mirror of
https://github.com/primedigitaltech/market-assistant.git
synced 2026-07-21 23:41:39 +08:00
refactor(jd): 反馈分组、列表统计、矩阵行与报告片段拆至 competitor_report
新增 consumer_feedback、run_context、list_mix、matrix_md、report_md_helpers；jd_competitor_report 从配置块后直接进入 build_competitor_markdown；移除未再使用的 ingredients 直引。 Made-with: Cursor
This commit is contained in:
hub-gif
parent
8bbb921552
commit
b6a52c04d2
@ -0,0 +1,161 @@
|
||||
"""评价与合并表按细类矩阵对齐：带前缀行、SKU→细类映射、按细类消费者反馈分组。"""
|
||||
from __future__ import annotations
|
||||
|
||||
from pipeline.csv_schema import MERGED_FIELD_TO_CSV_HEADER
|
||||
|
||||
from .comment_sentiment import _iter_comment_text_units
|
||||
from .constants import (
|
||||
_COMMENT_CSV_BODY,
|
||||
_COMMENT_CSV_SKU,
|
||||
_MERGED_SHOP_CELL_KEYS,
|
||||
)
|
||||
from .csv_io import _cell, _md_cell
|
||||
from .matrix_group import _competitor_matrix_group_key, _merged_rows_grouped_for_matrix
|
||||
|
||||
|
||||
def _comment_lines_with_product_context(
|
||||
comment_rows: list[dict[str, str]],
|
||||
merged_rows: list[dict[str, str]],
|
||||
*,
|
||||
sku_header: str,
|
||||
title_h: str,
|
||||
) -> list[str]:
|
||||
"""与 ``comment_rows`` 顺序对齐：带细类/SKU/品名/店铺前缀，供 §8.2 大模型抽样。"""
|
||||
sku_meta: dict[str, tuple[str, str, str]] = {}
|
||||
for row in merged_rows:
|
||||
sku = _cell(row, sku_header).strip()
|
||||
if not sku:
|
||||
continue
|
||||
gk = _competitor_matrix_group_key(row)
|
||||
if not gk:
|
||||
continue
|
||||
sku_meta[sku] = (
|
||||
gk,
|
||||
_cell(row, title_h),
|
||||
_cell(row, *_MERGED_SHOP_CELL_KEYS),
|
||||
)
|
||||
out: list[str] = []
|
||||
for cr in comment_rows:
|
||||
txt = _cell(cr, _COMMENT_CSV_BODY, "tagCommentContent")
|
||||
if not txt:
|
||||
continue
|
||||
sku = _cell(cr, _COMMENT_CSV_SKU, "sku").strip()
|
||||
meta = sku_meta.get(sku)
|
||||
if meta:
|
||||
gname, tit, shop = meta
|
||||
prefix = (
|
||||
f"【细类：{gname}｜SKU：{sku}｜品名：{_md_cell(tit, 80)}｜"
|
||||
f"店铺：{_md_cell(shop, 40)}】"
|
||||
)
|
||||
out.append(prefix + txt)
|
||||
else:
|
||||
out.append(txt)
|
||||
return out
|
||||
|
||||
|
||||
def _sku_to_matrix_group_map(
|
||||
merged_rows: list[dict[str, str]], sku_header: str
|
||||
) -> dict[str, str]:
|
||||
m: dict[str, str] = {}
|
||||
for row in merged_rows:
|
||||
sku = _cell(row, sku_header).strip()
|
||||
if not sku:
|
||||
continue
|
||||
gk = _competitor_matrix_group_key(row)
|
||||
if gk:
|
||||
m[sku] = gk
|
||||
return m
|
||||
|
||||
|
||||
def _comment_text_units_for_matrix_group(
|
||||
gname: str,
|
||||
merged_rows: list[dict[str, str]],
|
||||
comment_rows_in_group: list[dict[str, str]],
|
||||
sku_header: str,
|
||||
) -> list[str]:
|
||||
"""某细类下的评价正文列表；无 flat 时用该细类合并行的 comment_preview。"""
|
||||
texts: list[str] = []
|
||||
for row in comment_rows_in_group:
|
||||
t = _cell(row, _COMMENT_CSV_BODY, "tagCommentContent")
|
||||
if t:
|
||||
texts.append(t)
|
||||
if texts:
|
||||
return texts
|
||||
for row in merged_rows:
|
||||
if _competitor_matrix_group_key(row) != gname:
|
||||
continue
|
||||
p = _cell(
|
||||
row,
|
||||
MERGED_FIELD_TO_CSV_HEADER["comment_preview"],
|
||||
"comment_preview",
|
||||
)
|
||||
if p:
|
||||
texts.append(p)
|
||||
return texts
|
||||
|
||||
|
||||
def _consumer_feedback_by_matrix_group(
|
||||
*,
|
||||
merged_rows: list[dict[str, str]],
|
||||
comment_rows: list[dict[str, str]],
|
||||
sku_header: str,
|
||||
) -> list[tuple[str, list[dict[str, str]], list[str]]]:
|
||||
"""
|
||||
与 §5 矩阵同序的细类列表；每项为 (细类名, 该类的 comments_flat 行, 用于场景统计的文本单元)。
|
||||
评价 SKU 不在深入样本时归入「未归类（评价 SKU 无对应深入样本）」。
|
||||
"""
|
||||
if not merged_rows:
|
||||
if not comment_rows:
|
||||
return []
|
||||
texts = _iter_comment_text_units(comment_rows, [])
|
||||
return [
|
||||
(
|
||||
"未归类（无深入合并表）",
|
||||
list(comment_rows),
|
||||
texts,
|
||||
)
|
||||
]
|
||||
|
||||
sku_map = _sku_to_matrix_group_map(merged_rows, sku_header)
|
||||
merged_by_sku: dict[str, dict[str, str]] = {}
|
||||
for row in merged_rows:
|
||||
s = _cell(row, sku_header).strip()
|
||||
if s:
|
||||
merged_by_sku[s] = row
|
||||
by_g: dict[str, list[dict[str, str]]] = {}
|
||||
for row in comment_rows:
|
||||
sku = _cell(row, _COMMENT_CSV_SKU, "sku").strip()
|
||||
g = sku_map.get(sku)
|
||||
if g:
|
||||
by_g.setdefault(g, []).append(row)
|
||||
continue
|
||||
if sku and sku in merged_by_sku:
|
||||
# 深入样本存在但缺 detail_category_path（或路径无法解析为可读细类）：不参与按细类分析
|
||||
continue
|
||||
by_g.setdefault("未归类（评价 SKU 无对应深入样本）", []).append(row)
|
||||
|
||||
out: list[tuple[str, list[dict[str, str]], list[str]]] = []
|
||||
used: set[str] = set()
|
||||
for gname, _ in _merged_rows_grouped_for_matrix(merged_rows):
|
||||
cr = by_g.get(gname, [])
|
||||
tu = _comment_text_units_for_matrix_group(
|
||||
gname, merged_rows, cr, sku_header
|
||||
)
|
||||
out.append((gname, cr, tu))
|
||||
used.add(gname)
|
||||
for gname, cr in sorted(by_g.items(), key=lambda x: (-len(x[1]), x[0])):
|
||||
if gname in used:
|
||||
continue
|
||||
tu = _comment_text_units_for_matrix_group(
|
||||
gname, merged_rows, cr, sku_header
|
||||
)
|
||||
out.append((gname, cr, tu))
|
||||
return out
|
||||
|
||||
|
||||
__all__ = [
|
||||
"_comment_lines_with_product_context",
|
||||
"_comment_text_units_for_matrix_group",
|
||||
"_consumer_feedback_by_matrix_group",
|
||||
"_sku_to_matrix_group_map",
|
||||
]
|
||||
138
backend/crawler_copy/jd_pc_search/competitor_report/list_mix.py
Normal file
138
backend/crawler_copy/jd_pc_search/competitor_report/list_mix.py
Normal file
@ -0,0 +1,138 @@
|
||||
"""列表可见度代理指标、品牌/店铺扇图用的名称列表与计数。"""
|
||||
from __future__ import annotations
|
||||
|
||||
from collections import Counter
|
||||
from typing import Any
|
||||
|
||||
from pipeline.csv_schema import JD_SEARCH_CSV_HEADERS, MERGED_FIELD_TO_CSV_HEADER
|
||||
|
||||
from .constants import (
|
||||
_LEGACY_LIST_BRAND_TITLE_KEY,
|
||||
_LEGACY_SHOP_NAME_KEY,
|
||||
_LIST_BRAND_TITLE_HEADER,
|
||||
_MERGED_SHOP_CELL_KEYS,
|
||||
)
|
||||
from .csv_io import _cell, _collect_prices
|
||||
from .price_stats import _price_stats_extended
|
||||
|
||||
|
||||
def _structure_names_for_pie_counter(row_names: list[str]) -> list[str]:
|
||||
"""
|
||||
与 ``_counter_mix_top_rows_with_remainder`` / 列表品牌·店铺扇图同一套规则：
|
||||
按 strip 后的名称逐行保留一条，便于 ``_brand_cr`` 与饼图 Counter 一致。
|
||||
"""
|
||||
return [(x or "").strip() for x in row_names if (x or "").strip()]
|
||||
|
||||
|
||||
def _brand_cr(cnames: list[str]) -> tuple[float | None, float | None, str, str]:
|
||||
"""按名称计数返回 (第一大主体份额, 前三合计份额, 头部标签, 头部占比展示字符串)。"""
|
||||
if not cnames:
|
||||
return None, None, "", ""
|
||||
cnt = Counter(cnames)
|
||||
total = sum(cnt.values())
|
||||
if total <= 0:
|
||||
return None, None, "", ""
|
||||
mc = cnt.most_common()
|
||||
top1_n = mc[0][1] if mc else 0
|
||||
top1 = mc[0][0] if mc else ""
|
||||
cr1 = top1_n / total
|
||||
top3_n = sum(n for _, n in mc[:3])
|
||||
cr3 = top3_n / total
|
||||
return cr1, cr3, top1, f"{100.0 * top1_n / total:.1f}%"
|
||||
|
||||
|
||||
def _counter_mix_top_rows_with_remainder(
|
||||
row_names: list[str], *, top_n: int, remainder_label: str
|
||||
) -> list[tuple[str, int]]:
|
||||
"""
|
||||
与列表品牌/店铺扇图一致：按 strip 后的名称计数；``most_common(top_n)`` 未覆盖的长尾合并为
|
||||
``remainder_label``，保证各块 count 之和等于可统计行数（与 ``_structure_names_for_pie_counter`` 总条数一致）。
|
||||
"""
|
||||
c = Counter((x or "").strip() for x in row_names if (x or "").strip())
|
||||
if not c:
|
||||
return []
|
||||
total = sum(c.values())
|
||||
common = c.most_common(top_n)
|
||||
accounted = sum(v for _, v in common)
|
||||
rest = total - accounted
|
||||
out: list[tuple[str, int]] = list(common)
|
||||
if rest > 0:
|
||||
out.append((remainder_label, rest))
|
||||
return out
|
||||
|
||||
|
||||
def _search_list_proxies(rows: list[dict[str, str]]) -> dict[str, Any]:
|
||||
"""
|
||||
基于 pc_search_export 的「列表可见度」指标，**不是**全渠道零售额或 TAM。
|
||||
"""
|
||||
sku_k = JD_SEARCH_CSV_HEADERS["sku_id"]
|
||||
shop_k = JD_SEARCH_CSV_HEADERS["shop_name"]
|
||||
page_k = JD_SEARCH_CSV_HEADERS["page"]
|
||||
cat_k = JD_SEARCH_CSV_HEADERS["leaf_category"]
|
||||
skus: set[str] = set()
|
||||
shops: set[str] = set()
|
||||
pages: set[str] = set()
|
||||
cats: set[str] = set()
|
||||
for r in rows:
|
||||
s = _cell(r, sku_k)
|
||||
if s:
|
||||
skus.add(s)
|
||||
sh = _cell(r, shop_k)
|
||||
if sh:
|
||||
shops.add(sh)
|
||||
pg = _cell(r, page_k)
|
||||
if pg:
|
||||
pages.add(pg)
|
||||
c = _cell(r, cat_k)
|
||||
if c:
|
||||
cats.add(c)
|
||||
prices = _collect_prices(rows)
|
||||
pst = _price_stats_extended(prices)
|
||||
return {
|
||||
"total_rows": len(rows),
|
||||
"unique_skus": len(skus),
|
||||
"unique_shops": len(shops),
|
||||
"unique_pages": len(pages),
|
||||
"page_span": (min((int(p) for p in pages if p.isdigit()), default=None), max((int(p) for p in pages if p.isdigit()), default=None)),
|
||||
"unique_leaf_cats": len(cats),
|
||||
"list_price_stats": pst,
|
||||
}
|
||||
|
||||
|
||||
def _structure_shops(rows: list[dict[str, str]], *, list_export: bool) -> list[str]:
|
||||
if list_export:
|
||||
return [
|
||||
_cell(r, JD_SEARCH_CSV_HEADERS["shop_name"], _LEGACY_SHOP_NAME_KEY)
|
||||
for r in rows
|
||||
if _cell(r, JD_SEARCH_CSV_HEADERS["shop_name"], _LEGACY_SHOP_NAME_KEY)
|
||||
]
|
||||
out: list[str] = []
|
||||
for r in rows:
|
||||
s = _cell(r, *_MERGED_SHOP_CELL_KEYS)
|
||||
if s:
|
||||
out.append(s)
|
||||
return out
|
||||
|
||||
|
||||
def _structure_brands(rows: list[dict[str, str]], *, list_export: bool) -> list[str]:
|
||||
if list_export:
|
||||
return [
|
||||
_cell(r, _LIST_BRAND_TITLE_HEADER, _LEGACY_LIST_BRAND_TITLE_KEY)
|
||||
for r in rows
|
||||
if _cell(r, _LIST_BRAND_TITLE_HEADER, _LEGACY_LIST_BRAND_TITLE_KEY)
|
||||
]
|
||||
return [
|
||||
_cell(r, MERGED_FIELD_TO_CSV_HEADER["detail_brand"], "detail_brand")
|
||||
for r in rows
|
||||
if _cell(r, MERGED_FIELD_TO_CSV_HEADER["detail_brand"], "detail_brand")
|
||||
]
|
||||
|
||||
|
||||
__all__ = [
|
||||
"_brand_cr",
|
||||
"_counter_mix_top_rows_with_remainder",
|
||||
"_search_list_proxies",
|
||||
"_structure_brands",
|
||||
"_structure_names_for_pie_counter",
|
||||
"_structure_shops",
|
||||
]
|
||||
@ -0,0 +1,87 @@
|
||||
"""竞品矩阵 Markdown 行：配料格与整行管道表单元。"""
|
||||
from __future__ import annotations
|
||||
|
||||
from pipeline.csv_schema import MERGED_FIELD_TO_CSV_HEADER, merged_csv_effective_total_sales
|
||||
|
||||
from .constants import (
|
||||
_COMMENT_FUZZ_KEYS,
|
||||
_DETAIL_PRICE_FINAL_CSV_KEYS,
|
||||
_LEGACY_RANK_TAGLINE_KEY,
|
||||
_LEGACY_SELLING_POINT_KEY,
|
||||
_LIST_SHOW_PRICE_CELL_KEYS,
|
||||
_MERGED_SHOP_CELL_KEYS,
|
||||
_RANK_TAGLINE_KEY,
|
||||
_SELLING_POINT_KEY,
|
||||
)
|
||||
from .csv_io import _cell, _detail_category_path_cell, _md_cell
|
||||
from .ingredients import (
|
||||
_ingredients_from_product_attributes,
|
||||
_ingredients_single_line,
|
||||
_is_ingredient_url_blob,
|
||||
)
|
||||
|
||||
|
||||
def _matrix_ingredients_cell(row: dict[str, str], *, max_len: int = 420) -> str:
|
||||
"""
|
||||
优先 ``detail_body_ingredients``（配料 OCR/文本）；旧合并表可能为 ``detail_body_image_urls``。
|
||||
若为 URL 串则尝试 ``detail_product_attributes`` 中的「配料/配料表：」片段。
|
||||
"""
|
||||
raw = _cell(
|
||||
row,
|
||||
MERGED_FIELD_TO_CSV_HEADER["detail_body_ingredients"],
|
||||
"detail_body_ingredients",
|
||||
"detail_body_image_urls",
|
||||
)
|
||||
if raw and not _is_ingredient_url_blob(raw):
|
||||
return _md_cell(_ingredients_single_line(raw), max_len)
|
||||
from_attr = _ingredients_from_product_attributes(
|
||||
_cell(
|
||||
row,
|
||||
MERGED_FIELD_TO_CSV_HEADER["detail_product_attributes"],
|
||||
"detail_product_attributes",
|
||||
)
|
||||
)
|
||||
if from_attr:
|
||||
return _md_cell(from_attr, max_len)
|
||||
if raw and _is_ingredient_url_blob(raw):
|
||||
return _md_cell(
|
||||
"（详情长图链接，无配料正文；可在采集侧开启配料识别后重新跑批次）",
|
||||
max_len,
|
||||
)
|
||||
return "—"
|
||||
|
||||
|
||||
def _competitor_matrix_md_line(
|
||||
row: dict[str, str], *, sku_header: str, title_h: str
|
||||
) -> str:
|
||||
sku = _md_cell(_cell(row, sku_header), 14)
|
||||
title = _md_cell(_cell(row, title_h), 56)
|
||||
brand = _md_cell(
|
||||
_cell(row, MERGED_FIELD_TO_CSV_HEADER["detail_brand"], "detail_brand"), 16
|
||||
)
|
||||
pj = _md_cell(_cell(row, *_LIST_SHOW_PRICE_CELL_KEYS), 10)
|
||||
df = _md_cell(_cell(row, *_DETAIL_PRICE_FINAL_CSV_KEYS), 10)
|
||||
shop = _md_cell(_cell(row, *_MERGED_SHOP_CELL_KEYS), 22)
|
||||
sell = _md_cell(_cell(row, _SELLING_POINT_KEY, _LEGACY_SELLING_POINT_KEY), 36)
|
||||
rank = _md_cell(
|
||||
_cell(row, _RANK_TAGLINE_KEY, _LEGACY_RANK_TAGLINE_KEY), 28
|
||||
)
|
||||
cat = _md_cell(_detail_category_path_cell(row), 24)
|
||||
ing = _matrix_ingredients_cell(row)
|
||||
ts_eff = merged_csv_effective_total_sales(row)
|
||||
cc = _md_cell(ts_eff or _cell(row, *_COMMENT_FUZZ_KEYS), 14)
|
||||
prev = _md_cell(
|
||||
_cell(
|
||||
row,
|
||||
MERGED_FIELD_TO_CSV_HEADER["comment_preview"],
|
||||
"comment_preview",
|
||||
),
|
||||
72,
|
||||
)
|
||||
return (
|
||||
f"| {sku} | {title} | {brand} | {pj} | {df} | {shop} | {sell} | {rank} | "
|
||||
f"{cat} | {ing} | {cc} | {prev} |"
|
||||
)
|
||||
|
||||
|
||||
__all__ = ["_competitor_matrix_md_line", "_matrix_ingredients_cell"]
|
||||
@ -0,0 +1,202 @@
|
||||
"""报告 Markdown 片段：Mermaid、场景摘要、规则策略提示、插图路径与解读段落。"""
|
||||
from __future__ import annotations
|
||||
|
||||
import re
|
||||
from collections import Counter
|
||||
from pathlib import Path
|
||||
from typing import Any
|
||||
|
||||
from .csv_io import _md_cell
|
||||
|
||||
|
||||
def _mermaid_pie_focus_keywords(hits: Counter[str], *, top_k: int = 8) -> str:
|
||||
"""关注词全局 Top 的 Mermaid pie（便于渲染或导出工具识别）。"""
|
||||
top = hits.most_common(top_k)
|
||||
if not top:
|
||||
return ""
|
||||
rest = list(hits.most_common())
|
||||
if len(rest) > top_k:
|
||||
others_n = sum(n for _, n in rest[top_k:])
|
||||
else:
|
||||
others_n = 0
|
||||
lines = ["```mermaid", 'pie title 关注词命中次数（全局 Top，子串计数）']
|
||||
for w, n in top:
|
||||
if n <= 0:
|
||||
continue
|
||||
label = (w or "?").replace('"', "'").replace("\n", " ")[:18]
|
||||
lines.append(f' "{label}（{n}）" : {n}')
|
||||
if others_n > 0:
|
||||
lines.append(f' "其余词合计" : {others_n}')
|
||||
lines.append("```")
|
||||
return "\n".join(lines)
|
||||
|
||||
|
||||
def _scenario_summary_bullets(counter: Counter[str], n_texts: int, top_k: int = 5) -> list[str]:
|
||||
if n_texts <= 0 or not counter:
|
||||
return []
|
||||
ordered = counter.most_common()
|
||||
lines: list[str] = []
|
||||
head = ordered[:top_k]
|
||||
parts = []
|
||||
for label, cnt in head:
|
||||
pct = 100.0 * cnt / n_texts
|
||||
parts.append(f"「{label}」约 **{cnt}** 条（占有效文本 **{pct:.0f}%**）")
|
||||
lines.append(
|
||||
"用户自述的用途/场景（基于预设词组，**非语义分类**）：" + "；".join(parts) + "。"
|
||||
)
|
||||
tail = [lbl for lbl, n in ordered[top_k:] if n > 0]
|
||||
if tail:
|
||||
lines.append(f"另有提及较少的场景标签：{'、'.join(tail)}。")
|
||||
return lines
|
||||
|
||||
|
||||
def _strategy_hints(
|
||||
*,
|
||||
cr1: float | None,
|
||||
pst: dict[str, Any],
|
||||
hits: Counter[str],
|
||||
n_comments: int,
|
||||
scen_counts: Counter[str],
|
||||
scen_n_texts: int,
|
||||
) -> list[str]:
|
||||
"""基于规则的「提示性」结论，均标注待验证。"""
|
||||
hints: list[str] = []
|
||||
if cr1 is not None and cr1 >= 0.45:
|
||||
hints.append(
|
||||
"样本内品牌集中度较高（第一大品牌份额偏高），头部玩家占据显著曝光；新原料/解决方案宜明确差异化价值主张（**需线下渠道与招商信息交叉验证**）。"
|
||||
)
|
||||
elif cr1 is not None and cr1 < 0.25:
|
||||
hints.append(
|
||||
"样本内品牌较分散，品类或关键词下竞争格局未固化，存在定位与叙事空间（**需扩大样本页数与关键词矩阵验证**）。"
|
||||
)
|
||||
if pst.get("stdev") and pst.get("mean") and pst["mean"] > 0:
|
||||
cv = pst["stdev"] / pst["mean"]
|
||||
if cv > 0.35:
|
||||
hints.append(
|
||||
"价格离散度较高，同时存在偏低价与偏高价陈列，可分别对标「性价比带」与「品质/功能带」竞品（**终端到手价受促销影响，非成本结构**）。"
|
||||
)
|
||||
if hits:
|
||||
top = hits.most_common(3)
|
||||
top_s = "、".join(w for w, _ in top)
|
||||
hints.append(
|
||||
f"评价文本中「{top_s}」等主题出现较多，可作为消费者沟通与产品卖点的假设输入（**非严格主题模型，建议人工抽样复核**）。"
|
||||
)
|
||||
if n_comments < 5:
|
||||
hints.append(
|
||||
"有效评价样本偏少，消费者洞察部分仅作方向参考，正式结论建议加大 SKU 数或评论分页。"
|
||||
)
|
||||
if scen_n_texts >= 5 and scen_counts:
|
||||
top_lbl, top_n = scen_counts.most_common(1)[0]
|
||||
share = top_n / scen_n_texts
|
||||
if share >= 0.25:
|
||||
hints.append(
|
||||
f"用途/场景中「{top_lbl}」在约 {100 * share:.0f}% 的有效评价自述中出现，可作为沟通场景与卖点的优先假设（**词组规则，建议抽样核对原句**）。"
|
||||
)
|
||||
if not hints:
|
||||
hints.append(
|
||||
"当前样本下自动规则未触发强信号；请结合业务目标人工解读对比矩阵与原始 CSV。"
|
||||
)
|
||||
return hints
|
||||
|
||||
|
||||
def _embed_chart(run_dir: Path, filename: str, caption: str = "") -> list[str]:
|
||||
"""若 ``report_assets/<filename>`` 存在则返回插图 Markdown 片段。"""
|
||||
if not (run_dir / "report_assets" / filename).is_file():
|
||||
return []
|
||||
cap = (caption or "").strip()
|
||||
out: list[str] = []
|
||||
if cap:
|
||||
out.append(f"*{cap}*")
|
||||
out.append("")
|
||||
out.append(f"")
|
||||
out.append("")
|
||||
return out
|
||||
|
||||
|
||||
def _scenario_group_asset_slug(group: str, index: int) -> str:
|
||||
"""与 ``pipeline.reporting.charts`` 中场景分组图文件名规则一致（勿改格式）。"""
|
||||
raw = (group or "").strip()
|
||||
core = re.sub(r"[^\w\u4e00-\u9fff-]", "", raw)[:20]
|
||||
if not core:
|
||||
core = "group"
|
||||
return f"i{index:02d}_{core}"
|
||||
|
||||
|
||||
def _focus_scenario_combo_bar_filename(group: str, index: int) -> str:
|
||||
"""关注词 + 使用场景并排条形图（与 ``pipeline.reporting.charts.save_combo_focus_scenario_bar`` 同源）。"""
|
||||
slug = _scenario_group_asset_slug(group, index)
|
||||
return f"chart_focus_and_scenarios_bar__{slug}.png"
|
||||
|
||||
|
||||
def _matrix_prices_sales_chart_filename(group: str, index: int) -> str:
|
||||
"""与 ``pipeline.reporting.charts.generate_report_charts`` 中 ``chart_matrix_prices_sales__*`` 一致。"""
|
||||
slug = _scenario_group_asset_slug(group, index)
|
||||
return f"chart_matrix_prices_sales__{slug}.png"
|
||||
|
||||
|
||||
def _lines_4_reading_brand(
|
||||
*,
|
||||
cr1: float | None,
|
||||
cr3: float | None,
|
||||
top: str,
|
||||
brand_rows_n: int,
|
||||
n_structure: int,
|
||||
) -> list[str]:
|
||||
if cr1 is None or not (top or "").strip():
|
||||
return []
|
||||
lines = [
|
||||
"",
|
||||
"**数据解读（规则摘要）**：",
|
||||
"",
|
||||
f"- 在含品牌字段的 **{brand_rows_n}** 条列表行（占本章结构样本 **{n_structure}** 行）中，"
|
||||
f"「{_md_cell(top.strip(), 36)}」曝光约占 **{100 * cr1:.1f}%**（按行计，同一 SKU 多行会重复计）。",
|
||||
]
|
||||
if cr3 is not None:
|
||||
lines.append(
|
||||
f"- 前三品牌合计约 **{100 * cr3:.1f}%**；若该比例偏高，说明搜索页品牌集中度高，"
|
||||
"新品需搭配清晰的差异定位与资源投放，避免与头部在泛词下正面撞车。"
|
||||
)
|
||||
lines.append("")
|
||||
return lines
|
||||
|
||||
|
||||
def _lines_4_reading_shop(
|
||||
*,
|
||||
cr1: float | None,
|
||||
cr3: float | None,
|
||||
top: str,
|
||||
shop_rows_n: int,
|
||||
n_structure: int,
|
||||
) -> list[str]:
|
||||
if not shop_rows_n:
|
||||
return []
|
||||
lines = [
|
||||
"",
|
||||
"**数据解读（规则摘要）**：",
|
||||
"",
|
||||
f"- 含店铺名的列表行共 **{shop_rows_n}** 条（结构样本 **{n_structure}** 行），反映搜索曝光下的店铺格局。",
|
||||
]
|
||||
if cr1 is not None and (top or "").strip():
|
||||
lines.append(
|
||||
f"- 第一大店铺「{_md_cell(top.strip(), 40)}」约占 **{100 * cr1:.1f}%**；"
|
||||
"该指标刻画的是**列表可见度**而非销量，适合用于判断货架被哪些店铺占据。"
|
||||
)
|
||||
if cr3 is not None:
|
||||
lines.append(
|
||||
f"- 前三店铺合计约 **{100 * cr3:.1f}%**；若集中度高，可考虑从店铺矩阵、旗舰店/专营店布局等角度拆解竞争。"
|
||||
)
|
||||
lines.append("")
|
||||
return lines
|
||||
|
||||
|
||||
__all__ = [
|
||||
"_embed_chart",
|
||||
"_focus_scenario_combo_bar_filename",
|
||||
"_lines_4_reading_brand",
|
||||
"_lines_4_reading_shop",
|
||||
"_matrix_prices_sales_chart_filename",
|
||||
"_mermaid_pie_focus_keywords",
|
||||
"_scenario_group_asset_slug",
|
||||
"_scenario_summary_bullets",
|
||||
"_strategy_hints",
|
||||
]
|
||||
@ -0,0 +1,91 @@
|
||||
"""运行目录解析、关键词推断、pc_search_raw 检索规模读取。"""
|
||||
from __future__ import annotations
|
||||
|
||||
import json
|
||||
import re
|
||||
from collections import Counter
|
||||
from pathlib import Path
|
||||
from typing import Any
|
||||
|
||||
|
||||
def _run_batch_label(run_dir: Path) -> str:
|
||||
name = run_dir.name
|
||||
m = re.match(r"^(\d{8})_(\d{6})_", name)
|
||||
if m:
|
||||
return f"{m.group(1)} {m.group(2)}"
|
||||
return name
|
||||
|
||||
|
||||
def _resolve_existing_run_dir(raw: str | Path | None) -> Path | None:
|
||||
if raw is None:
|
||||
return None
|
||||
s = str(raw).strip()
|
||||
if not s:
|
||||
return None
|
||||
p = Path(s).expanduser()
|
||||
if not p.is_absolute():
|
||||
p = (Path.cwd() / p).resolve()
|
||||
else:
|
||||
p = p.resolve()
|
||||
return p
|
||||
|
||||
|
||||
def _infer_keyword(run_dir: Path, meta: dict[str, Any] | None) -> str:
|
||||
if meta:
|
||||
k = str(meta.get("keyword") or "").strip()
|
||||
if k:
|
||||
return k
|
||||
m = re.match(r"^\d{8}_\d{6}_(.+)$", run_dir.name)
|
||||
if m:
|
||||
return m.group(1).strip()
|
||||
return ""
|
||||
|
||||
|
||||
def _pc_search_result_count_from_raw(
|
||||
run_dir: Path,
|
||||
) -> tuple[int | None, str, list[int], int, int]:
|
||||
"""
|
||||
从 ``pc_search_raw/*.json`` 读取 ``data.resultCount``（京东 PC 搜索接口返回的检索命中规模）。
|
||||
多文件时取众数；返回 (众数值, data.listKeyWord 首见值, 出现过的不同取值升序, 解析到的样本文件数)。
|
||||
"""
|
||||
raw_dir = run_dir / "pc_search_raw"
|
||||
if not raw_dir.is_dir():
|
||||
return None, "", [], 0, 0
|
||||
counts: list[int] = []
|
||||
list_kw = ""
|
||||
n_files = 0
|
||||
for p in sorted(raw_dir.glob("*.json")):
|
||||
try:
|
||||
obj = json.loads(p.read_text(encoding="utf-8"))
|
||||
except (json.JSONDecodeError, OSError, UnicodeError):
|
||||
continue
|
||||
n_files += 1
|
||||
if not isinstance(obj, dict):
|
||||
continue
|
||||
data = obj.get("data")
|
||||
if not isinstance(data, dict):
|
||||
continue
|
||||
rc = data.get("resultCount")
|
||||
val: int | None = None
|
||||
if isinstance(rc, int) and not isinstance(rc, bool) and rc >= 0:
|
||||
val = rc
|
||||
elif isinstance(rc, str) and rc.strip().isdigit():
|
||||
val = int(rc.strip())
|
||||
if val is not None:
|
||||
counts.append(val)
|
||||
lk = data.get("listKeyWord")
|
||||
if isinstance(lk, str) and lk.strip() and not list_kw:
|
||||
list_kw = lk.strip()
|
||||
if not counts:
|
||||
return None, list_kw, [], n_files, 0
|
||||
consensus_rc, _freq = Counter(counts).most_common(1)[0]
|
||||
uniques = sorted(set(counts))
|
||||
return consensus_rc, list_kw, uniques, n_files, len(counts)
|
||||
|
||||
|
||||
__all__ = [
|
||||
"_infer_keyword",
|
||||
"_pc_search_result_count_from_raw",
|
||||
"_resolve_existing_run_dir",
|
||||
"_run_batch_label",
|
||||
]
|
||||
@ -65,11 +65,6 @@ from competitor_report.comment_sentiment import ( # noqa: E402
|
||||
_merge_comment_previews,
|
||||
_parse_comment_score,
|
||||
)
|
||||
from competitor_report.ingredients import ( # noqa: E402
|
||||
_ingredients_from_product_attributes,
|
||||
_ingredients_single_line,
|
||||
_is_ingredient_url_blob,
|
||||
)
|
||||
from competitor_report.llm_group_payloads import ( # noqa: E402
|
||||
build_comment_groups_llm_payload,
|
||||
build_matrix_groups_llm_payload,
|
||||
@ -86,6 +81,40 @@ from competitor_report.matrix_group import ( # noqa: E402
|
||||
_merged_rows_grouped_for_matrix,
|
||||
)
|
||||
from competitor_report.price_stats import _price_stats_extended # noqa: E402
|
||||
from competitor_report.consumer_feedback import ( # noqa: E402
|
||||
_comment_lines_with_product_context,
|
||||
_consumer_feedback_by_matrix_group,
|
||||
_sku_to_matrix_group_map,
|
||||
)
|
||||
from competitor_report.list_mix import ( # noqa: E402
|
||||
_brand_cr,
|
||||
_counter_mix_top_rows_with_remainder,
|
||||
_search_list_proxies,
|
||||
_structure_brands,
|
||||
_structure_names_for_pie_counter,
|
||||
_structure_shops,
|
||||
)
|
||||
from competitor_report.matrix_md import ( # noqa: E402
|
||||
_competitor_matrix_md_line,
|
||||
_matrix_ingredients_cell,
|
||||
)
|
||||
from competitor_report.report_md_helpers import ( # noqa: E402
|
||||
_embed_chart,
|
||||
_focus_scenario_combo_bar_filename,
|
||||
_lines_4_reading_brand,
|
||||
_lines_4_reading_shop,
|
||||
_matrix_prices_sales_chart_filename,
|
||||
_mermaid_pie_focus_keywords,
|
||||
_scenario_group_asset_slug,
|
||||
_scenario_summary_bullets,
|
||||
_strategy_hints,
|
||||
)
|
||||
from competitor_report.run_context import ( # noqa: E402
|
||||
_infer_keyword,
|
||||
_pc_search_result_count_from_raw,
|
||||
_resolve_existing_run_dir,
|
||||
_run_batch_label,
|
||||
)
|
||||
|
||||
# ---------------------------------------------------------------------------
|
||||
# 运行配置（按需改这里；与 competitor_report.constants 中默认关注词等配合使用）
|
||||
@ -100,586 +129,6 @@ OVERRIDE_MAX_SKUS: int | None = None
|
||||
OVERRIDE_PAGE_START: int | None = None
|
||||
OVERRIDE_PAGE_TO: int | None = None
|
||||
|
||||
def _comment_lines_with_product_context(
|
||||
comment_rows: list[dict[str, str]],
|
||||
merged_rows: list[dict[str, str]],
|
||||
*,
|
||||
sku_header: str,
|
||||
title_h: str,
|
||||
) -> list[str]:
|
||||
"""与 ``comment_rows`` 顺序对齐：带细类/SKU/品名/店铺前缀，供 §8.2 大模型抽样。"""
|
||||
sku_meta: dict[str, tuple[str, str, str]] = {}
|
||||
for row in merged_rows:
|
||||
sku = _cell(row, sku_header).strip()
|
||||
if not sku:
|
||||
continue
|
||||
gk = _competitor_matrix_group_key(row)
|
||||
if not gk:
|
||||
continue
|
||||
sku_meta[sku] = (
|
||||
gk,
|
||||
_cell(row, title_h),
|
||||
_cell(row, *_MERGED_SHOP_CELL_KEYS),
|
||||
)
|
||||
out: list[str] = []
|
||||
for cr in comment_rows:
|
||||
txt = _cell(cr, _COMMENT_CSV_BODY, "tagCommentContent")
|
||||
if not txt:
|
||||
continue
|
||||
sku = _cell(cr, _COMMENT_CSV_SKU, "sku").strip()
|
||||
meta = sku_meta.get(sku)
|
||||
if meta:
|
||||
gname, tit, shop = meta
|
||||
prefix = (
|
||||
f"【细类：{gname}｜SKU：{sku}｜品名：{_md_cell(tit, 80)}｜"
|
||||
f"店铺：{_md_cell(shop, 40)}】"
|
||||
)
|
||||
out.append(prefix + txt)
|
||||
else:
|
||||
out.append(txt)
|
||||
return out
|
||||
|
||||
|
||||
|
||||
def _mermaid_pie_focus_keywords(hits: Counter[str], *, top_k: int = 8) -> str:
|
||||
"""关注词全局 Top 的 Mermaid pie（便于渲染或导出工具识别）。"""
|
||||
top = hits.most_common(top_k)
|
||||
if not top:
|
||||
return ""
|
||||
rest = list(hits.most_common())
|
||||
if len(rest) > top_k:
|
||||
others_n = sum(n for _, n in rest[top_k:])
|
||||
else:
|
||||
others_n = 0
|
||||
lines = ["```mermaid", 'pie title 关注词命中次数（全局 Top，子串计数）']
|
||||
for w, n in top:
|
||||
if n <= 0:
|
||||
continue
|
||||
label = (w or "?").replace('"', "'").replace("\n", " ")[:18]
|
||||
lines.append(f' "{label}（{n}）" : {n}')
|
||||
if others_n > 0:
|
||||
lines.append(f' "其余词合计" : {others_n}')
|
||||
lines.append("```")
|
||||
return "\n".join(lines)
|
||||
|
||||
|
||||
|
||||
|
||||
def _scenario_summary_bullets(counter: Counter[str], n_texts: int, top_k: int = 5) -> list[str]:
|
||||
if n_texts <= 0 or not counter:
|
||||
return []
|
||||
ordered = counter.most_common()
|
||||
lines: list[str] = []
|
||||
head = ordered[:top_k]
|
||||
parts = []
|
||||
for label, cnt in head:
|
||||
pct = 100.0 * cnt / n_texts
|
||||
parts.append(f"「{label}」约 **{cnt}** 条（占有效文本 **{pct:.0f}%**）")
|
||||
lines.append(
|
||||
"用户自述的用途/场景（基于预设词组，**非语义分类**）：" + "；".join(parts) + "。"
|
||||
)
|
||||
tail = [lbl for lbl, n in ordered[top_k:] if n > 0]
|
||||
if tail:
|
||||
lines.append(f"另有提及较少的场景标签：{'、'.join(tail)}。")
|
||||
return lines
|
||||
|
||||
|
||||
def _sku_to_matrix_group_map(
|
||||
merged_rows: list[dict[str, str]], sku_header: str
|
||||
) -> dict[str, str]:
|
||||
m: dict[str, str] = {}
|
||||
for row in merged_rows:
|
||||
sku = _cell(row, sku_header).strip()
|
||||
if not sku:
|
||||
continue
|
||||
gk = _competitor_matrix_group_key(row)
|
||||
if gk:
|
||||
m[sku] = gk
|
||||
return m
|
||||
|
||||
|
||||
def _comment_text_units_for_matrix_group(
|
||||
gname: str,
|
||||
merged_rows: list[dict[str, str]],
|
||||
comment_rows_in_group: list[dict[str, str]],
|
||||
sku_header: str,
|
||||
) -> list[str]:
|
||||
"""某细类下的评价正文列表；无 flat 时用该细类合并行的 comment_preview。"""
|
||||
texts: list[str] = []
|
||||
for row in comment_rows_in_group:
|
||||
t = _cell(row, _COMMENT_CSV_BODY, "tagCommentContent")
|
||||
if t:
|
||||
texts.append(t)
|
||||
if texts:
|
||||
return texts
|
||||
for row in merged_rows:
|
||||
if _competitor_matrix_group_key(row) != gname:
|
||||
continue
|
||||
p = _cell(
|
||||
row,
|
||||
MERGED_FIELD_TO_CSV_HEADER["comment_preview"],
|
||||
"comment_preview",
|
||||
)
|
||||
if p:
|
||||
texts.append(p)
|
||||
return texts
|
||||
|
||||
|
||||
|
||||
|
||||
def _consumer_feedback_by_matrix_group(
|
||||
*,
|
||||
merged_rows: list[dict[str, str]],
|
||||
comment_rows: list[dict[str, str]],
|
||||
sku_header: str,
|
||||
) -> list[tuple[str, list[dict[str, str]], list[str]]]:
|
||||
"""
|
||||
与 §5 矩阵同序的细类列表；每项为 (细类名, 该类的 comments_flat 行, 用于场景统计的文本单元)。
|
||||
评价 SKU 不在深入样本时归入「未归类（评价 SKU 无对应深入样本）」。
|
||||
"""
|
||||
if not merged_rows:
|
||||
if not comment_rows:
|
||||
return []
|
||||
texts = _iter_comment_text_units(comment_rows, [])
|
||||
return [
|
||||
(
|
||||
"未归类（无深入合并表）",
|
||||
list(comment_rows),
|
||||
texts,
|
||||
)
|
||||
]
|
||||
|
||||
sku_map = _sku_to_matrix_group_map(merged_rows, sku_header)
|
||||
merged_by_sku: dict[str, dict[str, str]] = {}
|
||||
for row in merged_rows:
|
||||
s = _cell(row, sku_header).strip()
|
||||
if s:
|
||||
merged_by_sku[s] = row
|
||||
by_g: dict[str, list[dict[str, str]]] = {}
|
||||
for row in comment_rows:
|
||||
sku = _cell(row, _COMMENT_CSV_SKU, "sku").strip()
|
||||
g = sku_map.get(sku)
|
||||
if g:
|
||||
by_g.setdefault(g, []).append(row)
|
||||
continue
|
||||
if sku and sku in merged_by_sku:
|
||||
# 深入样本存在但缺 detail_category_path（或路径无法解析为可读细类）：不参与按细类分析
|
||||
continue
|
||||
by_g.setdefault("未归类（评价 SKU 无对应深入样本）", []).append(row)
|
||||
|
||||
out: list[tuple[str, list[dict[str, str]], list[str]]] = []
|
||||
used: set[str] = set()
|
||||
for gname, _ in _merged_rows_grouped_for_matrix(merged_rows):
|
||||
cr = by_g.get(gname, [])
|
||||
tu = _comment_text_units_for_matrix_group(
|
||||
gname, merged_rows, cr, sku_header
|
||||
)
|
||||
out.append((gname, cr, tu))
|
||||
used.add(gname)
|
||||
for gname, cr in sorted(by_g.items(), key=lambda x: (-len(x[1]), x[0])):
|
||||
if gname in used:
|
||||
continue
|
||||
tu = _comment_text_units_for_matrix_group(
|
||||
gname, merged_rows, cr, sku_header
|
||||
)
|
||||
out.append((gname, cr, tu))
|
||||
return out
|
||||
|
||||
|
||||
def _run_batch_label(run_dir: Path) -> str:
|
||||
name = run_dir.name
|
||||
m = re.match(r"^(\d{8})_(\d{6})_", name)
|
||||
if m:
|
||||
return f"{m.group(1)} {m.group(2)}"
|
||||
return name
|
||||
|
||||
|
||||
def _resolve_existing_run_dir(raw: str | Path | None) -> Path | None:
|
||||
if raw is None:
|
||||
return None
|
||||
s = str(raw).strip()
|
||||
if not s:
|
||||
return None
|
||||
p = Path(s).expanduser()
|
||||
if not p.is_absolute():
|
||||
p = (Path.cwd() / p).resolve()
|
||||
else:
|
||||
p = p.resolve()
|
||||
return p
|
||||
|
||||
|
||||
def _infer_keyword(run_dir: Path, meta: dict[str, Any] | None) -> str:
|
||||
if meta:
|
||||
k = str(meta.get("keyword") or "").strip()
|
||||
if k:
|
||||
return k
|
||||
m = re.match(r"^\d{8}_\d{6}_(.+)$", run_dir.name)
|
||||
if m:
|
||||
return m.group(1).strip()
|
||||
return ""
|
||||
|
||||
|
||||
def _pc_search_result_count_from_raw(
|
||||
run_dir: Path,
|
||||
) -> tuple[int | None, str, list[int], int, int]:
|
||||
"""
|
||||
从 ``pc_search_raw/*.json`` 读取 ``data.resultCount``（京东 PC 搜索接口返回的检索命中规模）。
|
||||
多文件时取众数；返回 (众数值, data.listKeyWord 首见值, 出现过的不同取值升序, 解析到的样本文件数)。
|
||||
"""
|
||||
raw_dir = run_dir / "pc_search_raw"
|
||||
if not raw_dir.is_dir():
|
||||
return None, "", [], 0, 0
|
||||
counts: list[int] = []
|
||||
list_kw = ""
|
||||
n_files = 0
|
||||
for p in sorted(raw_dir.glob("*.json")):
|
||||
try:
|
||||
obj = json.loads(p.read_text(encoding="utf-8"))
|
||||
except (json.JSONDecodeError, OSError, UnicodeError):
|
||||
continue
|
||||
n_files += 1
|
||||
if not isinstance(obj, dict):
|
||||
continue
|
||||
data = obj.get("data")
|
||||
if not isinstance(data, dict):
|
||||
continue
|
||||
rc = data.get("resultCount")
|
||||
val: int | None = None
|
||||
if isinstance(rc, int) and not isinstance(rc, bool) and rc >= 0:
|
||||
val = rc
|
||||
elif isinstance(rc, str) and rc.strip().isdigit():
|
||||
val = int(rc.strip())
|
||||
if val is not None:
|
||||
counts.append(val)
|
||||
lk = data.get("listKeyWord")
|
||||
if isinstance(lk, str) and lk.strip() and not list_kw:
|
||||
list_kw = lk.strip()
|
||||
if not counts:
|
||||
return None, list_kw, [], n_files, 0
|
||||
consensus_rc, _freq = Counter(counts).most_common(1)[0]
|
||||
uniques = sorted(set(counts))
|
||||
return consensus_rc, list_kw, uniques, n_files, len(counts)
|
||||
|
||||
|
||||
def _structure_names_for_pie_counter(row_names: list[str]) -> list[str]:
|
||||
"""
|
||||
与 ``_counter_mix_top_rows_with_remainder`` / 列表品牌·店铺扇图同一套规则：
|
||||
按 strip 后的名称逐行保留一条，便于 ``_brand_cr`` 与饼图 Counter 一致。
|
||||
"""
|
||||
return [(x or "").strip() for x in row_names if (x or "").strip()]
|
||||
|
||||
|
||||
def _brand_cr(cnames: list[str]) -> tuple[float | None, float | None, str, str]:
|
||||
"""按名称计数返回 (第一大主体份额, 前三合计份额, 头部标签, 头部占比展示字符串)。"""
|
||||
if not cnames:
|
||||
return None, None, "", ""
|
||||
cnt = Counter(cnames)
|
||||
total = sum(cnt.values())
|
||||
if total <= 0:
|
||||
return None, None, "", ""
|
||||
mc = cnt.most_common()
|
||||
top1_n = mc[0][1] if mc else 0
|
||||
top1 = mc[0][0] if mc else ""
|
||||
cr1 = top1_n / total
|
||||
top3_n = sum(n for _, n in mc[:3])
|
||||
cr3 = top3_n / total
|
||||
return cr1, cr3, top1, f"{100.0 * top1_n / total:.1f}%"
|
||||
|
||||
|
||||
def _counter_mix_top_rows_with_remainder(
|
||||
row_names: list[str], *, top_n: int, remainder_label: str
|
||||
) -> list[tuple[str, int]]:
|
||||
"""
|
||||
与列表品牌/店铺扇图一致：按 strip 后的名称计数；``most_common(top_n)`` 未覆盖的长尾合并为
|
||||
``remainder_label``，保证各块 count 之和等于可统计行数（与 ``_structure_names_for_pie_counter`` 总条数一致）。
|
||||
"""
|
||||
c = Counter((x or "").strip() for x in row_names if (x or "").strip())
|
||||
if not c:
|
||||
return []
|
||||
total = sum(c.values())
|
||||
common = c.most_common(top_n)
|
||||
accounted = sum(v for _, v in common)
|
||||
rest = total - accounted
|
||||
out: list[tuple[str, int]] = list(common)
|
||||
if rest > 0:
|
||||
out.append((remainder_label, rest))
|
||||
return out
|
||||
|
||||
|
||||
|
||||
|
||||
def _search_list_proxies(rows: list[dict[str, str]]) -> dict[str, Any]:
|
||||
"""
|
||||
基于 pc_search_export 的「列表可见度」指标，**不是**全渠道零售额或 TAM。
|
||||
"""
|
||||
sku_k = JD_SEARCH_CSV_HEADERS["sku_id"]
|
||||
shop_k = JD_SEARCH_CSV_HEADERS["shop_name"]
|
||||
page_k = JD_SEARCH_CSV_HEADERS["page"]
|
||||
cat_k = JD_SEARCH_CSV_HEADERS["leaf_category"]
|
||||
skus: set[str] = set()
|
||||
shops: set[str] = set()
|
||||
pages: set[str] = set()
|
||||
cats: set[str] = set()
|
||||
for r in rows:
|
||||
s = _cell(r, sku_k)
|
||||
if s:
|
||||
skus.add(s)
|
||||
sh = _cell(r, shop_k)
|
||||
if sh:
|
||||
shops.add(sh)
|
||||
pg = _cell(r, page_k)
|
||||
if pg:
|
||||
pages.add(pg)
|
||||
c = _cell(r, cat_k)
|
||||
if c:
|
||||
cats.add(c)
|
||||
prices = _collect_prices(rows)
|
||||
pst = _price_stats_extended(prices)
|
||||
return {
|
||||
"total_rows": len(rows),
|
||||
"unique_skus": len(skus),
|
||||
"unique_shops": len(shops),
|
||||
"unique_pages": len(pages),
|
||||
"page_span": (min((int(p) for p in pages if p.isdigit()), default=None), max((int(p) for p in pages if p.isdigit()), default=None)),
|
||||
"unique_leaf_cats": len(cats),
|
||||
"list_price_stats": pst,
|
||||
}
|
||||
|
||||
|
||||
|
||||
|
||||
def _structure_shops(rows: list[dict[str, str]], *, list_export: bool) -> list[str]:
|
||||
if list_export:
|
||||
return [
|
||||
_cell(r, JD_SEARCH_CSV_HEADERS["shop_name"], _LEGACY_SHOP_NAME_KEY)
|
||||
for r in rows
|
||||
if _cell(r, JD_SEARCH_CSV_HEADERS["shop_name"], _LEGACY_SHOP_NAME_KEY)
|
||||
]
|
||||
out: list[str] = []
|
||||
for r in rows:
|
||||
s = _cell(r, *_MERGED_SHOP_CELL_KEYS)
|
||||
if s:
|
||||
out.append(s)
|
||||
return out
|
||||
|
||||
|
||||
def _structure_brands(rows: list[dict[str, str]], *, list_export: bool) -> list[str]:
|
||||
if list_export:
|
||||
return [
|
||||
_cell(r, _LIST_BRAND_TITLE_HEADER, _LEGACY_LIST_BRAND_TITLE_KEY)
|
||||
for r in rows
|
||||
if _cell(r, _LIST_BRAND_TITLE_HEADER, _LEGACY_LIST_BRAND_TITLE_KEY)
|
||||
]
|
||||
return [
|
||||
_cell(r, MERGED_FIELD_TO_CSV_HEADER["detail_brand"], "detail_brand")
|
||||
for r in rows
|
||||
if _cell(r, MERGED_FIELD_TO_CSV_HEADER["detail_brand"], "detail_brand")
|
||||
]
|
||||
|
||||
|
||||
|
||||
|
||||
def _matrix_ingredients_cell(row: dict[str, str], *, max_len: int = 420) -> str:
|
||||
"""
|
||||
优先 ``detail_body_ingredients``（配料 OCR/文本）；旧合并表可能为 ``detail_body_image_urls``。
|
||||
若为 URL 串则尝试 ``detail_product_attributes`` 中的「配料/配料表：」片段。
|
||||
"""
|
||||
raw = _cell(
|
||||
row,
|
||||
MERGED_FIELD_TO_CSV_HEADER["detail_body_ingredients"],
|
||||
"detail_body_ingredients",
|
||||
"detail_body_image_urls",
|
||||
)
|
||||
if raw and not _is_ingredient_url_blob(raw):
|
||||
return _md_cell(_ingredients_single_line(raw), max_len)
|
||||
from_attr = _ingredients_from_product_attributes(
|
||||
_cell(
|
||||
row,
|
||||
MERGED_FIELD_TO_CSV_HEADER["detail_product_attributes"],
|
||||
"detail_product_attributes",
|
||||
)
|
||||
)
|
||||
if from_attr:
|
||||
return _md_cell(from_attr, max_len)
|
||||
if raw and _is_ingredient_url_blob(raw):
|
||||
return _md_cell(
|
||||
"（详情长图链接，无配料正文；可在采集侧开启配料识别后重新跑批次）",
|
||||
max_len,
|
||||
)
|
||||
return "—"
|
||||
|
||||
|
||||
def _competitor_matrix_md_line(
|
||||
row: dict[str, str], *, sku_header: str, title_h: str
|
||||
) -> str:
|
||||
sku = _md_cell(_cell(row, sku_header), 14)
|
||||
title = _md_cell(_cell(row, title_h), 56)
|
||||
brand = _md_cell(
|
||||
_cell(row, MERGED_FIELD_TO_CSV_HEADER["detail_brand"], "detail_brand"), 16
|
||||
)
|
||||
pj = _md_cell(_cell(row, *_LIST_SHOW_PRICE_CELL_KEYS), 10)
|
||||
df = _md_cell(_cell(row, *_DETAIL_PRICE_FINAL_CSV_KEYS), 10)
|
||||
shop = _md_cell(_cell(row, *_MERGED_SHOP_CELL_KEYS), 22)
|
||||
sell = _md_cell(_cell(row, _SELLING_POINT_KEY, _LEGACY_SELLING_POINT_KEY), 36)
|
||||
rank = _md_cell(
|
||||
_cell(row, _RANK_TAGLINE_KEY, _LEGACY_RANK_TAGLINE_KEY), 28
|
||||
)
|
||||
cat = _md_cell(_detail_category_path_cell(row), 24)
|
||||
ing = _matrix_ingredients_cell(row)
|
||||
ts_eff = merged_csv_effective_total_sales(row)
|
||||
cc = _md_cell(ts_eff or _cell(row, *_COMMENT_FUZZ_KEYS), 14)
|
||||
prev = _md_cell(
|
||||
_cell(
|
||||
row,
|
||||
MERGED_FIELD_TO_CSV_HEADER["comment_preview"],
|
||||
"comment_preview",
|
||||
),
|
||||
72,
|
||||
)
|
||||
return (
|
||||
f"| {sku} | {title} | {brand} | {pj} | {df} | {shop} | {sell} | {rank} | "
|
||||
f"{cat} | {ing} | {cc} | {prev} |"
|
||||
)
|
||||
|
||||
|
||||
def _strategy_hints(
|
||||
*,
|
||||
cr1: float | None,
|
||||
pst: dict[str, Any],
|
||||
hits: Counter[str],
|
||||
n_comments: int,
|
||||
scen_counts: Counter[str],
|
||||
scen_n_texts: int,
|
||||
) -> list[str]:
|
||||
"""基于规则的「提示性」结论，均标注待验证。"""
|
||||
hints: list[str] = []
|
||||
if cr1 is not None and cr1 >= 0.45:
|
||||
hints.append(
|
||||
"样本内品牌集中度较高（第一大品牌份额偏高），头部玩家占据显著曝光；新原料/解决方案宜明确差异化价值主张（**需线下渠道与招商信息交叉验证**）。"
|
||||
)
|
||||
elif cr1 is not None and cr1 < 0.25:
|
||||
hints.append(
|
||||
"样本内品牌较分散，品类或关键词下竞争格局未固化，存在定位与叙事空间（**需扩大样本页数与关键词矩阵验证**）。"
|
||||
)
|
||||
if pst.get("stdev") and pst.get("mean") and pst["mean"] > 0:
|
||||
cv = pst["stdev"] / pst["mean"]
|
||||
if cv > 0.35:
|
||||
hints.append(
|
||||
"价格离散度较高，同时存在偏低价与偏高价陈列，可分别对标「性价比带」与「品质/功能带」竞品（**终端到手价受促销影响，非成本结构**）。"
|
||||
)
|
||||
if hits:
|
||||
top = hits.most_common(3)
|
||||
top_s = "、".join(w for w, _ in top)
|
||||
hints.append(
|
||||
f"评价文本中「{top_s}」等主题出现较多，可作为消费者沟通与产品卖点的假设输入（**非严格主题模型，建议人工抽样复核**）。"
|
||||
)
|
||||
if n_comments < 5:
|
||||
hints.append(
|
||||
"有效评价样本偏少，消费者洞察部分仅作方向参考，正式结论建议加大 SKU 数或评论分页。"
|
||||
)
|
||||
if scen_n_texts >= 5 and scen_counts:
|
||||
top_lbl, top_n = scen_counts.most_common(1)[0]
|
||||
share = top_n / scen_n_texts
|
||||
if share >= 0.25:
|
||||
hints.append(
|
||||
f"用途/场景中「{top_lbl}」在约 {100 * share:.0f}% 的有效评价自述中出现，可作为沟通场景与卖点的优先假设（**词组规则，建议抽样核对原句**）。"
|
||||
)
|
||||
if not hints:
|
||||
hints.append(
|
||||
"当前样本下自动规则未触发强信号；请结合业务目标人工解读对比矩阵与原始 CSV。"
|
||||
)
|
||||
return hints
|
||||
|
||||
|
||||
def _embed_chart(run_dir: Path, filename: str, caption: str = "") -> list[str]:
|
||||
"""若 ``report_assets/<filename>`` 存在则返回插图 Markdown 片段。"""
|
||||
if not (run_dir / "report_assets" / filename).is_file():
|
||||
return []
|
||||
cap = (caption or "").strip()
|
||||
out: list[str] = []
|
||||
if cap:
|
||||
out.append(f"*{cap}*")
|
||||
out.append("")
|
||||
out.append(f"")
|
||||
out.append("")
|
||||
return out
|
||||
|
||||
|
||||
def _scenario_group_asset_slug(group: str, index: int) -> str:
|
||||
"""与 ``pipeline.reporting.charts`` 中场景分组图文件名规则一致（勿改格式）。"""
|
||||
raw = (group or "").strip()
|
||||
core = re.sub(r"[^\w\u4e00-\u9fff-]", "", raw)[:20]
|
||||
if not core:
|
||||
core = "group"
|
||||
return f"i{index:02d}_{core}"
|
||||
|
||||
|
||||
def _focus_scenario_combo_bar_filename(group: str, index: int) -> str:
|
||||
"""关注词 + 使用场景并排条形图（与 ``pipeline.reporting.charts.save_combo_focus_scenario_bar`` 同源）。"""
|
||||
slug = _scenario_group_asset_slug(group, index)
|
||||
return f"chart_focus_and_scenarios_bar__{slug}.png"
|
||||
|
||||
|
||||
def _matrix_prices_sales_chart_filename(group: str, index: int) -> str:
|
||||
"""与 ``pipeline.reporting.charts.generate_report_charts`` 中 ``chart_matrix_prices_sales__*`` 一致。"""
|
||||
slug = _scenario_group_asset_slug(group, index)
|
||||
return f"chart_matrix_prices_sales__{slug}.png"
|
||||
|
||||
|
||||
def _lines_4_reading_brand(
|
||||
*,
|
||||
cr1: float | None,
|
||||
cr3: float | None,
|
||||
top: str,
|
||||
brand_rows_n: int,
|
||||
n_structure: int,
|
||||
) -> list[str]:
|
||||
if cr1 is None or not (top or "").strip():
|
||||
return []
|
||||
lines = [
|
||||
"",
|
||||
"**数据解读（规则摘要）**：",
|
||||
"",
|
||||
f"- 在含品牌字段的 **{brand_rows_n}** 条列表行（占本章结构样本 **{n_structure}** 行）中，"
|
||||
f"「{_md_cell(top.strip(), 36)}」曝光约占 **{100 * cr1:.1f}%**（按行计，同一 SKU 多行会重复计）。",
|
||||
]
|
||||
if cr3 is not None:
|
||||
lines.append(
|
||||
f"- 前三品牌合计约 **{100 * cr3:.1f}%**；若该比例偏高，说明搜索页品牌集中度高，"
|
||||
"新品需搭配清晰的差异定位与资源投放，避免与头部在泛词下正面撞车。"
|
||||
)
|
||||
lines.append("")
|
||||
return lines
|
||||
|
||||
|
||||
def _lines_4_reading_shop(
|
||||
*,
|
||||
cr1: float | None,
|
||||
cr3: float | None,
|
||||
top: str,
|
||||
shop_rows_n: int,
|
||||
n_structure: int,
|
||||
) -> list[str]:
|
||||
if not shop_rows_n:
|
||||
return []
|
||||
lines = [
|
||||
"",
|
||||
"**数据解读（规则摘要）**：",
|
||||
"",
|
||||
f"- 含店铺名的列表行共 **{shop_rows_n}** 条（结构样本 **{n_structure}** 行），反映搜索曝光下的店铺格局。",
|
||||
]
|
||||
if cr1 is not None and (top or "").strip():
|
||||
lines.append(
|
||||
f"- 第一大店铺「{_md_cell(top.strip(), 40)}」约占 **{100 * cr1:.1f}%**；"
|
||||
"该指标刻画的是**列表可见度**而非销量，适合用于判断货架被哪些店铺占据。"
|
||||
)
|
||||
if cr3 is not None:
|
||||
lines.append(
|
||||
f"- 前三店铺合计约 **{100 * cr3:.1f}%**；若集中度高，可考虑从店铺矩阵、旗舰店/专营店布局等角度拆解竞争。"
|
||||
)
|
||||
lines.append("")
|
||||
return lines
|
||||
|
||||
|
||||
def build_competitor_markdown(
|
||||
*,
|
||||
|
||||
Loading…
x
Reference in New Issue
Block a user