feat(pre-flight): канон-справочник + golden-нормализация, дизайн Nomad, питч

- contracts/service_canon.json — 15 канонов медуслуг RU/KZ/EN + синонимы
- fixtures/normalized/golden.json — 52 эталонные строки (оракул TDD)
- design/ — tailwind-токены + shadcn-тема Nomad
- pitch/ — структура слайдов + демо-сценарий

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
This commit is contained in:
2026-06-26 13:03:08 +05:00
parent fa2fbc59b0
commit 062ff5fc11
15 changed files with 1489 additions and 138 deletions
+137 -122
View File
@@ -1,149 +1,164 @@
#!/usr/bin/env python3
# -*- coding: utf-8 -*-
"""Генератор синтетических прайсов клиник (RU/KZ) в разных форматах для отладки pipeline."""
import csv, os, random, subprocess, sys
"""Генератор синтетических прайсов клиник (RU/KZ/EN) + эталонная нормализация + справочник канонов.
OUT = sys.argv[1] if len(sys.argv) > 1 else "fixtures/raw"
os.makedirs(OUT, exist_ok=True)
random.seed(42) # детерминированно
Выдаёт:
fixtures/raw/* — 6 форматов «сырых» прайсов (Excel/CSV/txt/PNG/PDF)
fixtures/normalized/golden.json — эталон: каждая строка → канон (оракул для TDD ETL)
contracts/service_canon.json — справочник канонических услуг (seed для нормализации)
Детерминированно (seed=42).
"""
import csv, json, os, random, subprocess, sys
ROOT = sys.argv[1] if len(sys.argv) > 1 else "."
RAW = os.path.join(ROOT, "fixtures/raw")
NORM = os.path.join(ROOT, "fixtures/normalized")
CONTRACTS = os.path.join(ROOT, "contracts")
for d in (RAW, NORM, CONTRACTS):
os.makedirs(d, exist_ok=True)
random.seed(42)
def ensure(pkg, mod=None):
try:
__import__(mod or pkg)
except ImportError:
subprocess.check_call([sys.executable, "-m", "pip", "install", "-q", pkg])
try: __import__(mod or pkg)
except ImportError: subprocess.check_call([sys.executable, "-m", "pip", "install", "-q", pkg])
# Базовый каталог: (канон RU, канон KZ, [варианты названий], мин, макс, единица)
CATALOG = [
("Общий анализ крови", "Қанның жалпы анализі",
["Общий анализ крови", "ОАК", "Общий анализ крови (развернутый)", "Клинический анализ крови", "Қанның жалпы анализі"],
1200, 4500, "1 иссл."),
("УЗИ органов брюшной полости", "Құрсақ қуысы ағзаларының УДЗ",
["УЗИ органов брюшной полости", "УЗИ ОБП", "УЗИ брюшной полости", "Құрсақ қуысы УДЗ"],
6000, 13000, "1 иссл."),
("Приём врача-терапевта", "Терапевт дәрігердің қабылдауы",
["Прием терапевта", "Консультация терапевта первичная", "Приём врача-терапевта", "Терапевт қабылдауы"],
4000, 9000, "1 приём"),
("Электрокардиография (ЭКГ)", "Электрокардиография (ЭКГ)",
["ЭКГ", "Электрокардиография", "ЭКГ с расшифровкой", "ЭКГ + описание"],
2000, 5000, "1 иссл."),
("Глюкоза крови", "Қандағы глюкоза",
["Глюкоза", "Глюкоза крови", "Сахар крови", "Қандағы глюкоза"],
800, 2200, "1 иссл."),
("Консультация кардиолога", "Кардиолог кеңесі",
["Консультация кардиолога", "Прием кардиолога", "Кардиолог консультация первичная"],
5000, 12000, "1 приём"),
("МРТ головного мозга", "Бас миының МРТ",
["МРТ головного мозга", "Магнитно-резонансная томография головного мозга", "МРТ г", "Бас миы МРТ"],
18000, 35000, "1 иссл."),
("Биохимический анализ крови", "Қанның биохимиялық анализі",
["Биохимия крови", "Биохимический анализ крови (базовый)", "Б/х крови", "Қан биохимиясы"],
3500, 9000, "1 иссл."),
("Удаление зуба", "Тісті жұлу",
["Удаление зуба", "Удаление зуба простое", "Экстракция зуба", "Тіс жұлу"],
6000, 15000, "1 зуб"),
("Рентген грудной клетки", "Кеуде клеткасының рентгені",
["Рентген грудной клетки", "Рентгенография ОГК", "Флюорография", "Кеуде рентгені"],
2500, 6000, "1 иссл."),
# Справочник канонических медуслуг. lang-тегированные синонимы — материал для нормализации.
CANONS = [
{"id":"cbc","ru":"Общий анализ крови","kz":"Қанның жалпы анализі","en":"Complete blood count (CBC)",
"category":"Лаборатория","code":"B03.016.003","unit":"1 иссл.","price":(1200,4500),
"syn":[("ОАК","ru"),("Общий анализ крови (развернутый)","ru"),("Клинический анализ крови","ru"),("КАК","ru"),("CBC","en"),("Қанның жалпы талдауы","kz")]},
{"id":"uzi_abd","ru":"УЗИ органов брюшной полости","kz":"Құрсақ қуысы ағзаларының УДЗ","en":"Abdominal ultrasound",
"category":"Диагностика/УЗИ","code":"A04.16.001","unit":"1 иссл.","price":(6000,13000),
"syn":[("УЗИ ОБП","ru"),("УЗИ брюшной полости","ru"),("Ультразвуковое исследование брюшной полости","ru"),("Құрсақ қуысы УДЗ","kz"),("Abdominal US","en")]},
{"id":"therapist","ru":"Приём врача-терапевта","kz":"Терапевт дәрігердің қабылдауы","en":"General practitioner consultation",
"category":"Приём специалистов","code":"B01.047.001","unit":"1 приём","price":(4000,9000),
"syn":[("Прием терапевта","ru"),("Консультация терапевта первичная","ru"),("Терапевт қабылдауы","kz"),("GP visit","en")]},
{"id":"ecg","ru":"Электрокардиография (ЭКГ)","kz":"Электрокардиография (ЭКГ)","en":"Electrocardiography (ECG)",
"category":"Диагностика","code":"A05.10.006","unit":"1 иссл.","price":(2000,5000),
"syn":[("ЭКГ","ru"),("ЭКГ с расшифровкой","ru"),("Электрокардиография","ru"),("ECG","en"),("ЭКГ + описание","ru")]},
{"id":"glucose","ru":"Глюкоза крови","kz":"Қандағы глюкоза","en":"Blood glucose",
"category":"Лаборатория","code":"B03.016.006","unit":"1 иссл.","price":(800,2200),
"syn":[("Глюкоза","ru"),("Сахар крови","ru"),("Қандағы глюкоза","kz"),("Glucose","en")]},
{"id":"cardiologist","ru":"Консультация кардиолога","kz":"Кардиолог кеңесі","en":"Cardiologist consultation",
"category":"Приём специалистов","code":"B01.015.001","unit":"1 приём","price":(5000,12000),
"syn":[("Прием кардиолога","ru"),("Кардиолог консультация первичная","ru"),("Кардиолог қабылдауы","kz")]},
{"id":"mri_brain","ru":"МРТ головного мозга","kz":"Бас миының МРТ","en":"Brain MRI",
"category":"Диагностика/МРТ","code":"A05.23.009","unit":"1 иссл.","price":(18000,35000),
"syn":[("МРТ г","ru"),("Магнитно-резонансная томография головного мозга","ru"),("Бас миы МРТ","kz"),("Brain MRI","en")]},
{"id":"biochem","ru":"Биохимический анализ крови","kz":"Қанның биохимиялық анализі","en":"Blood chemistry panel",
"category":"Лаборатория","code":"B03.016.004","unit":"1 иссл.","price":(3500,9000),
"syn":[("Биохимия крови","ru"),("Б/х крови","ru"),("Биохимический анализ крови (базовый)","ru"),("Қан биохимиясы","kz")]},
{"id":"tooth_ext","ru":"Удаление зуба","kz":"Тісті жұлу","en":"Tooth extraction",
"category":"Стоматология","code":"A16.07.001","unit":"1 зуб","price":(6000,15000),
"syn":[("Удаление зуба простое","ru"),("Экстракция зуба","ru"),("Тіс жұлу","kz"),("Tooth removal","en")]},
{"id":"xray_chest","ru":"Рентген грудной клетки","kz":"Кеуде клеткасының рентгені","en":"Chest X-ray",
"category":"Диагностика","code":"A06.09.007","unit":"1 иссл.","price":(2500,6000),
"syn":[("Рентгенография ОГК","ru"),("Флюорография","ru"),("Кеуде рентгені","kz"),("Chest X-ray","en")]},
{"id":"tsh","ru":"Тиреотропный гормон (ТТГ)","kz":"Тиреотропты гормон (ТТГ)","en":"Thyroid-stimulating hormone (TSH)",
"category":"Лаборатория","code":"B03.016.021","unit":"1 иссл.","price":(1500,4000),
"syn":[("ТТГ","ru"),("Тиреотропный гормон","ru"),("TSH","en"),("ТТГ гормоны","kz")]},
{"id":"urinalysis","ru":"Общий анализ мочи","kz":"Зәрдің жалпы анализі","en":"Urinalysis",
"category":"Лаборатория","code":"B03.016.010","unit":"1 иссл.","price":(900,2500),
"syn":[("ОАМ","ru"),("Общий анализ мочи","ru"),("Зәр анализі","kz"),("Urine test","en")]},
{"id":"gyn","ru":"Приём гинеколога","kz":"Гинеколог қабылдауы","en":"Gynecologist consultation",
"category":"Приём специалистов","code":"B01.001.001","unit":"1 приём","price":(5000,11000),
"syn":[("Прием гинеколога","ru"),("Консультация гинеколога первичная","ru"),("Гинеколог кеңесі","kz")]},
{"id":"covid_pcr","ru":"ПЦР на COVID-19","kz":"COVID-19 ПТР","en":"COVID-19 PCR test",
"category":"Лаборатория","code":"A26.08.027","unit":"1 иссл.","price":(7000,16000),
"syn":[("ПЦР COVID-19","ru"),("ПЦР-тест на коронавирус","ru"),("COVID PCR","en"),("COVID ПТР","kz")]},
{"id":"vit_d","ru":"Витамин D (25-OH)","kz":"D дәрумені (25-OH)","en":"Vitamin D (25-OH)",
"category":"Лаборатория","code":"B03.016.115","unit":"1 иссл.","price":(6000,12000),
"syn":[("25-OH витамин D","ru"),("Витамин Д","ru"),("Vitamin D","en"),("D витамині","kz")]},
]
CLINICS = ["Клиника «Аль-Шифа»","Медцентр «Сункар»","Diagnostic Lab Astana","Клиника «Береке»","Медцентр «Тенгри»"]
CLINICS = ["Клиника «Аль-Шифа»", "Медцентр «Сункар»", "Diagnostic Lab Astana", "Клиника «Береке»", "Медцентр «Тенгри»"]
def price(c): return random.randint(c["price"][0]//100, c["price"][1]//100) * 100
def variant(c): return random.choice([(c["ru"],"ru")] + c["syn"])
def price(lo, hi):
return round(random.randint(lo, hi), -2) # круглые сотни
def variants_for(clinic_idx, n):
"""Выбрать n услуг с разными вариантами написания и ценами."""
GOLDEN = []
def build(file_name, clinic, n):
rows = []
chosen = random.sample(CATALOG, k=min(n, len(CATALOG)))
for ru, kz, names, lo, hi, unit in chosen:
name = random.choice(names)
rows.append((name, price(lo, hi), unit))
for c in random.sample(CANONS, k=min(n, len(CANONS))):
raw_name, lang = variant(c); p = price(c)
rows.append({"raw_name":raw_name,"price":p,"unit":c["unit"]})
GOLDEN.append({"source_file":file_name,"clinic":clinic,"raw_name":raw_name,"lang":lang,
"canon_id":c["id"],"canonical_ru":c["ru"],"canonical_kz":c["kz"],"canonical_en":c["en"],
"category":c["category"],"code":c["code"],"price":p,"currency":"KZT","unit":c["unit"]})
return rows
ensure("openpyxl")
from openpyxl import Workbook
ensure("openpyxl"); from openpyxl import Workbook
# 1) Чистый Excel
wb = Workbook(); ws = wb.active; ws.title = "Прайс"
ws.append(["Наименование услуги", "Цена, тг"])
for name, p, _ in variants_for(0, 9):
ws.append([name, p])
wb.save(f"{OUT}/clinic_alpha_clean.xlsx")
# 1) чистый Excel
rows = build("clinic_alpha_clean.xlsx", CLINICS[0], 9)
wb=Workbook(); ws=wb.active; ws.title="Прайс"; ws.append(["Наименование услуги","Цена, тг"])
for r in rows: ws.append([r["raw_name"], r["price"]])
wb.save(f"{RAW}/clinic_alpha_clean.xlsx")
# 2) Грязный Excel: секции-заголовки, "от", разные колонки, KZ
wb = Workbook(); ws = wb.active; ws.title = "PRICE 2026"
ws.append(["Прайс-лист медцентра «Сункар» (с 01.06.2026)"])
ws.append([])
ws.append(["", "Услуга / Қызмет", "Стоимость (тенге)", "Ед."])
section = ["— Лаборатория / Зертхана —", "— Диагностика —", "— Приём специалистов —"]
i = 1
for s in section:
ws.append([s, None, None, None])
for name, p, unit in variants_for(1, 3):
val = f"от {p}" if random.random() < 0.3 else p
ws.append([i, name, val, unit]); i += 1
wb.save(f"{OUT}/clinic_beta_dirty.xlsx")
# 2) грязный Excel
rows = build("clinic_beta_dirty.xlsx", CLINICS[1], 9)
wb=Workbook(); ws=wb.active; ws.title="PRICE 2026"
ws.append(["Прайс-лист медцентра «Сункар» (с 01.06.2026)"]); ws.append([])
ws.append(["","Услуга / Қызмет","Стоимость (тенге)","Ед."])
secs=["— Лаборатория / Зертхана —","— Диагностика —","— Приём специалистов —"]; i=1
for k,r in enumerate(rows):
if k%3==0 and secs: ws.append([secs.pop(0),None,None,None])
val=f"от {r['price']}" if random.random()<0.3 else r["price"]
ws.append([i, r["raw_name"], val, r["unit"]]); i+=1
wb.save(f"{RAW}/clinic_beta_dirty.xlsx")
# 3) CSV с разделителем ; и смешанными языками
with open(f"{OUT}/clinic_gamma.csv", "w", encoding="utf-8", newline="") as f:
w = csv.writer(f, delimiter=";")
w.writerow(["service", "price_kzt", "unit"])
for name, p, unit in variants_for(2, 8):
w.writerow([name, p, unit])
# 3) CSV ;
rows = build("clinic_gamma.csv", CLINICS[2], 8)
with open(f"{RAW}/clinic_gamma.csv","w",encoding="utf-8",newline="") as f:
w=csv.writer(f,delimiter=";"); w.writerow(["service","price_kzt","unit"])
for r in rows: w.writerow([r["raw_name"], r["price"], r["unit"]])
# 4) Грязный текст (прайс абзацем)
with open(f"{OUT}/clinic_delta_messy.txt", "w", encoding="utf-8") as f:
# 4) грязный текст
rows = build("clinic_delta_messy.txt", CLINICS[4], 7)
with open(f"{RAW}/clinic_delta_messy.txt","w",encoding="utf-8") as f:
f.write("МЕДЦЕНТР «ТЕНГРИ» — действующие цены\n\n")
for name, p, unit in variants_for(3, 7):
sep = random.choice(["", " ... ", ": ", " "])
cur = random.choice(["тг", "тенге", "", "KZT"])
f.write(f"{name}{sep}{p} {cur} ({unit})\n")
for r in rows:
sep=random.choice([""," ... ",": "," "]); cur=random.choice(["тг","тенге","","KZT"])
f.write(f"{r['raw_name']}{sep}{r['price']} {cur} ({r['unit']})\n")
f.write("\n* Цены справочные, уточняйте в регистратуре. Бағалар анықтамалық.\n")
# 5) Картинка-«скан» прайса (имитация фото)
ensure("Pillow", "PIL")
from PIL import Image, ImageDraw, ImageFont
img = Image.new("RGB", (900, 700), (250, 248, 244))
d = ImageDraw.Draw(img)
# 5) PNG-скан
rows = build("clinic_epsilon_scan.png", CLINICS[3], 9)
ensure("Pillow","PIL"); from PIL import Image, ImageDraw, ImageFont
def font(sz):
for p in ["/System/Library/Fonts/Supplemental/Arial.ttf", "/Library/Fonts/Arial.ttf",
"/usr/share/fonts/truetype/dejavu/DejaVuSans.ttf"]:
if os.path.exists(p):
return ImageFont.truetype(p, sz)
for p in ["/System/Library/Fonts/Supplemental/Arial.ttf","/Library/Fonts/Arial.ttf","/usr/share/fonts/truetype/dejavu/DejaVuSans.ttf"]:
if os.path.exists(p): return ImageFont.truetype(p,sz)
return ImageFont.load_default()
d.text((40, 30), "Клиника «Береке» — прайс-лист", fill=(20, 20, 20), font=font(34))
d.line((40, 85, 860, 85), fill=(180, 180, 180), width=2)
y = 110
for name, p, unit in variants_for(4, 9):
d.text((50, y), name, fill=(30, 30, 30), font=font(24))
d.text((720, y), f"{p} тг", fill=(30, 30, 30), font=font(24))
y += 58
img = img.rotate(-1.4, expand=False, fillcolor=(250, 248, 244)) # лёгкий наклон как у фото
img.save(f"{OUT}/clinic_epsilon_scan.png")
img=Image.new("RGB",(900,720),(250,248,244)); d=ImageDraw.Draw(img)
d.text((40,30),"Клиника «Береке» — прайс-лист",fill=(20,20,20),font=font(34)); d.line((40,85,860,85),fill=(180,180,180),width=2)
y=110
for r in rows:
d.text((50,y),r["raw_name"][:42],fill=(30,30,30),font=font(23)); d.text((720,y),f"{r['price']} тг",fill=(30,30,30),font=font(23)); y+=62
img=img.rotate(-1.4,expand=False,fillcolor=(250,248,244)); img.save(f"{RAW}/clinic_epsilon_scan.png")
# 6) PDF (если есть reportlab)
# 6) PDF
rows = build("clinic_zeta_prices.pdf", CLINICS[2], 10)
pdf="skip"
try:
ensure("reportlab")
from reportlab.lib.pagesizes import A4
from reportlab.pdfgen import canvas
from reportlab.pdfbase import pdfmetrics
from reportlab.pdfbase.ttfonts import TTFont
fp = next((p for p in ["/Library/Fonts/Arial.ttf", "/System/Library/Fonts/Supplemental/Arial.ttf",
"/usr/share/fonts/truetype/dejavu/DejaVuSans.ttf"] if os.path.exists(p)), None)
from reportlab.lib.pagesizes import A4; from reportlab.pdfgen import canvas
from reportlab.pdfbase import pdfmetrics; from reportlab.pdfbase.ttfonts import TTFont
fp=next((p for p in ["/Library/Fonts/Arial.ttf","/System/Library/Fonts/Supplemental/Arial.ttf","/usr/share/fonts/truetype/dejavu/DejaVuSans.ttf"] if os.path.exists(p)),None)
if fp:
pdfmetrics.registerFont(TTFont("U", fp))
c = canvas.Canvas(f"{OUT}/clinic_zeta_prices.pdf", pagesize=A4)
c.setFont("U", 18); c.drawString(60, 790, "Diagnostic Lab Astana — Price list / Прайс")
c.setFont("U", 12); y = 750
for name, p, unit in variants_for(2, 10):
c.drawString(60, y, name[:55]); c.drawRightString(540, y, f"{p} KZT"); y -= 28
c.save()
pdf = "ok"
else:
pdf = "no font"
except Exception as e:
pdf = f"skip ({e})"
pdfmetrics.registerFont(TTFont("U",fp)); c=canvas.Canvas(f"{RAW}/clinic_zeta_prices.pdf",pagesize=A4)
c.setFont("U",18); c.drawString(60,790,"Diagnostic Lab Astana — Price list / Прайс"); c.setFont("U",12); y=750
for r in rows: c.drawString(60,y,r["raw_name"][:55]); c.drawRightString(540,y,f"{r['price']} KZT"); y-=28
c.save(); pdf="ok"
except Exception as e: pdf=f"skip ({e})"
# эталон нормализации + справочник
with open(f"{NORM}/golden.json","w",encoding="utf-8") as f:
json.dump(GOLDEN, f, ensure_ascii=False, indent=2)
canon_out=[{"canon_id":c["id"],"canonical_ru":c["ru"],"canonical_kz":c["kz"],"canonical_en":c["en"],
"category":c["category"],"code":c["code"],"unit":c["unit"],
"synonyms":[{"text":t,"lang":l} for t,l in c["syn"]]} for c in CANONS]
with open(f"{CONTRACTS}/service_canon.json","w",encoding="utf-8") as f:
json.dump(canon_out, f, ensure_ascii=False, indent=2)
print("PDF:", pdf)
print("Files:", sorted(os.listdir(OUT)))
print("raw:", sorted(os.listdir(RAW)))
print("golden rows:", len(GOLDEN), "| canons:", len(CANONS))