Start validating entries before indexing them.

4 files changed, 84 insertions, 4 deletions

diff --git a/lesana/collection.py b/lesana/collection.py
index 738f6ad..8f76ea7 100644
--- a/lesana/collection.py
+++ b/lesana/collection.py
@@ -64,6 +64,40 @@ class Entry(object):
     def idterm(self):
         return "Q"+self.uid
 
+    def validate(self):
+        print("validating", self)
+        errors = []
+        valid = True
+        for field in self.collection.settings['fields']:
+            value = self.data.get(field['name'], None)
+            if not value:
+                # empty fields are always fine
+                continue
+            t = field['type']
+            if t == 'integer':
+                try:
+                    int(value)
+                except ValueError:
+                    valid = False
+                    errors.append({
+                        'field': field['name'],
+                        'error': 'Invalid value for integer field: {}'.format(
+                            value
+                            ),
+                        })
+            elif t == 'float':
+                try:
+                    float(value)
+                except ValueError:
+                    valid = False
+                    errors.append({
+                        'field': field['name'],
+                        'error': 'Invalid value for float field: {}'.format(
+                            value
+                            ),
+                        })
+        return valid, errors
+
 
 class Collection(object):
     """
@@ -104,7 +138,13 @@ class Collection(object):
                 data = ruamel.yaml.safe_load(fp)
             else:
                 data = ruamel.yaml.load(fp, ruamel.yaml.RoundTripLoader)
-            entry = self.entry_class(self, data, fname)
+        entry = self.entry_class(self, data, fname)
+        valid, errors = entry.validate()
+        if not valid:
+            logging.warning(
+                "Not indexing {fname}: invalid data".format(fname=fname)
+                )
+            return False, errors
 
         doc = xapian.Document()
         self.indexer.set_document(doc)
@@ -134,6 +174,7 @@ class Collection(object):
         doc.add_value(0, entry.fname.encode('utf-8'))
 
         cache.replace_document(entry.idterm, doc)
+        return True, []
 
     @property
     def indexed_fields(self):
@@ -176,14 +217,21 @@ class Collection(object):
         updated = 0
         for fname in fnames:
             try:
-                self._index_file(fname, cache)
+                valid, errors = self._index_file(fname, cache)
             except IOError as e:
                 logging.warning("Could not load file {}: {}".format(
                     fname,
                     str(e))
                     )
             else:
-                updated += 1
+                if valid:
+                    updated += 1
+                else:
+                    logging.warning(
+                        "File {fname} could not be indexed: {errors}".format(
+                            fname=fname,
+                            errors=errors)
+                        )
         return updated
 
     def save_entries(self, entries=[]):
diff --git a/tests/data/wrong/items/b9a832309c984ada9f267471660c1313.yaml b/tests/data/wrong/items/b9a832309c984ada9f267471660c1313.yaml
new file mode 100644
index 0000000..ec44b7c
--- /dev/null
+++ b/tests/data/wrong/items/b9a832309c984ada9f267471660c1313.yaml
@@ -0,0 +1,5 @@
+name: 'Problematic entry'
+description: |
+  .
+position: 'somewhere'
+number: 'four'
diff --git a/tests/data/wrong/settings.yaml b/tests/data/wrong/settings.yaml
index 9871421..ef9ab74 100644
--- a/tests/data/wrong/settings.yaml
+++ b/tests/data/wrong/settings.yaml
@@ -9,3 +9,6 @@ fields:
     - name: position
       type: string
       index: facet
+    - name: number
+      type: integer
+      help: "Enter an integer here"
diff --git a/tests/test_collection.py b/tests/test_collection.py
index 869828a..9de681a 100644
--- a/tests/test_collection.py
+++ b/tests/test_collection.py
@@ -54,7 +54,7 @@ class testCollection(unittest.TestCase):
         self.assertIsNotNone(self.collection.settings)
         self.assertIsNotNone(self.collection.stemmer)
         # Fields with no "index" entry are not indexed
-        self.assertEqual(len(self.collection.settings['fields']), 3)
+        self.assertEqual(len(self.collection.settings['fields']), 4)
         self.assertEqual(len(self.collection.indexed_fields), 1)
 
     def test_load_safe(self):
@@ -240,6 +240,30 @@ class testComplexCollection(unittest.TestCase):
         self.collection.update_cache()
 
 
+class testCollectionWithErrors(unittest.TestCase):
+    @classmethod
+    def setUpClass(self):
+        self.collection = lesana.Collection('tests/data/wrong')
+
+    @classmethod
+    def tearDownClass(self):
+        shutil.rmtree(os.path.join(self.collection.basedir, '.lesana'))
+
+    def test_init(self):
+        self.assertIsNotNone(self.collection.settings)
+        self.assertEqual(
+            self.collection.settings['name'],
+            "Lesana collection with certain errors"
+            )
+        self.assertEqual(len(self.collection.settings['fields']), 4)
+        self.assertIsNotNone(self.collection.stemmer)
+        self.assertEqual(len(self.collection.indexed_fields), 1)
+
+    def test_index(self):
+        loaded = self.collection.update_cache()
+        self.assertEqual(loaded, 0)
+
+
 class testCollectionCreation(unittest.TestCase):
     def test_init(self):
         tmpdir = tempfile.mkdtemp()