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.
- data/.gitignore +46 -33
- data/.gitmodules +3 -0
- data/.rspec +1 -1
- data/.travis.yml +8 -1
- data/.yardopts +0 -13
- data/Guardfile +4 -6
- data/{LICENSE.textile → LICENSE.md} +43 -55
- data/README-old.md +422 -0
- data/README.md +279 -418
- data/Rakefile +21 -5
- data/TODO.md +6 -6
- data/bin/wu-clean-encoding +31 -0
- data/bin/wu-lign +2 -2
- data/bin/wu-local +69 -0
- data/bin/wu-server +70 -0
- data/examples/Gemfile +38 -0
- data/examples/README.md +9 -0
- data/examples/dataflow/apache_log_line.rb +64 -25
- data/examples/dataflow/fibonacci_series.rb +101 -0
- data/examples/dataflow/parse_apache_logs.rb +37 -7
- data/examples/{dataflow.rb → dataflow/scraper_macro_flow.rb} +0 -0
- data/examples/dataflow/simple.rb +4 -4
- data/examples/geo.rb +4 -0
- data/examples/geo/geo_grids.numbers +0 -0
- data/examples/geo/geolocated.rb +331 -0
- data/examples/geo/quadtile.rb +69 -0
- data/examples/geo/spec/geolocated_spec.rb +247 -0
- data/examples/geo/tile_fetcher.rb +77 -0
- data/examples/graph/minimum_spanning_tree.rb +61 -61
- data/examples/jabberwocky.txt +36 -0
- data/examples/models/wikipedia.rb +20 -0
- data/examples/munging/Gemfile +8 -0
- data/examples/munging/airline_flights/airline.rb +57 -0
- data/examples/munging/airline_flights/airline_flights.rake +83 -0
- data/{lib/wukong/settings.rb → examples/munging/airline_flights/airplane.rb} +0 -0
- data/examples/munging/airline_flights/airport.rb +211 -0
- data/examples/munging/airline_flights/airport_id_unification.rb +129 -0
- data/examples/munging/airline_flights/airport_ok_chars.rb +4 -0
- data/examples/munging/airline_flights/flight.rb +156 -0
- data/examples/munging/airline_flights/models.rb +4 -0
- data/examples/munging/airline_flights/parse.rb +26 -0
- data/examples/munging/airline_flights/reconcile_airports.rb +142 -0
- data/examples/munging/airline_flights/route.rb +35 -0
- data/examples/munging/airline_flights/tasks.rake +83 -0
- data/examples/munging/airline_flights/timezone_fixup.rb +62 -0
- data/examples/munging/airline_flights/topcities.rb +167 -0
- data/examples/munging/airports/40_wbans.txt +40 -0
- data/examples/munging/airports/filter_weather_reports.rb +37 -0
- data/examples/munging/airports/join.pig +31 -0
- data/examples/munging/airports/to_tsv.rb +33 -0
- data/examples/munging/airports/usa_wbans.pig +19 -0
- data/examples/munging/airports/usa_wbans.txt +2157 -0
- data/examples/munging/airports/wbans.pig +19 -0
- data/examples/munging/airports/wbans.txt +2310 -0
- data/examples/munging/geo/geo_json.rb +54 -0
- data/examples/munging/geo/geo_models.rb +69 -0
- data/examples/munging/geo/geonames_models.rb +78 -0
- data/examples/munging/geo/iso_codes.rb +172 -0
- data/examples/munging/geo/reconcile_countries.rb +124 -0
- data/examples/munging/geo/tasks.rake +71 -0
- data/examples/munging/rake_helper.rb +62 -0
- data/examples/munging/weather/.gitignore +1 -0
- data/examples/munging/weather/Gemfile +4 -0
- data/examples/munging/weather/Rakefile +28 -0
- data/examples/munging/weather/extract_ish.rb +13 -0
- data/examples/munging/weather/models/weather.rb +119 -0
- data/examples/munging/weather/utils/noaa_downloader.rb +46 -0
- data/examples/munging/wikipedia/README.md +34 -0
- data/examples/munging/wikipedia/Rakefile +193 -0
- data/examples/munging/wikipedia/articles/extract_articles-parsed.rb +79 -0
- data/examples/munging/wikipedia/articles/extract_articles-templated.rb +136 -0
- data/examples/munging/wikipedia/articles/textualize_articles.rb +54 -0
- data/examples/munging/wikipedia/articles/verify_structure.rb +43 -0
- data/examples/munging/wikipedia/articles/wp2txt-LICENSE.txt +22 -0
- data/examples/munging/wikipedia/articles/wp2txt_article.rb +259 -0
- data/examples/munging/wikipedia/articles/wp2txt_utils.rb +452 -0
- data/examples/munging/wikipedia/dbpedia/dbpedia_common.rb +4 -0
- data/examples/munging/wikipedia/dbpedia/dbpedia_extract_geocoordinates.rb +78 -0
- data/examples/munging/wikipedia/dbpedia/extract_links.rb +193 -0
- data/examples/munging/wikipedia/dbpedia/sameas_extractor.rb +20 -0
- data/examples/munging/wikipedia/n1_subuniverse/n1_nodes.pig +18 -0
- data/examples/munging/wikipedia/page_metadata/extract_page_metadata.rb +21 -0
- data/examples/munging/wikipedia/page_metadata/extract_page_metadata.rb.old +27 -0
- data/examples/munging/wikipedia/pagelinks/augment_pagelinks.pig +29 -0
- data/examples/munging/wikipedia/pagelinks/extract_pagelinks.rb +14 -0
- data/examples/munging/wikipedia/pagelinks/extract_pagelinks.rb.old +25 -0
- data/examples/munging/wikipedia/pagelinks/undirect_pagelinks.pig +29 -0
- data/examples/munging/wikipedia/pageviews/augment_pageviews.pig +32 -0
- data/examples/munging/wikipedia/pageviews/extract_pageviews.rb +85 -0
- data/examples/munging/wikipedia/pig_style_guide.md +25 -0
- data/examples/munging/wikipedia/redirects/redirects_page_metadata.pig +19 -0
- data/examples/munging/wikipedia/subuniverse/sub_articles.pig +23 -0
- data/examples/munging/wikipedia/subuniverse/sub_page_metadata.pig +24 -0
- data/examples/munging/wikipedia/subuniverse/sub_pagelinks_from.pig +22 -0
- data/examples/munging/wikipedia/subuniverse/sub_pagelinks_into.pig +22 -0
- data/examples/munging/wikipedia/subuniverse/sub_pagelinks_within.pig +26 -0
- data/examples/munging/wikipedia/subuniverse/sub_pageviews.pig +29 -0
- data/examples/munging/wikipedia/subuniverse/sub_undirected_pagelinks_within.pig +24 -0
- data/examples/munging/wikipedia/utils/get_namespaces.rb +86 -0
- data/examples/munging/wikipedia/utils/munging_utils.rb +68 -0
- data/examples/munging/wikipedia/utils/namespaces.json +1 -0
- data/examples/rake_helper.rb +85 -0
- data/examples/server_logs/geo_ip_mapping/munge_geolite.rb +82 -0
- data/examples/server_logs/logline.rb +95 -0
- data/examples/server_logs/models.rb +66 -0
- data/examples/server_logs/page_counts.pig +48 -0
- data/examples/server_logs/server_logs-01-parse-script.rb +13 -0
- data/examples/server_logs/server_logs-02-histograms-full.rb +33 -0
- data/examples/server_logs/server_logs-02-histograms-mapper.rb +14 -0
- data/{old/examples/server_logs/breadcrumbs.rb → examples/server_logs/server_logs-03-breadcrumbs-full.rb} +26 -30
- data/examples/server_logs/server_logs-04-page_page_edges-full.rb +40 -0
- data/examples/string_reverser.rb +26 -0
- data/examples/text/pig_latin.rb +2 -2
- data/examples/text/regional_flavor/README.md +14 -0
- data/examples/text/regional_flavor/article_wordbags.pig +39 -0
- data/examples/text/regional_flavor/j01-article_wordbags.rb +4 -0
- data/examples/text/regional_flavor/simple_pig_script.pig +27 -0
- data/examples/word_count/accumulator.rb +26 -0
- data/examples/word_count/tokenizer.rb +13 -0
- data/examples/word_count/word_count.rb +6 -0
- data/examples/workflow/cherry_pie.dot +97 -0
- data/examples/workflow/cherry_pie.png +0 -0
- data/examples/workflow/cherry_pie.rb +61 -26
- data/lib/hanuman.rb +34 -7
- data/lib/hanuman/graph.rb +55 -31
- data/lib/hanuman/graphvizzer.rb +199 -178
- data/lib/hanuman/graphvizzer/gv_models.rb +161 -0
- data/lib/hanuman/graphvizzer/gv_presenter.rb +97 -0
- data/lib/hanuman/link.rb +35 -0
- data/lib/hanuman/registry.rb +46 -0
- data/lib/hanuman/stage.rb +76 -32
- data/lib/wukong.rb +23 -24
- data/lib/wukong/boot.rb +87 -0
- data/lib/wukong/configuration.rb +8 -0
- data/lib/wukong/dataflow.rb +45 -78
- data/lib/wukong/driver.rb +99 -0
- data/lib/wukong/emitter.rb +22 -0
- data/lib/wukong/model/faker.rb +24 -24
- data/lib/wukong/model/flatpack_parser/flat.rb +60 -0
- data/lib/wukong/model/flatpack_parser/flatpack.rb +4 -0
- data/lib/wukong/model/flatpack_parser/lang.rb +46 -0
- data/lib/wukong/model/flatpack_parser/parser.rb +55 -0
- data/lib/wukong/model/flatpack_parser/tokens.rb +130 -0
- data/lib/wukong/processor.rb +60 -114
- data/lib/wukong/spec_helpers.rb +81 -0
- data/lib/wukong/spec_helpers/integration_driver.rb +144 -0
- data/lib/wukong/spec_helpers/integration_driver_matchers.rb +219 -0
- data/lib/wukong/spec_helpers/processor_helpers.rb +95 -0
- data/lib/wukong/spec_helpers/processor_methods.rb +108 -0
- data/lib/wukong/spec_helpers/shared_examples.rb +15 -0
- data/lib/wukong/spec_helpers/spec_driver.rb +28 -0
- data/lib/wukong/spec_helpers/spec_driver_matchers.rb +195 -0
- data/lib/wukong/version.rb +2 -1
- data/lib/wukong/widget/filters.rb +311 -0
- data/lib/wukong/widget/processors.rb +156 -0
- data/lib/wukong/widget/reducers.rb +7 -0
- data/lib/wukong/widget/reducers/accumulator.rb +73 -0
- data/lib/wukong/widget/reducers/bin.rb +318 -0
- data/lib/wukong/widget/reducers/count.rb +61 -0
- data/lib/wukong/widget/reducers/group.rb +85 -0
- data/lib/wukong/widget/reducers/group_concat.rb +70 -0
- data/lib/wukong/widget/reducers/moments.rb +72 -0
- data/lib/wukong/widget/reducers/sort.rb +130 -0
- data/lib/wukong/widget/serializers.rb +287 -0
- data/lib/wukong/widget/sink.rb +10 -52
- data/lib/wukong/widget/source.rb +7 -113
- data/lib/wukong/widget/utils.rb +46 -0
- data/lib/wukong/widgets.rb +6 -0
- data/spec/examples/dataflow/fibonacci_series_spec.rb +18 -0
- data/spec/examples/dataflow/parsing_spec.rb +12 -11
- data/spec/examples/dataflow/simple_spec.rb +32 -6
- data/spec/examples/dataflow/telegram_spec.rb +36 -36
- data/spec/examples/graph/minimum_spanning_tree_spec.rb +30 -31
- data/spec/examples/munging/airline_flights/identifiers_spec.rb +16 -0
- data/spec/examples/munging/airline_flights_spec.rb +202 -0
- data/spec/examples/text/pig_latin_spec.rb +13 -16
- data/spec/examples/workflow/cherry_pie_spec.rb +34 -4
- data/spec/hanuman/graph_spec.rb +27 -2
- data/spec/hanuman/hanuman_spec.rb +10 -0
- data/spec/hanuman/registry_spec.rb +123 -0
- data/spec/hanuman/stage_spec.rb +61 -7
- data/spec/spec_helper.rb +29 -19
- data/spec/support/hanuman_test_helpers.rb +14 -12
- data/spec/support/shared_context_for_reducers.rb +37 -0
- data/spec/support/shared_examples_for_builders.rb +101 -0
- data/spec/support/shared_examples_for_shortcuts.rb +57 -0
- data/spec/support/wukong_test_helpers.rb +37 -11
- data/spec/wukong/dataflow_spec.rb +77 -55
- data/spec/wukong/local_runner_spec.rb +24 -24
- data/spec/wukong/model/faker_spec.rb +132 -131
- data/spec/wukong/runner_spec.rb +8 -8
- data/spec/wukong/widget/filters_spec.rb +61 -0
- data/spec/wukong/widget/processors_spec.rb +126 -0
- data/spec/wukong/widget/reducers/bin_spec.rb +92 -0
- data/spec/wukong/widget/reducers/count_spec.rb +11 -0
- data/spec/wukong/widget/reducers/group_spec.rb +20 -0
- data/spec/wukong/widget/reducers/moments_spec.rb +36 -0
- data/spec/wukong/widget/reducers/sort_spec.rb +26 -0
- data/spec/wukong/widget/serializers_spec.rb +92 -0
- data/spec/wukong/widget/sink_spec.rb +15 -15
- data/spec/wukong/widget/source_spec.rb +65 -41
- data/spec/wukong/wukong_spec.rb +10 -0
- data/wukong.gemspec +17 -10
- metadata +359 -335
- data/.document +0 -5
- data/VERSION +0 -1
- data/bin/hdp-bin +0 -44
- data/bin/hdp-bzip +0 -23
- data/bin/hdp-cat +0 -3
- data/bin/hdp-catd +0 -3
- data/bin/hdp-cp +0 -3
- data/bin/hdp-du +0 -86
- data/bin/hdp-get +0 -3
- data/bin/hdp-kill +0 -3
- data/bin/hdp-kill-task +0 -3
- data/bin/hdp-ls +0 -11
- data/bin/hdp-mkdir +0 -2
- data/bin/hdp-mkdirp +0 -12
- data/bin/hdp-mv +0 -3
- data/bin/hdp-parts_to_keys.rb +0 -77
- data/bin/hdp-ps +0 -3
- data/bin/hdp-put +0 -3
- data/bin/hdp-rm +0 -32
- data/bin/hdp-sort +0 -40
- data/bin/hdp-stream +0 -40
- data/bin/hdp-stream-flat +0 -22
- data/bin/hdp-stream2 +0 -39
- data/bin/hdp-sync +0 -17
- data/bin/hdp-wc +0 -67
- data/bin/wu-flow +0 -10
- data/bin/wu-map +0 -17
- data/bin/wu-red +0 -17
- data/bin/wukong +0 -17
- data/data/CREDITS.md +0 -355
- data/data/graph/airfares.tsv +0 -2174
- data/data/text/gift_of_the_magi.txt +0 -225
- data/data/text/jabberwocky.txt +0 -36
- data/data/text/rectification_of_names.txt +0 -33
- data/data/twitter/a_atsigns_b.tsv +0 -64
- data/data/twitter/a_follows_b.tsv +0 -53
- data/data/twitter/tweet.tsv +0 -167
- data/data/twitter/twitter_user.tsv +0 -55
- data/data/wikipedia/dbpedia-sentences.tsv +0 -1000
- data/docpages/INSTALL.textile +0 -92
- data/docpages/LICENSE.textile +0 -107
- data/docpages/README-elastic_map_reduce.textile +0 -377
- data/docpages/README-performance.textile +0 -90
- data/docpages/README-wulign.textile +0 -65
- data/docpages/UsingWukong-part1-get_ready.textile +0 -17
- data/docpages/UsingWukong-part2-ThinkingBigData.textile +0 -75
- data/docpages/UsingWukong-part3-parsing.textile +0 -138
- data/docpages/_config.yml +0 -39
- data/docpages/avro/avro_notes.textile +0 -56
- data/docpages/avro/performance.textile +0 -36
- data/docpages/avro/tethering.textile +0 -19
- data/docpages/bigdata-tips.textile +0 -143
- data/docpages/code/api_response_example.txt +0 -20
- data/docpages/code/parser_skeleton.rb +0 -38
- data/docpages/diagrams/MapReduceDiagram.graffle +0 -0
- data/docpages/favicon.ico +0 -0
- data/docpages/gem.css +0 -16
- data/docpages/hadoop-tips.textile +0 -83
- data/docpages/index.textile +0 -92
- data/docpages/intro.textile +0 -8
- data/docpages/moreinfo.textile +0 -174
- data/docpages/news.html +0 -24
- data/docpages/pig/PigLatinExpressionsList.txt +0 -122
- data/docpages/pig/PigLatinReferenceManual.txt +0 -1640
- data/docpages/pig/commandline_params.txt +0 -26
- data/docpages/pig/cookbook.html +0 -481
- data/docpages/pig/images/hadoop-logo.jpg +0 -0
- data/docpages/pig/images/instruction_arrow.png +0 -0
- data/docpages/pig/images/pig-logo.gif +0 -0
- data/docpages/pig/piglatin_ref1.html +0 -1103
- data/docpages/pig/piglatin_ref2.html +0 -14340
- data/docpages/pig/setup.html +0 -505
- data/docpages/pig/skin/basic.css +0 -166
- data/docpages/pig/skin/breadcrumbs.js +0 -237
- data/docpages/pig/skin/fontsize.js +0 -166
- data/docpages/pig/skin/getBlank.js +0 -40
- data/docpages/pig/skin/getMenu.js +0 -45
- data/docpages/pig/skin/images/chapter.gif +0 -0
- data/docpages/pig/skin/images/chapter_open.gif +0 -0
- data/docpages/pig/skin/images/current.gif +0 -0
- data/docpages/pig/skin/images/external-link.gif +0 -0
- data/docpages/pig/skin/images/header_white_line.gif +0 -0
- data/docpages/pig/skin/images/page.gif +0 -0
- data/docpages/pig/skin/images/pdfdoc.gif +0 -0
- data/docpages/pig/skin/images/rc-b-l-15-1body-2menu-3menu.png +0 -0
- data/docpages/pig/skin/images/rc-b-r-15-1body-2menu-3menu.png +0 -0
- data/docpages/pig/skin/images/rc-b-r-5-1header-2tab-selected-3tab-selected.png +0 -0
- data/docpages/pig/skin/images/rc-t-l-5-1header-2searchbox-3searchbox.png +0 -0
- data/docpages/pig/skin/images/rc-t-l-5-1header-2tab-selected-3tab-selected.png +0 -0
- data/docpages/pig/skin/images/rc-t-l-5-1header-2tab-unselected-3tab-unselected.png +0 -0
- data/docpages/pig/skin/images/rc-t-r-15-1body-2menu-3menu.png +0 -0
- data/docpages/pig/skin/images/rc-t-r-5-1header-2searchbox-3searchbox.png +0 -0
- data/docpages/pig/skin/images/rc-t-r-5-1header-2tab-selected-3tab-selected.png +0 -0
- data/docpages/pig/skin/images/rc-t-r-5-1header-2tab-unselected-3tab-unselected.png +0 -0
- data/docpages/pig/skin/print.css +0 -54
- data/docpages/pig/skin/profile.css +0 -181
- data/docpages/pig/skin/screen.css +0 -587
- data/docpages/pig/tutorial.html +0 -1059
- data/docpages/pig/udf.html +0 -1509
- data/docpages/tutorial.textile +0 -283
- data/docpages/usage.textile +0 -195
- data/docpages/wutils.textile +0 -263
- data/examples/dataflow/complex.rb +0 -11
- data/examples/dataflow/donuts.rb +0 -13
- data/examples/tiny_count/jabberwocky_output.tsv +0 -92
- data/examples/word_count.rb +0 -48
- data/examples/workflow/fiddle.rb +0 -24
- data/lib/away/escapement.rb +0 -129
- data/lib/away/exe.rb +0 -11
- data/lib/away/experimental.rb +0 -5
- data/lib/away/from_file.rb +0 -52
- data/lib/away/job.rb +0 -56
- data/lib/away/job/rake_compat.rb +0 -17
- data/lib/away/registry.rb +0 -79
- data/lib/away/runner.rb +0 -276
- data/lib/away/runner/execute.rb +0 -121
- data/lib/away/script.rb +0 -161
- data/lib/away/script/hadoop_command.rb +0 -240
- data/lib/away/source/file_list_source.rb +0 -15
- data/lib/away/source/looper.rb +0 -18
- data/lib/away/task.rb +0 -219
- data/lib/hanuman/action.rb +0 -21
- data/lib/hanuman/chain.rb +0 -4
- data/lib/hanuman/graphviz.rb +0 -74
- data/lib/hanuman/resource.rb +0 -6
- data/lib/hanuman/slot.rb +0 -87
- data/lib/hanuman/slottable.rb +0 -220
- data/lib/wukong/bad_record.rb +0 -15
- data/lib/wukong/event.rb +0 -44
- data/lib/wukong/local_runner.rb +0 -55
- data/lib/wukong/mapred.rb +0 -3
- data/lib/wukong/universe.rb +0 -48
- data/lib/wukong/widget/filter.rb +0 -81
- data/lib/wukong/widget/gibberish.rb +0 -123
- data/lib/wukong/widget/monitor.rb +0 -26
- data/lib/wukong/widget/reducer.rb +0 -66
- data/lib/wukong/widget/stringifier.rb +0 -50
- data/lib/wukong/workflow.rb +0 -22
- data/lib/wukong/workflow/command.rb +0 -42
- data/old/config/emr-example.yaml +0 -48
- data/old/examples/README.txt +0 -17
- data/old/examples/contrib/jeans/README.markdown +0 -165
- data/old/examples/contrib/jeans/data/normalized_sizes +0 -3
- data/old/examples/contrib/jeans/data/orders.tsv +0 -1302
- data/old/examples/contrib/jeans/data/sizes +0 -3
- data/old/examples/contrib/jeans/normalize.rb +0 -20
- data/old/examples/contrib/jeans/sizes.rb +0 -55
- data/old/examples/corpus/bnc_word_freq.rb +0 -44
- data/old/examples/corpus/bucket_counter.rb +0 -47
- data/old/examples/corpus/dbpedia_abstract_to_sentences.rb +0 -86
- data/old/examples/corpus/sentence_bigrams.rb +0 -53
- data/old/examples/corpus/sentence_coocurrence.rb +0 -66
- data/old/examples/corpus/stopwords.rb +0 -138
- data/old/examples/corpus/words_to_bigrams.rb +0 -53
- data/old/examples/emr/README.textile +0 -110
- data/old/examples/emr/dot_wukong_dir/credentials.json +0 -7
- data/old/examples/emr/dot_wukong_dir/emr.yaml +0 -69
- data/old/examples/emr/dot_wukong_dir/emr_bootstrap.sh +0 -33
- data/old/examples/emr/elastic_mapreduce_example.rb +0 -28
- data/old/examples/network_graph/adjacency_list.rb +0 -74
- data/old/examples/network_graph/breadth_first_search.rb +0 -72
- data/old/examples/network_graph/gen_2paths.rb +0 -68
- data/old/examples/network_graph/gen_multi_edge.rb +0 -112
- data/old/examples/network_graph/gen_symmetric_links.rb +0 -64
- data/old/examples/pagerank/README.textile +0 -6
- data/old/examples/pagerank/gen_initial_pagerank_graph.pig +0 -57
- data/old/examples/pagerank/pagerank.rb +0 -72
- data/old/examples/pagerank/pagerank_initialize.rb +0 -42
- data/old/examples/pagerank/run_pagerank.sh +0 -21
- data/old/examples/sample_records.rb +0 -33
- data/old/examples/server_logs/apache_log_parser.rb +0 -15
- data/old/examples/server_logs/nook.rb +0 -48
- data/old/examples/server_logs/nook/faraday_dummy_adapter.rb +0 -94
- data/old/examples/server_logs/user_agent.rb +0 -40
- data/old/examples/simple_word_count.rb +0 -82
- data/old/examples/size.rb +0 -61
- data/old/examples/stats/avg_value_frequency.rb +0 -86
- data/old/examples/stats/binning_percentile_estimator.rb +0 -140
- data/old/examples/stats/data/avg_value_frequency.tsv +0 -3
- data/old/examples/stats/rank_and_bin.rb +0 -173
- data/old/examples/stupidly_simple_filter.rb +0 -40
- data/old/examples/word_count.rb +0 -75
- data/old/graph/graphviz_builder.rb +0 -580
- data/old/graph_easy/Attributes.pm +0 -4181
- data/old/graph_easy/Graphviz.pm +0 -2232
- data/old/wukong.rb +0 -18
- data/old/wukong/and_pig.rb +0 -38
- data/old/wukong/bad_record.rb +0 -18
- data/old/wukong/datatypes.rb +0 -24
- data/old/wukong/datatypes/enum.rb +0 -127
- data/old/wukong/datatypes/fake_types.rb +0 -17
- data/old/wukong/decorator.rb +0 -28
- data/old/wukong/encoding/asciize.rb +0 -108
- data/old/wukong/extensions.rb +0 -16
- data/old/wukong/extensions/array.rb +0 -18
- data/old/wukong/extensions/blank.rb +0 -93
- data/old/wukong/extensions/class.rb +0 -189
- data/old/wukong/extensions/date_time.rb +0 -53
- data/old/wukong/extensions/emittable.rb +0 -69
- data/old/wukong/extensions/enumerable.rb +0 -79
- data/old/wukong/extensions/hash.rb +0 -167
- data/old/wukong/extensions/hash_keys.rb +0 -16
- data/old/wukong/extensions/hash_like.rb +0 -150
- data/old/wukong/extensions/hashlike_class.rb +0 -47
- data/old/wukong/extensions/module.rb +0 -2
- data/old/wukong/extensions/pathname.rb +0 -27
- data/old/wukong/extensions/string.rb +0 -65
- data/old/wukong/extensions/struct.rb +0 -17
- data/old/wukong/extensions/symbol.rb +0 -11
- data/old/wukong/filename_pattern.rb +0 -74
- data/old/wukong/helper.rb +0 -7
- data/old/wukong/helper/stopwords.rb +0 -195
- data/old/wukong/helper/tokenize.rb +0 -35
- data/old/wukong/logger.rb +0 -38
- data/old/wukong/periodic_monitor.rb +0 -72
- data/old/wukong/schema.rb +0 -269
- data/old/wukong/script.rb +0 -286
- data/old/wukong/script/avro_command.rb +0 -5
- data/old/wukong/script/cassandra_loader_script.rb +0 -40
- data/old/wukong/script/emr_command.rb +0 -168
- data/old/wukong/script/hadoop_command.rb +0 -237
- data/old/wukong/script/local_command.rb +0 -41
- data/old/wukong/store.rb +0 -10
- data/old/wukong/store/base.rb +0 -27
- data/old/wukong/store/cassandra.rb +0 -10
- data/old/wukong/store/cassandra/streaming.rb +0 -75
- data/old/wukong/store/cassandra/struct_loader.rb +0 -21
- data/old/wukong/store/cassandra_model.rb +0 -91
- data/old/wukong/store/chh_chunked_flat_file_store.rb +0 -37
- data/old/wukong/store/chunked_flat_file_store.rb +0 -48
- data/old/wukong/store/conditional_store.rb +0 -57
- data/old/wukong/store/factory.rb +0 -8
- data/old/wukong/store/flat_file_store.rb +0 -89
- data/old/wukong/store/key_store.rb +0 -51
- data/old/wukong/store/null_store.rb +0 -15
- data/old/wukong/store/read_thru_store.rb +0 -22
- data/old/wukong/store/tokyo_tdb_key_store.rb +0 -33
- data/old/wukong/store/tyrant_rdb_key_store.rb +0 -57
- data/old/wukong/store/tyrant_tdb_key_store.rb +0 -20
- data/old/wukong/streamer.rb +0 -30
- data/old/wukong/streamer/accumulating_reducer.rb +0 -83
- data/old/wukong/streamer/base.rb +0 -126
- data/old/wukong/streamer/counting_reducer.rb +0 -25
- data/old/wukong/streamer/filter.rb +0 -20
- data/old/wukong/streamer/instance_streamer.rb +0 -15
- data/old/wukong/streamer/json_streamer.rb +0 -21
- data/old/wukong/streamer/line_streamer.rb +0 -12
- data/old/wukong/streamer/list_reducer.rb +0 -31
- data/old/wukong/streamer/rank_and_bin_reducer.rb +0 -145
- data/old/wukong/streamer/record_streamer.rb +0 -14
- data/old/wukong/streamer/reducer.rb +0 -11
- data/old/wukong/streamer/set_reducer.rb +0 -14
- data/old/wukong/streamer/struct_streamer.rb +0 -48
- data/old/wukong/streamer/summing_reducer.rb +0 -29
- data/old/wukong/streamer/uniq_by_last_reducer.rb +0 -51
- data/old/wukong/typed_struct.rb +0 -12
- data/spec/away/encoding_spec.rb +0 -32
- data/spec/away/exe_spec.rb +0 -20
- data/spec/away/flow_spec.rb +0 -82
- data/spec/away/graph_spec.rb +0 -6
- data/spec/away/job_spec.rb +0 -15
- data/spec/away/rake_compat_spec.rb +0 -9
- data/spec/away/script_spec.rb +0 -81
- data/spec/hanuman/graphviz_spec.rb +0 -29
- data/spec/hanuman/slot_spec.rb +0 -2
- data/spec/support/examples_helper.rb +0 -10
- data/spec/support/streamer_test_helpers.rb +0 -6
- data/spec/support/wukong_widget_helpers.rb +0 -66
- data/spec/wukong/processor_spec.rb +0 -109
- data/spec/wukong/widget/filter_spec.rb +0 -99
- data/spec/wukong/widget/stringifier_spec.rb +0 -51
- data/spec/wukong/workflow/command_spec.rb +0 -5
@@ -1,189 +0,0 @@
|
|
1
|
-
#
|
2
|
-
# This is taken in whole from the extlib gem. Thanks y'all.
|
3
|
-
#
|
4
|
-
|
5
|
-
|
6
|
-
# Copyright (c) 2004-2008 David Heinemeier Hansson
|
7
|
-
#
|
8
|
-
# Permission is hereby granted, free of charge, to any person obtaining
|
9
|
-
# a copy of this software and associated documentation files (the
|
10
|
-
# "Software"), to deal in the Software without restriction, including
|
11
|
-
# without limitation the rights to use, copy, modify, merge, publish,
|
12
|
-
# distribute, sublicense, and/or sell copies of the Software, and to
|
13
|
-
# permit persons to whom the Software is furnished to do so, subject to
|
14
|
-
# the following conditions:
|
15
|
-
#
|
16
|
-
# The above copyright notice and this permission notice shall be
|
17
|
-
# included in all copies or substantial portions of the Software.
|
18
|
-
#
|
19
|
-
# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
|
20
|
-
# EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
|
21
|
-
# MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
|
22
|
-
# NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE
|
23
|
-
# LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
|
24
|
-
# OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
|
25
|
-
# WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
|
26
|
-
|
27
|
-
# Allows attributes to be shared within an inheritance hierarchy, but where
|
28
|
-
# each descendant gets a copy of their parents' attributes, instead of just a
|
29
|
-
# pointer to the same. This means that the child can add elements to, for
|
30
|
-
# example, an array without those additions being shared with either their
|
31
|
-
# parent, siblings, or children, which is unlike the regular class-level
|
32
|
-
# attributes that are shared across the entire hierarchy.
|
33
|
-
class Class
|
34
|
-
# Defines class-level and instance-level attribute reader.
|
35
|
-
#
|
36
|
-
# @param *syms<Array> Array of attributes to define reader for.
|
37
|
-
# @return <Array[#to_s]> List of attributes that were made into cattr_readers
|
38
|
-
#
|
39
|
-
# @api public
|
40
|
-
#
|
41
|
-
# @todo Is this inconsistent in that it does not allow you to prevent
|
42
|
-
# an instance_reader via :instance_reader => false
|
43
|
-
def cattr_reader(*syms)
|
44
|
-
syms.flatten.each do |sym|
|
45
|
-
next if sym.is_a?(Hash)
|
46
|
-
class_eval(<<-RUBY, __FILE__, __LINE__ + 1)
|
47
|
-
unless defined? @@#{sym}
|
48
|
-
@@#{sym} = nil
|
49
|
-
end
|
50
|
-
|
51
|
-
def self.#{sym}
|
52
|
-
@@#{sym}
|
53
|
-
end
|
54
|
-
|
55
|
-
def #{sym}
|
56
|
-
@@#{sym}
|
57
|
-
end
|
58
|
-
RUBY
|
59
|
-
end
|
60
|
-
end
|
61
|
-
|
62
|
-
# Defines class-level (and optionally instance-level) attribute writer.
|
63
|
-
#
|
64
|
-
# @param <Array[*#to_s, Hash{:instance_writer => Boolean}]> Array of attributes to define writer for.
|
65
|
-
# @option syms :instance_writer<Boolean> if true, instance-level attribute writer is defined.
|
66
|
-
# @return <Array[#to_s]> List of attributes that were made into cattr_writers
|
67
|
-
#
|
68
|
-
# @api public
|
69
|
-
def cattr_writer(*syms)
|
70
|
-
options = syms.last.is_a?(Hash) ? syms.pop : {}
|
71
|
-
syms.flatten.each do |sym|
|
72
|
-
class_eval(<<-RUBY, __FILE__, __LINE__ + 1)
|
73
|
-
unless defined? @@#{sym}
|
74
|
-
@@#{sym} = nil
|
75
|
-
end
|
76
|
-
|
77
|
-
def self.#{sym}=(obj)
|
78
|
-
@@#{sym} = obj
|
79
|
-
end
|
80
|
-
RUBY
|
81
|
-
|
82
|
-
unless options[:instance_writer] == false
|
83
|
-
class_eval(<<-RUBY, __FILE__, __LINE__ + 1)
|
84
|
-
def #{sym}=(obj)
|
85
|
-
@@#{sym} = obj
|
86
|
-
end
|
87
|
-
RUBY
|
88
|
-
end
|
89
|
-
end
|
90
|
-
end
|
91
|
-
|
92
|
-
# Defines class-level (and optionally instance-level) attribute accessor.
|
93
|
-
#
|
94
|
-
# @param *syms<Array[*#to_s, Hash{:instance_writer => Boolean}]> Array of attributes to define accessor for.
|
95
|
-
# @option syms :instance_writer<Boolean> if true, instance-level attribute writer is defined.
|
96
|
-
# @return <Array[#to_s]> List of attributes that were made into accessors
|
97
|
-
#
|
98
|
-
# @api public
|
99
|
-
def cattr_accessor(*syms)
|
100
|
-
cattr_reader(*syms)
|
101
|
-
cattr_writer(*syms)
|
102
|
-
end
|
103
|
-
|
104
|
-
# Defines class-level inheritable attribute reader. Attributes are available to subclasses,
|
105
|
-
# each subclass has a copy of parent's attribute.
|
106
|
-
#
|
107
|
-
# @param *syms<Array[#to_s]> Array of attributes to define inheritable reader for.
|
108
|
-
# @return <Array[#to_s]> Array of attributes converted into inheritable_readers.
|
109
|
-
#
|
110
|
-
# @api public
|
111
|
-
#
|
112
|
-
# @todo Do we want to block instance_reader via :instance_reader => false
|
113
|
-
# @todo It would be preferable that we do something with a Hash passed in
|
114
|
-
# (error out or do the same as other methods above) instead of silently
|
115
|
-
# moving on). In particular, this makes the return value of this function
|
116
|
-
# less useful.
|
117
|
-
def class_inheritable_reader(*ivars)
|
118
|
-
instance_reader = ivars.pop[:reader] if ivars.last.is_a?(Hash)
|
119
|
-
|
120
|
-
ivars.each do |ivar|
|
121
|
-
self.class_eval <<-RUBY, __FILE__, __LINE__ + 1
|
122
|
-
def self.#{ivar}
|
123
|
-
return @#{ivar} if defined?(@#{ivar})
|
124
|
-
return nil if self.object_id == #{self.object_id}
|
125
|
-
ivar = superclass.#{ivar}
|
126
|
-
return nil if ivar.nil?
|
127
|
-
@#{ivar} = ivar.try_dup
|
128
|
-
end
|
129
|
-
RUBY
|
130
|
-
|
131
|
-
unless instance_reader == false
|
132
|
-
self.class_eval <<-RUBY, __FILE__, __LINE__ + 1
|
133
|
-
def #{ivar}
|
134
|
-
self.class.#{ivar}
|
135
|
-
end
|
136
|
-
RUBY
|
137
|
-
end
|
138
|
-
end
|
139
|
-
end
|
140
|
-
|
141
|
-
# Defines class-level inheritable attribute writer. Attributes are available to subclasses,
|
142
|
-
# each subclass has a copy of parent's attribute.
|
143
|
-
#
|
144
|
-
# @param *syms<Array[*#to_s, Hash{:instance_writer => Boolean}]> Array of attributes to
|
145
|
-
# define inheritable writer for.
|
146
|
-
# @option syms :instance_writer<Boolean> if true, instance-level inheritable attribute writer is defined.
|
147
|
-
# @return <Array[#to_s]> An Array of the attributes that were made into inheritable writers.
|
148
|
-
#
|
149
|
-
# @api public
|
150
|
-
#
|
151
|
-
# @todo We need a style for class_eval <<-HEREDOC. I'd like to make it
|
152
|
-
# class_eval(<<-RUBY, __FILE__, __LINE__), but we should codify it somewhere.
|
153
|
-
def class_inheritable_writer(*ivars)
|
154
|
-
instance_writer = ivars.pop[:instance_writer] if ivars.last.is_a?(Hash)
|
155
|
-
ivars.each do |ivar|
|
156
|
-
self.class_eval <<-RUBY, __FILE__, __LINE__ + 1
|
157
|
-
def self.#{ivar}=(obj)
|
158
|
-
@#{ivar} = obj
|
159
|
-
end
|
160
|
-
RUBY
|
161
|
-
unless instance_writer == false
|
162
|
-
self.class_eval <<-RUBY, __FILE__, __LINE__ + 1
|
163
|
-
def #{ivar}=(obj) self.class.#{ivar} = obj end
|
164
|
-
RUBY
|
165
|
-
end
|
166
|
-
end
|
167
|
-
end
|
168
|
-
|
169
|
-
# Defines class-level inheritable attribute accessor. Attributes are available to subclasses,
|
170
|
-
# each subclass has a copy of parent's attribute.
|
171
|
-
#
|
172
|
-
# @param *syms<Array[*#to_s, Hash{:instance_writer => Boolean}]> Array of attributes to
|
173
|
-
# define inheritable accessor for.
|
174
|
-
# @option syms :instance_writer<Boolean> if true, instance-level inheritable attribute writer is defined.
|
175
|
-
# @return <Array[#to_s]> An Array of attributes turned into inheritable accessors.
|
176
|
-
#
|
177
|
-
# @api public
|
178
|
-
def class_inheritable_accessor(*syms)
|
179
|
-
class_inheritable_reader(*syms)
|
180
|
-
class_inheritable_writer(*syms)
|
181
|
-
end
|
182
|
-
end
|
183
|
-
|
184
|
-
|
185
|
-
class Fixnum
|
186
|
-
def try_dup
|
187
|
-
self
|
188
|
-
end
|
189
|
-
end
|
@@ -1,53 +0,0 @@
|
|
1
|
-
require 'time'
|
2
|
-
require 'date'
|
3
|
-
|
4
|
-
class Time
|
5
|
-
# strftime() format to flatten a date
|
6
|
-
FLAT_FORMAT = "%Y%m%d%H%M%S"
|
7
|
-
# Flatten
|
8
|
-
def to_flat
|
9
|
-
utc.strftime(FLAT_FORMAT)
|
10
|
-
end
|
11
|
-
|
12
|
-
#
|
13
|
-
# Parses the time but never fails.
|
14
|
-
# Return value is always in the UTC time zone.
|
15
|
-
#
|
16
|
-
# A flattened datetime -- a 14-digit YYYYmmddHHMMMSS -- is fixed to the UTC
|
17
|
-
# time zone by parsing it as YYYYmmddHHMMMSSZ <- 'Z' at end
|
18
|
-
#
|
19
|
-
def self.parse_safely dt
|
20
|
-
return nil if dt.blank?
|
21
|
-
begin
|
22
|
-
case
|
23
|
-
when dt.is_a?(Time) then dt.utc
|
24
|
-
when (dt.to_s =~ /\A\d{14}\z/) then parse(dt.to_s+'Z', true)
|
25
|
-
else parse(dt.to_s, true).utc
|
26
|
-
end
|
27
|
-
rescue StandardError => e
|
28
|
-
Log.debug e
|
29
|
-
end
|
30
|
-
end
|
31
|
-
|
32
|
-
def self.parse_and_flatten str
|
33
|
-
parse_safely(str).to_flat
|
34
|
-
end
|
35
|
-
end
|
36
|
-
|
37
|
-
class DateTime < Date
|
38
|
-
# strftime() format to flatten a date
|
39
|
-
FLAT_FORMAT = "%Y%m%d%H%M%S"
|
40
|
-
# Flatten
|
41
|
-
def to_flat
|
42
|
-
strftime(FLAT_FORMAT)
|
43
|
-
end
|
44
|
-
end
|
45
|
-
|
46
|
-
class Date
|
47
|
-
# strftime() format to flatten a date
|
48
|
-
FLAT_FORMAT = "%Y%m%d"
|
49
|
-
# Flatten
|
50
|
-
def to_flat
|
51
|
-
strftime(FLAT_FORMAT)
|
52
|
-
end
|
53
|
-
end
|
@@ -1,69 +0,0 @@
|
|
1
|
-
Object.class_eval do
|
2
|
-
def to_flat() [to_s] end
|
3
|
-
end
|
4
|
-
|
5
|
-
module Enumerable
|
6
|
-
alias_method :to_flat, :to_a
|
7
|
-
end
|
8
|
-
|
9
|
-
Struct.class_eval do
|
10
|
-
#
|
11
|
-
# The last portion of the class in underscored form
|
12
|
-
# note memoization
|
13
|
-
#
|
14
|
-
def self.resource_name
|
15
|
-
@resource_name ||= self.to_s.gsub(%r{.*::}, '').underscore.to_sym
|
16
|
-
end
|
17
|
-
#
|
18
|
-
# Flatten for packing as resource name followed by all fields
|
19
|
-
#
|
20
|
-
def to_flat include_key=false
|
21
|
-
if include_key.is_a? Proc
|
22
|
-
sort_key = include_key.call(self)
|
23
|
-
elsif (! include_key.blank?) && respond_to?(:key)
|
24
|
-
sort_key = [self.class.resource_name, key].flatten.join("-")
|
25
|
-
else
|
26
|
-
sort_key = self.class.resource_name
|
27
|
-
end
|
28
|
-
[sort_key, *to_a] # .map(&:to_flat).flatten
|
29
|
-
end
|
30
|
-
end
|
31
|
-
|
32
|
-
module HashLike
|
33
|
-
#
|
34
|
-
# Flatten for packing as resource name followed by all fields
|
35
|
-
#
|
36
|
-
def to_flat include_key=true
|
37
|
-
if include_key.is_a? Proc
|
38
|
-
sort_key = include_key.call(self)
|
39
|
-
elsif include_key && respond_to?(:key)
|
40
|
-
sort_key = [self.class.resource_name, key].flatten.join("-")
|
41
|
-
else
|
42
|
-
sort_key = self.class.resource_name
|
43
|
-
end
|
44
|
-
[sort_key, *to_a] # .map(&:to_flat).flatten
|
45
|
-
end
|
46
|
-
end
|
47
|
-
|
48
|
-
Hash.class_eval do
|
49
|
-
def to_flat
|
50
|
-
map do |k, v|
|
51
|
-
[k.to_flat, v.to_flat].join(":")
|
52
|
-
end
|
53
|
-
end
|
54
|
-
end
|
55
|
-
|
56
|
-
class Integer
|
57
|
-
#
|
58
|
-
# Express boolean as 1 (true) or 0 (false). In contravention of typical ruby
|
59
|
-
# semantics (but in a way that is more robust for wukong-like batch
|
60
|
-
# processing), the number 0, the string '0', nil and false are all considered
|
61
|
-
# false. (This also makes the method idempotent: repeated calls give same result.)
|
62
|
-
#
|
63
|
-
def self.unbooleanize bool
|
64
|
-
case bool
|
65
|
-
when 0, '0', false, nil then 0
|
66
|
-
else 1
|
67
|
-
end
|
68
|
-
end
|
69
|
-
end
|
@@ -1,79 +0,0 @@
|
|
1
|
-
module Enumerable
|
2
|
-
|
3
|
-
# Calculates a sum from the elements. Examples:
|
4
|
-
#
|
5
|
-
# payments.sum { |p| p.price * p.tax_rate }
|
6
|
-
# payments.sum(&:price)
|
7
|
-
#
|
8
|
-
# The latter is a shortcut for:
|
9
|
-
#
|
10
|
-
# payments.inject { |sum, p| sum + p.price }
|
11
|
-
#
|
12
|
-
# It can also calculate the sum without the use of a block.
|
13
|
-
#
|
14
|
-
# [5, 15, 10].sum # => 30
|
15
|
-
# ["foo", "bar"].sum # => "foobar"
|
16
|
-
# [[1, 2], [3, 1, 5]].sum => [1, 2, 3, 1, 5]
|
17
|
-
#
|
18
|
-
# The default sum of an empty list is zero. You can override this default:
|
19
|
-
#
|
20
|
-
# [].sum(Payment.new(0)) { |i| i.amount } # => Payment.new(0)
|
21
|
-
#
|
22
|
-
def sum(identity = 0, &block)
|
23
|
-
if block_given?
|
24
|
-
map(&block).sum(identity)
|
25
|
-
else
|
26
|
-
inject { |sum, element| sum + element } || identity
|
27
|
-
end
|
28
|
-
end
|
29
|
-
|
30
|
-
# Iterates over a collection, passing the current element *and* the
|
31
|
-
# +memo+ to the block. Handy for building up hashes or
|
32
|
-
# reducing collections down to one object. Examples:
|
33
|
-
#
|
34
|
-
# %w(foo bar).each_with_object({}) { |str, hsh| hsh[str] = str.upcase } #=> {'foo' => 'FOO', 'bar' => 'BAR'}
|
35
|
-
#
|
36
|
-
# *Note* that you can't use immutable objects like numbers, true or false as
|
37
|
-
# the memo. You would think the following returns 120, but since the memo is
|
38
|
-
# never changed, it does not.
|
39
|
-
#
|
40
|
-
# (1..5).each_with_object(1) { |value, memo| memo *= value } # => 1
|
41
|
-
#
|
42
|
-
def each_with_object(memo, &block)
|
43
|
-
returning memo do |m|
|
44
|
-
each do |element|
|
45
|
-
block.call(element, m)
|
46
|
-
end
|
47
|
-
end
|
48
|
-
end unless [].respond_to?(:each_with_object)
|
49
|
-
|
50
|
-
# Convert an enumerable to a hash. Examples:
|
51
|
-
#
|
52
|
-
# people.index_by(&:login)
|
53
|
-
# => { "nextangle" => <Person ...>, "chade-" => <Person ...>, ...}
|
54
|
-
# people.index_by { |person| "#{person.first_name} #{person.last_name}" }
|
55
|
-
# => { "Chade- Fowlersburg-e" => <Person ...>, "David Heinemeier Hansson" => <Person ...>, ...}
|
56
|
-
#
|
57
|
-
def index_by
|
58
|
-
inject({}) do |accum, elem|
|
59
|
-
accum[yield(elem)] = elem
|
60
|
-
accum
|
61
|
-
end
|
62
|
-
end
|
63
|
-
|
64
|
-
# Returns true if the collection has more than 1 element. Functionally equivalent to collection.size > 1.
|
65
|
-
# Works with a block too ala any?, so people.many? { |p| p.age > 26 } # => returns true if more than 1 person is over 26.
|
66
|
-
def many?(&block)
|
67
|
-
size = block_given? ? select(&block).size : self.size
|
68
|
-
size > 1
|
69
|
-
end
|
70
|
-
|
71
|
-
# Returns true if none of the elements match the given block.
|
72
|
-
#
|
73
|
-
# success = responses.none? {|r| r.status / 100 == 5 }
|
74
|
-
#
|
75
|
-
# This is a builtin method in Ruby 1.8.7 and later.
|
76
|
-
def none?(&block)
|
77
|
-
!any?(&block)
|
78
|
-
end unless [].respond_to?(:none?)
|
79
|
-
end
|
@@ -1,167 +0,0 @@
|
|
1
|
-
#
|
2
|
-
# h2. extensions/hash.rb -- hash extensions
|
3
|
-
#
|
4
|
-
|
5
|
-
require 'set'
|
6
|
-
class Hash
|
7
|
-
|
8
|
-
# Slice a hash to include only the given keys. This is useful for
|
9
|
-
# limiting an options hash to valid keys before passing to a method:
|
10
|
-
#
|
11
|
-
# def search(criteria = {})
|
12
|
-
# assert_valid_keys(:mass, :velocity, :time)
|
13
|
-
# end
|
14
|
-
#
|
15
|
-
# search(options.slice(:mass, :velocity, :time))
|
16
|
-
# Returns a new hash with only the given keys.
|
17
|
-
def slice(*keys)
|
18
|
-
allowed = Set.new(respond_to?(:convert_key) ? keys.map { |key| convert_key(key) } : keys)
|
19
|
-
reject{|key,| !allowed.include?(key) }
|
20
|
-
end
|
21
|
-
#
|
22
|
-
# Replace the hash with only the given keys.
|
23
|
-
#
|
24
|
-
def slice!(*keys)
|
25
|
-
replace(slice(*keys))
|
26
|
-
end
|
27
|
-
#
|
28
|
-
# #values_of is an alias for #values_at, but can be called on a Hash, a
|
29
|
-
# Struct, or an instance of a class that includes HashLike
|
30
|
-
#
|
31
|
-
alias_method :values_of, :values_at
|
32
|
-
|
33
|
-
#
|
34
|
-
# Create a hash from an array of keys and corresponding values.
|
35
|
-
#
|
36
|
-
def self.zip(keys, values, default=nil, &block)
|
37
|
-
hash = block_given? ? Hash.new(&block) : Hash.new(default)
|
38
|
-
keys.zip(values){|key,val| hash[key]=val }
|
39
|
-
hash
|
40
|
-
end
|
41
|
-
|
42
|
-
# lambda for recursive merges
|
43
|
-
::Hash::DEEP_MERGER = proc do |key,v1,v2|
|
44
|
-
(v1.respond_to?(:merge) && v2.respond_to?(:merge)) ? v1.merge(v2.compact, &Hash::DEEP_MERGER) : (v2.nil? ? v1 : v2)
|
45
|
-
end unless defined?(::Hash::DEEP_MERGER)
|
46
|
-
|
47
|
-
#
|
48
|
-
# Merge hashes recursively.
|
49
|
-
# Nothing special happens to array values
|
50
|
-
#
|
51
|
-
# x = { :subhash => { 1 => :val_from_x, 222 => :only_in_x, 333 => :only_in_x }, :scalar => :scalar_from_x}
|
52
|
-
# y = { :subhash => { 1 => :val_from_y, 999 => :only_in_y }, :scalar => :scalar_from_y }
|
53
|
-
# x.deep_merge y
|
54
|
-
# => {:subhash=>{1=>:val_from_y, 222=>:only_in_x, 333=>:only_in_x, 999=>:only_in_y}, :scalar=>:scalar_from_y}
|
55
|
-
# y.deep_merge x
|
56
|
-
# => {:subhash=>{1=>:val_from_x, 222=>:only_in_x, 333=>:only_in_x, 999=>:only_in_y}, :scalar=>:scalar_from_x}
|
57
|
-
#
|
58
|
-
# Nil values always lose.
|
59
|
-
#
|
60
|
-
# x = {:subhash=>{:nil_in_x=>nil, 1=>:val1,}, :nil_in_x=>nil}
|
61
|
-
# y = {:subhash=>{:nil_in_x=>5}, :nil_in_x=>5}
|
62
|
-
# y.deep_merge x
|
63
|
-
# => {:subhash=>{1=>:val1, :nil_in_x=>5}, :nil_in_x=>5}
|
64
|
-
# x.deep_merge y
|
65
|
-
# => {:subhash=>{1=>:val1, :nil_in_x=>5}, :nil_in_x=>5}
|
66
|
-
#
|
67
|
-
def deep_merge hsh2
|
68
|
-
merge hsh2, &Hash::DEEP_MERGER
|
69
|
-
end unless method_defined?(:deep_merge)
|
70
|
-
|
71
|
-
def deep_merge! hsh2
|
72
|
-
merge! hsh2, &Hash::DEEP_MERGER
|
73
|
-
end unless method_defined?(:deep_merge!)
|
74
|
-
|
75
|
-
#
|
76
|
-
# Treat hash as tree of hashes:
|
77
|
-
#
|
78
|
-
# x = { 1 => :val, :subhash => { 1 => :val1 } }
|
79
|
-
# x.deep_set(:subhash, :cat, :hat)
|
80
|
-
# # => { 1 => :val, :subhash => { 1 => :val1, :cat => :hat } }
|
81
|
-
# x.deep_set(:subhash, 1, :newval)
|
82
|
-
# # => { 1 => :val, :subhash => { 1 => :newval, :cat => :hat } }
|
83
|
-
#
|
84
|
-
#
|
85
|
-
def deep_set *args
|
86
|
-
val = args.pop
|
87
|
-
last_key = args.pop
|
88
|
-
# dig down to last subtree (building out if necessary)
|
89
|
-
hsh = args.empty? ? self : args.inject(self){|h, k| h[k] ||= {} }
|
90
|
-
# set leaf value
|
91
|
-
hsh[last_key] = val
|
92
|
-
end unless method_defined?(:deep_set)
|
93
|
-
|
94
|
-
#
|
95
|
-
# Treat hash as tree of hashes:
|
96
|
-
#
|
97
|
-
# x = { 1 => :val, :subhash => { 1 => :val1 } }
|
98
|
-
# x.deep_get(:subhash, 1)
|
99
|
-
# # => :val
|
100
|
-
# x.deep_get(:subhash, 2)
|
101
|
-
# # => nil
|
102
|
-
# x.deep_get(:subhash, 2, 3)
|
103
|
-
# # => nil
|
104
|
-
# x.deep_get(:subhash, 2)
|
105
|
-
# # => nil
|
106
|
-
#
|
107
|
-
def deep_get *args
|
108
|
-
last_key = args.pop
|
109
|
-
# dig down to last subtree (building out if necessary)
|
110
|
-
hsh = args.inject(self){|h, k| h[k] || {} }
|
111
|
-
# get leaf value
|
112
|
-
hsh[last_key]
|
113
|
-
end unless method_defined?(:deep_get)
|
114
|
-
|
115
|
-
|
116
|
-
#
|
117
|
-
# Treat hash as tree of hashes:
|
118
|
-
#
|
119
|
-
# x = { 1 => :val, :subhash => { 1 => :val1, 2 => :val2 } }
|
120
|
-
# x.deep_delete(:subhash, 1)
|
121
|
-
# #=> :val
|
122
|
-
# x
|
123
|
-
# #=> { 1 => :val, :subhash => { 2 => :val2 } }
|
124
|
-
#
|
125
|
-
def deep_delete *args
|
126
|
-
last_key = args.pop
|
127
|
-
last_hsh = args.empty? ? self : (deep_get(*args)||{})
|
128
|
-
last_hsh.delete(last_key)
|
129
|
-
end unless method_defined?(:deep_delete)
|
130
|
-
|
131
|
-
#
|
132
|
-
# remove all key-value pairs where the value is nil
|
133
|
-
#
|
134
|
-
def compact
|
135
|
-
reject{|key,val| val.nil? }
|
136
|
-
end unless method_defined?(:compact)
|
137
|
-
#
|
138
|
-
# Replace the hash with its compacted self
|
139
|
-
#
|
140
|
-
def compact!
|
141
|
-
replace(compact)
|
142
|
-
end unless method_defined?(:compact!)
|
143
|
-
|
144
|
-
#
|
145
|
-
# remove all key-value pairs where the value is blank
|
146
|
-
#
|
147
|
-
def compact_blank
|
148
|
-
reject{|key,val| val.blank? }
|
149
|
-
end
|
150
|
-
#
|
151
|
-
# Replace the hash with its compact_blank'ed self
|
152
|
-
#
|
153
|
-
def compact_blank!
|
154
|
-
replace(compact_blank)
|
155
|
-
end
|
156
|
-
|
157
|
-
# Stolen from ActiveSupport::CoreExtensions::Hash::ReverseMerge.
|
158
|
-
def reverse_merge(other_hash)
|
159
|
-
other_hash.merge(self)
|
160
|
-
end
|
161
|
-
|
162
|
-
# Stolen from ActiveSupport::CoreExtensions::Hash::ReverseMerge.
|
163
|
-
def reverse_merge!(other_hash)
|
164
|
-
replace(reverse_merge(other_hash))
|
165
|
-
end
|
166
|
-
|
167
|
-
end
|