Hermes Agent
2a06bcaee0
- artikel_*.html: 24 Dateien von <style>-Blöcken befreit (body, max-width zerstören Joomla) - journalistic-article-writing SKILL.md: falsche 'AUSNAHME 09.07.' entfernt - Ursache: Skill-Kommentar erlaubte Sub-Agenten <style> 'für Vorschau' — kein Gatekeeper existiert
97 lines
3.3 KiB
Python
97 lines
3.3 KiB
Python
#!/usr/bin/env python3
|
|
"""Check what items are being excluded by the filter."""
|
|
|
|
import re
|
|
import html as html_mod
|
|
|
|
with open('/home/hermes/workspace/fantasywelt_data/neuheiten.html', 'r') as f:
|
|
content = f.read()
|
|
|
|
neu_section = content.split('Wieder auf Lager')[0]
|
|
rest = content.split('Wieder auf Lager')[1] if 'Wieder auf Lager' in content else ''
|
|
wieder_section = rest.split('Neue Vorbestellungen')[0] if 'Neue Vorbestellungen' in rest else rest
|
|
|
|
EXCLUDE_KEYWORDS = [
|
|
'puzzle', 'sleeve', 'deck box', 'deckbox', 'boulder', 'portfolio',
|
|
'ultimate guard', 'sidewinder', 'omnihive', 'xenoskin',
|
|
'card binder', 'deck protector', 'toploader',
|
|
'perfect fit', 'inner sleeve', 'outer sleeve', 'matte sleeve',
|
|
'sammelalbum', 'playmat tube',
|
|
'spielmatte', 'playmat', 'game mat', 'neoprene', 'play mat',
|
|
'orderunit',
|
|
'upgrade pack', 'upgrade kit', 'insert', 'organizer', 'organiser',
|
|
'ausgrabung', 'kristall', 'farblabor',
|
|
'dice set', 'wuerfelset', 'wuerfel set', 'dice pack', 'token set',
|
|
'miniatures pack', 'playerboard',
|
|
'storage box', 'token box', 'token tray', 'bit box',
|
|
'feldherr', 'schaumstoff', 'plastic tray',
|
|
'deck box', 'deckbox',
|
|
'spiel knobelblock', 'knobelblock',
|
|
'anatomiepuzzle',
|
|
'spotlight deck', 'spotlight-deck',
|
|
'gaming tokens', 'gamegenic', 'alcove flip box',
|
|
'play mat', 'premium game mat',
|
|
'minifiguren', 'blind bag',
|
|
'decals set', 'decal set', 'propeller set',
|
|
'leather folder',
|
|
'global map',
|
|
'tarot deck',
|
|
'organizer',
|
|
'würfel blind box', 'wuerfel blind box',
|
|
'deluxe upgrade',
|
|
'abdeckung',
|
|
'exclusive extras',
|
|
]
|
|
|
|
def is_excluded(name):
|
|
name_lower = name.lower().replace('-', ' ').replace(' ', ' ')
|
|
for kw in EXCLUDE_KEYWORDS:
|
|
if kw in name_lower:
|
|
return True
|
|
return False
|
|
|
|
def parse_all_cards(section):
|
|
cards = re.split(
|
|
r'<div style="display:flex;gap:0\.8rem;padding:1rem 0;'
|
|
r'border-bottom:1px solid #eee;align-items:flex-start;">',
|
|
section
|
|
)
|
|
cards = cards[1:]
|
|
parsed = []
|
|
for card in cards:
|
|
name_match = re.search(
|
|
r'<a href="(https://www\.fantasywelt\.de/[^"]+)"[^>]*>([^<]+)</a>',
|
|
card
|
|
)
|
|
if not name_match:
|
|
continue
|
|
name = html_mod.unescape(name_match.group(2).strip())
|
|
if 'Kaufen' in name or '\u2192' in name or len(name) < 3:
|
|
continue
|
|
parsed.append(name)
|
|
return parsed
|
|
|
|
neu_all = parse_all_cards(neu_section)
|
|
wieder_all = parse_all_cards(wieder_section)
|
|
|
|
print("=== EXCLUDED from NEU IM SHOP ===")
|
|
for name in neu_all:
|
|
if is_excluded(name):
|
|
name_lower = name.lower().replace('-', ' ').replace(' ', ' ')
|
|
matching = [kw for kw in EXCLUDE_KEYWORDS if kw in name_lower]
|
|
print(f" EXCLUDED: {name}")
|
|
print(f" matched: {matching}")
|
|
|
|
print(f"\n=== EXCLUDED from WIEDER AUF LAGER (first 30) ===")
|
|
count = 0
|
|
for name in wieder_all:
|
|
if is_excluded(name):
|
|
name_lower = name.lower().replace('-', ' ').replace(' ', ' ')
|
|
matching = [kw for kw in EXCLUDE_KEYWORDS if kw in name_lower]
|
|
print(f" EXCLUDED: {name}")
|
|
print(f" matched: {matching}")
|
|
count += 1
|
|
if count >= 30:
|
|
print(f" ... and more")
|
|
break
|