Merge pull request #143 from wiki-ai/new_page_error

Fixes #121

revscoring/datasources/tests/test_diff.py

@@ -42,6 +42,13 @@ def test_operations():
 
     eq_(b, REVISION_TOKENS)
 
+    # Make sure we don't error when there is no parent revision
+    cache = {
+        parent_revision.text: None,
+        revision.text: REVISION_TEXT
+    }
+
+    operations, a, b = solve(diff.operations, cache=cache)
 
 def test_added_tokens():
     cache = {

revscoring/datasources/tests/test_parent_revision.py

@@ -10,6 +10,7 @@ def test_words():
                   cache={parent_revision.text: "Some text words 55."})
     eq_(words, ["Some", "text", "words"])
 
+    # Make sure we don't error when there is no parent revision
     words = solve(parent_revision.words,
                   cache={parent_revision.text: None})
     eq_(words, [])

revscoring/extractors/api.py

@@ -57,7 +57,7 @@ def __init__(self, session, language=None, context=None, cache=None):
             Datasource("parent_revision.doc", self.process_parent_revision_doc,
                        depends_on=[revision.metadata]),
             Datasource("parent_revision.metadata",
-                       self.process_revision_metadata,
+                       self.process_revision_metadata_if_exists,
                        depends_on=[parent_revision_doc]),
             Datasource("parent_revision.text",
                        self.process_revision_text,
@@ -66,7 +66,7 @@ def __init__(self, session, language=None, context=None, cache=None):
                        self.process_previous_user_revision_doc,
                        depends_on=[revision.metadata]),
             Datasource("previous_user_revision.metadata",
-                       self.process_revision_metadata,
+                       self.process_revision_metadata_if_exists,
                        depends_on=[previous_user_revision_doc]),
             Datasource("page_creation.doc",
                        self.process_page_creation_doc,
@@ -298,6 +298,13 @@ def process_revision_metadata(cls, revision_doc):
             raise RevisionDocumentNotFound()
         return cls.revision_metadata_from_doc(revision_doc)
 
+    @classmethod
+    def process_revision_metadata_if_exists(cls, revision_doc):
+        if revision_doc is None:
+            return None
+        else:
+            return cls.revision_metadata_from_doc(revision_doc)
+
     @classmethod
     def process_user_info(cls, user_doc):
         return cls.user_info_from_doc(user_doc)

revscoring/features/tests/test_parent_revision.py

@@ -34,6 +34,13 @@ def test_was_same_user():
     }
     eq_(solve(was_same_user, cache=cache), True)
 
+    # Make sure we don't error when there is no parent revision
+    cache = {
+        revision.metadata: FakeRevisionMetadata(None, "127.4.5.6"),
+        parent_revision.metadata: None
+    }
+    eq_(solve(was_same_user, cache=cache), False)
+
 def test_seconds_since():
     FakeRevisionMetadata = namedtuple("FakeRevisionMetadata",
                                       ['timestamp'])
@@ -44,6 +51,13 @@ def test_seconds_since():
     }
     eq_(solve(seconds_since, cache=cache), 9)
 
+    # Make sure we don't error when there is no parent revision
+    cache = {
+        revision.metadata: FakeRevisionMetadata(Timestamp(10)),
+        parent_revision.metadata: None
+    }
+    eq_(solve(seconds_since, cache=cache), 0)
+
 def test_bytes():
     FakeRevisionMetadata = namedtuple("FakeRevisionMetadata",
                                       ['bytes'])
@@ -53,18 +67,31 @@ def test_bytes():
     }
     eq_(solve(bytes, cache=cache), 25)
 
+    # Make sure we don't error when there is no parent revision
+    cache = {
+        parent_revision.metadata: None
+    }
+    eq_(solve(bytes, cache=cache), 0)
+
 def test_chars():
     cache = {
         parent_revision.text: "Twelve chars"
     }
     eq_(solve(chars, cache=cache), 12)
 
+    # Make sure we don't error when there is no parent revision
+    cache = {
+        parent_revision.text: None
+    }
+    eq_(solve(chars, cache=cache), 0)
+
 def test_markup_chars():
     cache = {
         parent_revision.text: "Twelve {{chars}}"
     }
     eq_(solve(markup_chars, cache=cache), 4)
 
+    # Make sure we don't error when there is no parent revision
     cache = {
         parent_revision.text: None
     }
@@ -76,6 +103,7 @@ def test_numeric_chars():
     }
     eq_(solve(numeric_chars, cache=cache), 2)
 
+    # Make sure we don't error when there is no parent revision
     cache = {
         parent_revision.text: None
     }
@@ -87,6 +115,7 @@ def test_symbolic_chars():
     }
     eq_(solve(symbolic_chars, cache=cache), 4)
 
+    # Make sure we don't error when there is no parent revision
     cache = {
         parent_revision.text: None
     }
@@ -98,6 +127,7 @@ def test_uppercase_chars():
     }
     eq_(solve(uppercase_chars, cache=cache), 2)
 
+    # Make sure we don't error when there is no parent revision
     cache = {
         parent_revision.text: None
     }

revscoring/features/tests/test_previous_user_revision.py

@@ -17,3 +17,10 @@ def test_seconds_since():
         previous_user_revision.metadata: FakeRevisionMetadata(Timestamp(1))
     }
     eq_(solve(seconds_since, cache=cache), 9)
+
+    # Makes sure we don't crash when there was no previous user revision
+    cache = {
+        revision.metadata: FakeRevisionMetadata(Timestamp(10)),
+        previous_user_revision.metadata: None
+    }
+    eq_(solve(seconds_since, cache=cache), 0)