jfox-cli 1.2.2__tar.gz → 1.2.3__tar.gz

{jfox_cli-1.2.2 → jfox_cli-1.2.3}/CHANGELOG.md

@@ -2,6 +2,13 @@
 
 All notable changes to jfox-cli will be documented in this file.
 
+## [1.2.3] - 2026-07-07
+
+### Fixes
+- **gem-synth**: find_anchors 在 SQL 层排除已处理，修循环空转（#290） (#291)
+
+[1.2.3]: https://github.com/zhuxixi/jfox/compare/v1.2.2...v1.2.3
+
 ## [1.2.2] - 2026-06-28
 
 ### Fixes

{jfox_cli-1.2.2 → jfox_cli-1.2.3}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: jfox-cli
-Version: 1.2.2
+Version: 1.2.3
 Summary: JFox - Zettelkasten 知识管理 CLI 工具
 Project-URL: Homepage, https://github.com/zhuxixi/jfox
 Project-URL: Repository, https://github.com/zhuxixi/jfox

{jfox_cli-1.2.2 → jfox_cli-1.2.3}/jfox/__init__.py

@@ -1,5 +1,5 @@
 """JFox - Zettelkasten 知识管理工具"""
 
-__version__ = "1.2.2"
+__version__ = "1.2.3"
 __author__ = "User"
 __email__ = "user@example.com"

{jfox_cli-1.2.2 → jfox_cli-1.2.3}/jfox/daemon/server.py

@@ -352,6 +352,18 @@ def list_fragments(
 def main():
     from . import DEFAULT_HOST, DEFAULT_PORT
 
+    # 配置 jfox 日志：daemon 由 `python -m jfox.daemon.server` 启动，不走 cli.py 的
+    # basicConfig；不配则用默认 lastResort handler（WARNING 级），gem_synth 等模块的
+    # INFO 日志（tick / 后台循环已启动）全被吞 → 循环活动看不见（#290 日志盲点）。
+    # process.py 已把本进程 stderr 重定向到 ~/.jfox_daemon.log，故 StreamHandler 即落盘。
+    _jfox = logging.getLogger("jfox")
+    if not _jfox.handlers:  # 幂等：同进程多次调 main()（测试/手动）不重复挂 handler（#291 CR）
+        _h = logging.StreamHandler()
+        _h.setFormatter(logging.Formatter("%(asctime)s %(name)s %(levelname)s %(message)s"))
+        _jfox.addHandler(_h)
+    if _jfox.level == logging.NOTSET:  # 不覆盖调用方/测试自定义级别（kimi#6）
+        _jfox.setLevel(logging.INFO)
+
     parser = argparse.ArgumentParser(description="JFox Embedding Daemon")
     parser.add_argument("--host", default=DEFAULT_HOST, help="监听地址")
     parser.add_argument("--port", type=int, default=DEFAULT_PORT, help="监听端口")

{jfox_cli-1.2.2 → jfox_cli-1.2.3}/jfox/gem_synth/anchors.py

@@ -47,30 +47,36 @@ def find_anchors(
     where = _anchor_where(anchor_types)
     if not where:
         return []
-    sql = (
-        f"SELECT fragment_id, session_id, fragment_type, timestamp, content, metadata_json "
-        f"FROM session_fragments WHERE {where}"
-    )
-    params: List = []
-    if session_id:
-        sql += " AND session_id = ?"
-        params.append(session_id)
-    sql += " ORDER BY fragment_id LIMIT ?"
-    params.append(limit * 3)  # 多取再在 Python 侧按精确条件过滤
 
     conn = sqlite3.connect(str(fragments_db))
     conn.row_factory = sqlite3.Row
     try:
+        # ATTACH synthesis_log + NOT EXISTS 子查询：SQL 层排除已处理锚点（修 #290），
+        # 让 LIMIT 直接作用在未处理锚点上，无 NOT IN 参数上限（#291 CR kimi#3）。
+        # ATTACH 正常不失败（db_path 由 SynthesisLog 必建）；失败即真异常上抛，由 loop
+        # find_error 接管+记日志——不静默降级（静默会重复合成、产重复 candidate）。
+        conn.execute("ATTACH DATABASE ? AS syn", (str(log.db_path),))
+        exclude_processed = (
+            " AND NOT EXISTS (SELECT 1 FROM syn.synthesis_log sl "
+            "WHERE sl.anchor_fragment_id = session_fragments.fragment_id)"
+        )
+        sql = (
+            f"SELECT fragment_id, session_id, fragment_type, timestamp, content, "
+            f"metadata_json FROM session_fragments WHERE {where}{exclude_processed}"
+        )
+        params: List = []
+        if session_id:
+            sql += " AND session_id = ?"
+            params.append(session_id)
+        sql += " ORDER BY fragment_id LIMIT ?"
+        # 多取以容 Python 侧 ask_user_question 精确复核的少量剔除
+        params.append(limit * 3)
         rows = conn.execute(sql, params).fetchall()
     finally:
-        conn.close()
-
-    unprocessed = set(log.filter_unprocessed([r["fragment_id"] for r in rows]))
+        conn.close()  # 兜底覆盖 ATTACH + SELECT，避免 ATTACH 抛异常时连接泄漏（#291 CR R3）
 
     result: List[Dict] = []
     for r in rows:
-        if r["fragment_id"] not in unprocessed:
-            continue
         try:
             md = json.loads(r["metadata_json"] or "{}")
         except (json.JSONDecodeError, TypeError, ValueError):

{jfox_cli-1.2.2 → jfox_cli-1.2.3}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "hatchling.build"
 
 [project]
 name = "jfox-cli"
-version = "1.2.2"
+version = "1.2.3"
 description = "JFox - Zettelkasten 知识管理 CLI 工具"
 readme = "README.md"
 license = {text = "MIT"}

{jfox_cli-1.2.2 → jfox_cli-1.2.3}/tests/unit/test_gem_synth_anchors.py

@@ -106,6 +106,36 @@ def test_count_anchors_empty_anchor_types(tmp_path):
     assert count_anchors(fragments_db=tmp_path / "f.db", anchor_types=[]) == 0
 
 
+def test_find_anchors_skips_clustered_processed_and_returns_unprocessed(tmp_path):
+    """已处理锚点集中在低 fragment_id 时，find_anchors(limit=1) 仍须返回未处理的（#290）。
+
+    回归根因：旧实现 `LIMIT limit*3` 只取前几条，若前 3 条全是已处理的 → Python 侧
+    filter_unprocessed 全滤掉 → 返回 0，哪怕后面还有未处理锚点。daemon 循环用 limit=1，
+    每 tick 都取到这 3 条已处理的 → 永远返回 0 → 空转、不合成。
+    修法：把"已处理"过滤下推到 SQL（NOT IN），让 LIMIT 直接作用在未处理锚点上。
+    """
+    store = FragmentStore(db_path=tmp_path / "f.db")
+    # 前 3 条锚点（低 fragment_id）会被标记已处理；第 4 条未处理
+    a1 = store.insert("s", "correction", "UserPromptSubmit", "纠1", {})
+    a2 = store.insert("s", "correction", "UserPromptSubmit", "纠2", {})
+    a3 = store.insert("s", "correction", "UserPromptSubmit", "纠3", {})
+    a4 = store.insert("s", "correction", "UserPromptSubmit", "纠4", {})
+    store.close()
+
+    log = SynthesisLog(db_path=tmp_path / "syn.db")
+    log.mark_processed(a1, "c1")
+    log.mark_processed(a2, "c2")
+    log.mark_processed(a3, "c3")
+
+    # limit=1：旧实现 LIMIT 3 取 a1/a2/a3 全已处理 → 0；修后须跳过它们返回 a4
+    anchors = find_anchors(
+        fragments_db=tmp_path / "f.db", log=log, anchor_types=["correction"], limit=1
+    )
+    assert len(anchors) == 1
+    assert anchors[0]["fragment_id"] == a4
+    log.close()
+
+
 def test_find_anchors_tolerates_malformed_metadata_json(tmp_path):
     """session_fragments 含非法 JSON 的 metadata_json 行时，find_anchors/count_anchors
     都不应崩——坏行跳过、合法 ask_user_question 照常返回/计数。

{jfox_cli-1.2.2 → jfox_cli-1.2.3}/uv.lock

@@ -867,7 +867,7 @@ wheels = [
 
 [[package]]
 name = "jfox-cli"
-version = "1.2.2"
+version = "1.2.3"
 source = { editable = "." }
 dependencies = [
     { name = "chromadb" },