From 7e98746d785b6f0ba461865ed0873bb35daca8ad Mon Sep 17 00:00:00 2001
From: Morgan Carr <256248948+morgan-coded@users.noreply.github.com>
Date: Fri, 4 Sep 2026 07:11:47 -0500
Subject: [PATCH] Delete every index entry for an id in Terms (#1219)

---
 src/python/txtai/scoring/terms.py      | 21 ++++++++++++++-------
 test/python/testscoring/testkeyword.py | 25 +++++++++++++++++++++++++
 2 files changed, 39 insertions(+), 7 deletions(-)

diff --git a/src/python/txtai/scoring/terms.py b/src/python/txtai/scoring/terms.py
index ea7d1314..ed7e5cda 100644
--- a/src/python/txtai/scoring/terms.py
+++ b/src/python/txtai/scoring/terms.py
@@ -129,14 +129,21 @@ class Terms:
             ids: ids to delete
         """
 
-        # Set index ids as deleted, ignoring ids that were never indexed and ids
-        # already marked deleted. Skipping duplicates keeps count() accurate when
-        # the same id is deleted more than once (e.g. repeated upsert/delete flows).
-        for i in ids:
-            if i in self.ids:
-                indexid = self.ids.index(i)
-                if indexid not in self.deletes:
+        # Walk each requested id from its last match to mark every index id, including re-added ids
+        deletes = set(self.deletes)
+        for uid in ids:
+            start = 0
+            while True:
+                try:
+                    indexid = self.ids.index(uid, start)
+                except ValueError:
+                    break
+
+                if indexid not in deletes:
                     self.deletes.append(indexid)
+                    deletes.add(indexid)
+
+                start = indexid + 1
 
     def index(self):
         """
diff --git a/test/python/testscoring/testkeyword.py b/test/python/testscoring/testkeyword.py
index 7a3ba3b6..6220f6cb 100644
--- a/test/python/testscoring/testkeyword.py
+++ b/test/python/testscoring/testkeyword.py
@@ -186,6 +186,31 @@ class TestKeyword(unittest.TestCase):
         scoring.delete([0])
         self.assertEqual(scoring.count(), total - 1)
 
+    def testDeleteReinsert(self):
+        """
+        Test that an id deleted, re-added and deleted again is removed from count() and search()
+        """
+
+        # Terms index: Terms.delete resolved each id with self.ids.index(), which only found the first
+        # (already deleted) position. The re-added copy survived, count() was one too high and search()
+        # raised a KeyError for the re-added text since the content had been removed.
+        for method in ["bm25", "tfidf", "sif"]:
+            scoring = ScoringFactory.create({"method": method, "terms": True, "content": True})
+            scoring.index(self.data)
+
+            total = scoring.count()
+
+            scoring.delete([0])
+            self.assertEqual(scoring.count(), total - 1)
+
+            # Re-add the same id with new text, then delete it again
+            scoring.upsert([(0, "Lunar eclipse visible across North America", None)])
+            self.assertEqual(scoring.count(), total)
+
+            scoring.delete([0])
+            self.assertFalse(scoring.search("lunar eclipse", 1))
+            self.assertEqual(scoring.count(), total - 1)
+
     def testTFIDF(self):
         """
         Test tfidf