rbbt-util 5.27.1 → 5.27.6

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: f9a8f4d4bbf80d9f7ad26f1867efe048bad08e1f0bd42b32ac731744380ba571
-  data.tar.gz: 9692dd68f20cec3b8688cc9d75e1bcd4d75a024924c56c191a87897162d34ef4
+  metadata.gz: fe7c3dec4d2333e77ea396a2d6e572512bf7033999ff4715f1a1c6c9d3d1b9d5
+  data.tar.gz: 32a3d905c1a95f9c7e5cb359855f8781509cfeecdd8db7b77ee29eaea528d030
 SHA512:
-  metadata.gz: cb9672eec11a5184795c1e85283f0ed948f34c77f91d31fd6852c389820abdadb555dc99acd823adefd8b6afbb3c4589988599f3095e2ba10e7bdb633404db98
-  data.tar.gz: e28718c7c60c25a4316997f477b2fca33b067865965bff51a423b1a84124e3abe20fdce50543e23e11852256376d4caaf6932a399d404e27823db2d577e543e8
+  metadata.gz: d009024642898f775d423e1ddede6aea538e8ed5ff4675eedadd64768e67b0a46f6519038378a221db76b0e196cfc713ca8dc7553788cdb9a51014c14a97de4c
+  data.tar.gz: 5c0612feca84b00e6d402f7a68ca1cebd6f4b7a785b40899bdaed3385a3f68853824dcc98d9cfd2e641768ef5ca3d52e38e41bdbc1d7a3751ae288e8f8d7ed2d

data/lib/rbbt/annotations/util.rb

@@ -145,10 +145,9 @@ module Annotated
 
              end
 
-    fields = fields.collect{|f| f.to_s}
+    fields = fields.collect{|f| f.to_s}.uniq
 
     case
-
     when (Annotated === annotations and not (AnnotatedArray === annotations and annotations.double_array))
       tsv = TSV.setup({}, :key_field => "List", :fields => fields, :type => :list, :unnamed => true)
 

data/lib/rbbt/entity.rb

@@ -194,13 +194,16 @@ module Entity
           multi_name = "_multiple_" << name
 
           define_method multi_name do
-            if self.instance_variable_get("@multiple_result")
-              return self.instance_variable_get("@multiple_result")
+            if self.instance_variable_get("@multiple_result_" + name.to_s)
+              return self.instance_variable_get("@multiple_result_" + name.to_s)
             end
-            raise MultipleEntity, "Entity #{name} runs with multiple entities: #{self}"
+            raise MultipleEntity, "Entity #{name} runs with multiple entities"
           end
 
           define_method name do |*args|
+            if self.instance_variable_get("@multiple_result_" + name.to_s)
+              return self.instance_variable_get("@multiple_result_" + name.to_s)
+            end
             obj = if Array === self
                     self
                   elsif self.respond_to?(:container) && Array === self.container
@@ -222,11 +225,11 @@ module Entity
             case res
             when Array
               missing.zip(res).each do |o,res|
-                o.instance_variable_set("@multiple_result", res)
+                o.instance_variable_set("@multiple_result_" + name.to_s, res)
               end
             when Hash
               res.each do |o,res|
-                o.instance_variable_set("@multiple_result", res)
+                o.instance_variable_set("@multiple_result_" + name.to_s, res)
               end
             end
 

data/lib/rbbt/fix_width_table.rb

@@ -4,20 +4,21 @@ class FixWidthTable
   def initialize(filename, value_size = nil, range = nil, update = false, in_memory = true)
     @filename = filename
 
-    if update or %w(memory stringio).include?(filename.to_s.downcase) or not File.exist?(filename)
+    if update || %w(memory stringio).include?(filename.to_s.downcase) || ! File.exist?(filename)
       Log.debug "FixWidthTable create: #{ filename }"
       @value_size  = value_size
       @range       = range
       @record_size = @value_size + (@range ? 16 : 8)
       @write = true
 
-      if %w(memory stringio).include? filename.to_s.downcase
+      if %w(memory stringio).include?(filename.to_s.downcase)
         @filename = :memory
         @file     = StringIO.new
       else
         FileUtils.rm @filename if File.exist? @filename
         FileUtils.mkdir_p File.dirname(@filename) unless File.exist? @filename
-        @file = File.open(@filename, 'wb')
+        #@file = File.open(@filename, 'wb')
+        @file = File.open(@filename, 'w:ASCII-8BIT')
       end
 
       @file.write [value_size].pack("L")
@@ -25,9 +26,9 @@ class FixWidthTable
 
       @size = 0
     else
-      Log.debug "FixWidthTable up-to-date: #{ filename }"
+      Log.debug "FixWidthTable up-to-date: #{ filename } - (in_memory:#{in_memory})"
       if in_memory
-        @file        = StringIO.new(Open.read(@filename, :mode => 'r:ASCII-8BIT'), 'r')
+        @file        = Open.open(@filename, :mode => 'r:ASCII-ASCII'){|f| StringIO.new f.read}
       else
         @file        = File.open(@filename, 'r:ASCII-8BIT')
       end

data/lib/rbbt/knowledge_base/entity.rb

@@ -5,14 +5,15 @@ class KnowledgeBase
 
   def select_entities(name, entities, options = {})
     index = get_index(name, options)
+
     source_field = index.source_field
     target_field = index.target_field
 
     source_type = Entity.formats[source_field] 
     target_type = Entity.formats[target_field]
 
-    source_entities = entities[:source] || entities[source_field] || entities[Entity.formats[source_field].to_s] 
-    target_entities = entities[:target] || entities[target_field] || entities[Entity.formats[target_field].to_s]
+    source_entities = entities[:source] || entities[source_field] || entities[Entity.formats[source_field].to_s] || entities[:both]
+    target_entities = entities[:target] || entities[target_field] || entities[Entity.formats[target_field].to_s] || entities[:both]
 
     [source_entities, target_entities]
   end

data/lib/rbbt/knowledge_base/query.rb

@@ -10,6 +10,7 @@ class KnowledgeBase
 
   def subset(name, entities, options = {}, &block)
     entities, options = options, entities if entities.nil? and Hash === options
+
     entities = case entities
                when :all
                  {:target => :all, :source => :all}
@@ -23,7 +24,15 @@ class KnowledgeBase
                  raise "Entities are not a Hash or an AnnotatedArray: #{Misc.fingerprint entities}"
                end
 
+    identify, identify_source, identify_target = entities.merge(options || {}).values_at :identify, :identify_source, :identify_target
+
     source, target = select_entities(name, entities, options)
+    
+    source = identify_source(name, source) if identify_source
+    target = identify_target(name, target) if identify_target
+
+    source = identify(name, source) if identify && !identify_source
+    target = identify(name, target) if identify && !identify_target
 
     return [] if source.nil? or target.nil?
     return [] if Array === target and target.empty?

data/lib/rbbt/persist.rb

@@ -407,7 +407,6 @@ module Persist
           repo.read_and_close do
             keys = repo.range subkey + 0.chr, true, subkey + 254.chr, true
           end
-          repo.close
         else
           repo.read_and_close do
             keys = repo.range subkey + 0.chr, true, subkey + 254.chr, true
@@ -461,13 +460,13 @@ module Persist
               entities.each_with_index do |e,i|
                 next if e.nil?
                 tsv_values = e.tsv_values("literal", "annotation_types", "JSON") 
-                repo[subkey + e.id << ":ANNOTATED_DOUBLE_ARRAY:" << i.to_s] = tsv_values
+                repo[subkey + "ANNOTATED_DOUBLE_ARRAY:" << i.to_s] = tsv_values
               end
             else
               entities.each_with_index do |e,i|
                 next if e.nil?
                 tsv_values = e.tsv_values("literal", "annotation_types", "JSON") 
-                repo[subkey + e.id << ":" << i.to_s] = tsv_values
+                repo[subkey + i.to_s] = tsv_values
               end
             end
           end

data/lib/rbbt/persist/tsv.rb

@@ -148,13 +148,13 @@ module Persist
 
         data.write_and_read do
           yield data
-        end
 
-        FileUtils.mv data.persistence_path, path if File.exist? data.persistence_path and not File.exist? path
-        tsv = CONNECTIONS[path] = CONNECTIONS.delete tmp_path
-        tsv.persistence_path = path
+          FileUtils.mv data.persistence_path, path if File.exist? data.persistence_path and not File.exist? path
+          tsv = CONNECTIONS[path] = CONNECTIONS.delete tmp_path
+          tsv.persistence_path = path
 
-        tsv.fix_io if tsv.respond_to? :fix_io
+          tsv.fix_io if tsv.respond_to? :fix_io
+        end
 
         data
       rescue Exception

data/lib/rbbt/persist/tsv/adapter.rb

@@ -9,43 +9,40 @@ module Persist
       @mutex ||= Mutex.new
     end
 
-    def prefix(key)
-      range(key, 1, key + MAX_CHAR, 1)
-    end
-
-    def get_prefix(key)
-      keys = prefix(key)
-      select(:key => keys)
-    end
-
     def closed?
       @closed
     end
 
-    def close
-      @closed = true
-      super
-      self
-    end
-
     def write?
       @writable
     end
 
     def read?
-      ! write?
+      ! (write? || closed?)
     end
 
-    def collect
-      res = []
-      each do |key, value|
-        res << if block_given?
-                 yield key, value
-        else
-          [key, value]
-        end
+    def write(*args)
+      begin
+        super(*args)
+        @writable = true
+      rescue NoMethodError
+      end
+    end
+
+    def close(*args)
+      begin
+        super(*args)
+        @closed = true
+      rescue NoMethodError
+      end
+      self
+    end
+
+    def read(*args)
+      begin
+        super(*args)
+      rescue NoMethodError
       end
-      res
     end
 
     def delete(key)
@@ -53,34 +50,59 @@ module Persist
     end
 
     def lock
+      return yield if @locked
       lock_filename = Persist.persistence_path(persistence_path, {:dir => TSV.lock_dir})
+      Log.stack caller if $LOG
       Misc.lock(lock_filename) do
-        yield
+        begin
+          @locked = true
+          yield
+        ensure
+          @locked = false
+        end
+      end
+    end
+
+    def lock_and_close
+      lock do
+        begin
+          yield
+        ensure
+          close
+        end
       end
     end
 
     def write_and_read
-      #lock_filename = Persist.persistence_path(persistence_path, {:dir => TSV.lock_dir})
-      #mutex.synchronize do
-      #Misc.lock(lock_filename) do
+      if write?
+        begin
+          return yield
+        ensure
+          read
+        end
+      end
+
       lock do
-        write true if closed? or not write?
+        write(true) if closed? || !write?
         begin
           yield
         ensure
           read
         end
       end
-      #end
-      #end
     end
 
     def write_and_close
-      #lock_filename = Persist.persistence_path(persistence_path + '.write', {:dir => TSV.lock_dir})
-      #mutex.synchronize do
-      #Misc.lock(lock_filename, true) do
+      if write?
+        begin
+          return yield
+        ensure
+          close unless @locked
+        end
+      end
+
       lock do
-        write true if closed? || ! write?
+        write(true) if closed? || ! write?
         res = begin
                 yield
               rescue Exception
@@ -91,11 +113,18 @@ module Persist
               end
         res
       end
-      #end
-      #end
     end
 
     def read_and_close
+      #return yield if @locked
+      if read? || write?
+        begin
+          return yield
+        ensure
+          close unless @locked
+        end
+      end
+
       lock do
         read true if closed? || ! read?
         begin
@@ -106,21 +135,85 @@ module Persist
       end
     end
 
-
     def merge!(hash)
       hash.each do |key,values|
         self[key] = values
       end
     end
 
-
     def range(*args)
-      super(*args) #- TSV::ENTRY_KEYS.to_a
+      self.read_and_close do
+        super(*args)
+      end
     end
 
     def include?(*args)
-      read if closed?
-      super(*args) #- TSV::ENTRY_KEYS.to_a
+      self.read_and_close do
+        super(*args) #- TSV::ENTRY_KEYS.to_a
+      end
+    end
+
+    def [](*args)
+      self.read_and_close do
+        super(*args) #- TSV::ENTRY_KEYS.to_a
+      end
+    end
+
+    def []=(*args)
+      self.write_and_close do
+        super(*args) #- TSV::ENTRY_KEYS.to_a
+      end
+    end
+
+    def keys(*args)
+      self.read_and_close do
+        super(*args)
+      end
+    end
+
+
+    def prefix(key)
+      self.read_and_close do
+        range(key, 1, key + MAX_CHAR, 1)
+      end
+    end
+
+    def get_prefix(key)
+      keys = prefix(key)
+      select(:key => keys)
+    end
+
+
+    def size(*args)
+      self.read_and_close do
+        super(*args)
+      end
+    end
+
+    def each(*args, &block)
+      self.read_and_close do
+        super(*args, &block)
+      end
+    end
+
+    def collect
+      res = []
+      each do |key, value|
+        res << if block_given?
+                 yield key, value
+        else
+          [key, value]
+        end
+      end
+      res
+    end
+
+    def values_at(*keys)
+      self.read_and_close do
+        keys.collect do |k|
+          self[k]
+        end
+      end
     end
   end
 end

data/lib/rbbt/persist/tsv/tokyocabinet.rb

@@ -16,7 +16,7 @@ module Persist
       end
 
       dir = File.dirname(File.expand_path(path))
-      Open.mkdir(dir) unless Open.exists?(dir)
+      File.mkdir(dir) unless File.exists?(dir)
 
       tokyocabinet_class = TokyoCabinet::HDB if tokyocabinet_class == "HDB" or tokyocabinet_class.nil?
       tokyocabinet_class = TokyoCabinet::BDB if tokyocabinet_class == "BDB"
@@ -87,13 +87,15 @@ module Persist
   def self.open_tokyocabinet(path, write, serializer = nil, tokyocabinet_class = TokyoCabinet::HDB)
     write = true unless File.exist? path
 
-    FileUtils.mkdir_p File.dirname(path) unless Open.exist?(File.dirname(path))
+    FileUtils.mkdir_p File.dirname(path) unless File.exist?(File.dirname(path))
 
     database = Persist::TCAdapter.open(path, write, tokyocabinet_class)
 
     unless serializer == :clean
       TSV.setup database
-      database.serializer = serializer || database.serializer
+      database.write_and_read do
+        database.serializer = serializer
+      end if serializer && database.serializer != serializer
     end
 
     database