rbbt-util 5.26.92 → 5.26.93

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 2f6ffa8a8e6f7aae286ddb3a7c755ffe28e8af7415e1ed47340f80841a04e9b4
-  data.tar.gz: 6da506ba7e686cb1d6a450107a30eaf242ebfe09019f0441f410eaa0d79c892c
+  metadata.gz: 45fa23a77186b3e5f47cf3ccf754a4b2906329639ca1d188e2b9f347850abd27
+  data.tar.gz: 49b4d7b48339260536aebbd8c9f35e59155bf253a7a63685683db54967ca5359
 SHA512:
-  metadata.gz: 6da79d73a730e2fa80bb7c99d17955dd016ddc06a3a2f5a2ded36dcc9fb1b9d301941a0b10c9ac55b85298eca73a361845befb8a038e1f115b43bc0ed18a533f
-  data.tar.gz: 4e0340ba7dccc41c1c3a3f6997e20caf31676fcd915329f07a8e6a5640cd51b96458cc9a3ba307503bcb55478b383d842420027a3eb3d3f58f3db16b4fce26b5
+  metadata.gz: 37d1a5ac5167e6f4612413c86b03bb96babf71164c9c7433d14d841d3fee489ecc4abc6a613ea49f2580876cd78bd32e8a40e7437953efcbcf0e5464d1f11589
+  data.tar.gz: db18ad88a2c5eae7b478eeebde69ea498ffd8f123b8e1113e4e7dc91a21ed7689147436ac690d5b5f4777bf7cd2efa1227250fa3a477ef76ebca88c8add80a8c

data/lib/rbbt/tsv/stream.rb

@@ -66,21 +66,27 @@ module TSV
           sfields = sfields.collect{|f| [prefix, f] * ":" }
         end
 
-        lines         << parser.first_line
-        empty         << stream               if parser.first_line.nil?
+        first_line = parser.first_line
+        first_line = nil if first_line.empty?
+
+        lines         << first_line
         key_fields    << parser.key_field
         fields        << sfields
         sizes         << sfields.length if sfields
         input_options << parser.options
         preambles     << parser.preamble      if preamble and not parser.preamble.empty?
 
-        if fix_flat and parser.type == :flat and parser.first_line
-          parts = lines[-1].split("\t")
-          lines[-1] = [parts[0], (parts[1..-1] || [])*"|"] * "\t"
-          TSV.stream_flat2double(parser.stream, :noheader => true).stream
-        else
-          parser.stream
-        end
+        stream = if fix_flat and parser.type == :flat and first_line
+                   parts = lines[-1].nil? ? [] : lines[-1].split("\t")
+                   lines[-1] = [parts[0], (parts[1..-1] || [])*"|"] * "\t"
+                   TSV.stream_flat2double(parser.stream, :noheader => true).stream
+                 else
+                   parser.stream
+                 end
+
+        empty         << stream               if parser.first_line.nil? || parser.first_line.empty?
+
+        stream
       end
 
       key_field = key_fields.compact.first
@@ -121,7 +127,7 @@ module TSV
         keys = []
         parts = []
         lines.each_with_index do |line,i|
-          if line.nil?
+          if line.nil? || line.empty?
             keys[i] = nil
             parts[i] = nil
           else

data/lib/rbbt/workflow/accessor.rb

@@ -367,14 +367,15 @@ module Workflow
                        if Hash === d
                          d[:workflow] ||= wf 
                          d[:task] ||= task_name
-                         if override_dependencies[d[:workflow].to_s] && value = override_dependencies[d[:workflow].to_s][d[:task]]
-                           setup_override_dependency(value, d[:workflow], d[:task])
-                         else
-                           task_info = d[:workflow].task_info(d[:task])
-
-                           inputs = assign_dep_inputs({}, options.merge(d[:inputs] || {}), real_dependencies, task_info) 
-                           d = d[:workflow]._job(d[:task], d[:jobname], inputs) 
-                         end
+                         _override_dependencies = override_dependencies.merge(override_dependencies(d[:inputs] || {}))
+                         d = if _override_dependencies[d[:workflow].to_s] && value = _override_dependencies[d[:workflow].to_s][d[:task]]
+                               setup_override_dependency(value, d[:workflow], d[:task])
+                             else
+                               task_info = d[:workflow].task_info(d[:task])
+
+                               inputs = assign_dep_inputs({}, options.merge(d[:inputs] || {}), real_dependencies, task_info) 
+                               d[:workflow]._job(d[:task], d[:jobname], inputs) 
+                             end
                        end
                        ComputeDependency.setup(d, compute) if compute
                        new_ << d
@@ -385,7 +386,8 @@ module Workflow
                      dep = dependency.call jobname, _inputs, real_dependencies
                      if Hash === dep
                        dep[:workflow] ||= wf || self
-                       if override_dependencies[dep[:workflow].to_s] && value = override_dependencies[dep[:workflow].to_s][dep[:task]]
+                       _override_dependencies = override_dependencies.merge(override_dependencies(dep[:inputs] || {}))
+                       if _override_dependencies[dep[:workflow].to_s] && value = _override_dependencies[dep[:workflow].to_s][dep[:task]]
                          setup_override_dependency(value, dep[:workflow], dep[:task])
                        else
                          task_info = (dep[:task] && dep[:workflow]) ? dep[:workflow].task_info(dep[:task]) : nil

data/test/rbbt/tsv/test_attach.rb

@@ -567,5 +567,73 @@ row2,CC
 
     assert_equal tsv1.attach(tsv2, :complete => true).attach(tsv3, :complete => true)["row1"], [nil, "B", nil]
   end
+
+  def test_attach_index_both_non_key
+    content1 =<<-EOF
+#: :sep=/\\s+/
+#Id    ValueA    ValueB
+row1    a|aa|aaa    b
+row2    A    B
+    EOF
+
+    content2 =<<-EOF
+#: :sep=/\\s+/
+#ValueE    OtherID
+e    Id1|Id2
+E    Id3
+    EOF
+
+    content_index =<<-EOF
+#: :sep=/\\s+/
+#ValueA    OtherID
+a    Id1
+A    Id3
+    EOF
+
+    Rbbt.claim Rbbt.tmp.test.test1.data, :string, content1
+    Rbbt.claim Rbbt.tmp.test.test2.data, :string, content2
+    Rbbt.claim Rbbt.tmp.test.test2.identifiers, :string, content_index
+
+    tsv1 = tsv2 = nil
+
+    tsv1 = Rbbt.tmp.test.test1.data.tsv :double,  :sep => /\s+/
+    tsv2 = Rbbt.tmp.test.test2.data.tsv :double,  :sep => /\s+/
+
+    tsv2.identifiers = Rbbt.tmp.test.test2.identifiers.produce.find #.to_s
+
+    tsv1.attach tsv2, :fields => ["ValueE"] #, :persist_input => true
+    Log.tsv tsv1
+    ppp tsv1
+    
+  end
+
+  def test_attach_both_non_key
+    content1 =<<-EOF
+#: :sep=/\\s+/
+#Id    ValueA    ValueB
+row1    a|aa|aaa    b
+row2    A    B
+    EOF
+
+    content2 =<<-EOF
+#: :sep=/\\s+/
+#ValueE    ValueB
+e    b
+E    B
+    EOF
+    Rbbt.claim Rbbt.tmp.test.test1.data, :string, content1
+    Rbbt.claim Rbbt.tmp.test.test2.data, :string, content2
+
+    tsv1 = tsv2 = nil
+
+    tsv1 = Rbbt.tmp.test.test1.data.tsv :double,  :sep => /\s+/
+    tsv2 = Rbbt.tmp.test.test2.data.tsv :double,  :sep => /\s+/
+
+    Log.severity = 0
+    tsv1.attach tsv2, :fields => ["ValueE"] #, :persist_input => true
+    Log.tsv tsv1
+    ppp tsv1
+    
+  end
 end
 

data/test/rbbt/tsv/test_index.rb

@@ -36,6 +36,22 @@ row2    A    B    Id3
 
   end
 
+  def test_index_fields_key
+    content =<<-EOF
+#Id    ValueA    ValueB    OtherID
+row1    a|aa|aaa    b    Id1|Id2
+row2    A    B    Id3
+    EOF
+
+    TmpFile.with_file(content) do |filename|
+      tsv = TSV.open(File.open(filename), :sep => /\s+/, :key_field => "OtherID", :persistence => false)
+      Log.tsv tsv
+      index = tsv.index(:case_insensitive => true, :persistence => false, :fields => ["Id"], :target => "ValueA")
+      Log.tsv index
+    end
+
+  end
+
   def test_best_index
     content =<<-EOF
 #Id    ValueA    ValueB    OtherID

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: rbbt-util
 version: !ruby/object:Gem::Version
-  version: 5.26.92
+  version: 5.26.93
 platform: ruby
 authors:
 - Miguel Vazquez
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2019-11-29 00:00:00.000000000 Z
+date: 2019-12-16 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: rake
@@ -535,93 +535,93 @@ required_rubygems_version: !ruby/object:Gem::Requirement
     - !ruby/object:Gem::Version
       version: '0'
 requirements: []
-rubygems_version: 3.0.4
+rubygems_version: 3.0.6
 signing_key: 
 specification_version: 4
 summary: Utilities for the Ruby Bioinformatics Toolkit (rbbt)
 test_files:
-- test/test_helper.rb
-- test/rbbt/test_annotations.rb
-- test/rbbt/association/test_database.rb
-- test/rbbt/association/test_item.rb
-- test/rbbt/association/test_index.rb
-- test/rbbt/association/test_open.rb
-- test/rbbt/association/test_util.rb
-- test/rbbt/test_association.rb
-- test/rbbt/knowledge_base/test_syndicate.rb
-- test/rbbt/knowledge_base/test_enrichment.rb
-- test/rbbt/knowledge_base/test_query.rb
-- test/rbbt/knowledge_base/test_entity.rb
-- test/rbbt/knowledge_base/test_registry.rb
-- test/rbbt/knowledge_base/test_traverse.rb
+- test/rbbt/test_entity.rb
+- test/rbbt/workflow/test_doc.rb
+- test/rbbt/workflow/test_step.rb
+- test/rbbt/workflow/remote/test_client.rb
+- test/rbbt/workflow/step/test_dependencies.rb
+- test/rbbt/workflow/test_task.rb
+- test/rbbt/resource/test_path.rb
+- test/rbbt/util/test_colorize.rb
+- test/rbbt/util/misc/test_omics.rb
+- test/rbbt/util/misc/test_pipes.rb
+- test/rbbt/util/misc/test_lock.rb
+- test/rbbt/util/misc/test_multipart_payload.rb
+- test/rbbt/util/misc/test_bgzf.rb
+- test/rbbt/util/test_concurrency.rb
+- test/rbbt/util/test_cmd.rb
+- test/rbbt/util/R/test_plot.rb
+- test/rbbt/util/R/test_eval.rb
+- test/rbbt/util/R/test_model.rb
 - test/rbbt/util/test_config.rb
-- test/rbbt/util/test_R.rb
+- test/rbbt/util/test_log.rb
+- test/rbbt/util/test_simpleDSL.rb
+- test/rbbt/util/log/test_progress.rb
 - test/rbbt/util/test_tmpfile.rb
-- test/rbbt/util/test_filecache.rb
-- test/rbbt/util/test_misc.rb
+- test/rbbt/util/test_R.rb
 - test/rbbt/util/test_excel2tsv.rb
-- test/rbbt/util/log/test_progress.rb
-- test/rbbt/util/misc/test_multipart_payload.rb
-- test/rbbt/util/misc/test_pipes.rb
-- test/rbbt/util/misc/test_bgzf.rb
-- test/rbbt/util/misc/test_lock.rb
-- test/rbbt/util/misc/test_omics.rb
-- test/rbbt/util/test_semaphore.rb
-- test/rbbt/util/simpleopt/test_get.rb
+- test/rbbt/util/test_misc.rb
+- test/rbbt/util/test_open.rb
+- test/rbbt/util/test_simpleopt.rb
 - test/rbbt/util/simpleopt/test_parse.rb
 - test/rbbt/util/simpleopt/test_setup.rb
-- test/rbbt/util/test_open.rb
+- test/rbbt/util/simpleopt/test_get.rb
+- test/rbbt/util/test_python.rb
+- test/rbbt/util/test_filecache.rb
 - test/rbbt/util/concurrency/test_processes.rb
 - test/rbbt/util/concurrency/test_threads.rb
 - test/rbbt/util/concurrency/processes/test_socket.rb
-- test/rbbt/util/test_log.rb
-- test/rbbt/util/test_colorize.rb
-- test/rbbt/util/test_cmd.rb
-- test/rbbt/util/test_concurrency.rb
-- test/rbbt/util/R/test_eval.rb
-- test/rbbt/util/R/test_plot.rb
-- test/rbbt/util/R/test_model.rb
-- test/rbbt/util/test_simpleopt.rb
+- test/rbbt/util/test_semaphore.rb
 - test/rbbt/util/test_chain_methods.rb
-- test/rbbt/util/test_simpleDSL.rb
-- test/rbbt/util/test_python.rb
-- test/rbbt/tsv/test_index.rb
-- test/rbbt/tsv/test_parser.rb
-- test/rbbt/tsv/test_field_index.rb
-- test/rbbt/tsv/test_stream.rb
-- test/rbbt/tsv/test_accessor.rb
+- test/rbbt/test_resource.rb
+- test/rbbt/test_packed_index.rb
 - test/rbbt/tsv/test_change_id.rb
-- test/rbbt/tsv/test_parallel.rb
-- test/rbbt/tsv/parallel/test_through.rb
-- test/rbbt/tsv/parallel/test_traverse.rb
-- test/rbbt/tsv/test_manipulate.rb
+- test/rbbt/tsv/test_attach.rb
 - test/rbbt/tsv/test_filter.rb
+- test/rbbt/tsv/test_parser.rb
+- test/rbbt/tsv/test_accessor.rb
 - test/rbbt/tsv/test_matrix.rb
+- test/rbbt/tsv/test_field_index.rb
 - test/rbbt/tsv/test_util.rb
-- test/rbbt/tsv/test_attach.rb
+- test/rbbt/tsv/test_index.rb
+- test/rbbt/tsv/test_parallel.rb
+- test/rbbt/tsv/test_manipulate.rb
 - test/rbbt/tsv/test_excel.rb
-- test/rbbt/test_monitor.rb
-- test/rbbt/test_workflow.rb
-- test/rbbt/test_hpc.rb
-- test/rbbt/annotations/test_util.rb
-- test/rbbt/test_packed_index.rb
-- test/rbbt/workflow/test_task.rb
-- test/rbbt/workflow/test_doc.rb
-- test/rbbt/workflow/remote/test_client.rb
-- test/rbbt/workflow/test_step.rb
-- test/rbbt/workflow/step/test_dependencies.rb
-- test/rbbt/entity/test_identifiers.rb
-- test/rbbt/test_entity.rb
-- test/rbbt/test_resource.rb
-- test/rbbt/test_persist.rb
-- test/rbbt/persist/tsv/test_tokyocabinet.rb
-- test/rbbt/persist/tsv/test_cdb.rb
-- test/rbbt/persist/tsv/test_leveldb.rb
+- test/rbbt/tsv/parallel/test_through.rb
+- test/rbbt/tsv/parallel/test_traverse.rb
+- test/rbbt/tsv/test_stream.rb
+- test/rbbt/test_association.rb
+- test/rbbt/association/test_database.rb
+- test/rbbt/association/test_item.rb
+- test/rbbt/association/test_open.rb
+- test/rbbt/association/test_util.rb
+- test/rbbt/association/test_index.rb
+- test/rbbt/test_knowledge_base.rb
 - test/rbbt/persist/tsv/test_kyotocabinet.rb
-- test/rbbt/persist/tsv/test_sharder.rb
+- test/rbbt/persist/tsv/test_cdb.rb
 - test/rbbt/persist/tsv/test_lmdb.rb
+- test/rbbt/persist/tsv/test_sharder.rb
+- test/rbbt/persist/tsv/test_leveldb.rb
+- test/rbbt/persist/tsv/test_tokyocabinet.rb
 - test/rbbt/persist/test_tsv.rb
 - test/rbbt/test_tsv.rb
+- test/rbbt/test_annotations.rb
 - test/rbbt/test_fix_width_table.rb
-- test/rbbt/test_knowledge_base.rb
-- test/rbbt/resource/test_path.rb
+- test/rbbt/test_workflow.rb
+- test/rbbt/entity/test_identifiers.rb
+- test/rbbt/annotations/test_util.rb
+- test/rbbt/test_hpc.rb
+- test/rbbt/test_monitor.rb
+- test/rbbt/test_persist.rb
+- test/rbbt/knowledge_base/test_entity.rb
+- test/rbbt/knowledge_base/test_registry.rb
+- test/rbbt/knowledge_base/test_syndicate.rb
+- test/rbbt/knowledge_base/test_query.rb
+- test/rbbt/knowledge_base/test_enrichment.rb
+- test/rbbt/knowledge_base/test_traverse.rb
+- test/test_helper.rb