From 708a5d9331b2156a78b80dcc3dddcfc01c885d64 Mon Sep 17 00:00:00 2001
From: Elon-Jiang <191222665+Elon-Jiang@users.noreply.github.com>
Date: Sat, 31 Oct 2026 01:25:59 +0800
Subject: [PATCH] fix(memory): keep distinct topic pointers during
 consolidation

---
 CHANGELOG.md                |  6 ++++++
 CHANGELOG_ZH.md             |  5 +++++
 core/harness/memory.py      | 20 +++++++++++--------
 tests/test_memory_layers.py | 40 +++++++++++++++++++++++++++++++++++++
 4 files changed, 63 insertions(+), 8 deletions(-)

diff --git a/CHANGELOG.md b/CHANGELOG.md
index 27740df1e..bb294eab1 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -5,6 +5,12 @@ Released versions are named in the entry title; the version on PyPI is
 [deepcode-hku](https://pypi.org/project/deepcode-hku/). 中文版见
 [CHANGELOG_ZH.md](CHANGELOG_ZH.md).
 
+## 2026-10-31 · Keep distinct memory topics in the index
+
+- **Preserve different topic links during memory consolidation.** A filename
+  prefix or a `|` inside a title or target no longer makes separate pointers
+  collapse into one; repeated links still merge and prose still deduplicates.
+
 ## 2026-10-29 · Lighter compaction, checked Goal completion, and folder browsing on the Web
 
 - **Clear old tool output first.** With `agents.defaults.microcompact` on,
diff --git a/CHANGELOG_ZH.md b/CHANGELOG_ZH.md
index 88b980e32..183fa8610 100644
--- a/CHANGELOG_ZH.md
+++ b/CHANGELOG_ZH.md
@@ -4,6 +4,11 @@ DeepCode 的主要变化，最新的在前，每天一条。正式版本号写
 PyPI 上的包名是 [deepcode-hku](https://pypi.org/project/deepcode-hku/)。
 English: [CHANGELOG.md](CHANGELOG.md).
 
+## 2026-10-31 · 整理记忆索引时保留不同主题
+
+- **不再误合并不同主题链接。** 文件名前缀相同、标题或目标中包含 `|` 时，
+  不同的记忆指针仍会保留；重复指针和正文去重规则保持不变。
+
 ## 2026-10-29 · 更轻的上下文压缩、目标完成前先检查、Web 端可浏览文件夹
 
 - **先清掉旧的工具输出。** 打开 `agents.defaults.microcompact` 后，长会话接近
diff --git a/core/harness/memory.py b/core/harness/memory.py
index a09ca0dde..6947ec6f2 100644
--- a/core/harness/memory.py
+++ b/core/harness/memory.py
@@ -764,27 +764,31 @@ def _line_key(line: str) -> str:
     return text.rstrip(" .,;:!—-")
 
 
-def _candidate_key(line: str) -> str:
+_CandidateKey = str | tuple[str, str]
+
+
+def _candidate_key(line: str) -> _CandidateKey:
     """Dedupe key: pointer identity for pointers, normalized text otherwise."""
     pointer = parse_memory_pointer(line)
     if pointer is not None:
         # Same title + target with a different hook is one topic, not two.
-        return f"{pointer.title.casefold()}|{pointer.target.casefold()}"
+        return (pointer.title.casefold(), pointer.target.casefold())
     return _line_key(line)
 
 
-def _drop_subsumed(order: Sequence[str]) -> list[str]:
+def _drop_subsumed(order: Sequence[_CandidateKey]) -> list[_CandidateKey]:
     """Keys whose normalized form is a strict prefix of a longer entry.
 
     Safe by construction: a strict prefix is literally contained in the longer
     line, so folding the stub into it cannot lose content.
     """
-    dropped: set[str] = set()
+    dropped: set[_CandidateKey] = set()
     for key in order:
-        if len(key) < _MIN_USEFUL_CHARS:
+        # Prefix subsumption applies to prose, never to distinct topic links.
+        if not isinstance(key, str) or len(key) < _MIN_USEFUL_CHARS:
             continue
         for other in order:
-            if other == key or other in dropped:
+            if not isinstance(other, str) or other == key or other in dropped:
                 continue
             if len(other) > len(key) and other.startswith(key):
                 dropped.add(key)
@@ -801,8 +805,8 @@ def consolidate(lines: Sequence[str]) -> list[str]:
     order is first-seen, and first-seen is decided solely by the input
     sequence: no set or dict iteration decides anything.
     """
-    unique: dict[str, str] = {}
-    order: list[str] = []
+    unique: dict[_CandidateKey, str] = {}
+    order: list[_CandidateKey] = []
     for raw in lines:
         text = " ".join(str(raw).split())
         key = _candidate_key(text)
diff --git a/tests/test_memory_layers.py b/tests/test_memory_layers.py
index 80588f443..67cff50bf 100644
--- a/tests/test_memory_layers.py
+++ b/tests/test_memory_layers.py
@@ -30,6 +30,7 @@
     MemoryPointer,
     consolidate,
     consolidate_memory_index,
+    consolidate_pointer_index,
     fetch_memory_topic,
     is_pointer_index,
     memory_dir,
@@ -256,6 +257,45 @@ def test_consolidate_dedupes_and_merges_prefixes():
     ]
 
 
+@pytest.mark.parametrize(
+    "lines",
+    [
+        [
+            "- [API reference](api.md) — current endpoint contract",
+            "- [API reference](api.md.v2.md) — next version contract",
+        ],
+        [
+            "- [A|B](c.md) — first topic",
+            "- [A](B|c.md) — separate topic",
+        ],
+    ],
+)
+@pytest.mark.parametrize("reverse", [False, True])
+def test_consolidate_keeps_distinct_pointer_identities(lines, reverse):
+    if reverse:
+        lines = list(reversed(lines))
+    assert consolidate(lines) == lines
+
+
+def test_pointer_index_rewrite_keeps_distinct_topics(tmp_path):
+    first = "- [API reference](api.md) — current endpoint contract"
+    second = "- [API reference](api.md.v2.md) — next version contract"
+    index = _write(tmp_path, "MEMORY.md", f"{first}\n{second}\n{first}\n")
+    _write(tmp_path, "api.md", "Current endpoint: /v1/search\n")
+    _write(tmp_path, "api.md.v2.md", "Next endpoint: /v2/search\n")
+
+    assert consolidate_pointer_index(tmp_path)
+
+    assert index.read_text(encoding="utf-8") == f"{first}\n{second}\n"
+    assert (
+        fetch_memory_topic(tmp_path, "api.md").text == "Current endpoint: /v1/search\n"
+    )
+    assert (
+        fetch_memory_topic(tmp_path, "api.md.v2.md").text
+        == "Next endpoint: /v2/search\n"
+    )
+
+
 def test_prune_line_cap_at_the_boundary():
     lines = ["first line", "second line", "third line"]
     assert prune(lines, max_lines=3) == lines  # exactly at the cap