Merge pull request #1452 from fishtown-analytics/feature/strip-seed-bom

Handle seeds with utf-8 BOM (#1177)

core/dbt/clients/agate_helper.py

@@ -1,6 +1,11 @@
+from codecs import BOM_UTF8
+
+import dbt.compat
 
 import agate
 
+BOM = BOM_UTF8.decode('utf-8')  # '\ufeff'
+
 DEFAULT_TYPE_TESTER = agate.TypeTester(types=[
     agate.data_types.Number(null_values=('null', '')),
     agate.data_types.TimeDelta(null_values=('null', '')),
@@ -41,4 +46,7 @@ def as_matrix(table):
 
 
 def from_csv(abspath):
-    return agate.Table.from_csv(abspath, column_types=DEFAULT_TYPE_TESTER)
+    with dbt.compat.open_file(abspath) as fp:
+        if fp.read(1) != BOM:
+            fp.seek(0)
+        return agate.Table.from_csv(fp, column_types=DEFAULT_TYPE_TESTER)

core/dbt/compat.py

@@ -82,11 +82,20 @@ def to_native_string(s):
 
 def write_file(path, s):
     if WHICH_PYTHON == 2:
-        with codecs.open(path, 'w', encoding='utf-8') as f:
-            return f.write(to_string(s))
+        open = codecs.open
     else:
-        with open(path, 'w') as f:
-            return f.write(to_string(s))
+        open = builtins.open
+    with open(path, 'w', encoding='utf-8') as f:
+        return f.write(to_string(s))
+
+
+def open_file(path):
+    """Open the path for reading. It must be utf-8 encoded."""
+    if WHICH_PYTHON == 2:
+        open = codecs.open
+    else:
+        open = builtins.open
+    return open(path, encoding='utf-8')
 
 
 if WHICH_PYTHON == 2: