refactor: Phase 3 拆分 generate 流水线并补全测试

提取 collect/formatters/themes 等 pipeline 模块，新增 wecom 分条、RSS、delta、Agent 工作流测试。

Co-authored-by: Cursor <cursoragent@cursor.com>

tests/test_agent_workflow.py

@@ -0,0 +1,55 @@
+from __future__ import annotations
+
+import json
+
+
+def test_analyze_trends_parses_json(monkeypatch, isolated_output):
+    import daily.agent_workflow as agent
+    import daily.config as config
+
+    monkeypatch.setattr(config, "OUTPUT_DIR", isolated_output)
+    monkeypatch.setattr(agent, "OUTPUT_DIR", isolated_output)
+
+    monkeypatch.setattr(
+        agent,
+        "llm_chat",
+        lambda system, user: json.dumps(
+            {
+                "headline": "Skills 视频工具升温",
+                "opening": "今日 remotion 相关技能继续走强。",
+                "themes": [{"name": "视频", "summary": "Remotion 生态活跃"}],
+                "top_picks": [],
+                "signals": [],
+            },
+            ensure_ascii=False,
+        ),
+    )
+
+    llm_input = {"date": "2026-07-03", "skills_trending": []}
+    trends = agent.analyze_trends(llm_input, date_str="2026-07-03")
+
+    assert trends is not None
+    assert trends["headline"] == "Skills 视频工具升温"
+    assert (isolated_output / "2026-07-03.trends.json").exists()
+
+
+def test_write_wecom_report_extracts_markdown_block(monkeypatch):
+    import daily.agent_workflow as agent
+
+    monkeypatch.setattr(
+        agent,
+        "llm_chat",
+        lambda system, user: "```markdown\n📰 **早报 · 2026-07-03**\n\n正文\n```",
+    )
+
+    md = agent.write_wecom_report(
+        {"date": "2026-07-03"},
+        {"headline": "test", "opening": "open"},
+        date_str="2026-07-03",
+        time_str="09:30 (UTC+8)",
+        updated="2026-07-02",
+    )
+
+    assert md is not None
+    assert md.startswith("📰")
+    assert "正文" in md

tests/test_delta.py

@@ -0,0 +1,67 @@
+from __future__ import annotations
+
+import json
+
+import pytest
+
+
+def test_build_movement_context_detects_new_skill(isolated_output, monkeypatch):
+    import daily.config as config
+    import daily.delta as delta
+
+    monkeypatch.setattr(config, "OUTPUT_DIR", isolated_output)
+    monkeypatch.setattr(delta, "OUTPUT_DIR", isolated_output)
+
+    prev_path = isolated_output / "2026-07-02.data.json"
+    prev_path.write_text(
+        json.dumps(
+            {
+                "data": {
+                    "date": "2026-07-02",
+                    "movement_baseline": {
+                        "skills_trending": [{"id": "a/old", "title": "old"}],
+                        "skills_hot": [],
+                        "github_trending": [],
+                        "github_emerging": [],
+                        "github_topic": [],
+                    },
+                }
+            },
+            ensure_ascii=False,
+        ),
+        encoding="utf-8",
+    )
+
+    trending = [
+        {"id": "a/new", "title": "new-skill", "source": "a", "installs": 10, "link": "https://x", "description": ""},
+        {"id": "a/old", "title": "old", "source": "a", "installs": 9, "link": "https://y", "description": ""},
+    ]
+    movement = delta.build_movement_context(
+        date_str="2026-07-03",
+        trending=trending,
+        hot=[],
+        github_trending=[],
+        github_emerging=[],
+        github_topic=[],
+    )
+
+    assert movement["baseline_date"] == "2026-07-02"
+    assert len(movement["skills_trending_moves"]) == 1
+    assert movement["skills_trending_moves"][0]["id"] == "a/new"
+
+
+def test_find_previous_data_skips_missing_days(isolated_output, monkeypatch):
+    import daily.config as config
+    import daily.delta as delta
+
+    monkeypatch.setattr(config, "OUTPUT_DIR", isolated_output)
+    monkeypatch.setattr(delta, "OUTPUT_DIR", isolated_output)
+
+    (isolated_output / "2026-07-01.data.json").write_text(
+        json.dumps({"data": {"date": "2026-07-01", "skills_trending": []}}),
+        encoding="utf-8",
+    )
+
+    found = delta.find_previous_data("2026-07-03")
+    assert found is not None
+    assert found[0] == "2026-07-01"

tests/test_rss_parse.py

@@ -0,0 +1,54 @@
+from __future__ import annotations
+
+from pathlib import Path
+
+import pytest
+
+FIXTURES = Path(__file__).parent / "fixtures"
+
+
+def test_parse_sample_rss_fixture():
+    from daily.news.feeds import NEWS_CATEGORIES
+    from daily.news.fetch import _parse_feed
+
+    category = NEWS_CATEGORIES[0]
+    feed_name = category.feeds[0].name
+    xml = (FIXTURES / "sample-rss.xml").read_text(encoding="utf-8")
+    items = _parse_feed(xml, feed_name, category)
+
+    assert len(items) == 1
+    assert items[0]["title"] == "Sample AI headline for smoke tests"
+    assert items[0]["link"] == "https://example.com/ai-news/1"
+    assert items[0]["source_name"] == feed_name
+    assert "minimal RSS item" in items[0]["summary"]
+
+
+def test_fetch_ai_news_offline(monkeypatch):
+    from daily.news import fetch as news_fetch
+
+    sample = {
+        "title": "Sample AI headline for smoke tests",
+        "link": "https://example.com/ai-news/1",
+        "summary": "A minimal RSS item used by offline tests.",
+        "published": "Wed, 02 Jul 2026 08:00:00 GMT",
+        "source_name": "OpenAI",
+        "category_id": "official",
+        "category_name": "厂商官方",
+        "category_icon": "🏢",
+    }
+
+    monkeypatch.setattr(
+        news_fetch,
+        "_fetch_news",
+        lambda categories: {
+            "enabled": True,
+            "hours": 72,
+            "categories": [{"id": "official", "name": "厂商官方", "icon": "🏢", "items": [sample]}],
+            "flat": [sample],
+            "stats": {"feeds_total": 1, "feeds_ok": 1, "items_raw": 1, "feeds_failed": []},
+        },
+    )
+
+    payload = news_fetch.fetch_ai_news()
+    assert payload["enabled"] is True
+    assert payload["flat"][0]["title"].startswith("Sample AI")

tests/test_smoke.py

@@ -36,11 +36,12 @@ def test_generate_report_offline(
     monkeypatch,
 ):
     import daily.generate as generate
+    import daily.pipeline.collect as collect
 
-    monkeypatch.setattr(generate, "load_feed", lambda force=False: skills_feed)
-    monkeypatch.setattr(generate, "fetch_github_trending", lambda n: [_sample_github_repo()])
-    monkeypatch.setattr(generate, "fetch_emerging_repos", lambda n, exclude=None: [])
-    monkeypatch.setattr(generate, "fetch_topic_hot_repos", lambda n, exclude=None: ("llm", []))
+    monkeypatch.setattr(collect, "load_feed", lambda force=False: skills_feed)
+    monkeypatch.setattr(collect, "fetch_github_trending", lambda n: [_sample_github_repo()])
+    monkeypatch.setattr(collect, "fetch_emerging_repos", lambda n, exclude=None: [])
+    monkeypatch.setattr(collect, "fetch_topic_hot_repos", lambda n, exclude=None: ("llm", []))
     monkeypatch.setattr(generate, "_now_cst", lambda: fixed_cst)
 
     _, wecom_md, out_md, out_wecom = generate.generate_report()

tests/test_wecom_split.py

@@ -0,0 +1,34 @@
+from __future__ import annotations
+
+from daily.wecom_split import split_wecom_messages
+
+
+def test_split_wecom_messages_keeps_short_text():
+    text = "📰 **早报 · 2026-07-03**\n\n短内容"
+    parts = split_wecom_messages(text, limit=4096)
+    assert parts == [text]
+
+
+def test_split_wecom_messages_adds_part_footer():
+    section = "📰 **早报**\n\n" + ("正文行\n" * 200)
+    parts = split_wecom_messages(section, limit=600)
+    assert len(parts) > 1
+    assert all(len(part.encode("utf-8")) <= 600 for part in parts)
+    assert parts[0].endswith("1/" + str(len(parts)))
+
+
+def test_split_sections_on_emoji_headers():
+    from daily.wecom_split import _split_sections
+
+    text = "\n\n".join(
+        [
+            "📰 **早报 · 2026-07-03**",
+            "💡 **今日速览**\n- item one\n- item two",
+            "🌍 **国际 AI 时讯**\n1. [Headline](https://example.com)",
+        ]
+    )
+    sections = _split_sections(text)
+    assert len(sections) == 3
+    assert sections[0].startswith("📰")
+    assert sections[1].startswith("💡")
+    assert sections[2].startswith("🌍")