feat(pipeline): LLM-extend usage scenario groups from comments

Add suggest_scenario_groups_llm: single call with excerpt corpus + existing
label/triggers, parse scenarios JSON, merge into comment_scenario_groups
(up to 40) before build_competitor_brief. Record in keyword_suggest_llm.json
(schema_version 3). Skip with MA_SKIP_LLM_SCENARIO_SUGGEST. Update tests
and analysis view hint.

Made-with: Cursor

backend/pipeline/jd_runner.py

@@ -18,7 +18,7 @@ from .models import PipelineJob
 
 def merge_llm_supplement_with_rules_report(llm_md: str, rules_md: str) -> str:
     """
-    **以规则引擎全文为正文**（含 §5 完整竞品矩阵、各章内嵌统计图与表格）。
+    **以规则引擎全文为正文**（含 §5 竞品矩阵——默认仅按细类价/声量条形图、无 Markdown 明细表，见 ``matrix_compact_section``；各章内嵌统计图与表格）。
 
     大模型稿作为 **§8.5** 嵌入在 **第八章末、第九章策略** 之前，与 §8.2～8.4 等具体分析同卷连贯，
     **不再**插在篇首「## 一、」之前。
@@ -166,8 +166,12 @@ def get_default_report_config() -> dict[str, Any]:
     """与 ``jd_competitor_report`` 模块常量一致的默认报告调参（供前端回填）。"""
     jcr, _ = _jd_crawler_modules()
     return {
-        "llm_comment_sentiment": False,
-        "llm_section_bridges": False,
+        "llm_comment_sentiment": True,
+        "llm_section_bridges": True,
+        "llm_matrix_group_summaries": True,
+        "llm_comment_group_summaries": True,
+        "llm_price_group_summaries": True,
+        "matrix_compact_section": True,
         "comment_focus_words": list(jcr.COMMENT_FOCUS_WORDS),
         "comment_scenario_groups": [
             {"label": lbl, "triggers": list(trs)}
@@ -211,13 +215,13 @@ def write_competitor_analysis_for_run_dir(
         except json.JSONDecodeError:
             meta = None
 
-    eff_rc: dict[str, Any] = (
-        dict(report_config) if isinstance(report_config, dict) else {}
-    )
+    eff_rc: dict[str, Any] = dict(get_default_report_config())
+    if isinstance(report_config, dict) and report_config:
+        eff_rc.update(report_config)
     all_tx = _flat_comment_texts(comment_rows)
     suggest_path = run_dir / "keyword_suggest_llm.json"
     suggest_record: dict[str, Any] = {
-        "schema_version": 2,
+        "schema_version": 3,
         "total_comment_texts": len(all_tx),
     }
     skip_kw = os.environ.get("MA_SKIP_LLM_KEYWORD_SUGGEST", "").strip().lower() in (
@@ -267,6 +271,54 @@ def write_competitor_analysis_for_run_dir(
         suggest_record["skipped"] = True
         suggest_record["suggested_focus_keywords"] = []
 
+    skip_scen = os.environ.get("MA_SKIP_LLM_SCENARIO_SUGGEST", "").strip().lower() in (
+        "1",
+        "true",
+        "yes",
+    )
+    if not skip_scen:
+        try:
+            from .llm_keyword_suggest import suggest_scenario_groups_llm
+
+            scen_base = [x for x in (eff_rc.get("comment_scenario_groups") or []) if isinstance(x, dict)]
+            scen_out = suggest_scenario_groups_llm(
+                keyword=kw,
+                existing_groups=scen_base,
+                all_comment_texts=all_tx,
+            )
+            suggest_record["suggested_scenario_groups"] = scen_out.get(
+                "suggested_scenario_groups"
+            ) or []
+            suggest_record["scenario_rationale"] = scen_out.get("scenario_rationale") or ""
+            exist_labels = {
+                str(x.get("label") or "").strip().lower()
+                for x in scen_base
+                if str(x.get("label") or "").strip()
+            }
+            merged_scen = list(scen_base)
+            for g in suggest_record["suggested_scenario_groups"]:
+                if not isinstance(g, dict):
+                    continue
+                lab = str(g.get("label") or "").strip()
+                tr_in = g.get("triggers")
+                triggers: list[str] = []
+                if isinstance(tr_in, list):
+                    for t in tr_in[:48]:
+                        s = str(t).strip()
+                        if 2 <= len(s) <= 48:
+                            triggers.append(s)
+                if not lab or lab.lower() in exist_labels or len(triggers) < 2:
+                    continue
+                merged_scen.append({"label": lab[:80], "triggers": triggers[:48]})
+                exist_labels.add(lab.lower())
+            eff_rc["comment_scenario_groups"] = merged_scen[:40]
+        except Exception as e:
+            suggest_record["scenario_error"] = str(e)
+            suggest_record["suggested_scenario_groups"] = []
+    else:
+        suggest_record["scenario_skipped"] = True
+        suggest_record["suggested_scenario_groups"] = []
+
     suggest_path.write_text(
         json.dumps(suggest_record, ensure_ascii=False, indent=2),
         encoding="utf-8",
@@ -304,7 +356,9 @@ def write_competitor_analysis_for_run_dir(
     )
     want_sent = bool(eff_rc.get("llm_comment_sentiment")) or env_on
     if want_sent and not skip_sent:
-        comment_units = jcr._iter_comment_text_units(comment_rows, merged_rows)
+        comment_units = jcr._comment_lines_with_product_context(
+            comment_rows, merged_rows
+        )
         if len(comment_units) >= 2:
             sentiment_llm_record["attempted"] = True
             try:
@@ -336,6 +390,127 @@ def write_competitor_analysis_for_run_dir(
         encoding="utf-8",
     )
 
+    matrix_llm_record: dict[str, Any] = {
+        "schema_version": 1,
+        "attempted": False,
+    }
+    llm_matrix_groups_md = ""
+    skip_mat_llm = os.environ.get(
+        "MA_SKIP_LLM_MATRIX_SUMMARIES", ""
+    ).strip().lower() in ("1", "true", "yes")
+    want_mat_llm = bool(eff_rc.get("llm_matrix_group_summaries"))
+    matrix_compact = bool(eff_rc.get("matrix_compact_section", True))
+    if want_mat_llm and matrix_compact and not skip_mat_llm:
+        pl_groups = jcr.build_matrix_groups_llm_payload(merged_rows)
+        if pl_groups:
+            matrix_llm_record["attempted"] = True
+            try:
+                from .llm_generate import generate_matrix_group_summaries_llm
+
+                llm_matrix_groups_md = generate_matrix_group_summaries_llm(
+                    pl_groups, keyword=kw
+                )
+                matrix_llm_record["ok"] = True
+                matrix_llm_record["chars"] = len(llm_matrix_groups_md)
+            except Exception as e:
+                matrix_llm_record["ok"] = False
+                matrix_llm_record["error"] = str(e)
+        else:
+            matrix_llm_record["skipped"] = "no_matrix_groups"
+    elif skip_mat_llm:
+        matrix_llm_record["skipped"] = "MA_SKIP_LLM_MATRIX_SUMMARIES"
+    elif not want_mat_llm:
+        matrix_llm_record["skipped"] = "not_enabled"
+    elif not matrix_compact:
+        matrix_llm_record["skipped"] = "matrix_not_compact"
+
+    (run_dir / "matrix_section_llm.json").write_text(
+        json.dumps(matrix_llm_record, ensure_ascii=False, indent=2),
+        encoding="utf-8",
+    )
+
+    comment_groups_llm_record: dict[str, Any] = {
+        "schema_version": 1,
+        "attempted": False,
+    }
+    llm_comment_groups_md = ""
+    skip_cg_llm = os.environ.get(
+        "MA_SKIP_LLM_COMMENT_GROUP_SUMMARIES", ""
+    ).strip().lower() in ("1", "true", "yes")
+    want_cg_llm = bool(eff_rc.get("llm_comment_group_summaries", True))
+    if want_cg_llm and not skip_cg_llm:
+        fw_cg, _, _ = jcr.resolve_report_tuning(eff_rc)
+        fb_cg = jcr._consumer_feedback_by_matrix_group(
+            merged_rows=merged_rows,
+            comment_rows=comment_rows,
+            sku_header="SKU(skuId)",
+        )
+        pl_cg = jcr.build_comment_groups_llm_payload(
+            feedback_groups=fb_cg,
+            focus_words=fw_cg,
+            merged_rows=merged_rows,
+        )
+        if pl_cg:
+            comment_groups_llm_record["attempted"] = True
+            try:
+                from .llm_generate import generate_comment_group_summaries_llm
+
+                llm_comment_groups_md = generate_comment_group_summaries_llm(
+                    pl_cg, keyword=kw
+                )
+                comment_groups_llm_record["ok"] = True
+                comment_groups_llm_record["chars"] = len(llm_comment_groups_md)
+            except Exception as e:
+                comment_groups_llm_record["ok"] = False
+                comment_groups_llm_record["error"] = str(e)
+        else:
+            comment_groups_llm_record["skipped"] = "no_feedback_groups"
+    elif skip_cg_llm:
+        comment_groups_llm_record["skipped"] = "MA_SKIP_LLM_COMMENT_GROUP_SUMMARIES"
+    elif not want_cg_llm:
+        comment_groups_llm_record["skipped"] = "not_enabled"
+
+    (run_dir / "comment_groups_llm.json").write_text(
+        json.dumps(comment_groups_llm_record, ensure_ascii=False, indent=2),
+        encoding="utf-8",
+    )
+
+    price_section_llm_record: dict[str, Any] = {
+        "schema_version": 1,
+        "attempted": False,
+    }
+    llm_price_groups_md = ""
+    skip_pg_llm = os.environ.get(
+        "MA_SKIP_LLM_PRICE_GROUP_SUMMARIES", ""
+    ).strip().lower() in ("1", "true", "yes")
+    want_pg_llm = bool(eff_rc.get("llm_price_group_summaries", True))
+    if want_pg_llm and not skip_pg_llm:
+        pl_pg = jcr.build_price_groups_llm_payload(merged_rows)
+        if pl_pg:
+            price_section_llm_record["attempted"] = True
+            try:
+                from .llm_generate import generate_price_group_summaries_llm
+
+                llm_price_groups_md = generate_price_group_summaries_llm(
+                    pl_pg, keyword=kw
+                )
+                price_section_llm_record["ok"] = True
+                price_section_llm_record["chars"] = len(llm_price_groups_md)
+            except Exception as e:
+                price_section_llm_record["ok"] = False
+                price_section_llm_record["error"] = str(e)
+        else:
+            price_section_llm_record["skipped"] = "no_price_groups"
+    elif skip_pg_llm:
+        price_section_llm_record["skipped"] = "MA_SKIP_LLM_PRICE_GROUP_SUMMARIES"
+    elif not want_pg_llm:
+        price_section_llm_record["skipped"] = "not_enabled"
+
+    (run_dir / "price_section_llm.json").write_text(
+        json.dumps(price_section_llm_record, ensure_ascii=False, indent=2),
+        encoding="utf-8",
+    )
+
     md = jcr.build_competitor_markdown(
         run_dir=run_dir,
         keyword=kw,
@@ -345,6 +520,9 @@ def write_competitor_analysis_for_run_dir(
         meta=meta,
         report_config=eff_rc,
         llm_sentiment_section_md=llm_sentiment_md or None,
+        llm_matrix_groups_md=llm_matrix_groups_md or None,
+        llm_comment_groups_md=llm_comment_groups_md or None,
+        llm_price_groups_md=llm_price_groups_md or None,
     )
 
     bridge_record: dict[str, Any] = {
@@ -472,17 +650,18 @@ def build_competitor_brief_for_job(
         except json.JSONDecodeError:
             meta = None
 
-    eff: dict[str, Any] | None = None
-    if isinstance(report_config, dict):
-        eff = dict(report_config)
+    eff: dict[str, Any] = dict(get_default_report_config())
     eff_path = base / "effective_report_config.json"
     if eff_path.is_file():
         try:
             loaded = json.loads(eff_path.read_text(encoding="utf-8"))
             if isinstance(loaded, dict) and loaded:
-                eff = loaded
+                eff.update(loaded)
         except json.JSONDecodeError:
             pass
+    # 任务上显式保存的 report_config 优先于目录内快照（避免旧 effective 覆盖用户 PATCH）。
+    if isinstance(report_config, dict) and report_config:
+        eff.update(report_config)
 
     return jcr.build_competitor_brief(
         run_dir=base,
@@ -510,6 +689,10 @@ def run_jd_keyword_and_report(
     report_config: dict[str, Any] | None = None,
     cancel_check: Any | None = None,
 ) -> Path:
+    """
+    执行京东关键词流水线至 **CSV / run_meta 落盘** 为止；**不写** ``competitor_analysis.md``。
+    ``report_config`` 保留与调用方兼容，采集阶段不使用；报告请用 ``regenerate_competitor_report`` 或 API 生成。
+    """
     _, kpl = _jd_crawler_modules()
 
     kw = (keyword or "").strip()
@@ -561,21 +744,13 @@ def run_jd_keyword_and_report(
             kpl.SCENARIO_FILTER_ENABLED = bool(scenario_filter_enabled)
 
         run_dir = kpl.main(keyword=kw)
-    except kpl.PipelineCancelled as e:
-        run_dir_path = Path(e.run_dir).resolve()
-        merged = run_dir_path / kpl.FILE_MERGED_CSV
-        if merged.is_file():
-            try:
-                write_competitor_analysis_for_run_dir(
-                    run_dir_path, kw, report_config=report_config
-                )
-            except Exception:
-                pass
+    except kpl.PipelinePausedForCookie:
+        raise
+    except kpl.PipelineCancelled:
         raise
     finally:
         for name, val in backup.items():
             setattr(kpl, name, val)
 
-    return write_competitor_analysis_for_run_dir(
-        Path(run_dir).resolve(), kw, report_config=report_config
-    )
+    # 竞品 Markdown 不在采集任务内生成；由前端「重新生成报告」或 ``regenerate_competitor_report`` 触发。
+    return Path(run_dir).resolve()

backend/pipeline/llm_keyword_suggest.py

@@ -1,4 +1,4 @@
-"""在报告生成前：基于**全量**评价文本分块调用大模型，联想补充关注词（参与后续统计与报告）。"""
+"""在报告生成前：基于评价正文调用大模型，联想补充**关注词**与**使用场景触发组**（写入 effective_report_config）。"""
 from __future__ import annotations
 
 import json
@@ -11,6 +11,8 @@ from django.conf import settings
 
 MAX_CHUNK_CHARS = 24_000
 MAX_CHUNKS = 12
+# 场景联想单次送入模型的评价摘录上限（字符）；过大易顶上下文
+SCENARIO_CORPUS_MAX_CHARS = 18_000
 
 _CHUNK_SYSTEM = """你是电商评价挖掘助手。输入 JSON 含 keyword、excerpt_index、excerpts（一段用户评价正文合集）。
 任务：从 excerpts 中抽取值得纳入「关注词/卖点监测」的**中文短语**（2～12 字为主，可为词组）。
@@ -90,6 +92,127 @@ def _parse_phrases_object(raw: str) -> list[str]:
     return []
 
 
+_SCENARIO_SYSTEM = """你是电商用户研究助手。输入 JSON 含：
+- ``keyword``：监测词；
+- ``existing_scenarios``：数组，每项为 ``{"label": "展示名", "triggers": ["子串1", ...]}``。统计时若评价正文**包含任一 trigger 子串**，则计入该 label（与宿主系统规则一致）。
+- ``excerpts``：多条用户评价正文摘录（已截断拼接）。
+
+任务：在**不重复** ``existing_scenarios`` 中已有 ``label``（逐字比较，勿改写字）的前提下，从 excerpts 归纳 **4～12 条**新的「用途/场景」监测组，覆盖评论里**明显出现但未被现有组覆盖**的消费情境（如「下午茶」「露营」「宿舍」等，须确有文本依据）。
+
+硬性规则：
+- **仅输出**一段 JSON：``{"scenarios": [{"label": "展示名", "triggers": ["子串1", "子串2", ...]}, ...]}``；
+- 每条 ``label`` 2～16 字；每组 ``triggers`` 3～10 条，每条 trigger 为 **2～12 字中文**子串，用于**子串命中**计数；
+- 不要医疗功效、治愈、降血糖承诺；不要与 existing 的 label 同名或仅差空格；
+- 不要输出 JSON 以外的文字。"""
+
+
+def _sample_corpus_for_scenarios(texts: list[str], *, max_chars: int) -> str:
+    """取评价正文前部拼接至 max_chars，供单次场景联想。"""
+    parts: list[str] = []
+    n = 0
+    for t in texts:
+        s = (t or "").strip()
+        if not s:
+            continue
+        extra = len(s) + 1
+        if n + extra > max_chars:
+            remain = max_chars - n - 1
+            if remain > 40:
+                parts.append(s[:remain])
+            break
+        parts.append(s)
+        n += extra
+    return "\n".join(parts)
+
+
+def _parse_scenarios_object(raw: str) -> list[dict[str, Any]]:
+    t = (raw or "").strip()
+    t = re.sub(r"^```(?:json)?\s*", "", t, flags=re.IGNORECASE)
+    t = re.sub(r"\s*```$", "", t)
+    try:
+        obj = json.loads(t)
+    except json.JSONDecodeError:
+        obj = None
+    if not isinstance(obj, dict):
+        m = re.search(r"\{[\s\S]*\}", t)
+        if not m:
+            return []
+        try:
+            obj = json.loads(m.group(0))
+        except json.JSONDecodeError:
+            return []
+    arr = obj.get("scenarios")
+    if not isinstance(arr, list):
+        return []
+    out: list[dict[str, Any]] = []
+    for item in arr:
+        if not isinstance(item, dict):
+            continue
+        label = str(item.get("label") or "").strip()[:80]
+        tr_raw = item.get("triggers")
+        triggers: list[str] = []
+        if isinstance(tr_raw, list):
+            seen_t: set[str] = set()
+            for x in tr_raw[:24]:
+                s = str(x).strip()
+                if len(s) < 2 or len(s) > 24:
+                    continue
+                if s in seen_t:
+                    continue
+                seen_t.add(s)
+                triggers.append(s)
+        if label and len(triggers) >= 1:
+            out.append({"label": label, "triggers": triggers[:12]})
+    return out
+
+
+def suggest_scenario_groups_llm(
+    *,
+    keyword: str,
+    existing_groups: list[dict[str, Any]],
+    all_comment_texts: list[str],
+) -> dict[str, Any]:
+    """
+    单次调用模型，基于评价摘录扩展 ``comment_scenario_groups`` 形态的新组（label + triggers）。
+    """
+    if not all_comment_texts:
+        return {
+            "suggested_scenario_groups": [],
+            "scenario_rationale": "无评价正文可分析。",
+        }
+    existing_compact: list[dict[str, Any]] = []
+    for g in (existing_groups or [])[:36]:
+        if not isinstance(g, dict):
+            continue
+        lab = str(g.get("label") or "").strip()
+        tr = g.get("triggers")
+        ts: list[str] = []
+        if isinstance(tr, list):
+            for x in tr[:16]:
+                s = str(x).strip()
+                if s:
+                    ts.append(s[:48])
+        if lab and ts:
+            existing_compact.append({"label": lab[:80], "triggers": ts})
+    excerpts = _sample_corpus_for_scenarios(
+        all_comment_texts, max_chars=SCENARIO_CORPUS_MAX_CHARS
+    )
+    payload = {
+        "keyword": keyword,
+        "existing_scenarios": existing_compact,
+        "excerpts": excerpts,
+    }
+    raw = _call_llm(_SCENARIO_SYSTEM, json.dumps(payload, ensure_ascii=False))
+    scenarios = _parse_scenarios_object(raw)
+    return {
+        "suggested_scenario_groups": scenarios[:14],
+        "scenario_rationale": (
+            f"基于约 {len(excerpts)} 字评价摘录单次调用模型；"
+            f"在 {len(existing_compact)} 组既有场景之外补充 {len(scenarios[:14])} 组候选。"
+        ),
+    }
+
+
 def suggest_focus_keywords_from_all_comments(
     *,
     keyword: str,
@@ -102,7 +225,6 @@ def suggest_focus_keywords_from_all_comments(
     if not all_comment_texts:
         return {
             "suggested_focus_keywords": [],
-            "suggested_scenario_hints": [],
             "rationale": "无评价正文可分析。",
             "chunks_processed": 0,
             "total_comment_texts": 0,
@@ -140,7 +262,6 @@ def suggest_focus_keywords_from_all_comments(
     out_kw = merged[:22]
     return {
         "suggested_focus_keywords": out_kw,
-        "suggested_scenario_hints": [],
         "rationale": (
             f"基于全量 {len(all_comment_texts)} 条评价文本，分 {len(chunks)} 段调用模型抽取短语并去重；"
             f"已排除与当前关注词统计表完全相同的词。"

backend/pipeline/tests/test_llm_keyword_suggest.py

@@ -21,7 +21,9 @@ from pipeline.llm_keyword_suggest import (
     MAX_CHUNKS,
     _chunk_comment_texts,
     _parse_phrases_object,
+    _parse_scenarios_object,
     suggest_focus_keywords_from_all_comments,
+    suggest_scenario_groups_llm,
 )
 
 
@@ -67,6 +69,20 @@ class ParsePhrasesObjectTests(SimpleTestCase):
         self.assertEqual(_parse_phrases_object("not json"), [])
 
 
+class ParseScenariosObjectTests(SimpleTestCase):
+    def test_plain_json(self) -> None:
+        raw = '{"scenarios": [{"label": "下午茶", "triggers": ["下午茶", "配咖啡"]}]}'
+        out = _parse_scenarios_object(raw)
+        self.assertEqual(len(out), 1)
+        self.assertEqual(out[0]["label"], "下午茶")
+        self.assertEqual(out[0]["triggers"], ["下午茶", "配咖啡"])
+
+    def test_fenced(self) -> None:
+        raw = '```\n{"scenarios": [{"label": "A", "triggers": ["x", "y"]}]}\n```'
+        out = _parse_scenarios_object(raw)
+        self.assertEqual(out[0]["label"], "A")
+
+
 class SuggestFocusKeywordsTests(SimpleTestCase):
     def test_no_comments_returns_empty(self) -> None:
         out = suggest_focus_keywords_from_all_comments(
@@ -108,3 +124,31 @@ class SuggestFocusKeywordsLiveLLMTests(SimpleTestCase):
             self.assertGreaterEqual(len(p), 2)
             self.assertLessEqual(len(p), 24)
         self.assertNotIn("甜度", kws)
+
+
+@unittest.skipUnless(
+    _llm_configured(),
+    "需要 OPENAI_* 或 LLM_* 密钥与网关地址。",
+)
+class SuggestScenarioGroupsLiveLLMTests(SimpleTestCase):
+    def test_live_suggests_new_scenario_groups(self) -> None:
+        existing = [
+            {"label": "早餐/代餐", "triggers": ["早餐", "代餐"]},
+        ]
+        comments = [
+            "下午配咖啡当下午茶还不错，办公室同事分着吃。",
+            "周末露营带了一盒，孩子当零食。",
+        ]
+        out = suggest_scenario_groups_llm(
+            keyword="饼干",
+            existing_groups=existing,
+            all_comment_texts=comments,
+        )
+        groups = out.get("suggested_scenario_groups") or []
+        self.assertIsInstance(groups, list)
+        self.assertGreater(len(groups), 0, "应至少返回 1 组新场景")
+        labels = {str(g.get("label", "")).strip().lower() for g in groups if isinstance(g, dict)}
+        self.assertNotIn("早餐/代餐", labels)
+        for g in groups:
+            tr = g.get("triggers") or []
+            self.assertGreaterEqual(len(tr), 1)

frontend/src/views/jd/JdAnalysisView.vue

@@ -13,7 +13,7 @@ import {
   jobExportReportDocumentUrl,
 } from '../../composables/useJobs'
 import {
-  generationInFlightKey,
+  generationInFlightKeys,
   withGenerationInFlight,
 } from '../../composables/useGenerationInFlight'
 
@@ -40,26 +40,49 @@ const reportMdForPreview = computed(() =>
   reportMdWithAssetUrls(reportMd.value, selectedId.value),
 )
 
-const genInFlight = generationInFlightKey()
+const inflight = generationInFlightKeys()
 const K_PREVIEW = 'preview-report:'
 const K_BRIEF = 'competitor-brief:'
 const K_PACK = 'brief-pack:'
+const K_REGEN = 'regenerate-report:'
 function genKeyMatches(prefix) {
   const id = selectedId.value
   if (!id) return false
-  return genInFlight.value === `${prefix}${id}`
+  return inflight.value.includes(`${prefix}${id}`)
 }
+/** 与当前选中任务一致（用于文案） */
 const loading = computed(() => genKeyMatches(K_PREVIEW))
 const briefLoading = computed(() => genKeyMatches(K_BRIEF))
 const packLoading = computed(() => genKeyMatches(K_PACK))
+/** 仍有预览/摘要/打包请求在进行（不因换页签后选中 id 被重置而误判为空闲） */
+const previewBusyAny = computed(() => inflight.value.some((x) => x.startsWith(K_PREVIEW)))
+const briefBusyAny = computed(() => inflight.value.some((x) => x.startsWith(K_BRIEF)))
+const packBusyAny = computed(() => inflight.value.some((x) => x.startsWith(K_PACK)))
+const previewBusyJobId = computed(() => {
+  const k = inflight.value.find((x) => x.startsWith(K_PREVIEW))
+  return k ? k.slice(K_PREVIEW.length) : ''
+})
+const briefBusyJobId = computed(() => {
+  const k = inflight.value.find((x) => x.startsWith(K_BRIEF))
+  return k ? k.slice(K_BRIEF.length) : ''
+})
+const packBusyJobId = computed(() => {
+  const k = inflight.value.find((x) => x.startsWith(K_PACK))
+  return k ? k.slice(K_PACK.length) : ''
+})
 const viewInFlightOtherJobId = computed(() => {
-  const k = genInFlight.value
-  if (!k) return null
-  const i = k.lastIndexOf(':')
-  if (i < 0) return null
-  const jid = k.slice(i + 1)
-  if (jid === selectedId.value) return null
-  return jid
+  for (const k of inflight.value) {
+    const i = k.lastIndexOf(':')
+    if (i < 0) continue
+    const jid = k.slice(i + 1)
+    if (jid !== selectedId.value) return jid
+  }
+  return null
+})
+/** 从「报告生成」页发起的重新生成尚未结束（与预览/打包并行跟踪） */
+const reportRegenBusyJobId = computed(() => {
+  const k = inflight.value.find((x) => x.startsWith(K_REGEN))
+  return k ? k.slice(K_REGEN.length) : null
 })
 
 const successJobs = computed(() =>
@@ -206,11 +229,14 @@ watch(
       <h2>分析报告查看</h2>
       <p class="hint-top">
         选择<strong>已成功</strong>的任务，在线阅读报告或下载。
-        流水线生成报告时会<strong>自动</strong>基于<strong>全部评价正文</strong>分块调用大模型扩展关注词，并写入统计图（PNG，见「二点五」章与简报包 <code>report_assets</code>）。
+        流水线生成报告时会<strong>自动</strong>基于<strong>全部评价正文</strong>分块调用大模型扩展<strong>关注词</strong>，并单次调用归纳<strong>使用场景</strong>（在预设场景组之外追加 label+触发子串），再写入统计图（PNG，见「二点五」章与简报包 <code>report_assets</code>）。
         <strong>一键下载简报包</strong>含报告稿、统计图、结构化 JSON、要点摘录。
         需要改规则或重算，请至
         <RouterLink to="/jd/analysis-build">报告生成</RouterLink>。
       </p>
+      <p class="hint-top hint-distinguish">
+        <strong>状态说明：</strong>「任务列表」里的<strong>待执行 / 执行中</strong>是<strong>流水线采集</strong>；本页按钮若显示<strong>请求处理中</strong>，表示<strong>当前浏览器</strong>正在等待预览/摘要/打包等<strong>读接口</strong>，二者不要混为一谈。
+      </p>
 
       <div class="toolbar">
         <label class="sel-label">任务</label>
@@ -220,8 +246,19 @@ watch(
             #{{ j.id }} · {{ j.keyword }} · {{ j.run_dir?.split(/[/\\]/).pop() || '' }}
           </option>
         </select>
-        <button type="button" class="ma-btn ma-btn-secondary" :disabled="!selectedId || loading" @click="loadReport">
-          {{ loading ? '加载中…' : '重新加载报告' }}
+        <button
+          type="button"
+          class="ma-btn ma-btn-secondary"
+          :disabled="!selectedId || previewBusyAny"
+          @click="loadReport"
+        >
+          {{
+            loading
+              ? '请求处理中（加载报告预览）…'
+              : previewBusyAny
+                ? `请求处理中（任务 #${previewBusyJobId} 预览）…`
+                : '重新加载报告'
+          }}
         </button>
         <a
           class="ma-btn ma-btn-secondary dl-link"
@@ -256,24 +293,39 @@ watch(
         <button
           type="button"
           class="ma-btn ma-btn-secondary"
-          :disabled="!selectedId || briefLoading || loading"
+          :disabled="!selectedId || briefBusyAny || previewBusyAny || packBusyAny"
           title="生成与报告相同统计口径的结构化数据"
           @click="loadCompetitorBrief"
         >
-          {{ briefLoading ? '摘要加载中…' : '加载结构化摘要' }}
+          {{
+            briefLoading
+              ? '请求处理中（结构化摘要）…'
+              : briefBusyAny
+                ? `请求处理中（任务 #${briefBusyJobId} 摘要）…`
+                : '加载结构化摘要'
+          }}
         </button>
         <button
           type="button"
           class="ma-btn ma-btn-primary"
-          :disabled="!selectedId || packLoading || loading || briefLoading"
+          :disabled="!selectedId || packBusyAny || previewBusyAny || briefBusyAny"
           title="ZIP：报告稿、结构化数据、要点摘录、说明"
           @click="downloadBriefPack"
         >
-          {{ packLoading ? '打包中…' : '一键下载简报包' }}
+          {{
+            packLoading
+              ? '请求处理中（简报包）…'
+              : packBusyAny
+                ? `请求处理中（任务 #${packBusyJobId} 简报包）…`
+                : '一键下载简报包'
+          }}
         </button>
       </div>
+      <p v-if="reportRegenBusyJobId" class="ma-warn-banner">
+        任务 #{{ reportRegenBusyJobId }} 的<strong>重新生成报告</strong>仍在进行中（可能从「报告生成」页发起）；预览与下载可能在写入完成后才反映最新稿。
+      </p>
       <p v-if="viewInFlightOtherJobId" class="ma-warn-banner">
-        任务 #{{ viewInFlightOtherJobId }} 仍有请求进行中；当前页切换任务后若按钮已恢复，请等待该任务完成或返回对应任务查看。
+        本浏览器对任务 #{{ viewInFlightOtherJobId }} 的<strong>预览 / 摘要 / 打包</strong>请求尚未结束（仅本页读接口等待，<strong>不是</strong>「任务列表」里的流水线执行中）。请稍候再操作或切回该任务。
       </p>
 
       <p v-if="selectedJob?.run_dir" class="run-dir-note ma-muted">