From 319b8124be57fada18c1d3e785a97a8cf18438ad Mon Sep 17 00:00:00 2001
From: Ilya Kreymer <ikreymer@gmail.com>
Date: Mon, 22 Sep 2014 21:12:25 -0700
Subject: [PATCH] cdxobject: add ability to create empty CDXObject(), add tests
 for CDXObject/IDXObject checking for supported and unsupported number of
 fields

---
 pywb/cdx/cdxobject.py           |  8 +++++++-
 pywb/cdx/test/test_cdxobject.py | 32 ++++++++++++++++++++++++++++++++
 2 files changed, 39 insertions(+), 1 deletion(-)
 create mode 100644 pywb/cdx/test/test_cdxobject.py

diff --git a/pywb/cdx/cdxobject.py b/pywb/cdx/cdxobject.py
index b64b0861..7a190be4 100644
--- a/pywb/cdx/cdxobject.py
+++ b/pywb/cdx/cdxobject.py
@@ -46,10 +46,16 @@ class CDXObject(OrderedDict):
          "orig.length", "orig.offset", "orig.filename"]
         ]
 
-    def __init__(self, cdxline):
+    def __init__(self, cdxline=''):
         OrderedDict.__init__(self)
 
         cdxline = cdxline.rstrip()
+
+        # Allows for filling the fields later or in a custom way
+        if not cdxline:
+            self.cdxline = cdxline
+            return
+
         fields = cdxline.split(' ')
 
         cdxformat = None
diff --git a/pywb/cdx/test/test_cdxobject.py b/pywb/cdx/test/test_cdxobject.py
new file mode 100644
index 00000000..13e242a7
--- /dev/null
+++ b/pywb/cdx/test/test_cdxobject.py
@@ -0,0 +1,32 @@
+from pywb.cdx.cdxobject import CDXObject, IDXObject, CDXException
+from pytest import raises
+
+def test_empty_cdxobject():
+    x = CDXObject('')
+    assert len(x) == 0
+
+def test_invalid_cdx_format():
+    with raises(CDXException):
+        x = CDXObject('a b c')
+
+
+def _make_line(fields):
+    line = ' '.join(['-'] * fields)
+    x = CDXObject(line)
+    assert len(x) == fields
+    assert str(x) == line
+
+def test_valid_cdx_formats():
+    # Currently supported cdx formats, 9, 11, 12, 14 field
+    # See CDXObject for more details
+    _make_line(9)
+    _make_line(12)
+
+    _make_line(11)
+    _make_line(14)
+
+def test_invalid_idx_format():
+    with raises(CDXException):
+        x = IDXObject('a b c')
+
+