lib/log_analysis.rb in hadoop-rubydsl-0.0.4 vs lib/log_analysis.rb in hadoop-rubydsl-0.0.5

@@ -7,10 +7,12 @@
   PASS = nil
   MODEL_METHODS = [:column, :value]
 
   # controller
   class LogAnalysisMapper < HadoopDsl::BaseMapper
+    @@reg_cache = {}
+
     def initialize(script, key, value)
       super(script, LogAnalysisMapperModel.new(key, value))
     end
 
     # model methods
@@ -25,13 +27,19 @@
     def separate(sep)
       parts = value.split(sep)
       @model.create_or_replace_columns_with(parts) {|column, value| column.value = value}
     end
 
-    def pattern(re)
+    def pattern(reg_str)
+      # try to get RE from cache
+      cached = @@reg_cache[reg_str] 
+      re = cached ? @@reg_cache[reg_str] : Regexp.new(reg_str)
+      @@reg_cache[reg_str] ||= re # new cache
+
       if value =~ re
         md = Regexp.last_match
         @model.create_or_replace_columns_with(md.captures) {|column, value| column.value = value}
+      else throw :each_line # non-local exit
       end
     end
 
     # column names by String converted to Symbol
     def column_name(*names)