rbbt-util 5.18.0 → 5.18.1

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: ab39fdc1ea50f9747effe02d0c7008a288194a88
-  data.tar.gz: 4359eaf63e8c0592bb0d035099bd94e1e12fa3b9
+  metadata.gz: 7f4bb8ad71b9080b39bc7217d9f4a5dc0da2719d
+  data.tar.gz: 94b7dfcbd524c888c73f945cf95e223b318205a7
 SHA512:
-  metadata.gz: cccceae6eee4bbc8d64dcfc8e9791c7e7b00ee32bfec839ff3bb914b2511c29b1382acf626f641e1b38c022caabe347492ea3eb1315f5a5b06b01dbc3d2a312c
-  data.tar.gz: dc1dc72ef5f0f69f36d81b9b7bbfcb8f84c5259618aa9e466967a750e13192eeaed3a5d69589fbbbead6347d3df8306cb22e43c098de7414b5fb91f049dc5adb
+  metadata.gz: 17e4da0d2455c8c93de87d51b4166b2156552611c3bafc17ce561fcb6d278e3ad16df23ece0727fea9a783c66d97b528e889246cd5da51a183ba368ba8c7ecff
+  data.tar.gz: 2c981aee80405b0d9794c4568114da6c3d2916f1fac755357f081566ad9d40c17346d5fe14c3388cb66f3a490ede94ab261bb1cb5e2d79cc0f51584d8fbd7b23

data/lib/rbbt/util/misc/inspect.rb

@@ -113,6 +113,18 @@ module Misc
     Digest::MD5.hexdigest(text)
   end
 
+  def self.sample_large_obj(obj, max = 100)
+    length = obj.length
+    head = obj[0..max/2]
+    tail = obj[-max/2..-1]
+    middle = (1..9).to_a.collect{|i| pos = (length / 10) * i + i; obj[pos-1..pos+1]}.flatten 
+    if Array === obj 
+      head + middle + tail + ["LENGTH: #{obj.length}"]
+    else
+      head << "..." << middle*"," << "..." << tail << "(#{obj.length})"
+    end
+  end
+
   HASH2MD5_MAX_STRING_LENGTH = 1000
   HASH2MD5_MAX_ARRAY_LENGTH = 100
   def self.hash2md5(hash)
@@ -184,4 +196,61 @@ module Misc
     end
   end
 
+
+  def self.obj2str(obj)
+    _obj = obj
+    obj = Annotated.purge(obj) if Annotated === obj
+
+    str = case obj
+          when TrueClass
+            'true'
+          when FalseClass
+            'false'
+          when Hash
+            "{"<< obj.collect{|k,v| obj2str(k) << '=>' << obj2str(v)}*"," << "}"
+          when Symbol
+            obj.to_s
+          when String
+            if obj.length > HASH2MD5_MAX_STRING_LENGTH
+              sample_large_obj(obj, HASH2MD5_MAX_STRING_LENGTH)
+            else
+              obj
+            end
+          when Array
+            if obj.length > HASH2MD5_MAX_ARRAY_LENGTH
+              "[" << sample_large_obj(obj, HASH2MD5_MAX_ARRAY_LENGTH).collect{|v| obj2str(v)} * "," << "]"
+            else
+              "[" << obj.collect{|v| obj2str(v)} * "," << "]"
+            end
+          when TSV::Parser
+            remove_long_items(obj)
+          when File 
+            "[File:" << obj.path << "]"
+          else
+            obj_ins = obj.inspect
+            if obj_ins =~ /:0x0/
+              obj_ins.sub(/:0x[a-f0-9]+@/,'')
+            else
+              obj_ins
+            end
+          end
+
+    if defined? Annotated and Annotated === _obj and not (defined? AssociationItem and AssociationItem === _obj)
+      info = Annotated.purge(_obj.info)
+      str << "_" << hash2str(info) 
+    end
+
+    str
+  end
+
+
+  def self.obj2md5(obj)
+    str = obj2str(obj)
+
+    if str.empty?
+      ""
+    else
+      digest(str)
+    end
+  end
 end

data/lib/rbbt/workflow/accessor.rb

@@ -586,8 +586,7 @@ module Workflow
       if inputs.any? or dependencies.any?
         tagged_jobname = case TAG
                          when :hash
-                           input_hash = Hash[*(1..inputs.length).zip(inputs).flatten]
-                           hash_str = Misc.hash2md5(input_hash)
+                           hash_str = Misc.obj2md5(inputs)
                            jobname + '_' << hash_str
                          else
                            jobname

data/test/rbbt/util/test_misc.rb

@@ -466,4 +466,50 @@ eum fugiat quo voluptas nulla pariatur?"
     end
   end
 
+  def test_obj2md5
+    str = "string" *  1000000
+    obj1 = [1,2,"test", 0.3, [1,2,3], {:a => [1,2], :b => str}]
+    obj2 = [1,2,"test", 0.3, [1,2,3], {:a => [1,2,3], :b => str}]
+    obj3 = [1,2,:test, 0.3, [1,2,3], {:a => [1,2,"3"], :b => str}]
+    assert_not_equal Misc.obj2md5(obj1), Misc.obj2md5(obj2)
+    assert_equal Misc.obj2md5(obj2), Misc.obj2md5(obj3)
+    hash = Hash[*(0..obj1.length).zip(obj1).flatten]
+
+    Misc.benchmark(10000) do
+      Misc.hash2md5(hash)
+    end
+    Misc.benchmark(10000) do
+      Misc.obj2md5(obj1)
+    end
+    Misc.profile do
+      10.times do
+        Misc.hash2md5(hash)
+      end
+    end
+    Misc.profile do
+      10.times do
+        Misc.obj2md5(obj1)
+      end
+    end
+  end
+
+  def test_sample_large_string
+    str = "string" *  1000000
+
+    max = 100
+    assert_equal Misc.sample_large_obj(str, max).length, max+51
+
+    max = 1000
+    assert_equal Misc.sample_large_obj(str, max).length, max+51
+  end
+
+  def test_sample_large_array
+    str = (0..1000000).to_a
+
+    max = 100
+    assert_equal Misc.sample_large_obj(str, max).length, max+29
+
+    max = 1000
+    assert_equal Misc.sample_large_obj(str, max).length, max+29
+  end
 end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: rbbt-util
 version: !ruby/object:Gem::Version
-  version: 5.18.0
+  version: 5.18.1
 platform: ruby
 authors:
 - Miguel Vazquez