Skip UTF BOM from first data event on UTF-8 decoded stream

adaltas · Oct 18, 2012 · 2127daa · 2127daa
1 parent 5d6e262
commit 2127daa
Show file tree

Hide file tree

Showing 4 changed files with 25 additions and 3 deletions.
diff --git a/lib/from.js b/lib/from.js
diff --git a/src/from.coffee b/src/from.coffee
@@ -170,10 +170,14 @@ module.exports = (csv) ->
   ###
   from.stream = (stream, options) ->
     @options options
+    first = true
     stream.on 'data', (data) ->
       if csv.writable
-        if false is csv.write data.toString()
+        strip = first and typeof data is 'string' and stream.encoding is 'utf8' and 0xFEFF is data.charCodeAt 0
+        string = if strip then data.substring 1 else data.toString()
+        if false is csv.write string
           stream.pause()
+      first = false
     stream.on 'error', (e) ->
       csv.error e
     stream.on 'end', ->

diff --git a/test/from.coffee b/test/from.coffee
@@ -45,6 +45,17 @@ describe 'from', ->
       .on 'end', ->
         next()
 
+  describe 'path', ->
+
+    it 'should strip UTF-8 BOM', (next) ->
+      csv()
+      .from.path("#{__dirname}/from/file_bom.csv")
+      .on 'record', (record) ->
+        record.length.should.eql 5
+        record.should.eql ['1','2','3','4','5']
+      .on 'end', ->
+        next()
+
   describe 'stream', ->
 
     it 'should be able to pause', (next) ->

diff --git a/test/from/file_bom.csv b/test/from/file_bom.csv
@@ -0,0 +1 @@
+"1","2","3","4","5"