from cry to latin + 2000,30

config.py

@@ -51,7 +51,7 @@ DEFAULT_MODEL = "Gemini 2.5 Flash"
 CHUNK_SIZE = 1500
 CHUNK_OVERLAP = 128
 
-MAX_CHARS_TABLE = 3000
+MAX_CHARS_TABLE = 2000
 MAX_ROWS_TABLE = 30
 
 CUSTOM_PROMPT = """

documents_prep.py

@@ -26,33 +26,38 @@ import re
 
 def normalize_steel_designations(text):
     """
-    Normalize steel designations by converting Latin letters to Cyrillic.
-    Handles patterns like 08X18H10T, 12X18H9, 10H17N13M2T, etc.
+    Normalize steel designations by converting Cyrillic letters to Latin.
+    This improves search/retrieval since embedding models work better with Latin.
+    Handles patterns like 08Х18Н10Т → 08X18H10T
     Returns: (normalized_text, changes_count, changes_list)
     """
     if not text:
         return text, 0, []
 
+    import re
+    
     changes_count = 0
     changes_list = []
 
-    # Mapping of Latin to Cyrillic for steel designations
+    # Mapping of Cyrillic to Latin for steel designations
     replacements = {
-        'X': 'Х',
-        'H': 'Н',
-        'T': 'Т',
-        'C': 'С',
-        'B': 'В',
-        'K': 'К',
-        'M': 'М',
-        'A': 'А',
-        'P': 'Р',
+        'Х': 'X',  # Cyrillic Kha → Latin X
+        'Н': 'H',  # Cyrillic En → Latin H
+        'Т': 'T',  # Cyrillic Te → Latin T
+        'С': 'C',  # Cyrillic Es → Latin C
+        'В': 'B',  # Cyrillic Ve → Latin B
+        'К': 'K',  # Cyrillic Ka → Latin K
+        'М': 'M',  # Cyrillic Em → Latin M
+        'А': 'A',  # Cyrillic A → Latin A
+        'Р': 'P',  # Cyrillic Er → Latin P
     }
 
-    # Regex to match steel designations like 08X18H10T, 10H17N13M2T, etc.
-    # \b\d{1,3} — starts with 1–3 digits
-    # (?:[A-ZА-Я]\d*)+ — then one or more groups of a letter + optional digits
-    pattern = r'\b\d{1,3}(?:[A-ZА-Я]\d*)+\b'
+    # Pattern: starts with digits, then letters+digits (steel grade pattern)
+    # Examples: 08Х18Н10Т, 12Х18Н9, 10Н17Н13М2Т, СВ-08Х19Н10
+    pattern = r'\b\d{1,3}(?:[A-ZА-ЯЁ]\d*)+\b'
+    
+    # Also match welding wire patterns like СВ-08Х19Н10
+    pattern_wire = r'\b[СC][ВB]-\d{1,3}(?:[A-ZА-ЯЁ]\d*)+\b'
 
     def replace_in_steel_grade(match):
         nonlocal changes_count, changes_list
@@ -63,7 +68,9 @@ def normalize_steel_designations(text):
             changes_list.append(f"{original} → {converted}")
         return converted
 
+    # Apply both patterns
     normalized_text = re.sub(pattern, replace_in_steel_grade, text)
+    normalized_text = re.sub(pattern_wire, replace_in_steel_grade, normalized_text)
 
     return normalized_text, changes_count, changes_list