new 2 custom tables + metadata repetition fixed

table_prep.py

@@ -19,7 +19,7 @@ CUSTOM_TABLE_CONFIGS = {
             "№ Б.2": {"method": "split_by_rows"}
         }
     },
-    "НП-104-18": { 
+    "НП-104-18": {
         "tables": {
             "*": {"method": "group_entire_table"}  # All tables
         }
@@ -38,34 +38,57 @@ CUSTOM_TABLE_CONFIGS = {
             "№ 3": {"method": "split_by_rows"}
         }
     },
-    "НП-089-15": {  # New addition
+    "НП-089-15": {
         "tables": {
-            "-": {"method": "split_by_rows"}  
+            "-": {"method": "split_by_rows"}
+        }
+    },
+    "НП-105-18": {  # New addition for problematic table
+        "tables": {
+            "№ 4.8": {"method": "group_entire_table"}
+        }
+    },
+    "ГОСТ Р 50.05.23-2020": {  # New addition for problematic table
+        "tables": {
+            "№8": {"method": "group_entire_table"}
+        }
+    },
+    "ГОСТ Р 50.03.01-2017": {  # New addition for А.8 table
+        "tables": {
+            "А.8": {"method": "group_entire_table"}
         }
     }
 }
 
 def create_meta_info(document_name, section, table_number, table_title, extra_info=""):
-    """Create standard meta information string"""
-    base_info = f'Документ "{document_name}", Раздел: {section}, Номер таблицы: {table_number}, Название таблицы: {table_title}'
+    base_info = f'Документ "{document_name}", Раздел: {section}, Таблица: {table_number}'
+    if table_title and table_title.strip():
+        base_info += f', Название: {table_title}'
     if extra_info:
         base_info += f', {extra_info}'
-    return base_info + '\n'
+    return base_info
 
 def create_chunk_text(meta_info, headers, rows, add_row_numbers=False):
-    """Create chunk text with headers and rows"""
+    chunk_lines = [meta_info.rstrip()]  # Remove trailing newline from meta_info
+    
+    # Add headers only once
     header_line = " | ".join(headers)
-    chunk_lines = [meta_info + "Заголовки: " + header_line]
+    chunk_lines.append(f"Заголовки: {header_line}")
     
+    # Add rows without redundant formatting
     for i, row in enumerate(rows, start=1):
-        row_text = " | ".join([f"{h}: {row.get(h, '')}" for h in headers])
+        row_parts = []
+        for h in headers:
+            value = row.get(h, '')
+            if value:  # Only add non-empty values
+                row_parts.append(f"{h}: {value}")
+        
         if add_row_numbers:
-            chunk_lines.append(f"Строка {i}: {row_text}")
+            chunk_lines.append(f"Строка {i}: {' | '.join(row_parts)}")
         else:
-            chunk_lines.append(row_text)
+            chunk_lines.append(' | '.join(row_parts))
     
     return "\n".join(chunk_lines)
-
 def group_by_column_method(table_data, document_name, group_column):
     """Group rows by specified column value"""
     documents = []
@@ -194,19 +217,16 @@ def process_table_with_custom_method(table_data, document_name, method_config):
 def table_to_document(table_data, document_id=None):
     if isinstance(table_data, dict):
         doc_id = document_id or table_data.get('document_id', table_data.get('document', 'Неизвестно'))
-        table_num = table_data.get('table_number', 'Неизвестно')
-        
-        # Check if this table should use custom processing
+        table_num = table_data.get('table_number', 'Неизвестно')        
         use_custom, doc_pattern, method_config = should_use_custom_processing(doc_id, table_num)
         
         if use_custom:
             log_message(f"Using custom processing for table {table_num} in document {doc_id}")
             custom_docs = process_table_with_custom_method(table_data, doc_id, method_config)
             if custom_docs:
-                # Return custom processed documents and skip default processing
                 return custom_docs
         
-        # Default processing for tables not in custom config
+        # DEFAULT PROCESSING (only if NOT using custom)
         table_title = table_data.get('table_title', 'Неизвестно')
         section = table_data.get('section', 'Неизвестно')