lib/masticate/mender.rb in masticate-0.1.0 vs lib/masticate/mender.rb in masticate-0.1.1

@@ -11,23 +11,40 @@
   end
 
   def mend(opts)
     @output = opts[:output] ? File.open(opts[:output], "w") : $stdout
     @col_sep = opts[:col_sep] || ','
+    @quote_char = opts[:quote_char] || "\0"
 
     expected_field_count = nil
     headers = nil
     @output_count = 0
     with_input do |input|
       while (line = get) do
         unless line =~ /^\s*$/
-          if !expected_field_count
+          if opts[:inlined]
+            row = explode(line)
+            ncells = row.count/2-1
+            if !expected_field_count
+              headers = row[0..ncells]
+              expected_field_count = headers.count
+              emit(headers.to_csv(:col_sep => @col_sep))
+            else
+              if row[0..ncells] != headers
+                raise "Header mismatch on line #{@input_count}\n  Expected: #{headers.join(',')}\n     Found: #{row[0..ncells].join(',')}"
+              end
+            end
+            row = row[ncells+1..-1]
+            emit(row.to_csv(:col_sep => @col_sep))
+          elsif !expected_field_count
             # trust the first row
-            headers = explode(line)
+            headers = explode(line).map(&:strip)
             case opts[:snip]
             when Fixnum
               headers.shift(opts[:snip])
+            when String
+              raise "TODO: snip named header. Multiple?"
             when nil
               # do nothing
             else
               raise "Do not understand snip instruction [#{opts[:snip].inspect}]"
             end
@@ -41,11 +58,11 @@
                 line = line + ' ' + nextbit
                 running_count = fieldcount(line)
               end
             end
 
-            if line.count(col_sep) > 2
+            unless opts[:dejunk] && junky?(line)
               emit(line)
             end
           end
         end
       end
@@ -58,20 +75,17 @@
       :headers => headers
     }
   end
 
   def fieldcount(line)
-    if col_sep == ','
-      CSV.parse_line(line).count
-    else
-      line.count(col_sep)+1
-    end
+    explode(line).count
   end
 
   def explode(line)
-    if col_sep == ','
-      CSV.parse_line(line).map(&:strip)
-    else
-      line.split(col_sep).map(&:strip)
-    end
+    CSV.parse_line(line, :col_sep => col_sep, :quote_char => @quote_char)
+  end
+
+  # a line is "junky" if it has 2 or fewer fields with any content
+  def junky?(line)
+    explode(line).select {|s| s && !s.strip.empty?}.count <= 2
   end
 end