apachecrunch 0.5 → 0.6

data/lib/procedure_dsl.rb

@@ -52,6 +52,22 @@ class ApacheCrunch
     end
 
 
+    # DSL routine that calculates the sum of the quantities given by the block
+    #
+    # All values produced by the block must have a to_f method, and this routine
+    # returns a float.
+    class Sum < ProcedureRoutine
+        def execute(&blk)
+            rslt = 0.0
+            while @_current_entry = @_log_parser.next_entry
+                rslt += instance_eval(&blk).to_f
+            end
+
+            rslt
+        end
+    end
+
+
     # DSL routine(s) that filter(s) for entries for which the given block evaluates to true
     #
     # This can be called as 'filter()', which means the filtering happens in a temporary file, or
@@ -191,14 +207,17 @@ class ApacheCrunch
     #         time_to_serve
     #     end
     #
-    # would return two numbers, the lower and upper bound of a 95% confidence interval for the values
-    # of time_to_serve.
+    # would return two numbers, the lower and upper bound of a 95% confidence interval for the
+    # values of time_to_serve.
+    #
+    # This routine returns a float, and the block must always evaluate to something with a to_f
+    # method.
     class ConfidenceInterval < ProcedureRoutine
         def execute(confidence, &blk)
             # Build a list of all the values found
             values = []
             while @_current_entry = @_log_parser.next_entry
-                values << instance_eval(&blk)
+                values << instance_eval(&blk).to_f
             end
             values.sort!
 
@@ -211,6 +230,26 @@ class ApacheCrunch
     end
 
 
+    # DSL routine that determines the nth percentile for the values to which the block evaluates
+    #
+    # This routine returns a float, and the block must always evaluate to something with a to_f
+    # method.
+    class Percentile < ProcedureRoutine
+        def execute(n, &blk)
+            # Build a list of all the values found
+            values = []
+            while @_current_entry = @_log_parser.next_entry
+                values << instance_eval(&blk).to_f
+            end
+            values.sort!
+
+            puts "values.length: #{values.length}"
+            puts "n/100.0*values.length: #{n/100.0*values.length}"
+            return values[((n/100.0)*values.length).to_i]
+        end
+    end
+
+
     # DSL routine that finds the most common n values for the given block.
     #
     # Returns a list of lists, each of which is [value, count].  This list is sorted by count.
@@ -275,6 +314,14 @@ class ApacheCrunch
             nil
         end
 
+        # DSL routine 'sum'
+        def sum(&blk)
+            routine = Sum.new(@_log_parser)
+            rv = routine.execute(&blk)
+            routine.finish
+            rv
+        end
+
         # DSL routine 'count_by'
         def count_by(&blk)
             routine = CountBy.new(@_log_parser)
@@ -307,6 +354,14 @@ class ApacheCrunch
             rv
         end
 
+        # DSL routine 'percentile'
+        def percentile(n, &blk)
+            routine = Percentile.new(@_log_parser)
+            rv = routine.execute(n, &blk)
+            routine.finish
+            rv
+        end
+
         # DSL routine 'most_common'
         def most_common(n, &blk)
             routine = MostCommon.new(@_log_parser)

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification 
 name: apachecrunch
 version: !ruby/object:Gem::Version 
-  version: "0.5"
+  version: "0.6"
 platform: ruby
 authors: 
 - Dan Slimmon
@@ -9,7 +9,7 @@ autorequire:
 bindir: bin
 cert_chain: []
 
-date: 2011-09-17 00:00:00 -04:00
+date: 2011-09-26 00:00:00 -04:00
 default_executable: 
 dependencies: []