget/filters_spec.rb +79 -0
  244. data/spec/wukong/widget/logger_spec.rb +23 -0
  245. data/spec/wukong/widget/operators_spec.rb +25 -0
  246. data/spec/wukong/widget/reducers/bin_spec.rb +92 -0
  247. data/spec/wukong/widget/reducers/count_spec.rb +11 -0
  248. data/spec/wukong/widget/reducers/group_spec.rb +21 -0
  249. data/spec/wukong/widget/reducers/join_xml_spec.rb +25 -0
  250. data/spec/wukong/widget/reducers/moments_spec.rb +36 -0
  251. data/spec/wukong/widget/reducers/sort_spec.rb +26 -0
  252. data/spec/wukong/widget/reducers/uniq_spec.rb +14 -0
  253. data/spec/wukong/widget/serializers_spec.rb +114 -0
  254. data/spec/wukong/widget/sink_spec.rb +19 -0
  255. data/spec/wukong/widget/source_spec.rb +65 -0
  256. data/spec/wukong/wu-local_spec.rb +109 -0
  257. data/spec/wukong/wu-source_spec.rb +32 -0
  258. data/spec/wukong/wu_spec.rb +14 -0
  259. data/spec/wukong/wukong_spec.rb +10 -0
  260. data/wukong.gemspec +35 -0
  261. metadata +465 -0
@@ -0,0 +1,161 @@
1
+ module Hanuman
2
+ module Graphvizzer
3
+
4
+ COL_1_WIDTH = 47
5
+
6
+ class Item
7
+ include Gorillib::Builder
8
+
9
+ field :name, Symbol
10
+ field :label, String, :default => ->{ name }
11
+ field :owner, Item
12
+
13
+ def indent(adj=0)
14
+ " " * (depth + adj)
15
+ end
16
+
17
+ def depth
18
+ owner.depth + 1
19
+ end
20
+
21
+ def quote(str) str.to_s.include?('"') ? str : %Q{"#{str}"} ; end
22
+ def attrib(attr, val) "#{attr}=#{val}" ; end
23
+ def brace(str) "#{indent}#{str} {" ; end
24
+ def close_brace() "#{indent}}" ; end
25
+
26
+ def line(str, attrs={}, term=';')
27
+ if attrs.empty?
28
+ attr_strs = ''
29
+ else
30
+ width = COL_1_WIDTH - indent.length
31
+ str = "%-#{width}s" % str
32
+ attr_strs = attrs.map{|attr, val| attrib(attr, val) }
33
+ attr_strs = "\t[ #{attr_strs.join(", ")} ]"
34
+ end
35
+ [indent, str, attr_strs, term].join
36
+ end
37
+ end
38
+
39
+ class Graph < Item
40
+ field :items, Array, :default => []
41
+ field :edges, Array, :default => []
42
+
43
+ def to_s
44
+ str = []
45
+ str << brace("subgraph #{quote("cluster_#{name}")}") ## subgraph "cluster_crust" {
46
+ str << line(attrib(" label", quote(label))) ## label="crust";
47
+ items.each do |item| ##
48
+ str << item.to_s ## "cherry_pie.crust.small_bowl" [ shape=Mrecord, label="{small\nbowl}" ];
49
+ end ## "cherry_pie.crust.flour" [ shape=Mrecord, label="{flour}" ];
50
+ edges.each do |edge| ##
51
+ str << edge.to_s ## "cherry_pie.crust.small_bowl" -> "cherry_pie.crust.add_to_4";
52
+ end ## "cherry_pie.crust.flour" -> "cherry_pie.crust.add_to_4";
53
+ str << close_brace ## }
54
+ str.join("\n")
55
+ end
56
+
57
+ def graph(name, attrs={})
58
+ obj = Graph.new(attrs.merge(:name => name, :owner => self))
59
+ items << obj
60
+ yield(obj) if block_given?
61
+ obj
62
+ end
63
+
64
+ def node(name, attrs={})
65
+ obj = Node.new(attrs.merge(:name => name, :owner => self))
66
+ items << obj
67
+ yield(obj) if block_given?
68
+ obj
69
+ end
70
+
71
+ def edge(from, into, from_slot=nil, into_slot=nil)
72
+ obj = Edge.new(
73
+ :name => name, :owner => self,
74
+ :from => from, :into => into,
75
+ :from_slot => from_slot, :into_slot => into_slot)
76
+ edges << obj
77
+ yield(obj) if block_given?
78
+ obj
79
+ end
80
+ end
81
+
82
+ class Universe < Graph
83
+ field :orient, Symbol, :doc => 'one of :TB, :BT, :LR, :RL', :default => :TB
84
+ field :engine, Symbol, :default => :dot
85
+
86
+ def to_s
87
+ str = []
88
+ str << brace("digraph #{name}") ## digraph Wukong {
89
+ str << line(" rankdir = #{orient}") ## rankdir = TB;
90
+ items.each do |item| ## subgraph "cluster_cherry_pie" {
91
+ str << item.to_s ## # ...
92
+ end ## }
93
+ str << close_brace ## }
94
+ str.join("\n")
95
+ end
96
+
97
+ def depth() 0; end
98
+
99
+ def save(path, type=nil)
100
+ File.open "#{path}.dot", "w" do |f|
101
+ f.puts self.to_s
102
+ end
103
+ system "#{engine} -T#{type} #{path}.dot > #{path}.#{type}" if type
104
+ end
105
+ end
106
+
107
+ class Node < Item
108
+ field :inslots, Array, :default => []
109
+ field :outslots, Array, :default => []
110
+ field :shape, Symbol, :default => :Mrecord
111
+
112
+ def to_s
113
+ line(
114
+ quote(name), ## "cherry_pie.crust.small_bowl" [
115
+ :shape => shape, ## shape=Mrecord,
116
+ :label => quote(structured_label), ## label="{{<in>sb}|small\nbowl}"
117
+ ) ## ];
118
+ end
119
+
120
+ def abbreviate(word)
121
+ word.to_s.
122
+ gsub(/-\d+$/,''). # remove a trailing index (-0...)
123
+ split(/[\W_]+/). # split into _ separated segment
124
+ reject(&:empty?).map{|str| str[0] }.join
125
+ end
126
+
127
+ def inslots_str
128
+ inslots.map{|slot| "<#{slot}>#{abbreviate(slot)}"}.join("|")
129
+ end
130
+
131
+ def outslots_str
132
+ outslots.map{|slot| "<_#{slot}>#{abbreviate(slot)}"}.join("|")
133
+ end
134
+
135
+ def label
136
+ super.to_s.gsub(/_\d+$/, '').gsub(/[_\.]+/, "\\n")
137
+ end
138
+
139
+ def structured_label
140
+ return label unless shape =~ /record/
141
+ str = "{"
142
+ str << "{" << (inslots.empty? ? "<i>" : inslots_str) << "}|"
143
+ str << label
144
+ str << "|{" << (outslots.empty? ? "<_o>" : outslots_str) << "}"
145
+ str << "}"
146
+ str
147
+ end
148
+ end
149
+
150
+ class Edge < Item
151
+ field :from, Whatever
152
+ field :into, Whatever
153
+
154
+ def to_s
155
+ width = COL_1_WIDTH - indent.length
156
+ "#{indent}%-#{width}s\t-> %-s;" % [ from, into ] ## "cherry_pie.crust.small_bowl" -> "cherry_pie.crust.add_to_4";
157
+ end
158
+ end
159
+
160
+ end
161
+ end
@@ -0,0 +1,97 @@
1
+ require 'hanuman/graphvizzer/gv_models'
2
+
3
+ module Hanuman
4
+
5
+ Stage.class_eval do
6
+ class_attribute :draw_shape
7
+ self.draw_shape = :record
8
+
9
+ def graphviz_id
10
+ graph_id
11
+ end
12
+
13
+
14
+ def gv_into_label() warn [self, self.class.ancestors]; %Q{"#{graphviz_id}":"i"} ; end
15
+ def gv_from_label() warn [self, self.class.ancestors]; %Q{"#{graphviz_id}":"_o"} ; end
16
+
17
+ def to_graphviz(gv)
18
+ gv.node(self.graphviz_id,
19
+ :label => name,
20
+ :shape => draw_shape,
21
+ )
22
+ end
23
+ end
24
+
25
+ Product.class_eval do
26
+ self.draw_shape = :Mrecord
27
+ end
28
+
29
+ module ::Wukong::Universe
30
+ def to_graphviz
31
+ gv = Hanuman::Graphvizzer::Universe.new(:name => self.name)
32
+ @workflows.each_value do |workflow|
33
+ workflow.to_graphviz(gv)
34
+ end
35
+ @dataflows.each_value do |dataflow|
36
+ dataflow.to_graphviz(gv)
37
+ end
38
+ gv
39
+ end
40
+ end
41
+
42
+ Graph.class_eval do
43
+ self.draw_shape = :record
44
+
45
+ def to_graphviz(gv)
46
+ gv.graph(graphviz_id, :label => name) do |gv2|
47
+ inslots.each_value{|slot| slot.to_graphviz(gv2) if slot.wired? }
48
+
49
+ stages.each_value{|stage| stage.to_graphviz(gv2) }
50
+
51
+ outslots.each_value{|slot| slot.to_graphviz(gv2) if slot.wired? }
52
+ #
53
+ edges.each_value do |edge|
54
+ gv2.edge(edge[:from].gv_from_label, edge[:into].gv_into_label)
55
+ end
56
+ end
57
+ end
58
+ end
59
+
60
+ module Hanuman::Slottable
61
+ def to_graphviz(gv)
62
+ super.tap{|node| node.receive!(
63
+ :inslots => inslots.to_a.map{|slot| slot.name },
64
+ :outslots => outslots.to_a.map{|slot| slot.name },
65
+ ) }
66
+ end
67
+ end
68
+
69
+ module InputSlotted
70
+ def gv_into_label() %Q{"#{graphviz_id}":"i"} ; end
71
+ end
72
+ module OutputSlotted
73
+ def gv_from_label() %Q{"#{graphviz_id}":"_o"} ; end
74
+ end
75
+
76
+ Slot.class_eval do
77
+ def to_graphviz(gv)
78
+ gv.node(self.graphviz_id, label: name, shape: :Mrecord)
79
+ end
80
+ def graphviz_id() (stage.is_a?(Hanuman::Graph)||stage.is_a?(Wukong::Universe)) ? graph_id : stage.graph_id ; end
81
+ end
82
+
83
+ InputSlot.class_eval do
84
+ def gv_into_label() %Q{"#{graphviz_id}":"#{name}"} ; end
85
+ def to_graphviz(gv)
86
+ super.tap{|node| node.receive!(inslots: [name] )}
87
+ end
88
+ end
89
+
90
+ OutputSlot.class_eval do
91
+ def gv_from_label() %Q{"#{graphviz_id}":"_#{name}"} ; end
92
+ def to_graphviz(gv)
93
+ super.tap{|node| node.receive!(outslots: [name] )}
94
+ end
95
+ end
96
+
97
+ end
@@ -0,0 +1,35 @@
1
+ module Hanuman
2
+ class LinkFactory
3
+
4
+ Registry = {
5
+ simple: ->(from_stage, *into_stage){ DirectedLink.new(from_stage, *into_stage) }
6
+ }
7
+
8
+ class << self
9
+
10
+ def connect(label, from_stage, *into_stage)
11
+ Registry[label].call(from_stage, *into_stage)
12
+ end
13
+
14
+ def register(label, factory_method)
15
+ Registry[label] = factory_method
16
+ end
17
+
18
+ end
19
+ end
20
+
21
+ class DirectedLink
22
+
23
+ attr_accessor :from, :into
24
+
25
+ def initialize(from, into)
26
+ @from = from
27
+ @into = into
28
+ end
29
+
30
+ def to_s
31
+ "#<#{self.class}(#{from.to_s} -> #{into.to_s})>"
32
+ end
33
+
34
+ end
35
+ end
@@ -0,0 +1,46 @@
1
+ module Hanuman
2
+ class Registry
3
+
4
+ REGISTRY = {} unless defined? REGISTRY
5
+
6
+ def create_or_update(label, builder)
7
+ create(label, builder) ? true : update(label, builder)
8
+ end
9
+
10
+ def create(label, builder)
11
+ return false if registered?(label)
12
+ REGISTRY[label] = builder
13
+ true
14
+ end
15
+
16
+ def update(label, new_definition)
17
+ return false unless registered?(label)
18
+ REGISTRY[label].merge!(new_definition)
19
+ true
20
+ end
21
+
22
+ def registered?(label)
23
+ REGISTRY.keys.include? label
24
+ end
25
+
26
+ def retrieve(label)
27
+ REGISTRY[label].clone rescue nil
28
+ end
29
+
30
+ def decorate_with_registry(graph_instance)
31
+ REGISTRY.each_pair do |label, definition|
32
+ graph_instance.define_singleton_method(label) do |*args, &blk|
33
+ builder = Hanuman::GlobalRegistry.retrieve(label)
34
+ builder = handle_dsl_arguments_for(builder, *args, &blk)
35
+ stages[builder.label] = builder
36
+ end
37
+ end
38
+ end
39
+
40
+ def show() REGISTRY.dup ; end
41
+ def clear!() REGISTRY.clear ; end
42
+ def merge!(contents) REGISTRY.merge!(contents) ; end
43
+ end
44
+
45
+ GlobalRegistry = Registry.new unless defined? GlobalRegistry
46
+ end
@@ -0,0 +1,128 @@
1
+ module Hanuman
2
+
3
+ module StageInstanceMethods
4
+
5
+ attr_accessor :graph
6
+
7
+ def linkable_name(direction) ; self.label ; end
8
+
9
+ def add_stage_link(type, other_stage)
10
+ self.links << Hanuman::LinkFactory.connect(type, linkable_name(:in), other_stage.linkable_name(:out))
11
+ graph.add_link(type, self, other_stage) if graph
12
+ end
13
+
14
+ def root
15
+ graph ? graph.root(self) : self
16
+ end
17
+
18
+ end
19
+
20
+ module StageClassMethods
21
+
22
+ def label() self.to_s.demodulize.underscore.to_sym ; end
23
+
24
+ def builder
25
+ return @builder if @builder
26
+ set_builder(StageBuilder.new(label: label))
27
+ end
28
+
29
+ def set_builder(builder)
30
+ @builder = builder unless builder.nil?
31
+ builder.for_class = self
32
+ @builder
33
+ end
34
+
35
+ def register(new_label = nil)
36
+ builder.label = new_label || label
37
+ Hanuman::GlobalRegistry.create_or_update(new_label || label, builder)
38
+ self
39
+ end
40
+ end
41
+
42
+ class Stage
43
+ include Gorillib::Model
44
+ include StageInstanceMethods
45
+ extend StageClassMethods
46
+
47
+ field :label, Symbol, :doc => false
48
+ field :links, Array, :default => [], doc: false
49
+
50
+ def clone
51
+ cloned_attrs = Hash[ attributes.map{ |key, val| dup_key = key.dup rescue key ; dup_val = val.dup rescue val ; [ dup_key, dup_val ] } ]
52
+ cloned_links = links.map{ |link| link.dup }
53
+ self.class.receive(cloned_attrs.merge(links: cloned_links))
54
+ end
55
+ end
56
+
57
+ class StageBuilder
58
+
59
+ include Gorillib::Model
60
+ include StageInstanceMethods
61
+
62
+ field :args, Hash, :default => {}
63
+ field :for_class, Class
64
+ field :label, Symbol
65
+ field :links, Array, :default => []
66
+
67
+ def define(*args, &blk)
68
+ stage = for_class || define_class(label, *args)
69
+ stage.class_eval(&blk) if block_given?
70
+ stage.register
71
+ end
72
+
73
+ def build(options = {})
74
+ for_class.receive self.serialize.merge(options).merge(options[label] || {})
75
+ end
76
+
77
+ def handle_extra_attributes(attrs)
78
+ args.merge!(attrs)
79
+ end
80
+
81
+ def merge!(other_builder = {})
82
+ attrs = other_builder.attributes rescue other_builder
83
+ self.receive!(attrs)
84
+ self
85
+ end
86
+
87
+ def namespace(*args) Hanuman::Stage ; end
88
+
89
+ def define_class(name, *args)
90
+ klass = namespace(*args).const_get(name.to_s.camelize, Class.new(namespace(*args))) rescue nil
91
+ klass ||= namespace(*args).const_set(name.to_s.camelize, Class.new(namespace(*args)))
92
+ klass.set_builder(self)
93
+ klass
94
+ end
95
+
96
+ def serialize()
97
+ attrs = attributes
98
+ args = attrs.delete(:args)
99
+ attrs.delete(:links) ; attrs.delete(:for_class)
100
+ attrs.merge(args)
101
+ end
102
+
103
+ # This is a hacky method to clone a Stage ; probably could be merged into serialize?
104
+ def clone
105
+ cloned_attrs = Hash[ serialize.map{ |key, val| dup_key = key.dup rescue key ; dup_val = val.dup rescue val ; [ dup_key, dup_val ] } ]
106
+ cloned_links = links.map{ |link| link.dup }
107
+ self.class.receive(cloned_attrs.merge(links: cloned_links).merge(for_class: for_class))
108
+ end
109
+
110
+ def into(stage_or_stages)
111
+ return stage_or_stages if stage_or_stages.nil?
112
+ if stage_or_stages.is_a?(Array)
113
+ stage_or_stages.each do |other_stage_or_stages|
114
+ while other_stage_or_stages.is_a?(Array)
115
+ other_stage_or_stages = other_stage_or_stages.first
116
+ end
117
+ other_stage = other_stage_or_stages
118
+ self.into(other_stage)
119
+ end
120
+ else
121
+ self.add_stage_link(:simple, stage_or_stages.root)
122
+ end
123
+ stage_or_stages
124
+ end
125
+ alias_method :|, :into
126
+
127
+ end
128
+ end