')
article_end = raw.find("", article_start)
article = raw[article_start:article_end] if article_start >= 0 else raw
official_url = ""
for url, anchor_text in re.findall(r'
]*>(.*?)', article, re.S | re.I):
if "view website" in _text_from_html(anchor_text).lower():
official_url = secure_url(url, href)
break
official_url = official_url or href
for block in re.findall(r'(.*?)', raw, re.S | re.I):
def field(class_name: str) -> str:
match = re.search(r'class="' + re.escape(class_name) + r'"[^>]*>(.*?) horizon:
continue
venue = field("venue_name") or "UFI 公布会场"
summary = meta["description"] or "UFI 认证并公布未来场次、会场与主办官网的专业贸易展会。"
items.append(_dynamic_item(
now=now,
source=source,
result=result,
event_key=urlparse(href).query or meta["title"],
title=meta["title"],
summary=summary,
start=start,
end=end,
venue=venue,
official_url=official_url,
source_url=href,
image=meta["image"],
opportunity="先用 UFI 认证核验展会资质，再通过主办官网完成报名、观展、签证、住宿与买家会面安排。",
make_id=make_id,
poster_for=poster_for,
))
return items
def _event_alias(item: dict[str, Any]) -> str:
value = f"{item.get('title_original') or item.get('title') or ''} {item.get('venue') or ''}".lower()
normalized = unicodedata.normalize("NFKD", value).encode("ascii", "ignore").decode("ascii")
aliases = (
("intertextile shanghai home", "intertextile-home"),
("intertextile shanghai apparel", "intertextile-apparel"),
("home textiles sourcing", "home-textiles-sourcing"),
("texworld paris", "texworld-paris"),
("heimtextil" , "heimtextil"),
("techtextil" , "techtextil"),
("premiere vision", "premiere-vision"),
("maison objet", "maison-objet"),
)
if "texworld" in normalized and "paris" in normalized:
return "texworld-paris"
if normalized.strip().startswith("fim ") and "valencia" in normalized:
return "feria-habitat-valencia"
if "feria habitat valencia" in normalized:
return "feria-habitat-valencia"
for token, alias in aliases:
if token in normalized:
if alias in ("heimtextil", "techtextil"):
venue = normalize_text(str(item.get("venue") or "")).lower()
city = "frankfurt" if "frankfurt" in venue else venue.split(",")[0]
return f"{alias}-{city}"
return alias
normalized = re.sub(r"\b(autumn|spring|summer|winter|edition|international|trade|fair|exhibition|licensed)\b", " ", normalized)
return re.sub(r"[^a-z0-9]+", " ", normalized).strip()
def dedupe_exhibitions(items: list[dict[str, Any]]) -> list[dict[str, Any]]:
output: list[dict[str, Any]] = []
source_priority = {"expo_texpertise_network": 1, "expo_ufi_approved": 2}
for item in sorted(items, key=lambda row: (row["start_date"], source_priority.get(row["source_id"], 0), row["title"])):
alias = _event_alias(item)
duplicate = False
for existing in output:
if existing["start_date"] != item["start_date"]:
continue
existing_alias = _event_alias(existing)
if alias == existing_alias or SequenceMatcher(None, alias, existing_alias).ratio() >= 0.84:
duplicate = True
break
if not duplicate:
output.append(item)
return sorted(output, key=lambda row: (row["start_date"], row["title"]))
def build_exhibitions(now: date, horizon: date, sources: Iterable[Any], results: dict[str, Any], base_item: Callable[..., dict[str, Any]], make_id: Callable[..., str], poster_for: Callable[..., dict[str, str]]) -> list[dict[str, Any]]:
source_list = list(sources)
items = _curated_exhibitions(now, horizon, source_list, results, base_item)
by_id = {source.id: source for source in source_list}
texpertise = results.get("expo_texpertise_network")
if texpertise and texpertise.ok:
items.extend(_texpertise_exhibitions(now, horizon, by_id["expo_texpertise_network"], texpertise, make_id, poster_for))
ufi = results.get("expo_ufi_approved")
if ufi and ufi.ok:
items.extend(_ufi_exhibitions(now, horizon, by_id["expo_ufi_approved"], ufi, make_id, poster_for))
deduped = dedupe_exhibitions(items)
relevance_order = {"核心家纺展": 0, "纺织面料展": 1, "家居软装展": 2, "邻近行业机会": 3}
for item in deduped:
focus = set(item.get("event_focus") or [])
if focus & {"床品与套件", "窗帘与遮阳", "酒店与工程"}:
item["industry_relevance"] = "核心家纺展"
elif focus & {"纺织面料", "功能性材料", "可持续材料"}:
item["industry_relevance"] = "纺织面料展"
elif focus & {"空间与装饰面料", "地毯与地材", "家具与整体软装"}:
item["industry_relevance"] = "家居软装展"
else:
item["industry_relevance"] = "邻近行业机会"
return sorted(deduped, key=lambda row: (relevance_order[row["industry_relevance"]], row["start_date"], row["title"]))
def _refine_translation(value: str) -> str:
replacements = {
"家纺采购": "家用纺织品采购",
"服装采购": "服装与面料采购",
"家具陈设": "家具与软装",
"法兰克福/主要": "法兰克福",
"纺织技术": "纺织技术与材料",
}
output = value
for source, target in replacements.items():
output = output.replace(source, target)
return output
TITLE_OVERRIDES = {
"Zhejiang Export Fair (Vietnam)": "浙江出口商品（越南）交易会",
"Creativ": "Creativ 萨尔茨堡创意手作展",
"Habitare": "芬兰 Habitare 家居设计展",
"Ambienta": "克罗地亚 Ambienta 家居与室内设计展",
"FIM": "瓦伦西亚 FIM 家具与室内设计展",
"TEXTILHOGAR": "西班牙 TEXTILHOGAR 家用纺织品展",
"SICAM": "意大利 SICAM 家具部件与室内材料展",
"MASTERSHOES": "土耳其 MASTERSHOES 鞋材与时尚展",
"Ispo": "ISPO 运动用品与功能面料展",
"Furniture Design Now": "贝尔格莱德 Furniture Design Now 家具设计展",
"HOMETEX": "伊斯坦布尔 HOMETEX 家用纺织品展",
"Texcare International": "Texcare International 法兰克福纺织护理展",
"Texhibition": "伊斯坦布尔 Texhibition 面料与纺织辅料展",
"ZUCHEX": "伊斯坦布尔 ZUCHEX 家庭用品展",
"Egypt Stitch and Tex": "埃及 Stitch & Tex 纺织服装展",
"Feria Habitat Valencia": "瓦伦西亚 Habitat 家具与室内设计展",
"ANAMOB": "土耳其 ANAMOB 家具展",
"China International Occupational Safety & Health Goods Expo (CIOSH)": "中国国际劳动保护用品交易会 CIOSH",
"Cinte Techtextil China": "中国国际产业用纺织品及非织造布展 Cinte Techtextil China",
"Neonyt Paris (Sommer) licensed to WSN Développement SAS": "巴黎 Neonyt 可持续时尚展（夏季）",
"Val:ue Milan - Spring": "米兰 Val:ue 面料采购展（春季）",
"Texprocess Americas": "Texprocess Americas 美洲纺织加工技术展",
"Apparel Sourcing Tashkent licensed to LLC BMC-LLP": "塔什干服装与面料采购展 Apparel Sourcing",
"Heimtextil Uzbekistan licensed to LLC BMC-LLP": "乌兹别克斯坦 Heimtextil 家用纺织品展",
"Texworld Tashkent licesed to LLC BMC-LLP": "塔什干 Texworld 国际纺织面料展",
"Warsaw Home & Contracts": "华沙 Home & Contract 家居与软装展",
"Design Shanghai": "上海国际设计展 Design Shanghai",
"Techtextil North America": "Techtextil North America 北美产业用纺织品展",
"Techtextil India": "Techtextil India 印度产业用纺织品展",
"Textile Asia": "巴基斯坦 Textile Asia 纺织展",
"Heimtextil": "Heimtextil 法兰克福家用及商用纺织品展",
"Techtextil": "Techtextil 法兰克福产业用纺织品展",
"Texprocess": "Texprocess 法兰克福纺织加工技术展",
"Home Textiles Sourcing": "纽约 Home Textiles Sourcing 家用纺织品采购展",
}
def translate_exhibitions(items: list[dict[str, Any]], output_dir: Path, max_new: int = 200) -> None:
cache = load_zh_cache(output_dir)
session = make_session()
translated_count = 0
cache_changed = False
for item in items:
if item.get("source_id") not in ("expo_texpertise_network", "expo_ufi_approved"):
continue
translated_any = False
for original_key, target_key in (("title_original", "title"), ("summary_original", "summary")):
original = normalize_text(str(item.get(original_key) or ""))
if not original:
continue
if original_key == "title_original" and original in TITLE_OVERRIDES:
item[target_key] = TITLE_OVERRIDES[original]
translated_any = True
continue
translated = cache.get(original)
if not translated and translated_count < max_new:
translated = translate_to_zh_cn(session, original)
if translated:
cache[original] = translated
translated_count += 1
cache_changed = True
if translated:
item[target_key] = _refine_translation(translated)
translated_any = True
item["translation_status"] = "已翻译并保留官方原文" if translated_any else "保留官方原文"
if cache_changed:
save_zh_cache(output_dir, cache)