Merge pull request #54 from theodi/check-header-without-schema

Check header without schema

README.md

@@ -106,6 +106,8 @@ The following types of error can be reported:
 * `:whitespace` -- a quoted column has leading or trailing whitespace
 * `:line_breaks` -- line breaks were inconsistent or incorrectly specified
 * `:no_header` -- the CSV file doesn't include a header
+* `:empty_column_name` -- a column in the CSV header has an empty name
+* `:duplicate_column_name` -- a column in the CSV header has a duplicate name
 
 ## Warnings
 
@@ -190,8 +192,6 @@ Schema validation provides some additional types of error and warning messages:
 * `:extra_column` (warning) -- a row in the CSV file has extra column.
 * `:unique` (error) -- a column with a `unique` constraint contains non-unique values
 * `:out_of_range` (error) -- a column with a `minimum` or `maximum` constraint contains a value that is outside of the range
-* `:empty_column_name` (error) -- a column in the CSV header has an empty name
-* `:duplicate_column_name` (error) -- a column in the CSV header has a duplicate name
 
 ## Contributing
 

lib/csvlint/schema.rb

@@ -15,15 +15,9 @@ def initialize(uri, fields=[])
     end
 
     def validate_header(header)
-      names = Set.new
+      reset
       header.each_with_index do |name,i|
-        build_warnings(:header_name, :schema, nil, i+1) if fields[i].name.downcase != name.downcase
-        build_errors(:empty_column_name, :schema, nil, i+1) if name == ""
-        if names.include?(name)
-          build_errors(:duplicate_column_name, :schema, nil, i+1)
-        else
-          names << name
-        end
+        build_warnings(:header_name, :schema, nil, i+1) if fields[i].name != name
       end
       return valid?
     end

lib/csvlint/validate.rb

@@ -87,12 +87,8 @@ def parse_csv(io)
            row = csv.shift
            wrapper.finished
            if row             
-             if header? && current_line == 1 #header
-               if @schema
-                 @schema.validate_header(row)
-                 @errors += @schema.errors
-                 @warnings += @schema.warnings
-               end
+             if header? && current_line == 1
+               validate_header(row)
              else
 
                build_formats(row, current_line)
@@ -128,6 +124,23 @@ def parse_csv(io)
       return expected_columns        
     end          
 
+    def validate_header(header)
+      names = Set.new
+      header.each_with_index do |name,i|
+        build_errors(:empty_column_name, :schema, nil, i+1) if name == ""
+        if names.include?(name)
+          build_errors(:duplicate_column_name, :schema, nil, i+1)
+        else
+          names << name
+        end
+      end
+      if @schema
+        @schema.validate_header(header)
+        @errors += @schema.errors
+        @warnings += @schema.warnings
+      end
+      return valid?
+    end
 
     def header?
       return @csv_header

spec/schema_spec.rb

@@ -87,30 +87,11 @@
       expect( schema.warnings.size ).to eql(1)
       expect( schema.warnings.first.type).to eql(:header_name)
       expect( schema.warnings.first.category).to eql(:schema)
-    end
-
-    it "should error if column names aren't unique" do
-      minimum = Csvlint::Field.new("minimum", { "minLength" => 3 } )
-      required = Csvlint::Field.new("required", { "required" => true } )
-      schema = Csvlint::Schema.new("http://example.org", [minimum, required] )
 
-      expect( schema.validate_header(["minimum", "minimum"]) ).to eql(false)
-      expect( schema.errors.size ).to eql(1)
-      expect( schema.errors.first.type).to eql(:duplicate_column_name)
-      expect( schema.errors.first.category).to eql(:schema)
-    end
+      expect( schema.validate_header(["minimum", "Required"]) ).to eql(true)
+      expect( schema.warnings.size ).to eql(1)
 
-    it "should error if column names are blank" do
-      minimum = Csvlint::Field.new("minimum", { "minLength" => 3 } )
-      required = Csvlint::Field.new("required", { "required" => true } )
-      schema = Csvlint::Schema.new("http://example.org", [minimum, required] )
-
-      expect( schema.validate_header(["minimum", ""]) ).to eql(false)
-      expect( schema.errors.size ).to eql(1)
-      expect( schema.errors.first.type).to eql(:empty_column_name)
-      expect( schema.errors.first.category).to eql(:schema)
-    end
-
+    end        
   end  
 
   context "when parsing JSON Tables" do

spec/validator_spec.rb

@@ -66,6 +66,28 @@
 
   end
 
+  context "when validating headers" do
+    it "should error if column names aren't unique" do      
+      data = StringIO.new( "minimum, minimum" )
+      validator = Csvlint::Validator.new(data)
+      expect( validator.validate_header(["minimum", "minimum"]) ).to eql(false)
+      expect( validator.errors.size ).to eql(1)
+      expect( validator.errors.first.type).to eql(:duplicate_column_name)
+      expect( validator.errors.first.category).to eql(:schema)
+    end
+
+    it "should error if column names are blank" do
+      data = StringIO.new( "minimum," )
+      validator = Csvlint::Validator.new(data)
+
+      expect( validator.validate_header(["minimum", ""]) ).to eql(false)
+      expect( validator.errors.size ).to eql(1)
+      expect( validator.errors.first.type).to eql(:empty_column_name)
+      expect( validator.errors.first.category).to eql(:schema)
+    end
+
+  end
+
   context "build_formats" do
 
     it "should return the format of each column correctly" do