Spaces:
Sleeping
Sleeping
Update app.py
Browse files
app.py
CHANGED
|
@@ -18,21 +18,44 @@ class KDChecker:
|
|
| 18 |
self.known_docs = ["Э3", "В4", "ПЭ3", "ВО", "ТЭ5", "СБ", "С5", "ОЛ", "Э1", "Э4", "Э7", "Д3", "Э6"]
|
| 19 |
|
| 20 |
def load_excel_db(self, excel_path):
|
|
|
|
|
|
|
|
|
|
| 21 |
if excel_path is None:
|
| 22 |
-
return "Файл не
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 23 |
|
| 24 |
all_data = []
|
| 25 |
sheets_log = []
|
| 26 |
|
| 27 |
try:
|
| 28 |
-
|
|
|
|
| 29 |
xls = pd.read_excel(excel_path, sheet_name=None, header=None, engine='openpyxl')
|
|
|
|
| 30 |
|
| 31 |
for sheet_name, df_raw in xls.items():
|
| 32 |
header_row_index = -1
|
| 33 |
cab_col_idx = -1
|
| 34 |
rem_col_idx = -1
|
| 35 |
|
|
|
|
| 36 |
for i in range(min(20, len(df_raw))):
|
| 37 |
row_values = [str(x).lower().strip() for x in df_raw.iloc[i].values]
|
| 38 |
c_idx = -1
|
|
@@ -64,17 +87,20 @@ class KDChecker:
|
|
| 64 |
sheets_log.append(f"Лист '{sheet_name}': заголовки не найдены")
|
| 65 |
|
| 66 |
if not all_data:
|
| 67 |
-
|
|
|
|
| 68 |
|
| 69 |
self.excel_db = pd.concat(all_data, ignore_index=True)
|
| 70 |
self.cabinet_list = sorted(self.excel_db["Cabinet"].unique().tolist())
|
| 71 |
-
|
|
|
|
| 72 |
msg = f"✅ База знаний загружена успешно!\nВсего записей: {len(self.excel_db)}\nОбработаны листы: {', '.join(sheets_log)}"
|
| 73 |
return msg, gr.update(choices=self.cabinet_list, value=None, interactive=True)
|
| 74 |
|
| 75 |
except Exception as e:
|
| 76 |
-
#
|
| 77 |
-
|
|
|
|
| 78 |
|
| 79 |
def extract_text(self, pdf_path):
|
| 80 |
try:
|
|
@@ -114,7 +140,6 @@ class KDChecker:
|
|
| 114 |
|
| 115 |
def get_remarks(self, cabinet_key, is_clean_key=True):
|
| 116 |
if self.excel_db.empty: return {}
|
| 117 |
-
|
| 118 |
if is_clean_key:
|
| 119 |
target = cabinet_key.replace(" ", "")
|
| 120 |
mask = self.excel_db['Cabinet_Clean'].str.contains(re.escape(target), case=False, na=False)
|
|
@@ -123,43 +148,31 @@ class KDChecker:
|
|
| 123 |
|
| 124 |
rows = self.excel_db[mask]
|
| 125 |
if rows.empty: return {}
|
| 126 |
-
|
| 127 |
parsed = {}
|
| 128 |
-
|
| 129 |
for remark_cell in rows['Remark']:
|
| 130 |
cell_text = str(remark_cell)
|
| 131 |
cell_text = re.sub(r'(\d+)\.([А-ЯA-Z])', r'\1. \2', cell_text)
|
| 132 |
-
|
| 133 |
items = re.split(r'(?:^|\n)\s*(?=\d+[\.\)])', cell_text)
|
| 134 |
-
|
| 135 |
for item in items:
|
| 136 |
if len(item) < 3: continue
|
| 137 |
clean_item = item.strip()
|
| 138 |
clean_item_no_num = re.sub(r'^\d+[\.\)]\s*', '', clean_item)
|
| 139 |
-
|
| 140 |
doc_pattern = r'^(?:Документ\s+|В\s+)?([А-ЯA-Z0-9\s,\(\)\-]+?)(?:[\.\:\-]|\s+)(.*)'
|
| 141 |
match = re.match(doc_pattern, clean_item_no_num, re.IGNORECASE | re.DOTALL)
|
| 142 |
-
|
| 143 |
detected_docs = []
|
| 144 |
final_text = clean_item
|
| 145 |
-
|
| 146 |
if match:
|
| 147 |
potential_docs_str = match.group(1).upper()
|
| 148 |
cleaned_codes = potential_docs_str.replace("(", " ").replace(")", " ").replace(",", " ")
|
| 149 |
parts = cleaned_codes.split()
|
| 150 |
-
|
| 151 |
valid_parts = [p for p in parts if p in self.known_docs]
|
| 152 |
if valid_parts:
|
| 153 |
detected_docs = valid_parts
|
| 154 |
final_text = match.group(2).strip()
|
| 155 |
-
|
| 156 |
-
if not detected_docs:
|
| 157 |
-
detected_docs = ["ALL"]
|
| 158 |
-
|
| 159 |
for doc in detected_docs:
|
| 160 |
if doc not in parsed: parsed[doc] = []
|
| 161 |
parsed[doc].append(final_text)
|
| 162 |
-
|
| 163 |
return parsed
|
| 164 |
|
| 165 |
def check_files(self, files, manual_cabinet):
|
|
@@ -226,11 +239,11 @@ class KDChecker:
|
|
| 226 |
pdf_title = detected_cabinet
|
| 227 |
if is_manual: pdf_title += " (Выбор вручную)"
|
| 228 |
|
| 229 |
-
#
|
| 230 |
try:
|
| 231 |
pdf = self.create_pdf(pdf_title, checklist)
|
| 232 |
except Exception as e:
|
| 233 |
-
return f"❌ Ошибка создания PDF: {e}
|
| 234 |
|
| 235 |
total = sum(len(v) for v in checklist.values())
|
| 236 |
method_str = "Ручной выбор" if is_manual else (
|
|
@@ -245,29 +258,26 @@ class KDChecker:
|
|
| 245 |
form = c.acroForm
|
| 246 |
width, height = A4
|
| 247 |
|
| 248 |
-
#
|
| 249 |
font_name = 'Helvetica'
|
| 250 |
-
font_path = "arial.ttf"
|
| 251 |
|
| 252 |
if os.path.exists(font_path):
|
| 253 |
try:
|
| 254 |
pdfmetrics.registerFont(TTFont('Arial', font_path))
|
| 255 |
font_name = 'Arial'
|
| 256 |
except:
|
| 257 |
-
pass
|
| 258 |
-
|
| 259 |
-
# Если шрифт не найден, но мы на сервере - это проблема.
|
| 260 |
-
# Код попытается использовать Helvetica, но кириллица пропадет.
|
| 261 |
|
| 262 |
y = height - 50
|
| 263 |
c.setFont(font_name, 16)
|
| 264 |
|
| 265 |
-
#
|
| 266 |
try:
|
| 267 |
c.drawString(50, y, f"ЧЕК-ЛИСТ ПРОВЕРКИ КД")
|
| 268 |
except:
|
| 269 |
c.setFont("Helvetica", 16)
|
| 270 |
-
c.drawString(50, y, "CHECK-LIST
|
| 271 |
c.setFont(font_name, 16)
|
| 272 |
|
| 273 |
y -= 25
|
|
@@ -278,7 +288,7 @@ class KDChecker:
|
|
| 278 |
c.drawString(50, y, f"Шкаф: {disp_cab}")
|
| 279 |
except:
|
| 280 |
c.setFont("Helvetica", 12)
|
| 281 |
-
c.drawString(50, y, "Cabinet
|
| 282 |
c.setFont(font_name, 12)
|
| 283 |
|
| 284 |
c.drawString(400, y, f"Дата: {datetime.now().strftime('%d.%m.%Y')}")
|
|
@@ -300,10 +310,10 @@ class KDChecker:
|
|
| 300 |
try:
|
| 301 |
c.drawString(50, y, f"Файл: {filename}")
|
| 302 |
except:
|
| 303 |
-
|
| 304 |
-
|
| 305 |
-
|
| 306 |
-
|
| 307 |
c.setFillColor(colors.black)
|
| 308 |
y -= 15
|
| 309 |
c.setFont(font_name, 10)
|
|
@@ -339,13 +349,10 @@ class KDChecker:
|
|
| 339 |
c.showPage();
|
| 340 |
text_start_y = height - 50;
|
| 341 |
c.setFont(font_name, 10)
|
| 342 |
-
|
| 343 |
try:
|
| 344 |
c.drawString(65, text_start_y, l.strip())
|
| 345 |
except:
|
| 346 |
-
# Игнорируем
|
| 347 |
-
pass
|
| 348 |
-
|
| 349 |
text_start_y -= 12
|
| 350 |
|
| 351 |
y = text_start_y - 8
|
|
|
|
| 18 |
self.known_docs = ["Э3", "В4", "ПЭ3", "ВО", "ТЭ5", "СБ", "С5", "ОЛ", "Э1", "Э4", "Э7", "Д3", "Э6"]
|
| 19 |
|
| 20 |
def load_excel_db(self, excel_path):
|
| 21 |
+
# --- ЛОГГЕР ---
|
| 22 |
+
log = [f"[{datetime.now().strftime('%H:%M:%S')}] Старт функции загрузки"]
|
| 23 |
+
|
| 24 |
if excel_path is None:
|
| 25 |
+
return "Файл не передан (excel_path is None)", gr.update(choices=[], value=None)
|
| 26 |
+
|
| 27 |
+
log.append(f"Путь к файлу: {excel_path}")
|
| 28 |
+
|
| 29 |
+
# Проверка размера файла (чтобы убедиться, что он загрузился)
|
| 30 |
+
try:
|
| 31 |
+
size = os.path.getsize(excel_path)
|
| 32 |
+
log.append(f"Размер файла: {size} байт")
|
| 33 |
+
except Exception as e:
|
| 34 |
+
log.append(f"Ошибка доступа к файлу: {e}")
|
| 35 |
+
return "\n".join(log), gr.update(choices=[], value=None)
|
| 36 |
+
|
| 37 |
+
# Проверка наличия движка
|
| 38 |
+
try:
|
| 39 |
+
import openpyxl
|
| 40 |
+
log.append("Библиотека openpyxl найдена.")
|
| 41 |
+
except ImportError:
|
| 42 |
+
return "ОШИБКА: Библиотека openpyxl не установлена на сервере!", gr.update(choices=[], value=None)
|
| 43 |
|
| 44 |
all_data = []
|
| 45 |
sheets_log = []
|
| 46 |
|
| 47 |
try:
|
| 48 |
+
log.append("Попытка чтения Excel через pandas...")
|
| 49 |
+
# Явно указываем движок
|
| 50 |
xls = pd.read_excel(excel_path, sheet_name=None, header=None, engine='openpyxl')
|
| 51 |
+
log.append(f"Файл прочитан. Найдены листы: {list(xls.keys())}")
|
| 52 |
|
| 53 |
for sheet_name, df_raw in xls.items():
|
| 54 |
header_row_index = -1
|
| 55 |
cab_col_idx = -1
|
| 56 |
rem_col_idx = -1
|
| 57 |
|
| 58 |
+
# Поиск заголовков
|
| 59 |
for i in range(min(20, len(df_raw))):
|
| 60 |
row_values = [str(x).lower().strip() for x in df_raw.iloc[i].values]
|
| 61 |
c_idx = -1
|
|
|
|
| 87 |
sheets_log.append(f"Лист '{sheet_name}': заголовки не найдены")
|
| 88 |
|
| 89 |
if not all_data:
|
| 90 |
+
log.append("Не найдены данные ни на одном листе.")
|
| 91 |
+
return "\n".join(log), gr.update(choices=[], value=None)
|
| 92 |
|
| 93 |
self.excel_db = pd.concat(all_data, ignore_index=True)
|
| 94 |
self.cabinet_list = sorted(self.excel_db["Cabinet"].unique().tolist())
|
| 95 |
+
|
| 96 |
+
log.append("Успешно объединили данные.")
|
| 97 |
msg = f"✅ База знаний загружена успешно!\nВсего записей: {len(self.excel_db)}\nОбработаны листы: {', '.join(sheets_log)}"
|
| 98 |
return msg, gr.update(choices=self.cabinet_list, value=None, interactive=True)
|
| 99 |
|
| 100 |
except Exception as e:
|
| 101 |
+
# Возвращаем полный лог ошибки
|
| 102 |
+
log.append(f"КРИТИЧЕСКАЯ ОШИБКА: {str(e)}")
|
| 103 |
+
return "\n".join(log), gr.update(choices=[], value=None)
|
| 104 |
|
| 105 |
def extract_text(self, pdf_path):
|
| 106 |
try:
|
|
|
|
| 140 |
|
| 141 |
def get_remarks(self, cabinet_key, is_clean_key=True):
|
| 142 |
if self.excel_db.empty: return {}
|
|
|
|
| 143 |
if is_clean_key:
|
| 144 |
target = cabinet_key.replace(" ", "")
|
| 145 |
mask = self.excel_db['Cabinet_Clean'].str.contains(re.escape(target), case=False, na=False)
|
|
|
|
| 148 |
|
| 149 |
rows = self.excel_db[mask]
|
| 150 |
if rows.empty: return {}
|
|
|
|
| 151 |
parsed = {}
|
|
|
|
| 152 |
for remark_cell in rows['Remark']:
|
| 153 |
cell_text = str(remark_cell)
|
| 154 |
cell_text = re.sub(r'(\d+)\.([А-ЯA-Z])', r'\1. \2', cell_text)
|
|
|
|
| 155 |
items = re.split(r'(?:^|\n)\s*(?=\d+[\.\)])', cell_text)
|
|
|
|
| 156 |
for item in items:
|
| 157 |
if len(item) < 3: continue
|
| 158 |
clean_item = item.strip()
|
| 159 |
clean_item_no_num = re.sub(r'^\d+[\.\)]\s*', '', clean_item)
|
|
|
|
| 160 |
doc_pattern = r'^(?:Документ\s+|В\s+)?([А-ЯA-Z0-9\s,\(\)\-]+?)(?:[\.\:\-]|\s+)(.*)'
|
| 161 |
match = re.match(doc_pattern, clean_item_no_num, re.IGNORECASE | re.DOTALL)
|
|
|
|
| 162 |
detected_docs = []
|
| 163 |
final_text = clean_item
|
|
|
|
| 164 |
if match:
|
| 165 |
potential_docs_str = match.group(1).upper()
|
| 166 |
cleaned_codes = potential_docs_str.replace("(", " ").replace(")", " ").replace(",", " ")
|
| 167 |
parts = cleaned_codes.split()
|
|
|
|
| 168 |
valid_parts = [p for p in parts if p in self.known_docs]
|
| 169 |
if valid_parts:
|
| 170 |
detected_docs = valid_parts
|
| 171 |
final_text = match.group(2).strip()
|
| 172 |
+
if not detected_docs: detected_docs = ["ALL"]
|
|
|
|
|
|
|
|
|
|
| 173 |
for doc in detected_docs:
|
| 174 |
if doc not in parsed: parsed[doc] = []
|
| 175 |
parsed[doc].append(final_text)
|
|
|
|
| 176 |
return parsed
|
| 177 |
|
| 178 |
def check_files(self, files, manual_cabinet):
|
|
|
|
| 239 |
pdf_title = detected_cabinet
|
| 240 |
if is_manual: pdf_title += " (Выбор вручную)"
|
| 241 |
|
| 242 |
+
# Создание PDF с защитой от отсутствия шрифта
|
| 243 |
try:
|
| 244 |
pdf = self.create_pdf(pdf_title, checklist)
|
| 245 |
except Exception as e:
|
| 246 |
+
return f"❌ Ошибка создания PDF: {e}", None
|
| 247 |
|
| 248 |
total = sum(len(v) for v in checklist.values())
|
| 249 |
method_str = "Ручной выбор" if is_manual else (
|
|
|
|
| 258 |
form = c.acroForm
|
| 259 |
width, height = A4
|
| 260 |
|
| 261 |
+
# Ищем шрифт на сервере
|
| 262 |
font_name = 'Helvetica'
|
| 263 |
+
font_path = "arial.ttf" # Ожидаем в корне
|
| 264 |
|
| 265 |
if os.path.exists(font_path):
|
| 266 |
try:
|
| 267 |
pdfmetrics.registerFont(TTFont('Arial', font_path))
|
| 268 |
font_name = 'Arial'
|
| 269 |
except:
|
| 270 |
+
pass # Если шрифт битый, останется Helvetica
|
|
|
|
|
|
|
|
|
|
| 271 |
|
| 272 |
y = height - 50
|
| 273 |
c.setFont(font_name, 16)
|
| 274 |
|
| 275 |
+
# Безопасная печать заголовка
|
| 276 |
try:
|
| 277 |
c.drawString(50, y, f"ЧЕК-ЛИСТ ПРОВЕРКИ КД")
|
| 278 |
except:
|
| 279 |
c.setFont("Helvetica", 16)
|
| 280 |
+
c.drawString(50, y, "CHECK-LIST")
|
| 281 |
c.setFont(font_name, 16)
|
| 282 |
|
| 283 |
y -= 25
|
|
|
|
| 288 |
c.drawString(50, y, f"Шкаф: {disp_cab}")
|
| 289 |
except:
|
| 290 |
c.setFont("Helvetica", 12)
|
| 291 |
+
c.drawString(50, y, "Cabinet name error (font)")
|
| 292 |
c.setFont(font_name, 12)
|
| 293 |
|
| 294 |
c.drawString(400, y, f"Дата: {datetime.now().strftime('%d.%m.%Y')}")
|
|
|
|
| 310 |
try:
|
| 311 |
c.drawString(50, y, f"Файл: {filename}")
|
| 312 |
except:
|
| 313 |
+
c.setFont("Helvetica", 11)
|
| 314 |
+
c.drawString(50, y, f"File: {filename}")
|
| 315 |
+
c.setFont(font_name, 11)
|
| 316 |
+
|
| 317 |
c.setFillColor(colors.black)
|
| 318 |
y -= 15
|
| 319 |
c.setFont(font_name, 10)
|
|
|
|
| 349 |
c.showPage();
|
| 350 |
text_start_y = height - 50;
|
| 351 |
c.setFont(font_name, 10)
|
|
|
|
| 352 |
try:
|
| 353 |
c.drawString(65, text_start_y, l.strip())
|
| 354 |
except:
|
| 355 |
+
pass # Игнорируем непечатаемые символы
|
|
|
|
|
|
|
| 356 |
text_start_y -= 12
|
| 357 |
|
| 358 |
y = text_start_y - 8
|