From b4b6e6723a3b4fe107e607ce0ec5b81f72e72a5a Mon Sep 17 00:00:00 2001
From: Elena ``of Valhalla'' Grandi <valhalla@trueelena.org>
Date: Fri, 6 Jan 2017 20:42:43 +0100
Subject: Changed search interface to be closer to xapian expectations

---
 lesana/collection.py     | 28 ++++++++++++++++++++++++----
 lesana/command.py        |  6 ++++--
 tests/test_collection.py | 23 ++++++++++++++++++++++-
 3 files changed, 50 insertions(+), 7 deletions(-)

diff --git a/lesana/collection.py b/lesana/collection.py
index 81b7a66..e4898f6 100644
--- a/lesana/collection.py
+++ b/lesana/collection.py
@@ -91,6 +91,7 @@ class Collection(object):
                 self.stemmer = xapian.Stem('english')
         else:
             self.stemmer = xapian.Stem('english')
+        self._enquire = None
         # This selects whether to load all other yaml files with
         # safe_load or load + RoundTripLoader
         self.safe = True
@@ -207,7 +208,10 @@ class Collection(object):
                 )
         return cache
 
-    def search(self, querystring, offset=0, pagesize=12):
+    def start_search(self, querystring):
+        """
+        Prepare a search for querystring.
+        """
         cache = self._get_cache()
         queryparser = xapian.QueryParser()
         queryparser.set_stemmer(self.stemmer)
@@ -217,12 +221,28 @@ class Collection(object):
 
         query = queryparser.parse_query(querystring)
 
-        enquire = xapian.Enquire(cache)
-        enquire.set_query(query)
+        self._enquire = xapian.Enquire(cache)
+        self._enquire.set_query(query)
 
-        for match in enquire.get_mset(offset, pagesize):
+    def get_search_results(self, offset=0, pagesize=12):
+        if not self._enquire:
+            return
+        for match in self._enquire.get_mset(offset, pagesize):
             yield self._match_to_entry(match)
 
+    def get_all_search_results(self):
+        if not self._enquire:
+            return
+        offset = 0
+        pagesize = 100
+        while True:
+            mset = self._enquire.get_mset(offset, pagesize)
+            if mset.size() == 0:
+                break
+            for match in mset:
+                yield self._match_to_entry(match)
+            offset += pagesize
+
     def _match_to_entry(self, match):
         fname = match.document.get_value(0).decode('utf-8')
         if self.safe:
diff --git a/lesana/command.py b/lesana/command.py
index 9101e8e..9425bd2 100644
--- a/lesana/command.py
+++ b/lesana/command.py
@@ -142,8 +142,10 @@ class Search(guacamole.Command):
             help='Am',
             )),
         (['--offset'], dict(
+            type=int,
             )),
         (['--pagesize'], dict(
+            type=int,
             )),
         (['query'], dict(
             help='Xapian query to search in the collection',
@@ -170,9 +172,9 @@ class Search(guacamole.Command):
         offset = ctx.args.offset or 0
         pagesize = ctx.args.pagesize or 12
         collection = Collection(ctx.args.collection)
+        collection.start_search(' '.join(ctx.args.query))
         # TODO: pass the entries to a proper template
-        for entry in collection.search(
-                ' '.join(ctx.args.query),
+        for entry in collection.get_search_results(
                 offset,
                 pagesize):
             print(entry)
diff --git a/tests/test_collection.py b/tests/test_collection.py
index d8190ca..a2c1a87 100644
--- a/tests/test_collection.py
+++ b/tests/test_collection.py
@@ -61,14 +61,31 @@ class testCollectionLoading(unittest.TestCase):
         self.collection.safe = False
         self.collection.update_cache()
 
+    def test_full_search(self):
+        self.collection = lesana.Collection('tests/data/simple')
+        self.collection.start_search('Item')
+        res = self.collection.get_all_search_results()
+        matches = list(res)
+        self.assertEqual(len(matches), 2)
+        for m in matches:
+            self.assertIsInstance(m, lesana.Entry)
+
     def test_search(self):
         self.collection = lesana.Collection('tests/data/simple')
-        res = self.collection.search('Item')
+        self.collection.start_search('Item')
+        res = self.collection.get_search_results()
         matches = list(res)
         self.assertEqual(len(matches), 2)
         for m in matches:
             self.assertIsInstance(m, lesana.Entry)
 
+    def test_search_non_init(self):
+        self.collection = lesana.Collection('tests/data/simple')
+        matches = list(self.collection.get_search_results())
+        self.assertEqual(matches, [])
+        matches = list(self.collection.get_all_search_results())
+        self.assertEqual(matches, [])
+
     def test_entry_from_uid(self):
         self.collection = lesana.Collection('tests/data/simple')
         entry = self.collection.entry_from_uid(
@@ -197,3 +214,7 @@ class testCollectionCreation(unittest.TestCase):
         self.assertTrue(os.path.isfile(os.path.join(tmpdir, 'settings.yaml')))
         self.assertFalse(os.path.isfile(os.path.join(tmpdir, '.gitignore')))
         shutil.rmtree(tmpdir)
+
+
+if __name__ == '__main__':
+    unittest.main()
-- 
cgit v1.2.3