fix: review findings - OUTPUT_DIR centralised, no hardcoded vwts.ru, normal imports
This commit is contained in:
+7
-4
@@ -1,7 +1,9 @@
|
||||
"""Основной цикл: обход страниц раздела → темы → посты → сохранение."""
|
||||
|
||||
import sys, logging
|
||||
from .config import HEADERS
|
||||
from pathlib import Path
|
||||
from urllib.parse import urlparse
|
||||
from .config import OUTPUT_DIR
|
||||
from .fetch import get, session
|
||||
from .paginate import count, verify_last
|
||||
from .parse import parse_topics, parse_posts, parse_tags
|
||||
@@ -85,11 +87,13 @@ def run(section_url: str):
|
||||
posts, tags = [], parse_tags(topic_soup)
|
||||
|
||||
# ── Обход страниц темы ────────────────────────────────────
|
||||
base = f"{urlparse(section_url).scheme}://{urlparse(section_url).netloc}"
|
||||
|
||||
for tpp in range(1, topic_pages + 1):
|
||||
if tpp == 1:
|
||||
posts += parse_posts(topic_soup)
|
||||
else:
|
||||
tp2 = get(f"https://vwts.ru/forum/topic/{tid}/page-{tpp}")
|
||||
tp2 = get(f"{base}/forum/topic/{tid}/page-{tpp}")
|
||||
if tp2:
|
||||
posts += parse_posts(tp2)
|
||||
|
||||
@@ -106,7 +110,6 @@ def run(section_url: str):
|
||||
log.info("=" * 60)
|
||||
log.info(f"✅ '{section_name}' — {st['topic_count']} тем "
|
||||
f"в {st['file_index']+1} частях")
|
||||
for f in sorted((__import__('pathlib').Path('vwts_data') /
|
||||
section_slug).glob("part_*.jsonl")):
|
||||
for f in sorted((OUTPUT_DIR / section_slug).glob("part_*.jsonl")):
|
||||
log.info(f" 📄 {f.name} ({f.stat().st_size/1024/1024:.1f} MB)")
|
||||
log.info("=" * 60)
|
||||
|
||||
Reference in New Issue
Block a user