wukong 3.0.0.pre → 3.0.0.pre2

Sign up to get free protection for your applications and to get access to all the features.
Files changed (476) hide show
  1. data/.gitignore +46 -33
  2. data/.gitmodules +3 -0
  3. data/.rspec +1 -1
  4. data/.travis.yml +8 -1
  5. data/.yardopts +0 -13
  6. data/Guardfile +4 -6
  7. data/{LICENSE.textile → LICENSE.md} +43 -55
  8. data/README-old.md +422 -0
  9. data/README.md +279 -418
  10. data/Rakefile +21 -5
  11. data/TODO.md +6 -6
  12. data/bin/wu-clean-encoding +31 -0
  13. data/bin/wu-lign +2 -2
  14. data/bin/wu-local +69 -0
  15. data/bin/wu-server +70 -0
  16. data/examples/Gemfile +38 -0
  17. data/examples/README.md +9 -0
  18. data/examples/dataflow/apache_log_line.rb +64 -25
  19. data/examples/dataflow/fibonacci_series.rb +101 -0
  20. data/examples/dataflow/parse_apache_logs.rb +37 -7
  21. data/examples/{dataflow.rb → dataflow/scraper_macro_flow.rb} +0 -0
  22. data/examples/dataflow/simple.rb +4 -4
  23. data/examples/geo.rb +4 -0
  24. data/examples/geo/geo_grids.numbers +0 -0
  25. data/examples/geo/geolocated.rb +331 -0
  26. data/examples/geo/quadtile.rb +69 -0
  27. data/examples/geo/spec/geolocated_spec.rb +247 -0
  28. data/examples/geo/tile_fetcher.rb +77 -0
  29. data/examples/graph/minimum_spanning_tree.rb +61 -61
  30. data/examples/jabberwocky.txt +36 -0
  31. data/examples/models/wikipedia.rb +20 -0
  32. data/examples/munging/Gemfile +8 -0
  33. data/examples/munging/airline_flights/airline.rb +57 -0
  34. data/examples/munging/airline_flights/airline_flights.rake +83 -0
  35. data/{lib/wukong/settings.rb → examples/munging/airline_flights/airplane.rb} +0 -0
  36. data/examples/munging/airline_flights/airport.rb +211 -0
  37. data/examples/munging/airline_flights/airport_id_unification.rb +129 -0
  38. data/examples/munging/airline_flights/airport_ok_chars.rb +4 -0
  39. data/examples/munging/airline_flights/flight.rb +156 -0
  40. data/examples/munging/airline_flights/models.rb +4 -0
  41. data/examples/munging/airline_flights/parse.rb +26 -0
  42. data/examples/munging/airline_flights/reconcile_airports.rb +142 -0
  43. data/examples/munging/airline_flights/route.rb +35 -0
  44. data/examples/munging/airline_flights/tasks.rake +83 -0
  45. data/examples/munging/airline_flights/timezone_fixup.rb +62 -0
  46. data/examples/munging/airline_flights/topcities.rb +167 -0
  47. data/examples/munging/airports/40_wbans.txt +40 -0
  48. data/examples/munging/airports/filter_weather_reports.rb +37 -0
  49. data/examples/munging/airports/join.pig +31 -0
  50. data/examples/munging/airports/to_tsv.rb +33 -0
  51. data/examples/munging/airports/usa_wbans.pig +19 -0
  52. data/examples/munging/airports/usa_wbans.txt +2157 -0
  53. data/examples/munging/airports/wbans.pig +19 -0
  54. data/examples/munging/airports/wbans.txt +2310 -0
  55. data/examples/munging/geo/geo_json.rb +54 -0
  56. data/examples/munging/geo/geo_models.rb +69 -0
  57. data/examples/munging/geo/geonames_models.rb +78 -0
  58. data/examples/munging/geo/iso_codes.rb +172 -0
  59. data/examples/munging/geo/reconcile_countries.rb +124 -0
  60. data/examples/munging/geo/tasks.rake +71 -0
  61. data/examples/munging/rake_helper.rb +62 -0
  62. data/examples/munging/weather/.gitignore +1 -0
  63. data/examples/munging/weather/Gemfile +4 -0
  64. data/examples/munging/weather/Rakefile +28 -0
  65. data/examples/munging/weather/extract_ish.rb +13 -0
  66. data/examples/munging/weather/models/weather.rb +119 -0
  67. data/examples/munging/weather/utils/noaa_downloader.rb +46 -0
  68. data/examples/munging/wikipedia/README.md +34 -0
  69. data/examples/munging/wikipedia/Rakefile +193 -0
  70. data/examples/munging/wikipedia/articles/extract_articles-parsed.rb +79 -0
  71. data/examples/munging/wikipedia/articles/extract_articles-templated.rb +136 -0
  72. data/examples/munging/wikipedia/articles/textualize_articles.rb +54 -0
  73. data/examples/munging/wikipedia/articles/verify_structure.rb +43 -0
  74. data/examples/munging/wikipedia/articles/wp2txt-LICENSE.txt +22 -0
  75. data/examples/munging/wikipedia/articles/wp2txt_article.rb +259 -0
  76. data/examples/munging/wikipedia/articles/wp2txt_utils.rb +452 -0
  77. data/examples/munging/wikipedia/dbpedia/dbpedia_common.rb +4 -0
  78. data/examples/munging/wikipedia/dbpedia/dbpedia_extract_geocoordinates.rb +78 -0
  79. data/examples/munging/wikipedia/dbpedia/extract_links.rb +193 -0
  80. data/examples/munging/wikipedia/dbpedia/sameas_extractor.rb +20 -0
  81. data/examples/munging/wikipedia/n1_subuniverse/n1_nodes.pig +18 -0
  82. data/examples/munging/wikipedia/page_metadata/extract_page_metadata.rb +21 -0
  83. data/examples/munging/wikipedia/page_metadata/extract_page_metadata.rb.old +27 -0
  84. data/examples/munging/wikipedia/pagelinks/augment_pagelinks.pig +29 -0
  85. data/examples/munging/wikipedia/pagelinks/extract_pagelinks.rb +14 -0
  86. data/examples/munging/wikipedia/pagelinks/extract_pagelinks.rb.old +25 -0
  87. data/examples/munging/wikipedia/pagelinks/undirect_pagelinks.pig +29 -0
  88. data/examples/munging/wikipedia/pageviews/augment_pageviews.pig +32 -0
  89. data/examples/munging/wikipedia/pageviews/extract_pageviews.rb +85 -0
  90. data/examples/munging/wikipedia/pig_style_guide.md +25 -0
  91. data/examples/munging/wikipedia/redirects/redirects_page_metadata.pig +19 -0
  92. data/examples/munging/wikipedia/subuniverse/sub_articles.pig +23 -0
  93. data/examples/munging/wikipedia/subuniverse/sub_page_metadata.pig +24 -0
  94. data/examples/munging/wikipedia/subuniverse/sub_pagelinks_from.pig +22 -0
  95. data/examples/munging/wikipedia/subuniverse/sub_pagelinks_into.pig +22 -0
  96. data/examples/munging/wikipedia/subuniverse/sub_pagelinks_within.pig +26 -0
  97. data/examples/munging/wikipedia/subuniverse/sub_pageviews.pig +29 -0
  98. data/examples/munging/wikipedia/subuniverse/sub_undirected_pagelinks_within.pig +24 -0
  99. data/examples/munging/wikipedia/utils/get_namespaces.rb +86 -0
  100. data/examples/munging/wikipedia/utils/munging_utils.rb +68 -0
  101. data/examples/munging/wikipedia/utils/namespaces.json +1 -0
  102. data/examples/rake_helper.rb +85 -0
  103. data/examples/server_logs/geo_ip_mapping/munge_geolite.rb +82 -0
  104. data/examples/server_logs/logline.rb +95 -0
  105. data/examples/server_logs/models.rb +66 -0
  106. data/examples/server_logs/page_counts.pig +48 -0
  107. data/examples/server_logs/server_logs-01-parse-script.rb +13 -0
  108. data/examples/server_logs/server_logs-02-histograms-full.rb +33 -0
  109. data/examples/server_logs/server_logs-02-histograms-mapper.rb +14 -0
  110. data/{old/examples/server_logs/breadcrumbs.rb → examples/server_logs/server_logs-03-breadcrumbs-full.rb} +26 -30
  111. data/examples/server_logs/server_logs-04-page_page_edges-full.rb +40 -0
  112. data/examples/string_reverser.rb +26 -0
  113. data/examples/text/pig_latin.rb +2 -2
  114. data/examples/text/regional_flavor/README.md +14 -0
  115. data/examples/text/regional_flavor/article_wordbags.pig +39 -0
  116. data/examples/text/regional_flavor/j01-article_wordbags.rb +4 -0
  117. data/examples/text/regional_flavor/simple_pig_script.pig +27 -0
  118. data/examples/word_count/accumulator.rb +26 -0
  119. data/examples/word_count/tokenizer.rb +13 -0
  120. data/examples/word_count/word_count.rb +6 -0
  121. data/examples/workflow/cherry_pie.dot +97 -0
  122. data/examples/workflow/cherry_pie.png +0 -0
  123. data/examples/workflow/cherry_pie.rb +61 -26
  124. data/lib/hanuman.rb +34 -7
  125. data/lib/hanuman/graph.rb +55 -31
  126. data/lib/hanuman/graphvizzer.rb +199 -178
  127. data/lib/hanuman/graphvizzer/gv_models.rb +161 -0
  128. data/lib/hanuman/graphvizzer/gv_presenter.rb +97 -0
  129. data/lib/hanuman/link.rb +35 -0
  130. data/lib/hanuman/registry.rb +46 -0
  131. data/lib/hanuman/stage.rb +76 -32
  132. data/lib/wukong.rb +23 -24
  133. data/lib/wukong/boot.rb +87 -0
  134. data/lib/wukong/configuration.rb +8 -0
  135. data/lib/wukong/dataflow.rb +45 -78
  136. data/lib/wukong/driver.rb +99 -0
  137. data/lib/wukong/emitter.rb +22 -0
  138. data/lib/wukong/model/faker.rb +24 -24
  139. data/lib/wukong/model/flatpack_parser/flat.rb +60 -0
  140. data/lib/wukong/model/flatpack_parser/flatpack.rb +4 -0
  141. data/lib/wukong/model/flatpack_parser/lang.rb +46 -0
  142. data/lib/wukong/model/flatpack_parser/parser.rb +55 -0
  143. data/lib/wukong/model/flatpack_parser/tokens.rb +130 -0
  144. data/lib/wukong/processor.rb +60 -114
  145. data/lib/wukong/spec_helpers.rb +81 -0
  146. data/lib/wukong/spec_helpers/integration_driver.rb +144 -0
  147. data/lib/wukong/spec_helpers/integration_driver_matchers.rb +219 -0
  148. data/lib/wukong/spec_helpers/processor_helpers.rb +95 -0
  149. data/lib/wukong/spec_helpers/processor_methods.rb +108 -0
  150. data/lib/wukong/spec_helpers/shared_examples.rb +15 -0
  151. data/lib/wukong/spec_helpers/spec_driver.rb +28 -0
  152. data/lib/wukong/spec_helpers/spec_driver_matchers.rb +195 -0
  153. data/lib/wukong/version.rb +2 -1
  154. data/lib/wukong/widget/filters.rb +311 -0
  155. data/lib/wukong/widget/processors.rb +156 -0
  156. data/lib/wukong/widget/reducers.rb +7 -0
  157. data/lib/wukong/widget/reducers/accumulator.rb +73 -0
  158. data/lib/wukong/widget/reducers/bin.rb +318 -0
  159. data/lib/wukong/widget/reducers/count.rb +61 -0
  160. data/lib/wukong/widget/reducers/group.rb +85 -0
  161. data/lib/wukong/widget/reducers/group_concat.rb +70 -0
  162. data/lib/wukong/widget/reducers/moments.rb +72 -0
  163. data/lib/wukong/widget/reducers/sort.rb +130 -0
  164. data/lib/wukong/widget/serializers.rb +287 -0
  165. data/lib/wukong/widget/sink.rb +10 -52
  166. data/lib/wukong/widget/source.rb +7 -113
  167. data/lib/wukong/widget/utils.rb +46 -0
  168. data/lib/wukong/widgets.rb +6 -0
  169. data/spec/examples/dataflow/fibonacci_series_spec.rb +18 -0
  170. data/spec/examples/dataflow/parsing_spec.rb +12 -11
  171. data/spec/examples/dataflow/simple_spec.rb +32 -6
  172. data/spec/examples/dataflow/telegram_spec.rb +36 -36
  173. data/spec/examples/graph/minimum_spanning_tree_spec.rb +30 -31
  174. data/spec/examples/munging/airline_flights/identifiers_spec.rb +16 -0
  175. data/spec/examples/munging/airline_flights_spec.rb +202 -0
  176. data/spec/examples/text/pig_latin_spec.rb +13 -16
  177. data/spec/examples/workflow/cherry_pie_spec.rb +34 -4
  178. data/spec/hanuman/graph_spec.rb +27 -2
  179. data/spec/hanuman/hanuman_spec.rb +10 -0
  180. data/spec/hanuman/registry_spec.rb +123 -0
  181. data/spec/hanuman/stage_spec.rb +61 -7
  182. data/spec/spec_helper.rb +29 -19
  183. data/spec/support/hanuman_test_helpers.rb +14 -12
  184. data/spec/support/shared_context_for_reducers.rb +37 -0
  185. data/spec/support/shared_examples_for_builders.rb +101 -0
  186. data/spec/support/shared_examples_for_shortcuts.rb +57 -0
  187. data/spec/support/wukong_test_helpers.rb +37 -11
  188. data/spec/wukong/dataflow_spec.rb +77 -55
  189. data/spec/wukong/local_runner_spec.rb +24 -24
  190. data/spec/wukong/model/faker_spec.rb +132 -131
  191. data/spec/wukong/runner_spec.rb +8 -8
  192. data/spec/wukong/widget/filters_spec.rb +61 -0
  193. data/spec/wukong/widget/processors_spec.rb +126 -0
  194. data/spec/wukong/widget/reducers/bin_spec.rb +92 -0
  195. data/spec/wukong/widget/reducers/count_spec.rb +11 -0
  196. data/spec/wukong/widget/reducers/group_spec.rb +20 -0
  197. data/spec/wukong/widget/reducers/moments_spec.rb +36 -0
  198. data/spec/wukong/widget/reducers/sort_spec.rb +26 -0
  199. data/spec/wukong/widget/serializers_spec.rb +92 -0
  200. data/spec/wukong/widget/sink_spec.rb +15 -15
  201. data/spec/wukong/widget/source_spec.rb +65 -41
  202. data/spec/wukong/wukong_spec.rb +10 -0
  203. data/wukong.gemspec +17 -10
  204. metadata +359 -335
  205. data/.document +0 -5
  206. data/VERSION +0 -1
  207. data/bin/hdp-bin +0 -44
  208. data/bin/hdp-bzip +0 -23
  209. data/bin/hdp-cat +0 -3
  210. data/bin/hdp-catd +0 -3
  211. data/bin/hdp-cp +0 -3
  212. data/bin/hdp-du +0 -86
  213. data/bin/hdp-get +0 -3
  214. data/bin/hdp-kill +0 -3
  215. data/bin/hdp-kill-task +0 -3
  216. data/bin/hdp-ls +0 -11
  217. data/bin/hdp-mkdir +0 -2
  218. data/bin/hdp-mkdirp +0 -12
  219. data/bin/hdp-mv +0 -3
  220. data/bin/hdp-parts_to_keys.rb +0 -77
  221. data/bin/hdp-ps +0 -3
  222. data/bin/hdp-put +0 -3
  223. data/bin/hdp-rm +0 -32
  224. data/bin/hdp-sort +0 -40
  225. data/bin/hdp-stream +0 -40
  226. data/bin/hdp-stream-flat +0 -22
  227. data/bin/hdp-stream2 +0 -39
  228. data/bin/hdp-sync +0 -17
  229. data/bin/hdp-wc +0 -67
  230. data/bin/wu-flow +0 -10
  231. data/bin/wu-map +0 -17
  232. data/bin/wu-red +0 -17
  233. data/bin/wukong +0 -17
  234. data/data/CREDITS.md +0 -355
  235. data/data/graph/airfares.tsv +0 -2174
  236. data/data/text/gift_of_the_magi.txt +0 -225
  237. data/data/text/jabberwocky.txt +0 -36
  238. data/data/text/rectification_of_names.txt +0 -33
  239. data/data/twitter/a_atsigns_b.tsv +0 -64
  240. data/data/twitter/a_follows_b.tsv +0 -53
  241. data/data/twitter/tweet.tsv +0 -167
  242. data/data/twitter/twitter_user.tsv +0 -55
  243. data/data/wikipedia/dbpedia-sentences.tsv +0 -1000
  244. data/docpages/INSTALL.textile +0 -92
  245. data/docpages/LICENSE.textile +0 -107
  246. data/docpages/README-elastic_map_reduce.textile +0 -377
  247. data/docpages/README-performance.textile +0 -90
  248. data/docpages/README-wulign.textile +0 -65
  249. data/docpages/UsingWukong-part1-get_ready.textile +0 -17
  250. data/docpages/UsingWukong-part2-ThinkingBigData.textile +0 -75
  251. data/docpages/UsingWukong-part3-parsing.textile +0 -138
  252. data/docpages/_config.yml +0 -39
  253. data/docpages/avro/avro_notes.textile +0 -56
  254. data/docpages/avro/performance.textile +0 -36
  255. data/docpages/avro/tethering.textile +0 -19
  256. data/docpages/bigdata-tips.textile +0 -143
  257. data/docpages/code/api_response_example.txt +0 -20
  258. data/docpages/code/parser_skeleton.rb +0 -38
  259. data/docpages/diagrams/MapReduceDiagram.graffle +0 -0
  260. data/docpages/favicon.ico +0 -0
  261. data/docpages/gem.css +0 -16
  262. data/docpages/hadoop-tips.textile +0 -83
  263. data/docpages/index.textile +0 -92
  264. data/docpages/intro.textile +0 -8
  265. data/docpages/moreinfo.textile +0 -174
  266. data/docpages/news.html +0 -24
  267. data/docpages/pig/PigLatinExpressionsList.txt +0 -122
  268. data/docpages/pig/PigLatinReferenceManual.txt +0 -1640
  269. data/docpages/pig/commandline_params.txt +0 -26
  270. data/docpages/pig/cookbook.html +0 -481
  271. data/docpages/pig/images/hadoop-logo.jpg +0 -0
  272. data/docpages/pig/images/instruction_arrow.png +0 -0
  273. data/docpages/pig/images/pig-logo.gif +0 -0
  274. data/docpages/pig/piglatin_ref1.html +0 -1103
  275. data/docpages/pig/piglatin_ref2.html +0 -14340
  276. data/docpages/pig/setup.html +0 -505
  277. data/docpages/pig/skin/basic.css +0 -166
  278. data/docpages/pig/skin/breadcrumbs.js +0 -237
  279. data/docpages/pig/skin/fontsize.js +0 -166
  280. data/docpages/pig/skin/getBlank.js +0 -40
  281. data/docpages/pig/skin/getMenu.js +0 -45
  282. data/docpages/pig/skin/images/chapter.gif +0 -0
  283. data/docpages/pig/skin/images/chapter_open.gif +0 -0
  284. data/docpages/pig/skin/images/current.gif +0 -0
  285. data/docpages/pig/skin/images/external-link.gif +0 -0
  286. data/docpages/pig/skin/images/header_white_line.gif +0 -0
  287. data/docpages/pig/skin/images/page.gif +0 -0
  288. data/docpages/pig/skin/images/pdfdoc.gif +0 -0
  289. data/docpages/pig/skin/images/rc-b-l-15-1body-2menu-3menu.png +0 -0
  290. data/docpages/pig/skin/images/rc-b-r-15-1body-2menu-3menu.png +0 -0
  291. data/docpages/pig/skin/images/rc-b-r-5-1header-2tab-selected-3tab-selected.png +0 -0
  292. data/docpages/pig/skin/images/rc-t-l-5-1header-2searchbox-3searchbox.png +0 -0
  293. data/docpages/pig/skin/images/rc-t-l-5-1header-2tab-selected-3tab-selected.png +0 -0
  294. data/docpages/pig/skin/images/rc-t-l-5-1header-2tab-unselected-3tab-unselected.png +0 -0
  295. data/docpages/pig/skin/images/rc-t-r-15-1body-2menu-3menu.png +0 -0
  296. data/docpages/pig/skin/images/rc-t-r-5-1header-2searchbox-3searchbox.png +0 -0
  297. data/docpages/pig/skin/images/rc-t-r-5-1header-2tab-selected-3tab-selected.png +0 -0
  298. data/docpages/pig/skin/images/rc-t-r-5-1header-2tab-unselected-3tab-unselected.png +0 -0
  299. data/docpages/pig/skin/print.css +0 -54
  300. data/docpages/pig/skin/profile.css +0 -181
  301. data/docpages/pig/skin/screen.css +0 -587
  302. data/docpages/pig/tutorial.html +0 -1059
  303. data/docpages/pig/udf.html +0 -1509
  304. data/docpages/tutorial.textile +0 -283
  305. data/docpages/usage.textile +0 -195
  306. data/docpages/wutils.textile +0 -263
  307. data/examples/dataflow/complex.rb +0 -11
  308. data/examples/dataflow/donuts.rb +0 -13
  309. data/examples/tiny_count/jabberwocky_output.tsv +0 -92
  310. data/examples/word_count.rb +0 -48
  311. data/examples/workflow/fiddle.rb +0 -24
  312. data/lib/away/escapement.rb +0 -129
  313. data/lib/away/exe.rb +0 -11
  314. data/lib/away/experimental.rb +0 -5
  315. data/lib/away/from_file.rb +0 -52
  316. data/lib/away/job.rb +0 -56
  317. data/lib/away/job/rake_compat.rb +0 -17
  318. data/lib/away/registry.rb +0 -79
  319. data/lib/away/runner.rb +0 -276
  320. data/lib/away/runner/execute.rb +0 -121
  321. data/lib/away/script.rb +0 -161
  322. data/lib/away/script/hadoop_command.rb +0 -240
  323. data/lib/away/source/file_list_source.rb +0 -15
  324. data/lib/away/source/looper.rb +0 -18
  325. data/lib/away/task.rb +0 -219
  326. data/lib/hanuman/action.rb +0 -21
  327. data/lib/hanuman/chain.rb +0 -4
  328. data/lib/hanuman/graphviz.rb +0 -74
  329. data/lib/hanuman/resource.rb +0 -6
  330. data/lib/hanuman/slot.rb +0 -87
  331. data/lib/hanuman/slottable.rb +0 -220
  332. data/lib/wukong/bad_record.rb +0 -15
  333. data/lib/wukong/event.rb +0 -44
  334. data/lib/wukong/local_runner.rb +0 -55
  335. data/lib/wukong/mapred.rb +0 -3
  336. data/lib/wukong/universe.rb +0 -48
  337. data/lib/wukong/widget/filter.rb +0 -81
  338. data/lib/wukong/widget/gibberish.rb +0 -123
  339. data/lib/wukong/widget/monitor.rb +0 -26
  340. data/lib/wukong/widget/reducer.rb +0 -66
  341. data/lib/wukong/widget/stringifier.rb +0 -50
  342. data/lib/wukong/workflow.rb +0 -22
  343. data/lib/wukong/workflow/command.rb +0 -42
  344. data/old/config/emr-example.yaml +0 -48
  345. data/old/examples/README.txt +0 -17
  346. data/old/examples/contrib/jeans/README.markdown +0 -165
  347. data/old/examples/contrib/jeans/data/normalized_sizes +0 -3
  348. data/old/examples/contrib/jeans/data/orders.tsv +0 -1302
  349. data/old/examples/contrib/jeans/data/sizes +0 -3
  350. data/old/examples/contrib/jeans/normalize.rb +0 -20
  351. data/old/examples/contrib/jeans/sizes.rb +0 -55
  352. data/old/examples/corpus/bnc_word_freq.rb +0 -44
  353. data/old/examples/corpus/bucket_counter.rb +0 -47
  354. data/old/examples/corpus/dbpedia_abstract_to_sentences.rb +0 -86
  355. data/old/examples/corpus/sentence_bigrams.rb +0 -53
  356. data/old/examples/corpus/sentence_coocurrence.rb +0 -66
  357. data/old/examples/corpus/stopwords.rb +0 -138
  358. data/old/examples/corpus/words_to_bigrams.rb +0 -53
  359. data/old/examples/emr/README.textile +0 -110
  360. data/old/examples/emr/dot_wukong_dir/credentials.json +0 -7
  361. data/old/examples/emr/dot_wukong_dir/emr.yaml +0 -69
  362. data/old/examples/emr/dot_wukong_dir/emr_bootstrap.sh +0 -33
  363. data/old/examples/emr/elastic_mapreduce_example.rb +0 -28
  364. data/old/examples/network_graph/adjacency_list.rb +0 -74
  365. data/old/examples/network_graph/breadth_first_search.rb +0 -72
  366. data/old/examples/network_graph/gen_2paths.rb +0 -68
  367. data/old/examples/network_graph/gen_multi_edge.rb +0 -112
  368. data/old/examples/network_graph/gen_symmetric_links.rb +0 -64
  369. data/old/examples/pagerank/README.textile +0 -6
  370. data/old/examples/pagerank/gen_initial_pagerank_graph.pig +0 -57
  371. data/old/examples/pagerank/pagerank.rb +0 -72
  372. data/old/examples/pagerank/pagerank_initialize.rb +0 -42
  373. data/old/examples/pagerank/run_pagerank.sh +0 -21
  374. data/old/examples/sample_records.rb +0 -33
  375. data/old/examples/server_logs/apache_log_parser.rb +0 -15
  376. data/old/examples/server_logs/nook.rb +0 -48
  377. data/old/examples/server_logs/nook/faraday_dummy_adapter.rb +0 -94
  378. data/old/examples/server_logs/user_agent.rb +0 -40
  379. data/old/examples/simple_word_count.rb +0 -82
  380. data/old/examples/size.rb +0 -61
  381. data/old/examples/stats/avg_value_frequency.rb +0 -86
  382. data/old/examples/stats/binning_percentile_estimator.rb +0 -140
  383. data/old/examples/stats/data/avg_value_frequency.tsv +0 -3
  384. data/old/examples/stats/rank_and_bin.rb +0 -173
  385. data/old/examples/stupidly_simple_filter.rb +0 -40
  386. data/old/examples/word_count.rb +0 -75
  387. data/old/graph/graphviz_builder.rb +0 -580
  388. data/old/graph_easy/Attributes.pm +0 -4181
  389. data/old/graph_easy/Graphviz.pm +0 -2232
  390. data/old/wukong.rb +0 -18
  391. data/old/wukong/and_pig.rb +0 -38
  392. data/old/wukong/bad_record.rb +0 -18
  393. data/old/wukong/datatypes.rb +0 -24
  394. data/old/wukong/datatypes/enum.rb +0 -127
  395. data/old/wukong/datatypes/fake_types.rb +0 -17
  396. data/old/wukong/decorator.rb +0 -28
  397. data/old/wukong/encoding/asciize.rb +0 -108
  398. data/old/wukong/extensions.rb +0 -16
  399. data/old/wukong/extensions/array.rb +0 -18
  400. data/old/wukong/extensions/blank.rb +0 -93
  401. data/old/wukong/extensions/class.rb +0 -189
  402. data/old/wukong/extensions/date_time.rb +0 -53
  403. data/old/wukong/extensions/emittable.rb +0 -69
  404. data/old/wukong/extensions/enumerable.rb +0 -79
  405. data/old/wukong/extensions/hash.rb +0 -167
  406. data/old/wukong/extensions/hash_keys.rb +0 -16
  407. data/old/wukong/extensions/hash_like.rb +0 -150
  408. data/old/wukong/extensions/hashlike_class.rb +0 -47
  409. data/old/wukong/extensions/module.rb +0 -2
  410. data/old/wukong/extensions/pathname.rb +0 -27
  411. data/old/wukong/extensions/string.rb +0 -65
  412. data/old/wukong/extensions/struct.rb +0 -17
  413. data/old/wukong/extensions/symbol.rb +0 -11
  414. data/old/wukong/filename_pattern.rb +0 -74
  415. data/old/wukong/helper.rb +0 -7
  416. data/old/wukong/helper/stopwords.rb +0 -195
  417. data/old/wukong/helper/tokenize.rb +0 -35
  418. data/old/wukong/logger.rb +0 -38
  419. data/old/wukong/periodic_monitor.rb +0 -72
  420. data/old/wukong/schema.rb +0 -269
  421. data/old/wukong/script.rb +0 -286
  422. data/old/wukong/script/avro_command.rb +0 -5
  423. data/old/wukong/script/cassandra_loader_script.rb +0 -40
  424. data/old/wukong/script/emr_command.rb +0 -168
  425. data/old/wukong/script/hadoop_command.rb +0 -237
  426. data/old/wukong/script/local_command.rb +0 -41
  427. data/old/wukong/store.rb +0 -10
  428. data/old/wukong/store/base.rb +0 -27
  429. data/old/wukong/store/cassandra.rb +0 -10
  430. data/old/wukong/store/cassandra/streaming.rb +0 -75
  431. data/old/wukong/store/cassandra/struct_loader.rb +0 -21
  432. data/old/wukong/store/cassandra_model.rb +0 -91
  433. data/old/wukong/store/chh_chunked_flat_file_store.rb +0 -37
  434. data/old/wukong/store/chunked_flat_file_store.rb +0 -48
  435. data/old/wukong/store/conditional_store.rb +0 -57
  436. data/old/wukong/store/factory.rb +0 -8
  437. data/old/wukong/store/flat_file_store.rb +0 -89
  438. data/old/wukong/store/key_store.rb +0 -51
  439. data/old/wukong/store/null_store.rb +0 -15
  440. data/old/wukong/store/read_thru_store.rb +0 -22
  441. data/old/wukong/store/tokyo_tdb_key_store.rb +0 -33
  442. data/old/wukong/store/tyrant_rdb_key_store.rb +0 -57
  443. data/old/wukong/store/tyrant_tdb_key_store.rb +0 -20
  444. data/old/wukong/streamer.rb +0 -30
  445. data/old/wukong/streamer/accumulating_reducer.rb +0 -83
  446. data/old/wukong/streamer/base.rb +0 -126
  447. data/old/wukong/streamer/counting_reducer.rb +0 -25
  448. data/old/wukong/streamer/filter.rb +0 -20
  449. data/old/wukong/streamer/instance_streamer.rb +0 -15
  450. data/old/wukong/streamer/json_streamer.rb +0 -21
  451. data/old/wukong/streamer/line_streamer.rb +0 -12
  452. data/old/wukong/streamer/list_reducer.rb +0 -31
  453. data/old/wukong/streamer/rank_and_bin_reducer.rb +0 -145
  454. data/old/wukong/streamer/record_streamer.rb +0 -14
  455. data/old/wukong/streamer/reducer.rb +0 -11
  456. data/old/wukong/streamer/set_reducer.rb +0 -14
  457. data/old/wukong/streamer/struct_streamer.rb +0 -48
  458. data/old/wukong/streamer/summing_reducer.rb +0 -29
  459. data/old/wukong/streamer/uniq_by_last_reducer.rb +0 -51
  460. data/old/wukong/typed_struct.rb +0 -12
  461. data/spec/away/encoding_spec.rb +0 -32
  462. data/spec/away/exe_spec.rb +0 -20
  463. data/spec/away/flow_spec.rb +0 -82
  464. data/spec/away/graph_spec.rb +0 -6
  465. data/spec/away/job_spec.rb +0 -15
  466. data/spec/away/rake_compat_spec.rb +0 -9
  467. data/spec/away/script_spec.rb +0 -81
  468. data/spec/hanuman/graphviz_spec.rb +0 -29
  469. data/spec/hanuman/slot_spec.rb +0 -2
  470. data/spec/support/examples_helper.rb +0 -10
  471. data/spec/support/streamer_test_helpers.rb +0 -6
  472. data/spec/support/wukong_widget_helpers.rb +0 -66
  473. data/spec/wukong/processor_spec.rb +0 -109
  474. data/spec/wukong/widget/filter_spec.rb +0 -99
  475. data/spec/wukong/widget/stringifier_spec.rb +0 -51
  476. data/spec/wukong/workflow/command_spec.rb +0 -5
@@ -1,27 +1,9 @@
1
1
  #!/usr/bin/env ruby
2
- require 'rubygems'
3
- require 'wukong/script'
4
-
5
- class Logline < Struct.new(
6
- :ip, :date, :time, :http_method, :protocol, :path, :response_code, :duration, :referer, :ua, :tz)
7
-
8
- def page_type
9
- case
10
- when path =~ /\.(css|js)$/ then :asset
11
- when path =~ /\.(png|gif|ico)$/ then :image
12
- when path =~ /\.(pl|s?html?|asp|jsp|cgi)$/ then :page
13
- else :other
14
- end
15
- end
16
-
17
- def is_page?
18
- page_type == :page
19
- end
20
-
21
- def day_hr
22
- visit.date + visit.time[0..1]
23
- end
24
- end
2
+ $LOAD_PATH.unshift File.expand_path('../../lib', File.dirname(__FILE__))
3
+ require 'configliere'
4
+ Settings.define :page_types, type: Array, default: ['page', 'video'], description: "Acceptable page types"
5
+ require 'wukong/script'
6
+ require_relative './logline'
25
7
 
26
8
 
27
9
  #
@@ -38,9 +20,11 @@ end
38
20
  #
39
21
  # where the partition key is visitor_id, and we sort by visitor_id and datetime.
40
22
  #
41
- class VisitorDatePath < Wukong::Streamer::StructStreamer
23
+ class BreadcrumbsMapper < Wukong::Streamer::ModelStreamer
24
+ self.model_klass = Logline
42
25
  def process visit, *args
43
- yield [visit.ip, visit.day_hr, visit.path]
26
+ # return unless Settings.page_types.include?(visit.page_type)
27
+ yield [visit.ip, visit.day_hr, visit.visit_time.to_i, visit.path]
44
28
  end
45
29
  end
46
30
 
@@ -65,11 +49,23 @@ end
65
49
  # page_trails <pagen> <n_pages_in_visit> <duration> <timestamp> < page1,page2,... >
66
50
  #
67
51
  # to discover all trails passing through a given page.
68
- class VisitorDatePath < Wukong::Streamer::Reducer
69
- def get_key ip, day_hr, path, *args
70
- [ip, day_hr]
52
+ class BreadcrumbsReducer < Wukong::Streamer::Reducer
53
+ def get_key ip, day_hr, itime, path, *args
54
+ [ip]
55
+ end
56
+ def start!(*args)
57
+ @path_times = []
58
+ super
71
59
  end
72
- def process_group visit, *args
73
- yield [visit.ip, visit.day_hr, visit.path]
60
+ def accumulate ip, day_hr, itime, path, *args
61
+ # @path_times << "(#{itime},#{path})"
62
+ @path_times << "#{itime}:#{path}"
63
+ end
64
+ def finalize
65
+ # yield [key, "{" << @path_times.join(",") << "}"]
66
+ yield [key, @path_times.join("|")]
74
67
  end
75
68
  end
69
+
70
+
71
+ Wukong.run( BreadcrumbsMapper, BreadcrumbsReducer, :sort_fields => 2 )
@@ -0,0 +1,40 @@
1
+ #!/usr/bin/env ruby
2
+ $LOAD_PATH.unshift File.expand_path('../../lib', File.dirname(__FILE__))
3
+ require 'wukong/script'
4
+ require_relative './logline'
5
+
6
+ class BreadcrumbsMapper < Wukong::Streamer::ModelStreamer
7
+ self.model_klass = Logline
8
+ def process visit, *args
9
+ # return unless Settings.page_types.include?(visit.page_type)
10
+ yield [visit.ip, visit.visit_time.to_i, visit.path]
11
+ end
12
+ end
13
+
14
+ class BreadcrumbEdgesReducer < Wukong::Streamer::Reducer
15
+ def get_key ip, itime, path
16
+ [ip]
17
+ end
18
+ def start!(*args)
19
+ @paths = Set.new
20
+ super
21
+ end
22
+ def accumulate ip, itime, path
23
+ @paths << path
24
+ end
25
+
26
+ # for each pair of paths, emit the edge in both directions
27
+ def finalize
28
+ @paths = @paths.to_a
29
+ while @paths.present?
30
+ from = @paths.shift
31
+ @paths.each do |into|
32
+ yield [key, from, into]
33
+ yield [key, into, from]
34
+ end
35
+ end
36
+ end
37
+ end
38
+
39
+
40
+ Wukong.run( BreadcrumbsMapper, BreadcrumbEdgesReducer, :sort_fields => 2 )
@@ -0,0 +1,26 @@
1
+ Wukong.processor(:string_reverser) do
2
+
3
+ def setup
4
+ log.info("Inside the setup method")
5
+ end
6
+
7
+ def process(rec)
8
+ notify('metrics', bad_word: rec, level: :warn) if rec.match(/fuck|shit|piss/)
9
+ yield rec.values
10
+ end
11
+
12
+ def finalize
13
+ log.info("Finalizing flow")
14
+ end
15
+
16
+ def stop
17
+ log.info("Inside the stop method")
18
+ end
19
+
20
+ end
21
+
22
+ Wukong.dataflow(:chained) do
23
+
24
+ from_json > string_reverser > to_tsv > topic(topic: 'foobar')
25
+
26
+ end
@@ -28,8 +28,8 @@ Wukong.processor :pig_latinize do
28
28
  end
29
29
 
30
30
  ExampleUniverse.dataflow(:pig_latin) do
31
- input :default, file_source(Pathname.path_to(:data, 'text/gift_of_the_magi.txt'))
32
- output :default, file_sink( Pathname.path_to(:tmp, 'text/pig_latin/gift_of_the_magi.txt'))
31
+ set_input :default, file_source(Pathname.path_to(:data, 'text/gift_of_the_magi.txt'))
32
+ set_output :default, file_sink( Pathname.path_to(:tmp, 'text/pig_latin/gift_of_the_magi.txt'))
33
33
 
34
34
  input(:default) > pig_latinize > output(:default)
35
35
  end
@@ -0,0 +1,14 @@
1
+ # Find the Regional Flavor of topics using Geolocated Wikipedia Articles
2
+
3
+ (Chapter 1 of "Big Data for Chimps")
4
+
5
+ 1. article -> wordbag
6
+ 2. join on page data to get geolocation
7
+ 3. use pagelinks to get larger pool of implied geolocations
8
+ 4. turn geolocations into quadtile keys
9
+ 5. aggregate topics by quadtile
10
+ 6. take summary statistics aggregated over term and quadkey
11
+ 7. combine those statistics to identify terms that occur more frequently than the base rate would predict
12
+ 8. explore and validate the results
13
+ 9. filter to find strongly-flavored words, and other reductions of the data for visualization
14
+
@@ -0,0 +1,39 @@
1
+ %declare wukong_dir '/Users/flip/ics/core/wukong'
2
+ %declare data_dir '$wukong_dir/data'
3
+ %declare dsfp_dir '/Users/flip/ics/data_science_fun_pack'
4
+ ;
5
+
6
+ register '$dsfp_dir/pig/varaha/target/varaha-1.0-SNAPSHOT.jar';
7
+ register '$dsfp_dir/pig/varaha/lib/mallet-2.0.7-RC2.jar';
8
+ register '$dsfp_dir/pig/varaha/lib/trove-2.0.4.jar';
9
+ register '$dsfp_dir/pig/varaha/lib/lucene-core-3.1.0.jar';
10
+ register '$dsfp_dir/pig/varaha/lib/pygmalion-1.1.0-SNAPSHOT.jar';
11
+ register '$dsfp_dir/pig/pigsy/target/pigsy-2.1.0-SNAPSHOT.jar';
12
+ register '$dsfp_dir/pig/datafu/dist/datafu-0.0.6-SNAPSHOT.jar';
13
+
14
+ define JsonStrToString pigsy.pig.json.JsonStrToString();
15
+ define TokenizeText varaha.text.TokenizeText();
16
+ define LDATopics varaha.topic.LDATopics();
17
+ define RangeConcat org.pygmalion.udf.RangeBasedStringConcat('0', ' ');
18
+
19
+ -- Load the markup-stripped wikipedia text
20
+ torture_strings = LOAD '$data_dir/helpers/torture/string_handling_test.tsv' AS (
21
+ desc:chararray, len:int, bytesize:int, str:chararray, has_str:int, jsonized_str:chararray,
22
+ escaped_chars:chararray, escaped_bytes:chararray, chars_list:chararray, bytes_list:chararray
23
+ );
24
+
25
+ -- Generate a random integer between 0 and n
26
+ decoded = FOREACH torture_strings {
27
+ unjsonized_str = JsonStrToString(jsonized_str);
28
+ is_str_equal = (((str == unjsonized_str) OR (has_str == 0)) ? 1 : 0);
29
+ is_len_equal = ( len == SIZE(unjsonized_str) ? 1 : 0 );
30
+ GENERATE
31
+ is_str_equal,
32
+ is_len_equal,
33
+ str AS str,
34
+ unjsonized_str AS unjsonized_str
35
+ ;
36
+ };
37
+
38
+ -- take a dump on your terminal window
39
+ DUMP decoded;
@@ -0,0 +1,4 @@
1
+
2
+ #
3
+ # Summarize each article as a wordbag.
4
+ #
@@ -0,0 +1,27 @@
1
+ %declare wukong_dir '/Users/flip/ics/core/wukong'
2
+ %declare data_dir '$wukong_dir/data'
3
+ %declare dsfp_dir '/Users/flip/ics/data_science_fun_pack'
4
+ ;
5
+
6
+ register '$dsfp_dir/pig/datafu/dist/datafu-0.0.6-SNAPSHOT.jar';
7
+ register '$dsfp_dir/pig/pigsy/target/pigsy-2.1.0-SNAPSHOT.jar';
8
+
9
+ define RandInt datafu.pig.numbers.RandInt();
10
+ define ConcatBag com.infochimps.hadoop.pig.ConcatBag();
11
+
12
+ -- Load the data: the integers from 0 .. 1023
13
+ ones = LOAD '$data_dir/helpers/numbers/integers-1ki.tsv' AS (val:int);
14
+
15
+ -- Generate a random integer between 0 and n
16
+ rands = FOREACH ones GENERATE val, RandInt(0,val) AS rand_val;
17
+ rand_str_g = GROUP rands ALL;
18
+ rand_str_s = FOREACH rand_str_g {
19
+ joined_str = ConcatBag(rands.(rand_val));
20
+ len = SIZE(joined_str);
21
+ GENERATE joined_str AS joined_str, len AS len;
22
+ };
23
+
24
+ -- take a dump on your terminal window
25
+ STORE rand_str_s INTO '/tmp/dump';
26
+
27
+ DESCRIBE rand_str_s;
@@ -0,0 +1,26 @@
1
+ Wukong.processor(:accumulator) do
2
+ attr_accessor :count, :current
3
+
4
+ def reset!() @current = nil ; @count = 0 ; end
5
+
6
+ def report_then_reset!(&blk)
7
+ yield [current, count].join("\t") unless current.nil?
8
+ reset!
9
+ end
10
+
11
+ def accumulate(word, seen)
12
+ @current = word if @current.nil?
13
+ @count += seen
14
+ end
15
+
16
+ def process(pair, &blk)
17
+ word, seen = pair.split("\t")
18
+ report_then_reset!(&blk) unless word == current
19
+ accumulate(word, seen.to_i)
20
+ end
21
+
22
+ def finalize(&blk)
23
+ report_then_reset!(&blk)
24
+ end
25
+
26
+ end
@@ -0,0 +1,13 @@
1
+ Wukong.processor(:tokenizer) do
2
+
3
+ field :min_length, Integer, :default => 1
4
+
5
+ def process(record)
6
+ words = record.downcase.strip.split(/\W/)
7
+ lengthy = words.select{ |word| word.length >= min_length }
8
+ lengthy.each do |word|
9
+ yield [ word, 1 ].join("\t")
10
+ end
11
+ end
12
+
13
+ end
@@ -0,0 +1,6 @@
1
+ require_relative 'tokenizer'
2
+ require_relative 'accumulator'
3
+
4
+ Wukong.dataflow(:word_count) do
5
+ tokenizer(min_length: 3) > sort(label: 'first_sort') > accumulator > sort(on: 1, numeric: true, reverse: true)
6
+ end
@@ -0,0 +1,97 @@
1
+ digraph Warrant {
2
+ rankdir = TB;
3
+ subgraph "cluster_cherry_pie" {
4
+ label = "cherry_pie";
5
+ subgraph "cluster_cherry_pie.crust" {
6
+ label = "crust";
7
+ "cherry_pie.crust.add_to_0" [ shape = record,label = "{{<container>c|<flour>f|<salt>s|<shortening>s}|add\nto|{<out_crumbly_mixture>c}}" ];
8
+ "cherry_pie.crust.flour" [ shape = Mrecord,label = "{flour}" ];
9
+ "cherry_pie.crust.salt" [ shape = Mrecord,label = "{salt}" ];
10
+ "cherry_pie.crust.shortening" [ shape = Mrecord,label = "{shortening}" ];
11
+ "cherry_pie.crust.small_bowl" [ shape = Mrecord,label = "{small\nbowl}" ];
12
+ "cherry_pie.crust.crumbly_mixture"[ shape = Mrecord,label = "{crumbly\nmixture}" ];
13
+ "cherry_pie.crust.add_to_6" [ shape = record,label = "{{<container>c|<buttermilk>b}|add\nto|{<out_dough>d}}" ];
14
+ "cherry_pie.crust.buttermilk" [ shape = Mrecord,label = "{buttermilk}" ];
15
+ "cherry_pie.crust.dough" [ shape = Mrecord,label = "{dough}" ];
16
+ "cherry_pie.crust.split_9" [ shape = record,label = "{{<dough>d}|split|{<out_ball_for_top>b|<out_ball_for_bottom>b}}" ];
17
+ "cherry_pie.crust.ball_for_top" [ shape = Mrecord,label = "{ball\nfor\ntop}" ];
18
+ "cherry_pie.crust.ball_for_bottom"[ shape = Mrecord,label = "{ball\nfor\nbottom}" ];
19
+ "cherry_pie.crust.rolling_pin_12" [ shape = record,label = "{{<ball_for_bottom>b}|rolling\npin|{<out_combine_13>c}}" ];
20
+ "cherry_pie.crust.combine_13" [ shape = record,label = "{{<pie_tin>p|<rolling_pin_12>r}|combine|{<out_pie_tin_with_crust>p}}" ];
21
+ "cherry_pie.crust.pie_tin" [ shape = Mrecord,label = "{pie\ntin}" ];
22
+ "cherry_pie.crust.pie_tin_with_crust"[ shape = Mrecord,label = "{pie\ntin\nwith\ncrust}" ];
23
+ "cherry_pie.crust.flour" -> "cherry_pie.crust.add_to_0":flour;
24
+ "cherry_pie.crust.salt" -> "cherry_pie.crust.add_to_0":salt;
25
+ "cherry_pie.crust.shortening" -> "cherry_pie.crust.add_to_0":shortening;
26
+ "cherry_pie.crust.small_bowl" -> "cherry_pie.crust.add_to_0":container;
27
+ "cherry_pie.crust.add_to_0":out_crumbly_mixture -> "cherry_pie.crust.crumbly_mixture";
28
+ "cherry_pie.crust.crumbly_mixture" -> "cherry_pie.crust.add_to_6":container;
29
+ "cherry_pie.crust.buttermilk" -> "cherry_pie.crust.add_to_6":buttermilk;
30
+ "cherry_pie.crust.add_to_6":out_dough -> "cherry_pie.crust.dough";
31
+ "cherry_pie.crust.dough" -> "cherry_pie.crust.split_9":dough;
32
+ "cherry_pie.crust.split_9":out_ball_for_top -> "cherry_pie.crust.ball_for_top";
33
+ "cherry_pie.crust.split_9":out_ball_for_bottom -> "cherry_pie.crust.ball_for_bottom";
34
+ "cherry_pie.crust.ball_for_bottom" -> "cherry_pie.crust.rolling_pin_12":ball_for_bottom;
35
+ "cherry_pie.crust.pie_tin" -> "cherry_pie.crust.combine_13":pie_tin;
36
+ "cherry_pie.crust.rolling_pin_12":out_combine_13 -> "cherry_pie.crust.combine_13":rolling_pin_12;
37
+ "cherry_pie.crust.combine_13":out_pie_tin_with_crust -> "cherry_pie.crust.pie_tin_with_crust";
38
+ }
39
+ "cherry_pie.crust" [ shape = record,label = "{crust}" ];
40
+ subgraph "cluster_cherry_pie.filling" {
41
+ label = "filling";
42
+ "cherry_pie.filling.drain_0" [ shape = record,label = "{{<cherries>c}|drain|{<out_drained_cherries>d|<out_cherry_juice>c}}" ];
43
+ "cherry_pie.filling.cherries" [ shape = Mrecord,label = "{cherries}" ];
44
+ "cherry_pie.filling.drained_cherries"[ shape = Mrecord,label = "{drained\ncherries}" ];
45
+ "cherry_pie.filling.cherry_juice" [ shape = Mrecord,label = "{cherry\njuice}" ];
46
+ "cherry_pie.filling.add_to_4" [ shape = record,label = "{{<container>c|<corn_starch>c|<sugar>s|<salt>s}|add\nto|{<out_whisk_9>w}}" ];
47
+ "cherry_pie.filling.corn_starch" [ shape = Mrecord,label = "{corn\nstarch}" ];
48
+ "cherry_pie.filling.sugar" [ shape = Mrecord,label = "{sugar}" ];
49
+ "cherry_pie.filling.salt" [ shape = Mrecord,label = "{salt}" ];
50
+ "cherry_pie.filling.saucepan" [ shape = Mrecord,label = "{saucepan}" ];
51
+ "cherry_pie.filling.whisk_9" [ shape = record,label = "{{<cherry_juice>c|<add_to_4>a}|whisk|{<out_raw_goop>r}}" ];
52
+ "cherry_pie.filling.raw_goop" [ shape = Mrecord,label = "{raw\ngoop}" ];
53
+ "cherry_pie.filling.cook_11" [ shape = record,label = "{{<raw_goop>r}|cook|{<out_goop>g}}" ];
54
+ "cherry_pie.filling.goop" [ shape = Mrecord,label = "{goop}" ];
55
+ "cherry_pie.filling.add_to_13" [ shape = record,label = "{{<container>c|<drained_cherries>d|<butter>b}|add\nto|{<out_cool_15>c}}" ];
56
+ "cherry_pie.filling.butter" [ shape = Mrecord,label = "{butter}" ];
57
+ "cherry_pie.filling.cool_15" [ shape = record,label = "{{<add_to_13>a}|cool|{<out_filling>f}}" ];
58
+ "cherry_pie.filling.cherries" -> "cherry_pie.filling.drain_0":cherries;
59
+ "cherry_pie.filling.drain_0":out_drained_cherries -> "cherry_pie.filling.drained_cherries";
60
+ "cherry_pie.filling.drain_0":out_cherry_juice -> "cherry_pie.filling.cherry_juice";
61
+ "cherry_pie.filling.corn_starch" -> "cherry_pie.filling.add_to_4":corn_starch;
62
+ "cherry_pie.filling.sugar" -> "cherry_pie.filling.add_to_4":sugar;
63
+ "cherry_pie.filling.salt" -> "cherry_pie.filling.add_to_4":salt;
64
+ "cherry_pie.filling.saucepan" -> "cherry_pie.filling.add_to_4":container;
65
+ "cherry_pie.filling.cherry_juice" -> "cherry_pie.filling.whisk_9":cherry_juice;
66
+ "cherry_pie.filling.add_to_4":out_whisk_9 -> "cherry_pie.filling.whisk_9":add_to_4;
67
+ "cherry_pie.filling.whisk_9":out_raw_goop -> "cherry_pie.filling.raw_goop";
68
+ "cherry_pie.filling.raw_goop" -> "cherry_pie.filling.cook_11":raw_goop;
69
+ "cherry_pie.filling.cook_11":out_goop -> "cherry_pie.filling.goop";
70
+ "cherry_pie.filling.drained_cherries" -> "cherry_pie.filling.add_to_13":drained_cherries;
71
+ "cherry_pie.filling.butter" -> "cherry_pie.filling.add_to_13":butter;
72
+ "cherry_pie.filling.goop" -> "cherry_pie.filling.add_to_13":container;
73
+ "cherry_pie.filling.add_to_13":out_cool_15 -> "cherry_pie.filling.cool_15":add_to_13;
74
+ "cherry_pie.filling.cool_15":out_filling -> "cherry_pie.filling";
75
+ }
76
+ "cherry_pie.filling" [ shape = record,label = "{filling}" ];
77
+ "cherry_pie.rolling_pin_2" [ shape = record,label = "{{<ball_for_top>b}|rolling\npin|{<out_top_crust>t}}" ];
78
+ "cherry_pie.top_crust" [ shape = Mrecord,label = "{top\ncrust}" ];
79
+ "cherry_pie.add_to_4" [ shape = record,label = "{{<container>c|<filling>f|<top_crust>t}|add\nto|{<out_cook_5>c}}" ];
80
+ "cherry_pie.cook_5" [ shape = record,label = "{{<oven>o|<add_to_4>a}|cook|{<out_cool_7>c}}" ];
81
+ "cherry_pie.oven" [ shape = Mrecord,label = "{oven}" ];
82
+ "cherry_pie.cool_7" [ shape = record,label = "{{<wire_rack>w|<cook_5>c}|cool|{<out_cherry_pie>c}}" ];
83
+ "cherry_pie.wire_rack" [ shape = Mrecord,label = "{wire\nrack}" ];
84
+ "cherry_pie.crust.ball_for_bottom" -> "cherry_pie.crust";
85
+ "cherry_pie.crust.pie_tin_with_crust" -> "cherry_pie.add_to_4":container;
86
+ "cherry_pie.crust.ball_for_top" -> "cherry_pie.rolling_pin_2":ball_for_top;
87
+ "cherry_pie.rolling_pin_2":out_top_crust -> "cherry_pie.top_crust";
88
+ "cherry_pie.filling" -> "cherry_pie.add_to_4":filling;
89
+ "cherry_pie.top_crust" -> "cherry_pie.add_to_4":top_crust;
90
+ "cherry_pie.oven" -> "cherry_pie.cook_5":oven;
91
+ "cherry_pie.add_to_4":out_cook_5 -> "cherry_pie.cook_5":add_to_4;
92
+ "cherry_pie.wire_rack" -> "cherry_pie.cool_7":wire_rack;
93
+ "cherry_pie.cook_5":out_cool_7 -> "cherry_pie.cool_7":cook_5;
94
+ "cherry_pie.cool_7":out_cherry_pie -> "cherry_pie";
95
+ }
96
+ "cherry_pie" [ shape = record,label = "{cherry\npie}" ];
97
+ }
@@ -1,6 +1,15 @@
1
1
  class Wukong::Workflow < Hanuman::Graph
2
2
 
3
- class Container < Hanuman::Resource ; end
3
+ class Container < Hanuman::Product
4
+ register_stage
5
+ end
6
+
7
+ class Qty < Hanuman::Product
8
+ doc 'a quantity of an ingredient'
9
+ register_stage
10
+ field :amount, String, :position => 1
11
+ end
12
+ class Utensil < Hanuman::Product ; register_stage ; end
4
13
 
5
14
  class Cook < ActionWithInputs
6
15
  register_action
@@ -17,50 +26,76 @@ class Wukong::Workflow < Hanuman::Graph
17
26
 
18
27
  class AddTo < ActionWithInputs
19
28
  register_action
20
- consumes :container
21
29
 
22
- def self.make(workflow, container, *input_stages, &block)
23
- options = input_stages.extract_options!
24
- super(workflow, *input_stages, options.merge(:container => container), &block)
25
- end
30
+ magic :container, Hanuman::Stage
26
31
  end
32
+
27
33
  end
28
34
 
35
+ Hanuman::Graph.class_eval{ alias_method :product, :product }
36
+
29
37
  # TODO: make repeated calls not retrieve object again -- it seems lookup is the special case, not creation.
30
38
 
31
- class Warrant ; extend Wukong::Universe ; end
32
- Warrant.workflow(:cherry_pie) do
33
- graph(:crust) do
34
- add_to(:small_bowl, :flour, :salt, :shortening) > :crumbly_mixture
39
+ #
40
+ # Make Warrant happy
41
+ #
42
+ # class Warrant ; extend Wukong::Universe ; end
43
+ Wukong.workflow(:cherry_pie) do
44
+
45
+ subgraph(:crust) do
46
+
47
+ add_to(container(:small_bowl),
48
+ qty(:flour, '3 cups'),
49
+ qty(:salt, '1.5 tsp'),
50
+ qty(:shortening, '6 tbsp')
51
+ ) > product(:crumbly_mixture)
35
52
 
36
53
  # equvalently:
37
- # add_to(:crumbly_mixture, :buttermilk) > :dough
38
- add_to(:crumbly_mixture) << :buttermilk > :dough
54
+ # add_to(:crumbly_mixture, qty(:buttermilk)) > product(:dough)
55
+ add_to(:crumbly_mixture) << qty(:buttermilk) > product(:dough)
39
56
 
40
- split(:dough).into(:ball_for_top, :ball_for_bottom)
57
+ split(:dough) do
58
+ into owner.product(:ball_for_top)
59
+ into owner.product(:ball_for_bottom)
60
+ end
41
61
 
42
62
  # equvalently:
43
- # combine << :pie_tin << (rolling_pin << :ball_for_bottom) > :pie_tin_with_crust
44
- combine(:pie_tin, (rolling_pin << :ball_for_bottom)).into(:pie_tin_with_crust)
63
+ # combine << :pie_tin << (rolling_pin << :ball_for_bottom) > :pie_tin_w_crust
64
+ combine(container(:pie_tin),
65
+ (rolling_pin << :ball_for_bottom)
66
+ ).into(product(:pie_tin_w_crust))
45
67
 
46
- self << stage(:ball_for_bottom)
47
- self << stage(:pie_tin_with_crust)
48
68
  end
49
69
 
50
- graph(:filling) do
51
- drain(:cherries).into(:drained_cherries, :cherry_juice)
52
- add_to(:saucepan, :corn_starch, :sugar, :salt) >
70
+ subgraph(:filling) do
71
+
72
+ qty(:cherries, '4 cups')
73
+
74
+ drain(:cherries).into(
75
+ product(:drained_cherries),
76
+ ).into( # FIXME - should be a multiple-output
77
+ product(:cherry_juice)
78
+ )
79
+ qty(:butter, '2 tbsp, cut up')
80
+ add_to(container(:saucepan),
81
+ qty(:corn_starch, '1/3 cup'),
82
+ qty(:sugar, '1.5 cups'),
83
+ qty(:salt, '1 dash')) >
53
84
  whisk << :cherry_juice >
54
- :raw_goop
55
- cook(:raw_goop, :trigger => 'goop slightly thickened') > :goop
56
- add_to(:goop, :drained_cherries, :butter) > cool > self
85
+ product(:raw_goop)
86
+ cook(:raw_goop, :trigger => 'goop slightly thickened') > product(:goop)
87
+ add_to(:goop, :drained_cherries, :butter) > cool > product(:output)
57
88
  end
58
89
 
59
- rolling_pin << stage(:crust).stage(:ball_for_top) > :top_crust
90
+ rolling_pin << stage(:crust).stage(:ball_for_top) > product(:top_crust)
60
91
 
61
- raw_pie = add_to(stage(:crust).stage(:pie_tin_with_crust), :filling)
92
+ raw_pie = add_to(
93
+ stage(:crust).stage(:pie_tin_w_crust),
94
+ stage(:filling).product(:output))
62
95
  raw_pie << :top_crust
63
96
 
64
- cook(:oven, raw_pie) > cool(:wire_rack) > self
97
+ x = utensil(:oven)
98
+
99
+ cook(utensil(:oven), raw_pie) > cool(utensil(:wire_rack))
65
100
 
66
101
  end