Add --reset to delete the xapian cache before indexing

4 files changed, 23 insertions, 2 deletions

diff --git a/docs/source/man/lesana-index.rst b/docs/source/man/lesana-index.rst
index 0ba52f1..a4a7ae4 100644
--- a/docs/source/man/lesana-index.rst
+++ b/docs/source/man/lesana-index.rst
@@ -20,4 +20,6 @@ OPTIONS
    Prints an help message and exits.
 --collection COLLECTION, -c COLLECTION
    The collection to work on. Default is ``.``
+--reset
+   Delete the existing xapian cache before indexing.
 
diff --git a/lesana/collection.py b/lesana/collection.py
index af3977d..efd329e 100644
--- a/lesana/collection.py
+++ b/lesana/collection.py
@@ -1,6 +1,7 @@
 import io
 import logging
 import os
+import shutil
 import uuid
 
 import ruamel.yaml
@@ -228,7 +229,7 @@ class Collection(object):
                 )
         return fields
 
-    def update_cache(self, fnames=None):
+    def update_cache(self, fnames=None, reset=False):
         """
         Update the xapian db with the data in files.
 
@@ -236,8 +237,13 @@ class Collection(object):
 
         If no files have been passed, add everything.
 
+        if ``reset`` the existing xapian db is deleted before indexing
+
         Return the number of files that have been added to the cache.
         """
+        if reset:
+            shutil.rmtree(os.path.join(self.basedir, '.lesana'))
+        os.makedirs(os.path.join(self.basedir, '.lesana'), exist_ok=True)
         cache = xapian.WritableDatabase(
             os.path.join(self.basedir, '.lesana/xapian'),
             xapian.DB_CREATE_OR_OPEN,
diff --git a/lesana/command.py b/lesana/command.py
index 31a452b..c3d1ba9 100644
--- a/lesana/command.py
+++ b/lesana/command.py
@@ -214,6 +214,13 @@ class Index(Command):
             dict(help='The collection to work on (default .)'),
         ),
         (
+            ['--reset'],
+            dict(
+                action='store_true',
+                help='Delete the existing index and reindex from scratch.',
+            ),
+        ),
+        (
             ['files'],
             dict(
                 help='List of files to index (default: everything)',
@@ -229,7 +236,10 @@ class Index(Command):
             files = (os.path.basename(f) for f in self.args.files)
         else:
             files = None
-        indexed = collection.update_cache(fnames=files)
+        indexed = collection.update_cache(
+            fnames=files,
+            reset=self.args.reset
+        )
         print("Found and indexed {} entries".format(indexed))
 
 
diff --git a/tests/test_collection.py b/tests/test_collection.py
index 8d4cf88..6a65ff3 100644
--- a/tests/test_collection.py
+++ b/tests/test_collection.py
@@ -219,6 +219,9 @@ class testSimpleCollection(unittest.TestCase):
         self.assertEqual(len(cm.output), 1)
         self.assertIn("non_existing_file", cm.output[0])
 
+    def test_index_reset(self):
+        self.collection.update_cache(reset=True)
+
     def test_get_entry_missing_eid(self):
         entry = self.collection.entry_from_eid('this is not an eid')
         self.assertIsNone(entry)