106 lines
5.1 KiB
Python
106 lines
5.1 KiB
Python
# xlsx_recorder.py · расширен отправкой JSON
|
||
from openpyxl import Workbook
|
||
from os.path import isdir, abspath, join
|
||
from os import mkdir
|
||
import re, json, math, logging, requests, os
|
||
|
||
log = logging.getLogger("recorder")
|
||
|
||
# ─────────────────────── настройки ───────────────────────
|
||
SEND_JSON = False # отправка POST
|
||
SAVE_JSON = True # сохранять копию JSON
|
||
POST_URL = "http://localhost:3005/parser/data"
|
||
|
||
INVALID_CHARS = r'[<>:"/\\|*?]'
|
||
def sanitize_filename(name: str, repl: str = "_") -> str:
|
||
"""Удаляет/заменяет символы, запрещённые в именах файлов."""
|
||
clean = re.sub(INVALID_CHARS, repl, name)
|
||
return clean.split("?", 1)[0].strip()
|
||
|
||
# объект для сохранения спаршенных категорий
|
||
class Recorder:
|
||
def __init__(self, records_folder="records_folder"):
|
||
# создаём папку при необходимости
|
||
rf_abs = abspath(records_folder)
|
||
if not isdir(rf_abs):
|
||
mkdir(rf_abs)
|
||
self.record_folder = rf_abs
|
||
|
||
# запись таблицы + JSON/POST
|
||
def record(self, csv_name, table):
|
||
csv_name = sanitize_filename(csv_name)
|
||
|
||
# ─── 1) сохраняем XLSX ─────────────────────────────
|
||
wb = Workbook()
|
||
ws = wb.active
|
||
for row in table:
|
||
ws.append(row)
|
||
xlsx_path = join(self.record_folder, f"{csv_name}.xlsx")
|
||
wb.save(xlsx_path)
|
||
log.info("XLSX saved → %s", xlsx_path)
|
||
|
||
# ─── 2) формируем JSON (по утверждённым правилам) ──
|
||
headers = table[0]
|
||
idx = {h: i for i, h in enumerate(headers)}
|
||
|
||
items = []
|
||
for row in table[1:]:
|
||
# базовые поля
|
||
article = row[idx["Артикул"]]
|
||
partnumber = row[idx["PartNumber"]]
|
||
size_full = row[idx["Свойство: Размер"]].replace("\n", "<br/>")
|
||
price_raw = row[idx["Цена закупки"]]
|
||
price_int = math.ceil(float(price_raw))
|
||
clr_name_raw = row[idx["Свойство: Цвет"]]
|
||
clr_name = clr_name_raw.capitalize()
|
||
vis = row[idx["Наличие на сайте"]]
|
||
weight_gram = float(row[idx["Свойство: Вес(г)"]]) if row[idx["Свойство: Вес(г)"]] else 0.0
|
||
weight_kg = math.ceil(weight_gram / 1000) if weight_gram else 0
|
||
|
||
# составляем объект
|
||
variant = {
|
||
"status_id": 1,
|
||
"color": clr_name,
|
||
"sku": f"{article}-{partnumber}",
|
||
"size": size_full,
|
||
"cost": price_int,
|
||
"originalUrl": row[idx["Краткое описание"]], # url_full в таблице
|
||
"originalName": row[idx["Название товара или услуги"]].capitalize(),
|
||
"originalDescription": (
|
||
row[idx["Полное описание"]].replace("\n", "<br/>") + "<br/>" +
|
||
row[idx["Параметр: Уход"]].replace("\n", "<br/>") + "<br/>" +
|
||
row[idx["Параметр: Происхождение"]].replace("\n", "<br/>")
|
||
).strip("<br/>"),
|
||
"originalComposition": row[idx["Параметр: Состав"]].replace("\n", "<br/>"),
|
||
"images": [img for img in row[idx["Изображения варианта"]].split("\n") if img],
|
||
"inStock": vis == "SHOW",
|
||
"weight": weight_kg
|
||
}
|
||
|
||
cat_raw = row[idx["Размещение на сайте"]].replace("Каталог/ZaraHome/WOMEN/", "")
|
||
category_name = re.sub(r"[^\w/-]+|_+", "_", cat_raw)
|
||
|
||
items.append({
|
||
"category": {"name": category_name},
|
||
"variant": variant,
|
||
"brand": {"name": "zara-home"}
|
||
})
|
||
|
||
payload = {"items": items, "parserName": "zara-home"}
|
||
|
||
# ─── 3) сохраняем JSON при необходимости ───────────
|
||
if SAVE_JSON:
|
||
json_path = join(self.record_folder, f"{csv_name}.json")
|
||
with open(json_path, "w", encoding="utf-8") as fh:
|
||
json.dump(payload, fh, ensure_ascii=False, indent=2)
|
||
log.info("JSON saved → %s", json_path)
|
||
|
||
# ─── 4) POST на локальный сервис ───────────────────
|
||
if SEND_JSON:
|
||
try:
|
||
resp = requests.post(POST_URL, json=payload, timeout=20)
|
||
resp.raise_for_status()
|
||
log.info("POST %s OK (%s items)", csv_name, len(items))
|
||
except Exception as err:
|
||
log.warning("POST %s FAILED: %s", csv_name, err)
|