Spaces:
Running
Running
Update app.py
Browse files
app.py
CHANGED
|
@@ -8,55 +8,90 @@ from playwright.sync_api import sync_playwright
|
|
| 8 |
from curl_cffi import requests as curl_req
|
| 9 |
import gradio as gr
|
| 10 |
|
|
|
|
|
|
|
|
|
|
|
|
|
| 11 |
os.system("playwright install chromium")
|
| 12 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 13 |
def get_official_japanese_name(ingredient_en):
|
| 14 |
-
"""
|
| 15 |
-
利用維基百科 API 跨語言連結,精準取得官方日文藥名 (JAN)
|
| 16 |
-
"""
|
| 17 |
try:
|
| 18 |
-
# 1. 查詢英文維基百科,確認該藥品頁面是否存在
|
| 19 |
search_url = f"https://en.wikipedia.org/w/api.php?action=query&list=search&srsearch={quote(ingredient_en)}&utf8=&format=json"
|
| 20 |
res = requests.get(search_url, timeout=5).json()
|
| 21 |
if not res['query']['search']: return None
|
| 22 |
|
| 23 |
-
# 取得最吻合的英文頁面標題
|
| 24 |
en_title = res['query']['search'][0]['title']
|
| 25 |
-
|
| 26 |
-
# 2. 查詢該頁面是否有對應的「日文版」頁面,並抓取日文標題
|
| 27 |
lang_url = f"https://en.wikipedia.org/w/api.php?action=query&titles={quote(en_title)}&prop=langlinks&lllang=ja&format=json"
|
| 28 |
lang_res = requests.get(lang_url, timeout=5).json()
|
| 29 |
|
| 30 |
pages = lang_res['query']['pages']
|
| 31 |
for page_id in pages:
|
| 32 |
if 'langlinks' in pages[page_id]:
|
| 33 |
-
# 日文維基百科的標題,99% 就是日本官方一般名 (JAN)
|
| 34 |
ja_title = pages[page_id]['langlinks'][0]['*']
|
| 35 |
-
# 清除有時維基百科標題會帶有的後綴,例如 "フィネレノン (化合物)"
|
| 36 |
return ja_title.split(' ')[0].split('(')[0]
|
| 37 |
except Exception:
|
| 38 |
pass
|
| 39 |
return None
|
| 40 |
|
| 41 |
def translate_lang(text, target_lang):
|
| 42 |
-
"""
|
| 43 |
-
|
| 44 |
-
|
| 45 |
-
# 💡 針對日文,優先使用「醫學實體對照法」,這比 Google 翻譯準確一萬倍
|
| 46 |
if target_lang == 'ja':
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 47 |
official_ja = get_official_japanese_name(text)
|
| 48 |
if official_ja:
|
| 49 |
-
|
|
|
|
| 50 |
|
| 51 |
-
#
|
| 52 |
try:
|
| 53 |
url = f"https://translate.googleapis.com/translate_a/single?client=gtx&sl=en&tl={target_lang}&dt=t&q={quote(text)}"
|
| 54 |
res = requests.get(url, timeout=5)
|
| 55 |
if res.status_code == 200:
|
| 56 |
-
|
|
|
|
|
|
|
| 57 |
except Exception:
|
| 58 |
pass
|
| 59 |
-
|
| 60 |
return text
|
| 61 |
|
| 62 |
def is_generic(brand_name, company_name, ingredient):
|
|
@@ -64,7 +99,6 @@ def is_generic(brand_name, company_name, ingredient):
|
|
| 64 |
c_lower = company_name.lower()
|
| 65 |
i_lower = ingredient.lower()
|
| 66 |
|
| 67 |
-
# 💡 新增您在 FASS 發現的學名藥廠: tiefenbacher, cinfa, polpharma, pharmaceutical innovation...
|
| 68 |
generic_keywords = [
|
| 69 |
'sandoz', 'teva', 'apotex', 'ratiopharm', 'jamp', 'mint', 'pharmascience', 'sanis', 'sivem',
|
| 70 |
'auro', 'glenmark', 'taro', 'marcan', 'nora', 'mantra', 'reddy', 'mepha', 'axapharm',
|
|
@@ -82,63 +116,21 @@ def is_generic(brand_name, company_name, ingredient):
|
|
| 82 |
return False
|
| 83 |
|
| 84 |
def clean_brand_name(raw_name):
|
| 85 |
-
|
| 86 |
-
精準擷取商品名主體:
|
| 87 |
-
1. 遇到劑量數字 (\d+) 直接切斷
|
| 88 |
-
2. 遇到多國常見的劑型、包裝描述詞彙直接切斷
|
| 89 |
-
"""
|
| 90 |
-
# 擴充了德文、法文、英文、日文的常見劑型與包裝詞彙
|
| 91 |
-
# 加入 inj (注射), pen (注射筆), prefilled (預填) 來應對 Semaglutide 等生物製劑
|
| 92 |
pattern = r'(皮下注|錠|カプセル|顆粒|シロップ|OD|細粒|液|Augentropfen|Schmelztabletten|Tabletten|kids|Lingual|Sol|cp|inj|pen|prefilled|flex|\d+)'
|
| 93 |
-
|
| 94 |
-
# 透過正則表達式忽略大小寫進行切割,並只取最前面的主商標名
|
| 95 |
cleaned = re.split(pattern, raw_name, flags=re.IGNORECASE)[0]
|
| 96 |
-
|
| 97 |
-
# 清除 ®、™ 符號,並去除頭尾可能殘留的空白或連接號
|
| 98 |
return cleaned.replace('®', '').replace('™', '').strip(' -_')
|
| 99 |
|
|
|
|
| 100 |
# ==========================================
|
| 101 |
-
# 🚀 模組 A:使用 curl_cffi 抓取 (
|
| 102 |
# ==========================================
|
| 103 |
-
def get_usa_originator(ingredient):
|
| 104 |
-
log, brands, companies = [], set(), set()
|
| 105 |
-
try:
|
| 106 |
-
session = curl_req.Session(impersonate="chrome120")
|
| 107 |
-
log.append("1. 發送 GET 請求至 FDA Orange Book...")
|
| 108 |
-
# 改用標準 GET 參數直接查詢結果頁面
|
| 109 |
-
url = f"https://www.accessdata.fda.gov/scripts/cder/ob/results_product.cfm?Generic_Name={quote(ingredient)}&rx_otc=All"
|
| 110 |
-
res = session.get(url, timeout=30, verify=False)
|
| 111 |
-
log.append(f" -> 狀態碼: {res.status_code}, 內容長度: {len(res.text)}")
|
| 112 |
-
|
| 113 |
-
soup = BeautifulSoup(res.text, 'html.parser')
|
| 114 |
-
table = soup.find('table', id='example')
|
| 115 |
-
if table:
|
| 116 |
-
rows = table.find('tbody').find_all('tr') if table.find('tbody') else table.find_all('tr')
|
| 117 |
-
log.append(f"2. 找到結果表格,共 {len(rows)} 列資料。")
|
| 118 |
-
headers = [th.get_text(strip=True).lower() for th in table.find_all('th')]
|
| 119 |
-
brand_idx = next((i for i, h in enumerate(headers) if 'proprietary name' in h), 2)
|
| 120 |
-
rld_idx = next((i for i, h in enumerate(headers) if 'rld' in h), 8)
|
| 121 |
-
mfg_idx = next((i for i, h in enumerate(headers) if 'applicant holder' in h), 10)
|
| 122 |
-
|
| 123 |
-
for tr in rows:
|
| 124 |
-
tds = tr.find_all('td')
|
| 125 |
-
if len(tds) > max(rld_idx, brand_idx) and "RLD" in tds[rld_idx].get_text(strip=True).upper():
|
| 126 |
-
brands.add(tds[brand_idx].get_text(strip=True))
|
| 127 |
-
if len(tds) > mfg_idx: companies.add(tds[mfg_idx].get_text(strip=True))
|
| 128 |
-
else:
|
| 129 |
-
log.append("❌ HTML 中未發現 table#example,可能是無此藥或被阻擋。")
|
| 130 |
-
|
| 131 |
-
if brands: return ", ".join(brands), ", ".join(companies), "\n".join(log)
|
| 132 |
-
return "查無原廠", "-", "\n".join(log)
|
| 133 |
-
except Exception as e: return "執行失敗", "-", "\n".join(log) + f"\n錯誤: {str(e)}"
|
| 134 |
-
|
| 135 |
def get_australia_originator(ingredient):
|
| 136 |
log, brands = [], set()
|
| 137 |
try:
|
| 138 |
session = curl_req.Session(impersonate="chrome120")
|
| 139 |
log.append("1. 發送 GET 至澳洲 TGA 搜尋頁面...")
|
| 140 |
res = session.get(f"https://www.tga.gov.au/resources/artg?keywords={ingredient}", timeout=30, verify=False)
|
| 141 |
-
log.append(f" -> 狀態碼: {res.status_code}")
|
| 142 |
soup = BeautifulSoup(res.text, 'html.parser')
|
| 143 |
|
| 144 |
articles = soup.find_all('article', class_='node--artg')
|
|
@@ -152,78 +144,189 @@ def get_australia_originator(ingredient):
|
|
| 152 |
if len(parts) > 1 and parts[0].strip():
|
| 153 |
brand = parts[0].strip()
|
| 154 |
if not is_generic(brand, "", ingredient):
|
| 155 |
-
# 💡 取得該筆紀錄的內頁連結 (href)
|
| 156 |
a_tag = title_tag.find('a')
|
| 157 |
href = a_tag['href'] if a_tag else None
|
| 158 |
cands.append({"brand": brand, "date": time_tag.get('datetime'), "href": href})
|
| 159 |
|
| 160 |
if cands:
|
| 161 |
-
# 依日期排序,找出最古老的註冊紀錄
|
| 162 |
cands = sorted(cands, key=lambda x: x['date'])
|
| 163 |
target = cands[0]
|
| 164 |
brand = target['brand']
|
| 165 |
log.append(f"✅ 依日期排序鎖定最早註冊: {target['date'][:10]} ({brand})")
|
| 166 |
|
| 167 |
-
company = "TGA資料庫"
|
| 168 |
-
|
| 169 |
-
# 💡 新增:進入內頁抓取藥廠 (Sponsor)
|
| 170 |
if target['href']:
|
| 171 |
-
# 確保網址是絕對路徑
|
| 172 |
detail_url = f"https://www.tga.gov.au{target['href']}" if target['href'].startswith('/') else target['href']
|
| 173 |
log.append("3. 進入內頁抓取 Sponsor 藥廠資訊...")
|
| 174 |
-
|
| 175 |
res_detail = session.get(detail_url, timeout=30, verify=False)
|
| 176 |
if res_detail.status_code == 200:
|
| 177 |
detail_soup = BeautifulSoup(res_detail.text, 'html.parser')
|
| 178 |
-
|
| 179 |
-
# 依據您提供的 HTML 結構精準定位
|
| 180 |
sponsor_div = detail_soup.find('div', class_=re.compile(r'field--name-field-sponsor'))
|
| 181 |
if sponsor_div and sponsor_div.find('a'):
|
| 182 |
company = sponsor_div.find('a').get_text(strip=True)
|
| 183 |
log.append(f" -> 成功取得藥廠名稱: {company}")
|
| 184 |
-
else:
|
| 185 |
-
log.append(" -> 內頁未找到 Sponsor 標籤。")
|
| 186 |
-
else:
|
| 187 |
-
log.append(f" -> 進入內頁失敗,狀態碼: {res_detail.status_code}")
|
| 188 |
-
|
| 189 |
return brand, company, "\n".join(log)
|
| 190 |
-
|
| 191 |
return "查無原廠", "-", "\n".join(log)
|
| 192 |
except Exception as e:
|
| 193 |
return "執行失敗", "-", "\n".join(log) + f"\n錯誤: {str(e)}"
|
| 194 |
|
|
|
|
| 195 |
# ==========================================
|
| 196 |
-
# 🚀 模組 B:使用 Playwright 抓取 (比、英、加、日、德、瑞)
|
| 197 |
# ==========================================
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 198 |
def get_belgium_originator(ingredient, page):
|
| 199 |
log, brands = [], set()
|
| 200 |
try:
|
| 201 |
log.append(f"1. 前往 CBIP (比利時) 搜尋頁面 (搜尋: {ingredient})...")
|
| 202 |
page.goto(f"https://www.cbip.be/fr/search?q={ingredient}", timeout=45000, wait_until="domcontentloaded")
|
| 203 |
-
|
| 204 |
-
# 等待搜尋結果區塊渲染
|
| 205 |
page.wait_for_selector('ul.search-results, .no-results', timeout=15000)
|
| 206 |
page.wait_for_timeout(2000)
|
| 207 |
|
| 208 |
soup = BeautifulSoup(page.content(), 'html.parser')
|
| 209 |
-
|
| 210 |
-
# 💡 修正:直接抓取搜尋結果清單中的所有連結,不再嚴格限制 href 的結構
|
| 211 |
-
log.append("2. 掃描搜尋結果...")
|
| 212 |
results_list = soup.find('ul', class_=re.compile(r'search-results|results'))
|
| 213 |
|
| 214 |
if results_list:
|
| 215 |
links = results_list.find_all('a')
|
| 216 |
for link in links:
|
| 217 |
text = link.get_text(strip=True)
|
| 218 |
-
|
| 219 |
-
# 過濾掉太短的字或明顯不是藥名的系統文字
|
| 220 |
if len(text) > 3 and not text.lower().startswith('voir'):
|
| 221 |
-
# 依賴強大的黑名單過濾學名藥廠 (如 EG, Teva, Sandoz)
|
| 222 |
if not is_generic(text, "", ingredient):
|
| 223 |
brands.add(clean_brand_name(text))
|
| 224 |
-
else:
|
| 225 |
-
log.append("❌ 畫面上未出現搜尋結果清單。")
|
| 226 |
-
|
| 227 |
if brands: return ", ".join(brands), "CBIP 資料庫", "\n".join(log)
|
| 228 |
return "查無原廠", "-", "\n".join(log) + "\n❌ 查無資料或全為學名藥"
|
| 229 |
except Exception as e:
|
|
@@ -237,7 +340,6 @@ def get_uk_originator(ingredient, page):
|
|
| 237 |
page.wait_for_selector('.search-results-product-info-title-link', timeout=15000)
|
| 238 |
soup = BeautifulSoup(page.content(), 'html.parser')
|
| 239 |
links = soup.find_all('a', class_='search-results-product-info-title-link')
|
| 240 |
-
log.append(f"2. 找到 {len(links)} 筆標題,開始篩選。")
|
| 241 |
for link in links:
|
| 242 |
title = link.get_text(strip=True)
|
| 243 |
if not title.lower().startswith(ingredient.lower()):
|
|
@@ -256,18 +358,12 @@ def get_canada_originator(ingredient, page):
|
|
| 256 |
page.goto("https://health-products.canada.ca/dpd-bdpp/index-eng.jsp", timeout=45000, wait_until="domcontentloaded")
|
| 257 |
page.locator('input[id="activeIngredient"]').fill(ingredient)
|
| 258 |
page.keyboard.press("Enter")
|
| 259 |
-
|
| 260 |
-
log.append("2. 等待結果表格或查無資料訊息...")
|
| 261 |
page.wait_for_selector('table#results, .alert-info, .alert-warning', timeout=15000)
|
| 262 |
soup = BeautifulSoup(page.content(), 'html.parser')
|
| 263 |
table = soup.find('table', id='results')
|
| 264 |
|
| 265 |
-
if not table or not table.find('tbody'):
|
| 266 |
-
log.append("❌ 畫面上未出現表格,可能查無此藥。")
|
| 267 |
-
return "查無資料", "-", "\n".join(log)
|
| 268 |
-
|
| 269 |
rows = table.find('tbody').find_all('tr')
|
| 270 |
-
log.append(f"3. 找到 {len(rows)} 筆資料,進行黑名單過濾。")
|
| 271 |
all_cands = []
|
| 272 |
for tr in rows:
|
| 273 |
tds = tr.find_all('td')
|
|
@@ -279,12 +375,10 @@ def get_canada_originator(ingredient, page):
|
|
| 279 |
all_cands.append({"company": comp, "product": tds[3].get_text(strip=True), "din": int(m.group())})
|
| 280 |
|
| 281 |
if not all_cands: return "查無原廠", "-", "\n".join(log) + "\n❌ 剩餘皆為學名藥廠"
|
| 282 |
-
|
| 283 |
all_cands = sorted(all_cands, key=lambda x: x['din'])
|
| 284 |
orig_comp = all_cands[0]['company']
|
| 285 |
brands = set([c['product'] for c in all_cands if c['company'] == orig_comp])
|
| 286 |
-
|
| 287 |
-
return ", ".join(brands), orig_comp, "\n".join(log)
|
| 288 |
except Exception as e: return "執行失敗", "-", "\n".join(log) + f"\n錯誤: {str(e)}"
|
| 289 |
|
| 290 |
def get_japan_originator(ing_ja, page):
|
|
@@ -292,79 +386,55 @@ def get_japan_originator(ing_ja, page):
|
|
| 292 |
try:
|
| 293 |
log.append(f"1. 前往 PMDA (搜尋: {ing_ja})...")
|
| 294 |
page.goto("https://www.pmda.go.jp/PmdaSearch/iyakuSearch/", timeout=45000, wait_until="domcontentloaded")
|
| 295 |
-
|
| 296 |
-
log.append("2. 準確定位置首頁搜尋框 (id='txtName')...")
|
| 297 |
-
# 依照您提供的 HTML 結構,直接鎖定 txtName
|
| 298 |
search_input = page.locator('input#txtName')
|
| 299 |
search_input.wait_for(state="attached", timeout=15000)
|
| 300 |
search_input.fill(ing_ja, force=True)
|
| 301 |
|
| 302 |
-
log.append("3. 觸發搜尋並監聽彈出新視窗...")
|
| 303 |
with page.expect_popup() as popup_info:
|
| 304 |
search_input.press("Enter")
|
| 305 |
popup = popup_info.value
|
| 306 |
-
|
| 307 |
-
log.append("4. 等待新視窗表格並啟動自動翻頁掃描...")
|
| 308 |
current_page = 1
|
| 309 |
|
| 310 |
while True:
|
| 311 |
-
# 等待表格出現
|
| 312 |
popup.wait_for_selector('table#ResultList, .errormsg, .non-result', timeout=15000)
|
| 313 |
soup = BeautifulSoup(popup.content(), 'html.parser')
|
| 314 |
table = soup.find('table', id='ResultList')
|
| 315 |
|
| 316 |
if table:
|
| 317 |
rows = table.find_all('tr')
|
| 318 |
-
log.append(f" -> 第 {current_page} 頁找到 {len(rows)} 列資料。")
|
| 319 |
for tr in rows:
|
| 320 |
tds = tr.find_all('td')
|
| 321 |
if len(tds) >= 3:
|
| 322 |
title = tds[1].get_text(strip=True)
|
| 323 |
-
# 這裡會自動把帶有「括號」或廠商名稱的學名藥濾掉
|
| 324 |
if not is_generic(title, "", ing_ja):
|
| 325 |
brands.add(clean_brand_name(title))
|
| 326 |
companies.add(tds[2].get_text(separator=" ", strip=True).replace('製造販売元/', ''))
|
| 327 |
-
else:
|
| 328 |
-
log.append(f"❌ 第 {current_page} 頁未出現 ResultList。")
|
| 329 |
-
break
|
| 330 |
|
| 331 |
-
# 尋找下一頁的按鈕 (例如 javascript:changePg(2);)
|
| 332 |
current_page += 1
|
| 333 |
next_page_link = popup.locator(f'a[href="javascript:changePg({current_page});"]')
|
| 334 |
-
|
| 335 |
if next_page_link.count() > 0:
|
| 336 |
-
log.append(f" -> 發現第 {current_page} 頁,執行翻頁...")
|
| 337 |
next_page_link.first.click(force=True)
|
| 338 |
-
popup.wait_for_timeout(2000)
|
| 339 |
-
else:
|
| 340 |
-
log.append(" -> 已無下一頁,結束掃描。")
|
| 341 |
-
break
|
| 342 |
|
| 343 |
popup.close()
|
| 344 |
-
|
| 345 |
if brands: return ", ".join(brands), ", ".join(companies), "\n".join(log)
|
| 346 |
-
return "查無原廠", "-", "\n".join(log)
|
| 347 |
-
except Exception as e:
|
| 348 |
-
return "執行失敗", "-", "\n".join(log) + f"\n錯誤: {str(e)}"
|
| 349 |
|
| 350 |
def get_switzerland_originator(ing_de, page):
|
| 351 |
log, brands, companies = [], set(), set()
|
| 352 |
try:
|
| 353 |
log.append(f"1. 前往 Swissmedicinfo Pro 專業版 (搜尋: {ing_de})...")
|
| 354 |
page.goto("https://swissmedicinfo-pro.ch/?Lang=EN", timeout=45000, wait_until="domcontentloaded")
|
| 355 |
-
|
| 356 |
-
log.append("2. 精準定位「成分專用」搜尋框 (txtSubstance)...")
|
| 357 |
-
# 💡 直接使用您提供的精準 ID,這招絕對萬無一失!
|
| 358 |
search_input = page.locator('input#MainContent_ucSearch1_txtSubstance')
|
| 359 |
search_input.wait_for(state="attached", timeout=15000)
|
| 360 |
search_input.fill(ing_de, force=True)
|
| 361 |
-
|
| 362 |
-
log.append("3. 按下 Enter 並等待 ASP.NET 載入完整資料...")
|
| 363 |
search_input.press("Enter")
|
| 364 |
|
| 365 |
page.wait_for_load_state('networkidle', timeout=15000)
|
| 366 |
page.wait_for_timeout(3000)
|
| 367 |
-
|
| 368 |
page.wait_for_selector('table[id*="GVMonographies"], #MainContent_LabelNoResult', timeout=15000)
|
| 369 |
|
| 370 |
soup = BeautifulSoup(page.content(), 'html.parser')
|
|
@@ -372,37 +442,27 @@ def get_switzerland_originator(ing_de, page):
|
|
| 372 |
|
| 373 |
if table and table.find('tbody'):
|
| 374 |
rows = table.find('tbody').find_all('tr', class_=re.compile(r'clickable-row'))
|
| 375 |
-
log.append(f"4. 畫面解析完成,找到 {len(rows)} 筆表格資料。")
|
| 376 |
-
|
| 377 |
for tr in rows:
|
| 378 |
tds = tr.find_all('td')
|
| 379 |
if len(tds) >= 4:
|
| 380 |
title = tds[0].get_text(strip=True)
|
| 381 |
comp = tds[3].get_text(strip=True)
|
| 382 |
-
|
| 383 |
if not is_generic(title, comp, ing_de):
|
| 384 |
brands.add(clean_brand_name(title))
|
| 385 |
if comp != "-": companies.add(comp)
|
| 386 |
-
else:
|
| 387 |
-
log.append("❌ 畫面上未出現表格,查無資料。")
|
| 388 |
-
|
| 389 |
if brands: return ", ".join(brands), ", ".join(companies), "\n".join(log)
|
| 390 |
-
return "查無原廠", "-", "\n".join(log)
|
| 391 |
-
except Exception as e:
|
| 392 |
-
return "執行失敗", "-", "\n".join(log) + f"\n錯誤: {str(e)}"
|
| 393 |
|
| 394 |
def get_germany_originator(ing_de, page):
|
| 395 |
log, brands, companies = [], set(), set()
|
| 396 |
try:
|
| 397 |
-
log.append(f"1. 前往 Gelbe Liste (搜尋: {ing_de})...")
|
| 398 |
page.goto(f"https://www.gelbe-liste.de/profi-suche/results?substance={quote(ing_de)}", timeout=30000)
|
| 399 |
page.wait_for_selector('.product-list', timeout=15000)
|
| 400 |
-
|
| 401 |
soup = BeautifulSoup(page.content(), 'html.parser')
|
| 402 |
ul = soup.find('ul', class_='product-list')
|
| 403 |
if ul:
|
| 404 |
lis = ul.find_all('li')
|
| 405 |
-
log.append(f"2. 找到 {len(lis)} 筆結果,過濾平行輸入商。")
|
| 406 |
for li in lis:
|
| 407 |
h5, p_tag = li.find('h5'), li.find('p', class_='small')
|
| 408 |
if h5:
|
|
@@ -411,141 +471,25 @@ def get_germany_originator(ing_de, page):
|
|
| 411 |
brands.add(clean_brand_name(title))
|
| 412 |
companies.add(comp)
|
| 413 |
if brands: return ", ".join(brands), ", ".join(companies), "\n".join(log)
|
| 414 |
-
return "查無原廠", "-", "\n".join(log)
|
| 415 |
except Exception as e: return "執行失敗", "-", "\n".join(log) + f"\n錯誤: {str(e)}"
|
| 416 |
|
| 417 |
-
def get_france_originator(ingredient, page):
|
| 418 |
-
log, brands, companies = [], set(), set()
|
| 419 |
-
try:
|
| 420 |
-
log.append(f"1. 前往 Vidal 搜尋頁面 (搜尋: {ingredient})...")
|
| 421 |
-
page.goto(f"https://www.vidal.fr/recherche.html?query={quote(ingredient)}", timeout=45000, wait_until="domcontentloaded")
|
| 422 |
-
|
| 423 |
-
log.append("2. 等待搜尋結果渲染...")
|
| 424 |
-
page.wait_for_selector('.results, .searchbar', timeout=15000)
|
| 425 |
-
|
| 426 |
-
soup = BeautifulSoup(page.content(), 'html.parser')
|
| 427 |
-
divs = soup.find_all('div', class_=re.compile(r'result drug'))
|
| 428 |
-
log.append(f"3. 找到 {len(divs)} 筆 'result drug' 區塊。")
|
| 429 |
-
|
| 430 |
-
cands = []
|
| 431 |
-
for div in divs:
|
| 432 |
-
info_div = div.find('div', class_='infos')
|
| 433 |
-
if info_div and info_div.find('a'):
|
| 434 |
-
a_tag = info_div.find('a')
|
| 435 |
-
title = a_tag.get_text(strip=True)
|
| 436 |
-
href = a_tag.get('href', '')
|
| 437 |
-
|
| 438 |
-
# 第一階段:先用商品名初步過濾掉明顯的學名藥
|
| 439 |
-
if not is_generic(title, "", ingredient):
|
| 440 |
-
cands.append({"raw_title": title, "href": href})
|
| 441 |
-
|
| 442 |
-
if cands:
|
| 443 |
-
log.append(f"4. 篩選出 {len(cands)} 筆非學名藥候選,準備進入內頁抓取藥廠 (Laboratoire)...")
|
| 444 |
-
|
| 445 |
-
for cand in cands:
|
| 446 |
-
title = cand['raw_title']
|
| 447 |
-
href = cand['href']
|
| 448 |
-
|
| 449 |
-
# 組裝絕對網址
|
| 450 |
-
if href.startswith('http'):
|
| 451 |
-
detail_url = href
|
| 452 |
-
elif href.startswith('/'):
|
| 453 |
-
detail_url = f"https://www.vidal.fr{href}"
|
| 454 |
-
else:
|
| 455 |
-
detail_url = f"https://www.vidal.fr/{href}"
|
| 456 |
-
|
| 457 |
-
log.append(f" -> 進入內頁: {clean_brand_name(title)}")
|
| 458 |
-
page.goto(detail_url, timeout=30000, wait_until="domcontentloaded")
|
| 459 |
-
|
| 460 |
-
comp = "-"
|
| 461 |
-
try:
|
| 462 |
-
# 💡 依照您提供的 HTML 結構,等待並精準抓取 div.nomlab
|
| 463 |
-
page.wait_for_selector('div.nomlab', timeout=10000)
|
| 464 |
-
detail_soup = BeautifulSoup(page.content(), 'html.parser')
|
| 465 |
-
nomlab_div = detail_soup.find('div', class_='nomlab')
|
| 466 |
-
|
| 467 |
-
if nomlab_div:
|
| 468 |
-
comp = nomlab_div.get_text(strip=True)
|
| 469 |
-
log.append(f" ✅ 成功取得藥廠: {comp}")
|
| 470 |
-
except Exception:
|
| 471 |
-
log.append(" ❌ 取得藥廠失敗或 Timeout。")
|
| 472 |
-
|
| 473 |
-
# 第二階段:取得藥廠後,進行最終嚴格過濾 (確保萬無一失)
|
| 474 |
-
if not is_generic(title, comp, ingredient):
|
| 475 |
-
brands.add(clean_brand_name(title))
|
| 476 |
-
if comp != "-": companies.add(comp)
|
| 477 |
-
|
| 478 |
-
if brands:
|
| 479 |
-
return ", ".join(brands), ", ".join(companies), "\n".join(log)
|
| 480 |
-
|
| 481 |
-
return "查無原廠", "-", "\n".join(log) + "\n❌ 查無非學名藥資料"
|
| 482 |
-
except Exception as e:
|
| 483 |
-
return "執行失敗", "-", "\n".join(log) + f"\n錯誤: {str(e)}"
|
| 484 |
-
|
| 485 |
-
def get_sweden_originator(ingredient, page):
|
| 486 |
-
log, brands, companies = [], set(), set()
|
| 487 |
-
try:
|
| 488 |
-
log.append(f"1. 前往瑞典 FASS 搜尋頁面 (搜尋: {ingredient})...")
|
| 489 |
-
page.goto(f"https://fass.se/search?query={quote(ingredient)}", timeout=45000, wait_until="domcontentloaded")
|
| 490 |
-
|
| 491 |
-
log.append("2. 鎖定特定的搜尋結果區塊...")
|
| 492 |
-
try:
|
| 493 |
-
page.wait_for_selector('details.app-toggle-details-icon, .no-results', state="attached", timeout=15000)
|
| 494 |
-
except Exception:
|
| 495 |
-
log.append(" -> 查無 details 節點,判斷為無結果。")
|
| 496 |
-
|
| 497 |
-
page.wait_for_timeout(2000) # 給予前端框架充裕時間建構 DOM 樹
|
| 498 |
-
|
| 499 |
-
soup = BeautifulSoup(page.content(), 'html.parser')
|
| 500 |
-
items = soup.find_all('details', class_=re.compile(r'app-toggle-details-icon'))
|
| 501 |
-
log.append(f"3. 找到 {len(items)} 筆搜尋結果。")
|
| 502 |
-
|
| 503 |
-
for item in items:
|
| 504 |
-
# 1. 從 summary 標籤中抓取商品名
|
| 505 |
-
summary = item.find('summary')
|
| 506 |
-
if not summary: continue
|
| 507 |
-
|
| 508 |
-
title_span = summary.find('span', class_=re.compile(r'font-semibold'))
|
| 509 |
-
title = title_span.get_text(strip=True) if title_span else ""
|
| 510 |
-
|
| 511 |
-
# 2. 💡 修正:深入內層的 <ol> 列表,精準抓取真正的藥廠名稱
|
| 512 |
-
comp = "-"
|
| 513 |
-
ol = item.find('ol')
|
| 514 |
-
if ol:
|
| 515 |
-
# 這樣就不會抓到外層的 Bilastin 了
|
| 516 |
-
comp_span = ol.find('span', class_='text-label-md')
|
| 517 |
-
if comp_span:
|
| 518 |
-
comp = comp_span.get_text(strip=True)
|
| 519 |
-
|
| 520 |
-
# 3. 執行黑名單過濾 (此時 comp 會是 Tiefenbacher 等,過濾器將成功攔截)
|
| 521 |
-
if title:
|
| 522 |
-
if not is_generic(title, comp, ingredient):
|
| 523 |
-
brands.add(clean_brand_name(title))
|
| 524 |
-
if comp != "-": companies.add(comp)
|
| 525 |
-
|
| 526 |
-
if brands: return ", ".join(brands), ", ".join(companies), "\n".join(log)
|
| 527 |
-
return "查無原廠", "-", "\n".join(log) + "\n❌ 查無資料或全為學名藥"
|
| 528 |
-
except Exception as e:
|
| 529 |
-
return "執行失敗", "-", "\n".join(log) + f"\n錯誤: {str(e)}"
|
| 530 |
-
|
| 531 |
# ==========================================
|
| 532 |
# 🚀 主執行中樞
|
| 533 |
# ==========================================
|
| 534 |
def run_all_ten_countries(ing_en, ing_ja_manual, ing_de_manual):
|
| 535 |
if not ing_en: return [["錯誤", "請輸入英文成分名", "-", ""]]
|
| 536 |
|
|
|
|
| 537 |
ing_ja = ing_ja_manual if ing_ja_manual else translate_lang(ing_en, 'ja')
|
| 538 |
ing_de = ing_de_manual if ing_de_manual else translate_lang(ing_en, 'de')
|
| 539 |
results = []
|
| 540 |
|
| 541 |
-
#
|
| 542 |
-
usa_b, usa_c, usa_log = get_usa_originator(ing_en)
|
| 543 |
-
results.append(["🇺🇸 美國 (FDA)", usa_b, usa_c, usa_log])
|
| 544 |
-
|
| 545 |
au_b, au_c, au_log = get_australia_originator(ing_en)
|
| 546 |
results.append(["🇦🇺 澳洲 (TGA)", au_b, au_c, au_log])
|
| 547 |
|
| 548 |
-
#
|
| 549 |
with sync_playwright() as p:
|
| 550 |
browser = p.chromium.launch(headless=True, args=['--no-sandbox', '--disable-dev-shm-usage'])
|
| 551 |
context = browser.new_context(user_agent="Mozilla/5.0 (Windows NT 10.0; Win64; x64) Chrome/122.0.0.0 Safari/537.36")
|
|
@@ -556,14 +500,15 @@ def run_all_ten_countries(ing_en, ing_ja_manual, ing_de_manual):
|
|
| 556 |
page.close()
|
| 557 |
results.append([name, b, c, log])
|
| 558 |
|
| 559 |
-
run_pw(
|
| 560 |
run_pw(get_belgium_originator, ing_en, "🇧🇪 比利時 (CBIP)")
|
|
|
|
| 561 |
run_pw(get_uk_originator, ing_en, "🇬🇧 英國 (eMC)")
|
| 562 |
run_pw(get_canada_originator, ing_en, "🇨🇦 加拿大 (DPD)")
|
| 563 |
run_pw(get_japan_originator, ing_ja, "🇯🇵 日本 (PMDA)")
|
| 564 |
run_pw(get_switzerland_originator, ing_de, "🇨🇭 瑞士 (Swissmedicinfo)")
|
| 565 |
-
run_pw(get_germany_originator, ing_de, "🇩🇪 德國 (Gelbe Liste)")
|
| 566 |
run_pw(get_sweden_originator, ing_en, "🇸🇪 瑞典 (FASS)")
|
|
|
|
| 567 |
|
| 568 |
browser.close()
|
| 569 |
|
|
@@ -573,13 +518,13 @@ def run_all_ten_countries(ing_en, ing_ja_manual, ing_de_manual):
|
|
| 573 |
# 🎨 UI 介面
|
| 574 |
# ==========================================
|
| 575 |
with gr.Blocks(title="十國原廠商品名智能檢索器") as demo:
|
| 576 |
-
gr.Markdown("## 🌐 跨國原廠商品名檢索器 (
|
| 577 |
|
| 578 |
with gr.Row():
|
| 579 |
ing_en = gr.Textbox(label="🧪 英文成分名 (必填)", placeholder="例如: Semaglutide")
|
| 580 |
with gr.Row():
|
| 581 |
with gr.Accordion("⚙️ 手動覆寫翻譯 (進階)", open=False):
|
| 582 |
-
ing_ja = gr.Textbox(label="🇯🇵 日文成分名", placeholder="若空白則自動翻譯")
|
| 583 |
ing_de = gr.Textbox(label="🇩🇪 德文成分名", placeholder="若空白則自動翻譯")
|
| 584 |
|
| 585 |
search_btn = gr.Button("🚀 啟動十國查詢", variant="primary")
|
|
|
|
| 8 |
from curl_cffi import requests as curl_req
|
| 9 |
import gradio as gr
|
| 10 |
|
| 11 |
+
# 💡 匯入 Hugging Face 原生 AI 翻譯所需套件
|
| 12 |
+
import torch
|
| 13 |
+
from transformers import pipeline
|
| 14 |
+
|
| 15 |
os.system("playwright install chromium")
|
| 16 |
|
| 17 |
+
# ==========================================
|
| 18 |
+
# 🧠 載入原生 ElanMT 醫療翻譯模型 (依據您提供的程式碼)
|
| 19 |
+
# ==========================================
|
| 20 |
+
print("⏳ 系統啟動中:正在載入 ElanMT 醫療翻譯模型...")
|
| 21 |
+
device = "cuda" if torch.cuda.is_available() else "cpu"
|
| 22 |
+
try:
|
| 23 |
+
# 根據您的 app.py,直接呼叫 transformers pipeline
|
| 24 |
+
translator_en_ja = pipeline("translation", model="Mitsua/elan-mt-bt-en-ja", device=device)
|
| 25 |
+
print(f"✅ ElanMT 模型載入完成!(執行環境: {device})")
|
| 26 |
+
except Exception as e:
|
| 27 |
+
print(f"⚠️ ElanMT 模型載入失敗,將退回備用機制。錯誤: {e}")
|
| 28 |
+
translator_en_ja = None
|
| 29 |
+
|
| 30 |
+
|
| 31 |
+
# ==========================================
|
| 32 |
+
# 🛠️ 共用工具:四重智能翻譯與學名藥濾網
|
| 33 |
+
# ==========================================
|
| 34 |
def get_official_japanese_name(ingredient_en):
|
| 35 |
+
"""利用維基百科 API 跨語言連結,精準取得官方日文藥名 (JAN)"""
|
|
|
|
|
|
|
| 36 |
try:
|
|
|
|
| 37 |
search_url = f"https://en.wikipedia.org/w/api.php?action=query&list=search&srsearch={quote(ingredient_en)}&utf8=&format=json"
|
| 38 |
res = requests.get(search_url, timeout=5).json()
|
| 39 |
if not res['query']['search']: return None
|
| 40 |
|
|
|
|
| 41 |
en_title = res['query']['search'][0]['title']
|
|
|
|
|
|
|
| 42 |
lang_url = f"https://en.wikipedia.org/w/api.php?action=query&titles={quote(en_title)}&prop=langlinks&lllang=ja&format=json"
|
| 43 |
lang_res = requests.get(lang_url, timeout=5).json()
|
| 44 |
|
| 45 |
pages = lang_res['query']['pages']
|
| 46 |
for page_id in pages:
|
| 47 |
if 'langlinks' in pages[page_id]:
|
|
|
|
| 48 |
ja_title = pages[page_id]['langlinks'][0]['*']
|
|
|
|
| 49 |
return ja_title.split(' ')[0].split('(')[0]
|
| 50 |
except Exception:
|
| 51 |
pass
|
| 52 |
return None
|
| 53 |
|
| 54 |
def translate_lang(text, target_lang):
|
| 55 |
+
"""四重防護智能翻譯分發器"""
|
| 56 |
+
text_lower = text.lower().strip()
|
| 57 |
+
|
|
|
|
| 58 |
if target_lang == 'ja':
|
| 59 |
+
# 1. 字典防護
|
| 60 |
+
ja_overrides = {
|
| 61 |
+
"bilastine": "ビラスチン",
|
| 62 |
+
"semaglutide": "セマグルチド"
|
| 63 |
+
}
|
| 64 |
+
if text_lower in ja_overrides:
|
| 65 |
+
print(f"🌍 [翻譯] 命中自訂字典: {text} -> {ja_overrides[text_lower]}")
|
| 66 |
+
return ja_overrides[text_lower]
|
| 67 |
+
|
| 68 |
+
# 2. 💡 原生 ElanMT 模型防護 (不需對外連線)
|
| 69 |
+
if translator_en_ja is not None:
|
| 70 |
+
try:
|
| 71 |
+
# 呼叫 pipeline 進行推論
|
| 72 |
+
result = translator_en_ja(text)
|
| 73 |
+
translated_text = result[0]['translation_text']
|
| 74 |
+
print(f"🌍 [翻譯] 命中 ElanMT 醫療模型: {text} -> {translated_text}")
|
| 75 |
+
return translated_text
|
| 76 |
+
except Exception as e:
|
| 77 |
+
print(f"⚠️ ElanMT 推論錯誤 ({e}),降級至維基百科...")
|
| 78 |
+
|
| 79 |
+
# 3. 維基百科防護
|
| 80 |
official_ja = get_official_japanese_name(text)
|
| 81 |
if official_ja:
|
| 82 |
+
print(f"🌍 [翻譯] 命中維基百科: {text} -> {official_ja}")
|
| 83 |
+
return official_ja
|
| 84 |
|
| 85 |
+
# 4. Google 翻譯兜底 (德文或前述皆失敗時)
|
| 86 |
try:
|
| 87 |
url = f"https://translate.googleapis.com/translate_a/single?client=gtx&sl=en&tl={target_lang}&dt=t&q={quote(text)}"
|
| 88 |
res = requests.get(url, timeout=5)
|
| 89 |
if res.status_code == 200:
|
| 90 |
+
result = res.json()[0][0][0].strip()
|
| 91 |
+
print(f"🌍 [翻譯] 命中 Google 翻譯: {text} -> {result}")
|
| 92 |
+
return result
|
| 93 |
except Exception:
|
| 94 |
pass
|
|
|
|
| 95 |
return text
|
| 96 |
|
| 97 |
def is_generic(brand_name, company_name, ingredient):
|
|
|
|
| 99 |
c_lower = company_name.lower()
|
| 100 |
i_lower = ingredient.lower()
|
| 101 |
|
|
|
|
| 102 |
generic_keywords = [
|
| 103 |
'sandoz', 'teva', 'apotex', 'ratiopharm', 'jamp', 'mint', 'pharmascience', 'sanis', 'sivem',
|
| 104 |
'auro', 'glenmark', 'taro', 'marcan', 'nora', 'mantra', 'reddy', 'mepha', 'axapharm',
|
|
|
|
| 116 |
return False
|
| 117 |
|
| 118 |
def clean_brand_name(raw_name):
|
| 119 |
+
# 去劑型化:精準擷取商品名主體
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 120 |
pattern = r'(皮下注|錠|カプセル|顆粒|シロップ|OD|細粒|液|Augentropfen|Schmelztabletten|Tabletten|kids|Lingual|Sol|cp|inj|pen|prefilled|flex|\d+)'
|
|
|
|
|
|
|
| 121 |
cleaned = re.split(pattern, raw_name, flags=re.IGNORECASE)[0]
|
|
|
|
|
|
|
| 122 |
return cleaned.replace('®', '').replace('™', '').strip(' -_')
|
| 123 |
|
| 124 |
+
|
| 125 |
# ==========================================
|
| 126 |
+
# 🚀 模組 A:使用 curl_cffi 抓取 (澳)
|
| 127 |
# ==========================================
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 128 |
def get_australia_originator(ingredient):
|
| 129 |
log, brands = [], set()
|
| 130 |
try:
|
| 131 |
session = curl_req.Session(impersonate="chrome120")
|
| 132 |
log.append("1. 發送 GET 至澳洲 TGA 搜尋頁面...")
|
| 133 |
res = session.get(f"https://www.tga.gov.au/resources/artg?keywords={ingredient}", timeout=30, verify=False)
|
|
|
|
| 134 |
soup = BeautifulSoup(res.text, 'html.parser')
|
| 135 |
|
| 136 |
articles = soup.find_all('article', class_='node--artg')
|
|
|
|
| 144 |
if len(parts) > 1 and parts[0].strip():
|
| 145 |
brand = parts[0].strip()
|
| 146 |
if not is_generic(brand, "", ingredient):
|
|
|
|
| 147 |
a_tag = title_tag.find('a')
|
| 148 |
href = a_tag['href'] if a_tag else None
|
| 149 |
cands.append({"brand": brand, "date": time_tag.get('datetime'), "href": href})
|
| 150 |
|
| 151 |
if cands:
|
|
|
|
| 152 |
cands = sorted(cands, key=lambda x: x['date'])
|
| 153 |
target = cands[0]
|
| 154 |
brand = target['brand']
|
| 155 |
log.append(f"✅ 依日期排序鎖定最早註冊: {target['date'][:10]} ({brand})")
|
| 156 |
|
| 157 |
+
company = "TGA資料庫"
|
|
|
|
|
|
|
| 158 |
if target['href']:
|
|
|
|
| 159 |
detail_url = f"https://www.tga.gov.au{target['href']}" if target['href'].startswith('/') else target['href']
|
| 160 |
log.append("3. 進入內頁抓取 Sponsor 藥廠資訊...")
|
|
|
|
| 161 |
res_detail = session.get(detail_url, timeout=30, verify=False)
|
| 162 |
if res_detail.status_code == 200:
|
| 163 |
detail_soup = BeautifulSoup(res_detail.text, 'html.parser')
|
|
|
|
|
|
|
| 164 |
sponsor_div = detail_soup.find('div', class_=re.compile(r'field--name-field-sponsor'))
|
| 165 |
if sponsor_div and sponsor_div.find('a'):
|
| 166 |
company = sponsor_div.find('a').get_text(strip=True)
|
| 167 |
log.append(f" -> 成功取得藥廠名稱: {company}")
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 168 |
return brand, company, "\n".join(log)
|
|
|
|
| 169 |
return "查無原廠", "-", "\n".join(log)
|
| 170 |
except Exception as e:
|
| 171 |
return "執行失敗", "-", "\n".join(log) + f"\n錯誤: {str(e)}"
|
| 172 |
|
| 173 |
+
|
| 174 |
# ==========================================
|
| 175 |
+
# 🚀 模組 B:使用 Playwright 抓取 (美、比、法、英、加、日、德、瑞、瑞典)
|
| 176 |
# ==========================================
|
| 177 |
+
def get_usa_originator(ingredient, page):
|
| 178 |
+
log, brands, companies = [], set(), set()
|
| 179 |
+
try:
|
| 180 |
+
log.append(f"1. 前往 FDA Orange Book (搜尋: {ingredient})...")
|
| 181 |
+
url = f"https://www.accessdata.fda.gov/scripts/cder/ob/results_product.cfm?Generic_Name={quote(ingredient)}&rx_otc=All"
|
| 182 |
+
page.goto(url, timeout=45000, wait_until="domcontentloaded")
|
| 183 |
+
|
| 184 |
+
log.append("2. 等待 DataTables (table#example) 動態渲染...")
|
| 185 |
+
try:
|
| 186 |
+
page.wait_for_selector('table#example, .alert-warning', state="attached", timeout=15000)
|
| 187 |
+
except Exception:
|
| 188 |
+
pass
|
| 189 |
+
|
| 190 |
+
page.wait_for_timeout(2000)
|
| 191 |
+
soup = BeautifulSoup(page.content(), 'html.parser')
|
| 192 |
+
table = soup.find('table', id='example')
|
| 193 |
+
|
| 194 |
+
if table and table.find('tbody'):
|
| 195 |
+
rows = table.find('tbody').find_all('tr')
|
| 196 |
+
log.append(f"3. 找到表格,共 {len(rows)} 列資料。")
|
| 197 |
+
headers = [th.get_text(strip=True).lower() for th in table.find_all('th')]
|
| 198 |
+
brand_idx = next((i for i, h in enumerate(headers) if 'proprietary name' in h), 2)
|
| 199 |
+
rld_idx = next((i for i, h in enumerate(headers) if 'rld' in h), 8)
|
| 200 |
+
mfg_idx = next((i for i, h in enumerate(headers) if 'applicant holder' in h), 10)
|
| 201 |
+
|
| 202 |
+
for tr in rows:
|
| 203 |
+
tds = tr.find_all('td')
|
| 204 |
+
if len(tds) > max(rld_idx, brand_idx):
|
| 205 |
+
rld_text = tds[rld_idx].get_text(strip=True).upper()
|
| 206 |
+
if "RLD" in rld_text or "RS" in rld_text:
|
| 207 |
+
title = tds[brand_idx].get_text(strip=True)
|
| 208 |
+
comp = tds[mfg_idx].get_text(strip=True) if len(tds) > mfg_idx else "-"
|
| 209 |
+
|
| 210 |
+
if not is_generic(title, comp, ingredient):
|
| 211 |
+
brands.add(clean_brand_name(title))
|
| 212 |
+
if comp != "-": companies.add(comp)
|
| 213 |
+
else:
|
| 214 |
+
log.append("❌ 畫面上未出現 table#example,可能查無此藥或遭阻擋。")
|
| 215 |
+
|
| 216 |
+
if brands: return ", ".join(brands), ", ".join(companies), "\n".join(log)
|
| 217 |
+
return "查無原廠", "-", "\n".join(log) + "\n❌ 查無資料或全為學名藥"
|
| 218 |
+
except Exception as e:
|
| 219 |
+
return "執行失敗", "-", "\n".join(log) + f"\n錯誤: {str(e)}"
|
| 220 |
+
|
| 221 |
+
def get_france_originator(ingredient, page):
|
| 222 |
+
log, brands, companies = [], set(), set()
|
| 223 |
+
try:
|
| 224 |
+
log.append(f"1. 前往 Vidal 搜尋頁面 (搜尋: {ingredient})...")
|
| 225 |
+
page.goto(f"https://www.vidal.fr/recherche.html?query={quote(ingredient)}", timeout=45000, wait_until="domcontentloaded")
|
| 226 |
+
page.wait_for_selector('.results, .searchbar', timeout=15000)
|
| 227 |
+
|
| 228 |
+
soup = BeautifulSoup(page.content(), 'html.parser')
|
| 229 |
+
divs = soup.find_all('div', class_=re.compile(r'result drug'))
|
| 230 |
+
log.append(f"3. 找到 {len(divs)} 筆 'result drug' 區塊。")
|
| 231 |
+
|
| 232 |
+
cands = []
|
| 233 |
+
for div in divs:
|
| 234 |
+
info_div = div.find('div', class_='infos')
|
| 235 |
+
if info_div and info_div.find('a'):
|
| 236 |
+
a_tag = info_div.find('a')
|
| 237 |
+
title = a_tag.get_text(strip=True)
|
| 238 |
+
href = a_tag.get('href', '')
|
| 239 |
+
if not is_generic(title, "", ingredient):
|
| 240 |
+
cands.append({"raw_title": title, "href": href})
|
| 241 |
+
|
| 242 |
+
if cands:
|
| 243 |
+
log.append(f"4. 篩選出 {len(cands)} 筆非學名藥候選,準備進入內頁抓取藥廠...")
|
| 244 |
+
for cand in cands:
|
| 245 |
+
title = cand['raw_title']
|
| 246 |
+
href = cand['href']
|
| 247 |
+
detail_url = href if href.startswith('http') else (f"https://www.vidal.fr{href}" if href.startswith('/') else f"https://www.vidal.fr/{href}")
|
| 248 |
+
|
| 249 |
+
log.append(f" -> 進入內頁: {clean_brand_name(title)}")
|
| 250 |
+
page.goto(detail_url, timeout=30000, wait_until="domcontentloaded")
|
| 251 |
+
|
| 252 |
+
comp = "-"
|
| 253 |
+
try:
|
| 254 |
+
page.wait_for_selector('div.nomlab', timeout=10000)
|
| 255 |
+
detail_soup = BeautifulSoup(page.content(), 'html.parser')
|
| 256 |
+
nomlab_div = detail_soup.find('div', class_='nomlab')
|
| 257 |
+
if nomlab_div:
|
| 258 |
+
comp = nomlab_div.get_text(strip=True)
|
| 259 |
+
log.append(f" ✅ 成功取得藥廠: {comp}")
|
| 260 |
+
except Exception:
|
| 261 |
+
pass
|
| 262 |
+
|
| 263 |
+
if not is_generic(title, comp, ingredient):
|
| 264 |
+
brands.add(clean_brand_name(title))
|
| 265 |
+
if comp != "-": companies.add(comp)
|
| 266 |
+
|
| 267 |
+
if brands: return ", ".join(brands), ", ".join(companies), "\n".join(log)
|
| 268 |
+
return "查無原廠", "-", "\n".join(log) + "\n❌ 查無非學名藥資料"
|
| 269 |
+
except Exception as e:
|
| 270 |
+
return "執行失敗", "-", "\n".join(log) + f"\n錯誤: {str(e)}"
|
| 271 |
+
|
| 272 |
+
def get_sweden_originator(ingredient, page):
|
| 273 |
+
log, brands, companies = [], set(), set()
|
| 274 |
+
try:
|
| 275 |
+
log.append(f"1. 前往瑞典 FASS 搜尋頁面 (搜尋: {ingredient})...")
|
| 276 |
+
page.goto(f"https://fass.se/search?query={quote(ingredient)}", timeout=45000, wait_until="domcontentloaded")
|
| 277 |
+
|
| 278 |
+
try:
|
| 279 |
+
page.wait_for_selector('details.app-toggle-details-icon, .no-results', state="attached", timeout=15000)
|
| 280 |
+
except Exception:
|
| 281 |
+
pass
|
| 282 |
+
|
| 283 |
+
page.wait_for_timeout(2000)
|
| 284 |
+
soup = BeautifulSoup(page.content(), 'html.parser')
|
| 285 |
+
items = soup.find_all('details', class_=re.compile(r'app-toggle-details-icon'))
|
| 286 |
+
log.append(f"3. 找到 {len(items)} 筆搜尋結果。")
|
| 287 |
+
|
| 288 |
+
for item in items:
|
| 289 |
+
summary = item.find('summary')
|
| 290 |
+
if not summary: continue
|
| 291 |
+
|
| 292 |
+
title_span = summary.find('span', class_=re.compile(r'font-semibold'))
|
| 293 |
+
title = title_span.get_text(strip=True) if title_span else ""
|
| 294 |
+
|
| 295 |
+
comp = "-"
|
| 296 |
+
ol = item.find('ol')
|
| 297 |
+
if ol:
|
| 298 |
+
comp_span = ol.find('span', class_='text-label-md')
|
| 299 |
+
if comp_span:
|
| 300 |
+
comp = comp_span.get_text(strip=True)
|
| 301 |
+
|
| 302 |
+
if title:
|
| 303 |
+
if not is_generic(title, comp, ingredient):
|
| 304 |
+
brands.add(clean_brand_name(title))
|
| 305 |
+
if comp != "-": companies.add(comp)
|
| 306 |
+
|
| 307 |
+
if brands: return ", ".join(brands), ", ".join(companies), "\n".join(log)
|
| 308 |
+
return "查無原廠", "-", "\n".join(log) + "\n❌ 查無資料或全為學名藥"
|
| 309 |
+
except Exception as e:
|
| 310 |
+
return "執行失敗", "-", "\n".join(log) + f"\n錯誤: {str(e)}"
|
| 311 |
+
|
| 312 |
def get_belgium_originator(ingredient, page):
|
| 313 |
log, brands = [], set()
|
| 314 |
try:
|
| 315 |
log.append(f"1. 前往 CBIP (比利時) 搜尋頁面 (搜尋: {ingredient})...")
|
| 316 |
page.goto(f"https://www.cbip.be/fr/search?q={ingredient}", timeout=45000, wait_until="domcontentloaded")
|
|
|
|
|
|
|
| 317 |
page.wait_for_selector('ul.search-results, .no-results', timeout=15000)
|
| 318 |
page.wait_for_timeout(2000)
|
| 319 |
|
| 320 |
soup = BeautifulSoup(page.content(), 'html.parser')
|
|
|
|
|
|
|
|
|
|
| 321 |
results_list = soup.find('ul', class_=re.compile(r'search-results|results'))
|
| 322 |
|
| 323 |
if results_list:
|
| 324 |
links = results_list.find_all('a')
|
| 325 |
for link in links:
|
| 326 |
text = link.get_text(strip=True)
|
|
|
|
|
|
|
| 327 |
if len(text) > 3 and not text.lower().startswith('voir'):
|
|
|
|
| 328 |
if not is_generic(text, "", ingredient):
|
| 329 |
brands.add(clean_brand_name(text))
|
|
|
|
|
|
|
|
|
|
| 330 |
if brands: return ", ".join(brands), "CBIP 資料庫", "\n".join(log)
|
| 331 |
return "查無原廠", "-", "\n".join(log) + "\n❌ 查無資料或全為學名藥"
|
| 332 |
except Exception as e:
|
|
|
|
| 340 |
page.wait_for_selector('.search-results-product-info-title-link', timeout=15000)
|
| 341 |
soup = BeautifulSoup(page.content(), 'html.parser')
|
| 342 |
links = soup.find_all('a', class_='search-results-product-info-title-link')
|
|
|
|
| 343 |
for link in links:
|
| 344 |
title = link.get_text(strip=True)
|
| 345 |
if not title.lower().startswith(ingredient.lower()):
|
|
|
|
| 358 |
page.goto("https://health-products.canada.ca/dpd-bdpp/index-eng.jsp", timeout=45000, wait_until="domcontentloaded")
|
| 359 |
page.locator('input[id="activeIngredient"]').fill(ingredient)
|
| 360 |
page.keyboard.press("Enter")
|
|
|
|
|
|
|
| 361 |
page.wait_for_selector('table#results, .alert-info, .alert-warning', timeout=15000)
|
| 362 |
soup = BeautifulSoup(page.content(), 'html.parser')
|
| 363 |
table = soup.find('table', id='results')
|
| 364 |
|
| 365 |
+
if not table or not table.find('tbody'): return "查無資料", "-", "\n".join(log)
|
|
|
|
|
|
|
|
|
|
| 366 |
rows = table.find('tbody').find_all('tr')
|
|
|
|
| 367 |
all_cands = []
|
| 368 |
for tr in rows:
|
| 369 |
tds = tr.find_all('td')
|
|
|
|
| 375 |
all_cands.append({"company": comp, "product": tds[3].get_text(strip=True), "din": int(m.group())})
|
| 376 |
|
| 377 |
if not all_cands: return "查無原廠", "-", "\n".join(log) + "\n❌ 剩餘皆為學名藥廠"
|
|
|
|
| 378 |
all_cands = sorted(all_cands, key=lambda x: x['din'])
|
| 379 |
orig_comp = all_cands[0]['company']
|
| 380 |
brands = set([c['product'] for c in all_cands if c['company'] == orig_comp])
|
| 381 |
+
return ", ".join(clean_brand_name(b) for b in brands), orig_comp, "\n".join(log)
|
|
|
|
| 382 |
except Exception as e: return "執行失敗", "-", "\n".join(log) + f"\n錯誤: {str(e)}"
|
| 383 |
|
| 384 |
def get_japan_originator(ing_ja, page):
|
|
|
|
| 386 |
try:
|
| 387 |
log.append(f"1. 前往 PMDA (搜尋: {ing_ja})...")
|
| 388 |
page.goto("https://www.pmda.go.jp/PmdaSearch/iyakuSearch/", timeout=45000, wait_until="domcontentloaded")
|
|
|
|
|
|
|
|
|
|
| 389 |
search_input = page.locator('input#txtName')
|
| 390 |
search_input.wait_for(state="attached", timeout=15000)
|
| 391 |
search_input.fill(ing_ja, force=True)
|
| 392 |
|
|
|
|
| 393 |
with page.expect_popup() as popup_info:
|
| 394 |
search_input.press("Enter")
|
| 395 |
popup = popup_info.value
|
|
|
|
|
|
|
| 396 |
current_page = 1
|
| 397 |
|
| 398 |
while True:
|
|
|
|
| 399 |
popup.wait_for_selector('table#ResultList, .errormsg, .non-result', timeout=15000)
|
| 400 |
soup = BeautifulSoup(popup.content(), 'html.parser')
|
| 401 |
table = soup.find('table', id='ResultList')
|
| 402 |
|
| 403 |
if table:
|
| 404 |
rows = table.find_all('tr')
|
|
|
|
| 405 |
for tr in rows:
|
| 406 |
tds = tr.find_all('td')
|
| 407 |
if len(tds) >= 3:
|
| 408 |
title = tds[1].get_text(strip=True)
|
|
|
|
| 409 |
if not is_generic(title, "", ing_ja):
|
| 410 |
brands.add(clean_brand_name(title))
|
| 411 |
companies.add(tds[2].get_text(separator=" ", strip=True).replace('製造販売元/', ''))
|
| 412 |
+
else: break
|
|
|
|
|
|
|
| 413 |
|
|
|
|
| 414 |
current_page += 1
|
| 415 |
next_page_link = popup.locator(f'a[href="javascript:changePg({current_page});"]')
|
|
|
|
| 416 |
if next_page_link.count() > 0:
|
|
|
|
| 417 |
next_page_link.first.click(force=True)
|
| 418 |
+
popup.wait_for_timeout(2000)
|
| 419 |
+
else: break
|
|
|
|
|
|
|
| 420 |
|
| 421 |
popup.close()
|
|
|
|
| 422 |
if brands: return ", ".join(brands), ", ".join(companies), "\n".join(log)
|
| 423 |
+
return "查無原廠", "-", "\n".join(log)
|
| 424 |
+
except Exception as e: return "執行失敗", "-", "\n".join(log) + f"\n錯誤: {str(e)}"
|
|
|
|
| 425 |
|
| 426 |
def get_switzerland_originator(ing_de, page):
|
| 427 |
log, brands, companies = [], set(), set()
|
| 428 |
try:
|
| 429 |
log.append(f"1. 前往 Swissmedicinfo Pro 專業版 (搜尋: {ing_de})...")
|
| 430 |
page.goto("https://swissmedicinfo-pro.ch/?Lang=EN", timeout=45000, wait_until="domcontentloaded")
|
|
|
|
|
|
|
|
|
|
| 431 |
search_input = page.locator('input#MainContent_ucSearch1_txtSubstance')
|
| 432 |
search_input.wait_for(state="attached", timeout=15000)
|
| 433 |
search_input.fill(ing_de, force=True)
|
|
|
|
|
|
|
| 434 |
search_input.press("Enter")
|
| 435 |
|
| 436 |
page.wait_for_load_state('networkidle', timeout=15000)
|
| 437 |
page.wait_for_timeout(3000)
|
|
|
|
| 438 |
page.wait_for_selector('table[id*="GVMonographies"], #MainContent_LabelNoResult', timeout=15000)
|
| 439 |
|
| 440 |
soup = BeautifulSoup(page.content(), 'html.parser')
|
|
|
|
| 442 |
|
| 443 |
if table and table.find('tbody'):
|
| 444 |
rows = table.find('tbody').find_all('tr', class_=re.compile(r'clickable-row'))
|
|
|
|
|
|
|
| 445 |
for tr in rows:
|
| 446 |
tds = tr.find_all('td')
|
| 447 |
if len(tds) >= 4:
|
| 448 |
title = tds[0].get_text(strip=True)
|
| 449 |
comp = tds[3].get_text(strip=True)
|
|
|
|
| 450 |
if not is_generic(title, comp, ing_de):
|
| 451 |
brands.add(clean_brand_name(title))
|
| 452 |
if comp != "-": companies.add(comp)
|
|
|
|
|
|
|
|
|
|
| 453 |
if brands: return ", ".join(brands), ", ".join(companies), "\n".join(log)
|
| 454 |
+
return "查無原廠", "-", "\n".join(log)
|
| 455 |
+
except Exception as e: return "執行失敗", "-", "\n".join(log) + f"\n錯誤: {str(e)}"
|
|
|
|
| 456 |
|
| 457 |
def get_germany_originator(ing_de, page):
|
| 458 |
log, brands, companies = [], set(), set()
|
| 459 |
try:
|
|
|
|
| 460 |
page.goto(f"https://www.gelbe-liste.de/profi-suche/results?substance={quote(ing_de)}", timeout=30000)
|
| 461 |
page.wait_for_selector('.product-list', timeout=15000)
|
|
|
|
| 462 |
soup = BeautifulSoup(page.content(), 'html.parser')
|
| 463 |
ul = soup.find('ul', class_='product-list')
|
| 464 |
if ul:
|
| 465 |
lis = ul.find_all('li')
|
|
|
|
| 466 |
for li in lis:
|
| 467 |
h5, p_tag = li.find('h5'), li.find('p', class_='small')
|
| 468 |
if h5:
|
|
|
|
| 471 |
brands.add(clean_brand_name(title))
|
| 472 |
companies.add(comp)
|
| 473 |
if brands: return ", ".join(brands), ", ".join(companies), "\n".join(log)
|
| 474 |
+
return "查無原廠", "-", "\n".join(log)
|
| 475 |
except Exception as e: return "執行失敗", "-", "\n".join(log) + f"\n錯誤: {str(e)}"
|
| 476 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 477 |
# ==========================================
|
| 478 |
# 🚀 主執行中樞
|
| 479 |
# ==========================================
|
| 480 |
def run_all_ten_countries(ing_en, ing_ja_manual, ing_de_manual):
|
| 481 |
if not ing_en: return [["錯誤", "請輸入英文成分名", "-", ""]]
|
| 482 |
|
| 483 |
+
# 呼叫四重智能翻譯 (現在會自動優先使用您的 ElanMT 模型)
|
| 484 |
ing_ja = ing_ja_manual if ing_ja_manual else translate_lang(ing_en, 'ja')
|
| 485 |
ing_de = ing_de_manual if ing_de_manual else translate_lang(ing_en, 'de')
|
| 486 |
results = []
|
| 487 |
|
| 488 |
+
# API 模組 (澳洲)
|
|
|
|
|
|
|
|
|
|
| 489 |
au_b, au_c, au_log = get_australia_originator(ing_en)
|
| 490 |
results.append(["🇦🇺 澳洲 (TGA)", au_b, au_c, au_log])
|
| 491 |
|
| 492 |
+
# 瀏覽器模組 (Playwright 嚴格隔離)
|
| 493 |
with sync_playwright() as p:
|
| 494 |
browser = p.chromium.launch(headless=True, args=['--no-sandbox', '--disable-dev-shm-usage'])
|
| 495 |
context = browser.new_context(user_agent="Mozilla/5.0 (Windows NT 10.0; Win64; x64) Chrome/122.0.0.0 Safari/537.36")
|
|
|
|
| 500 |
page.close()
|
| 501 |
results.append([name, b, c, log])
|
| 502 |
|
| 503 |
+
run_pw(get_usa_originator, ing_en, "🇺🇸 美國 (FDA)")
|
| 504 |
run_pw(get_belgium_originator, ing_en, "🇧🇪 比利時 (CBIP)")
|
| 505 |
+
run_pw(get_france_originator, ing_en, "🇫🇷 法國 (Vidal)")
|
| 506 |
run_pw(get_uk_originator, ing_en, "🇬🇧 英國 (eMC)")
|
| 507 |
run_pw(get_canada_originator, ing_en, "🇨🇦 加拿大 (DPD)")
|
| 508 |
run_pw(get_japan_originator, ing_ja, "🇯🇵 日本 (PMDA)")
|
| 509 |
run_pw(get_switzerland_originator, ing_de, "🇨🇭 瑞士 (Swissmedicinfo)")
|
|
|
|
| 510 |
run_pw(get_sweden_originator, ing_en, "🇸🇪 瑞典 (FASS)")
|
| 511 |
+
run_pw(get_germany_originator, ing_de, "🇩🇪 德國 (Gelbe Liste)")
|
| 512 |
|
| 513 |
browser.close()
|
| 514 |
|
|
|
|
| 518 |
# 🎨 UI 介面
|
| 519 |
# ==========================================
|
| 520 |
with gr.Blocks(title="十國原廠商品名智能檢索器") as demo:
|
| 521 |
+
gr.Markdown("## 🌐 跨國原廠商品名檢索器 (搭載原生 ElanMT 醫療翻譯)")
|
| 522 |
|
| 523 |
with gr.Row():
|
| 524 |
ing_en = gr.Textbox(label="🧪 英文成分名 (必填)", placeholder="例如: Semaglutide")
|
| 525 |
with gr.Row():
|
| 526 |
with gr.Accordion("⚙️ 手動覆寫翻譯 (進階)", open=False):
|
| 527 |
+
ing_ja = gr.Textbox(label="🇯🇵 日文成分名", placeholder="若空白則自動啟動 ElanMT 翻譯")
|
| 528 |
ing_de = gr.Textbox(label="🇩🇪 德文成分名", placeholder="若空白則自動翻譯")
|
| 529 |
|
| 530 |
search_btn = gr.Button("🚀 啟動十國查詢", variant="primary")
|