From 2d55fd2f5a59149865cfc6fb14d202e797840eee Mon Sep 17 00:00:00 2001
From: BukeLy <bukely0119@foxmail.com>
Date: Sun, 31 May 2026 21:11:26 +0800
Subject: [PATCH] fix(pifs): clean partial add pageindex cache

---
 pageindex/filesystem/core.py   |  3 ++-
 tests/test_pifs_add_command.py | 42 ++++++++++++++++++++++++++++++++++
 2 files changed, 44 insertions(+), 1 deletion(-)

diff --git a/pageindex/filesystem/core.py b/pageindex/filesystem/core.py
index ce2c336..d8f6310 100644
--- a/pageindex/filesystem/core.py
+++ b/pageindex/filesystem/core.py
@@ -1756,11 +1756,12 @@ class PageIndexFileSystem:
         records: list[dict[str, Any]],
         preexisting_doc_ids: set[str],
     ) -> None:
-        doc_ids: list[str] = []
+        doc_ids = sorted(self._pageindex_cache_doc_ids() - preexisting_doc_ids)
         for record in records:
             doc_id = str(record.get("pageindex_doc_id") or "").strip()
             if doc_id and doc_id not in preexisting_doc_ids:
                 doc_ids.append(doc_id)
+        doc_ids = sorted(set(doc_ids))
         if not doc_ids:
             return
         workspace = self.pageindex_client_workspace
diff --git a/tests/test_pifs_add_command.py b/tests/test_pifs_add_command.py
index b221ae0..d2b8f9c 100644
--- a/tests/test_pifs_add_command.py
+++ b/tests/test_pifs_add_command.py
@@ -288,6 +288,48 @@ def test_add_markdown_insert_failure_removes_pageindex_cache(tmp_path, monkeypat
     assert not list((workspace / "artifacts" / "raw").glob("*.json"))
 
 
+def test_add_markdown_index_failure_removes_pageindex_cache_delta(tmp_path, monkeypatch):
+    from pageindex import PageIndexClient
+
+    def fake_index(self, file_path, mode="auto"):
+        doc_id = "doc_partial_before_raise"
+        doc = {
+            "id": doc_id,
+            "type": "md",
+            "path": str(Path(file_path).resolve()),
+            "doc_name": "partial.md",
+            "doc_description": "",
+            "line_count": 3,
+            "structure": [{"title": "Partial", "node_id": "0001", "nodes": []}],
+        }
+        self.documents[doc_id] = doc
+        self._save_doc(doc_id)
+        raise RuntimeError("index failed after cache write")
+
+    monkeypatch.setattr(PageIndexClient, "index", fake_index)
+    source = tmp_path / "partial.md"
+    source.write_text("# Partial\n\nbody", encoding="utf-8")
+    workspace = tmp_path / "workspace"
+    filesystem = make_filesystem(workspace)
+    pageindex_workspace = workspace / "artifacts" / "pageindex_client"
+
+    with pytest.raises(RuntimeError, match="failed to build PageIndex tree"):
+        filesystem.add_file(source, "/documents/reports")
+
+    assert not (pageindex_workspace / "doc_partial_before_raise.json").exists()
+    meta_path = pageindex_workspace / "_meta.json"
+    if meta_path.exists():
+        meta = json.loads(meta_path.read_text(encoding="utf-8"))
+        assert "doc_partial_before_raise" not in meta
+    listing = filesystem.browse("/", recursive=True)
+    assert listing["files"] == []
+    assert listing["folders"] == []
+    assert filesystem.summary_projection_indexer.index.info()["document_count"] == 0
+    assert not list((workspace / "artifacts" / "uploads").glob("**/*"))
+    assert not list((workspace / "artifacts" / "text").glob("*.txt"))
+    assert not list((workspace / "artifacts" / "raw").glob("*.json"))
+
+
 def test_add_markdown_failure_preserves_unrelated_pageindex_cache(tmp_path, monkeypatch):
     from pageindex import PageIndexClient