rbbt-util 5.21.19 → 5.21.20

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: c722ae1fd02431a1b5cbe726b12d8cd2af4e8871
-  data.tar.gz: ea254aa679c034bf5294f9f80f7d430d95c6058b
+  metadata.gz: 0df0c7ae6b0b29814a176c029ed4c9f20e13307c
+  data.tar.gz: 228920d67af31815ef453b4f05b0c1bc32cf1175
 SHA512:
-  metadata.gz: 1c76e07f15de03b2e49b5d21c04cdb358026a957b80ab1973ac4cd389b646fec1140806eb92598a9f6a66e0ada631ef6e41e58e376104ea5954e44b8c4479854
-  data.tar.gz: 8d7a4dec88f3f42f513e9ea1a009d4911f1e98548696e213c878da217fafd8bed1f56b297f472803fc2e13b70ab5cdff5618ced5ad0a6b6c19055015d62d7166
+  metadata.gz: 0af021e82639371b123a9b04ae7d1b89541bf19e7c446fd8ce38ecd25066d06f585d36872edb636e61856b82e3e45b67cc2aa67ffdb236904f81e7d93ca8f160
+  data.tar.gz: c1a0e4a399e442a2d8545d71f28636ffcccb0fb1c60422c84c47806cef041738fa22ed10fe9aeb7fd5a347c635e837a025c1df4c3f402c42e520d100482321fd

data/lib/rbbt/tsv/accessor.rb

@@ -695,7 +695,7 @@ Example:
     end
 
     if self.key_field and self.fields
-      new.key_field = [self.key_field, self.fields[field_pos]] * ":" 
+      new.key_field = [self.key_field, self.fields[field_pos]] * sep
       new_fields = self.fields.dup 
       new_fields.delete_at field_pos
       new.fields = new_fields
@@ -703,5 +703,39 @@ Example:
 
     new
   end
+  
+  def zip(field = 0, merge = false, sep = ":")
+    new = {}
+    self.annotate new
+
+    new.type = :double if merge
+
+    new.with_unnamed do
+      if merge
+        self.through do |key,values|
+          new_key, new_value = key.split(sep)
+          new_values = values + [[new_value]]
+          if new.include? new_key
+            new[new_key] = Misc.zip_fields(Misc.zip_fields(new[new_key]) << new_values)
+          else
+            new[new_key] = new_values
+          end
+        end
+      else
+        self.through do |key,values|
+          new_key, new_value = key.split(sep)
+          new_values = values + [new_value]
+          new[new_key] = new_values
+        end
+      end
+    end
+
+    if self.key_field and self.fields
+      new.key_field = self.key_field.partition(sep).first
+      new.fields = new.fields + [field]
+    end
+
+    new
+  end
 end
 

data/share/rbbt_commands/tsv/unzip

@@ -15,6 +15,7 @@ Display summary information. Works with Tokyocabinet HDB and BDB as well.
 -tch--tokyocabinet File is a TC HDB
 -tcb--tokyocabinet_bd File is a TC BDB
 -hh--header_hash* Change the character used to mark the header line (defaults to #)
+-zs--zip_sep* Separator between zipped fields
 -f--field* Field to unzip
 -m--merge Merge lines
 -h--help Help
@@ -28,6 +29,7 @@ file = STDIN if file == '-'
 
 field = options[:field] || 0
 merge = options[:merge] || false
+zs = options[:zip_sep] || ":"
 
 case
 when options[:tokyocabinet]
@@ -42,4 +44,4 @@ else
   tsv = TSV.open(file, {:type => :double}.merge(options))
 end
 
-puts tsv.unzip(field, merge).to_s
+puts tsv.unzip(field, merge, zs).to_s

data/share/rbbt_commands/tsv/zip

@@ -0,0 +1,47 @@
+#!/usr/bin/env ruby
+
+require 'rbbt-util'
+require 'rbbt/util/simpleopt'
+
+$0 = "rbbt #{$previous_commands*""} #{ File.basename(__FILE__) }" if $previous_commands
+
+options = SOPT.setup <<EOF
+Zip fields in a TSV file 
+
+$ rbbt tsv zip [options] file.tsv <new_field>
+
+Display summary information. Works with Tokyocabinet HDB and BDB as well.
+
+-tch--tokyocabinet File is a TC HDB
+-tcb--tokyocabinet_bd File is a TC BDB
+-hh--header_hash* Change the character used to mark the header line (defaults to #)
+-zs--zip_sep* Separator between zipped fields
+-f--field* Field to unzip
+-m--merge Merge lines
+-h--help Help
+EOF
+
+SOPT.usage if options[:help]
+
+file = ARGV.shift
+
+file = STDIN if file == '-'
+
+field = options[:field] || 0
+merge = options[:merge] || false
+zs = options[:zip_sep] || ":"
+
+case
+when options[:tokyocabinet]
+  tsv = Persist.open_tokyocabinet(file, false)
+when options[:tokyocabinet_bd]
+  tsv = Persist.open_tokyocabinet(file, false, nil, TokyoCabinet::BDB)
+else
+  if String === file
+    file = file.dup
+    Path.setup(File.expand_path(file))
+  end
+  tsv = TSV.open(file, {:type => :double}.merge(options))
+end
+
+puts tsv.zip(field, merge, zs).to_s

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: rbbt-util
 version: !ruby/object:Gem::Version
-  version: 5.21.19
+  version: 5.21.20
 platform: ruby
 authors:
 - Miguel Vazquez
@@ -364,6 +364,7 @@ files:
 - share/rbbt_commands/tsv/subset
 - share/rbbt_commands/tsv/unzip
 - share/rbbt_commands/tsv/values
+- share/rbbt_commands/tsv/zip
 - share/rbbt_commands/watch
 - share/rbbt_commands/workflow/cmd
 - share/rbbt_commands/workflow/example
@@ -486,82 +487,82 @@ signing_key:
 specification_version: 4
 summary: Utilities for the Ruby Bioinformatics Toolkit (rbbt)
 test_files:
-- test/rbbt/test_workflow.rb
+- test/test_helper.rb
 - test/rbbt/resource/test_path.rb
-- test/rbbt/util/test_cmd.rb
-- test/rbbt/util/simpleopt/test_setup.rb
-- test/rbbt/util/simpleopt/test_get.rb
-- test/rbbt/util/simpleopt/test_parse.rb
-- test/rbbt/util/test_chain_methods.rb
-- test/rbbt/util/test_simpleDSL.rb
+- test/rbbt/association/test_item.rb
+- test/rbbt/association/test_database.rb
+- test/rbbt/association/test_open.rb
+- test/rbbt/association/test_index.rb
+- test/rbbt/association/test_util.rb
+- test/rbbt/util/test_concurrency.rb
 - test/rbbt/util/test_log.rb
+- test/rbbt/util/test_chain_methods.rb
+- test/rbbt/util/test_simpleopt.rb
+- test/rbbt/util/simpleopt/test_parse.rb
+- test/rbbt/util/simpleopt/test_get.rb
+- test/rbbt/util/simpleopt/test_setup.rb
+- test/rbbt/util/test_cmd.rb
+- test/rbbt/util/test_semaphore.rb
+- test/rbbt/util/concurrency/test_threads.rb
+- test/rbbt/util/concurrency/processes/test_socket.rb
+- test/rbbt/util/concurrency/test_processes.rb
+- test/rbbt/util/test_tmpfile.rb
 - test/rbbt/util/test_open.rb
+- test/rbbt/util/test_filecache.rb
+- test/rbbt/util/R/test_eval.rb
+- test/rbbt/util/R/test_model.rb
+- test/rbbt/util/test_simpleDSL.rb
+- test/rbbt/util/log/test_progress.rb
+- test/rbbt/util/test_colorize.rb
+- test/rbbt/util/test_R.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/misc/test_pipes.rb
+- test/rbbt/util/misc/test_bgzf.rb
 - test/rbbt/util/misc/test_omics.rb
-- test/rbbt/util/test_concurrency.rb
-- test/rbbt/util/test_R.rb
-- test/rbbt/util/log/test_progress.rb
-- test/rbbt/util/test_colorize.rb
-- test/rbbt/util/test_simpleopt.rb
+- test/rbbt/util/misc/test_multipart_payload.rb
 - test/rbbt/util/test_excel2tsv.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_semaphore.rb
 - test/rbbt/util/test_misc.rb
-- test/rbbt/util/test_tmpfile.rb
-- test/rbbt/util/R/test_model.rb
-- test/rbbt/util/R/test_eval.rb
-- test/rbbt/test_packed_index.rb
-- test/rbbt/entity/test_identifiers.rb
-- test/rbbt/test_association.rb
-- test/rbbt/knowledge_base/test_traverse.rb
-- test/rbbt/knowledge_base/test_registry.rb
-- test/rbbt/knowledge_base/test_entity.rb
-- test/rbbt/knowledge_base/test_enrichment.rb
-- test/rbbt/knowledge_base/test_syndicate.rb
-- test/rbbt/knowledge_base/test_query.rb
-- test/rbbt/test_resource.rb
 - test/rbbt/test_entity.rb
-- test/rbbt/test_knowledge_base.rb
-- test/rbbt/annotations/test_util.rb
-- test/rbbt/association/test_index.rb
-- test/rbbt/association/test_item.rb
-- test/rbbt/association/test_open.rb
-- test/rbbt/association/test_util.rb
-- test/rbbt/association/test_database.rb
-- test/rbbt/test_tsv.rb
 - test/rbbt/workflow/step/test_dependencies.rb
-- test/rbbt/workflow/test_task.rb
-- test/rbbt/workflow/test_step.rb
 - test/rbbt/workflow/test_doc.rb
-- test/rbbt/test_monitor.rb
-- test/rbbt/test_persist.rb
-- test/rbbt/test_annotations.rb
-- test/rbbt/persist/test_tsv.rb
-- test/rbbt/persist/tsv/test_lmdb.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_tokyocabinet.rb
-- test/rbbt/persist/tsv/test_leveldb.rb
-- test/rbbt/tsv/test_field_index.rb
+- test/rbbt/workflow/test_step.rb
+- test/rbbt/workflow/test_task.rb
+- test/rbbt/test_association.rb
+- test/rbbt/test_knowledge_base.rb
+- test/rbbt/tsv/parallel/test_traverse.rb
+- test/rbbt/tsv/parallel/test_through.rb
 - test/rbbt/tsv/test_parallel.rb
-- test/rbbt/tsv/test_index.rb
-- test/rbbt/tsv/test_matrix.rb
+- test/rbbt/tsv/test_accessor.rb
 - test/rbbt/tsv/test_change_id.rb
-- test/rbbt/tsv/test_parser.rb
 - test/rbbt/tsv/test_stream.rb
-- test/rbbt/tsv/test_util.rb
-- test/rbbt/tsv/test_accessor.rb
 - test/rbbt/tsv/test_filter.rb
+- test/rbbt/tsv/test_matrix.rb
 - test/rbbt/tsv/test_attach.rb
 - test/rbbt/tsv/test_manipulate.rb
-- test/rbbt/tsv/parallel/test_through.rb
-- test/rbbt/tsv/parallel/test_traverse.rb
+- test/rbbt/tsv/test_field_index.rb
+- test/rbbt/tsv/test_index.rb
+- test/rbbt/tsv/test_util.rb
+- test/rbbt/tsv/test_parser.rb
+- test/rbbt/test_packed_index.rb
+- test/rbbt/test_persist.rb
 - test/rbbt/test_fix_width_table.rb
-- test/test_helper.rb
+- test/rbbt/knowledge_base/test_traverse.rb
+- test/rbbt/knowledge_base/test_entity.rb
+- test/rbbt/knowledge_base/test_query.rb
+- test/rbbt/knowledge_base/test_enrichment.rb
+- test/rbbt/knowledge_base/test_syndicate.rb
+- test/rbbt/knowledge_base/test_registry.rb
+- test/rbbt/entity/test_identifiers.rb
+- test/rbbt/test_monitor.rb
+- test/rbbt/test_workflow.rb
+- test/rbbt/test_annotations.rb
+- test/rbbt/annotations/test_util.rb
+- test/rbbt/test_resource.rb
+- test/rbbt/persist/tsv/test_tokyocabinet.rb
+- test/rbbt/persist/tsv/test_kyotocabinet.rb
+- test/rbbt/persist/tsv/test_lmdb.rb
+- test/rbbt/persist/tsv/test_leveldb.rb
+- test/rbbt/persist/tsv/test_cdb.rb
+- test/rbbt/persist/tsv/test_sharder.rb
+- test/rbbt/persist/test_tsv.rb
+- test/rbbt/test_tsv.rb