Avoid false garbage classification for UPD mentions
Deploy contracts-flask / validate (push) Canceled after 0s

This commit is contained in:
“Naeel”
2026-08-27 10:44:48 +03:00
parent 475efefb40
commit e5c7c88073
5 changed files with 75 additions and 10 deletions
+1 -1
View File
@@ -1,7 +1,7 @@
"""Конфигурация приложения — все настройки в одном месте."""
import os
VERSION = "2.0.19"
VERSION = "2.0.20"
LLM_URL = os.getenv("LLM_API_URL", "https://api.aillm.ru/v1/chat/completions")
LLM_KEY = os.getenv("LLM_API_KEY", "")
+6 -2
View File
@@ -47,9 +47,13 @@ _GARBAGE_HEADER_MARKERS = [
'СЧЕТ-ФАКТУРА', 'СЧЕТ НА ОПЛАТУ', 'АКТ СВЕРКИ',
'АКТ ОКАЗАННЫХ УСЛУГ', 'АКТ ВЫПОЛНЕННЫХ РАБОТ',
'ПЛАТЁЖНОЕ ПОРУЧЕНИЕ', 'УНИВЕРСАЛЬНЫЙ ПЕРЕДАТОЧНЫЙ',
'УПД', 'ПЛАТЕЖНОЕ ПОРУЧЕНИЕ',
'ПЛАТЕЖНОЕ ПОРУЧЕНИЕ',
]
_GARBAGE_HEADER_RE = re.compile(
r'(?m)^\s*(?:УПД|УНИВЕРСАЛЬНЫЙ ПЕРЕДАТОЧНЫЙ ДОКУМЕНТ)\b'
)
def _is_garbage_by_filename(filename: str) -> bool:
"""Stage 1: regex по имени файла — быстро, 0 токенов."""
@@ -59,7 +63,7 @@ def _is_garbage_by_filename(filename: str) -> bool:
def _is_garbage_by_header(text: str) -> bool:
"""Stage 2: ключевые слова в первых 2KB текста — быстро, 0 токенов."""
header = text[:2000].upper()
return any(marker in header for marker in _GARBAGE_HEADER_MARKERS)
return any(marker in header for marker in _GARBAGE_HEADER_MARKERS) or bool(_GARBAGE_HEADER_RE.search(header))
def _call_llm_classify(header_text, llm_client=None):
+7 -7
View File
@@ -73,7 +73,7 @@
<body>
<div class="topbar">
<img src="/static/logo.svg" alt="Nubes">
<span class="title">Сверка договоров — LLM AI-driven Event Sourcing <span style="font-weight:400;color:var(--muted);font-size:12px;">v2.0.19</span></span>
<span class="title">Сверка договоров — LLM AI-driven Event Sourcing <span style="font-weight:400;color:var(--muted);font-size:12px;">v2.0.20</span></span>
<div id="pipelineStepper" style="display:flex;gap:8px;font-size:11px;align-items:center;color:var(--muted);">
<span id="stepUpload">○ Загрузка</span><span></span>
<span id="stepClassify">○ Классификация</span><span></span>
@@ -224,11 +224,11 @@
import { listZipFiles } from '/upload/zip/list_zip_files.js';
window.listZipFiles = listZipFiles;
</script>
<script src="/static/state.js?v=2.0.19"></script>
<script src="/static/app_utils.js?v=2.0.19"></script>
<script src="/static/files.js?v=2.0.19"></script>
<script src="/static/groups.js?v=2.0.19"></script>
<script src="/static/compare.js?v=2.0.19"></script>
<script src="/static/app.js?v=2.0.19"></script>
<script src="/static/state.js?v=2.0.20"></script>
<script src="/static/app_utils.js?v=2.0.20"></script>
<script src="/static/files.js?v=2.0.20"></script>
<script src="/static/groups.js?v=2.0.20"></script>
<script src="/static/compare.js?v=2.0.20"></script>
<script src="/static/app.js?v=2.0.20"></script>
</body>
</html>