将 wecom-notifier 升至 0.3.2 并锁定超长分段预算

pyproject.toml

@@ -32,7 +32,7 @@ dependencies = [
     "watchdog>=6.0.0",
     "websocket-client==1.5.1",
     "websockets>=13.0",
-    "wecom-notifier>=0.3.1",
+    "wecom-notifier>=0.3.2",
     "youtube-transcript-api>=1.2.3",
     "yt-dlp[default]>=2025.12.8",
     "zlx-ops-sdk>=0.1,<0.2",

tests/unit/test_wecom_notifier_chunking.py

@@ -0,0 +1,33 @@
+"""Offline regression for wecom-notifier markdown_v2 chunk budget.
+
+Locks the 0.3.2 fix for zlxlabs/wecom-notifier#2: after short lines, a
+~4.5KB single line must still yield chunks within the WeCom markdown_v2
+UTF-8 byte budget (page markers included).
+"""
+
+from wecom_notifier import MessageSegmenter
+
+# Upstream closeout (wecom-notifier#2): final markdown_v2 budget includes
+# page indicator, fences, and moderation expansion.
+WECOM_MARKDOWN_V2_BUDGET_BYTES = 3800
+
+
+def test_markdown_v2_chunks_stay_within_wecom_budget():
+    short_lines = "short line\n" * 500
+    oversized_line = "X" * 4500
+    content = short_lines + oversized_line + "\ntail line\n" * 50
+    assert len(content.encode("utf-8")) >= 10_000
+    assert "\n" not in oversized_line
+    assert len(oversized_line.encode("utf-8")) >= 4500
+
+    segments = MessageSegmenter().segment(content, "markdown_v2")
+
+    assert segments
+    oversize = [
+        (index, len(segment.content.encode("utf-8")))
+        for index, segment in enumerate(segments)
+        if len(segment.content.encode("utf-8")) > WECOM_MARKDOWN_V2_BUDGET_BYTES
+    ]
+    assert oversize == [], (
+        f"chunks exceeded markdown_v2 budget {WECOM_MARKDOWN_V2_BUDGET_BYTES}: {oversize}"
+    )