main.py

@@ -26,8 +26,9 @@ DIRNAME = "excels"
 PARSED_DIR = "parsed"
 
 DEBUG_ONE_FAC = None #'fevt'
+DEBUG_NO_SAVE_STATES = False
 
-parser.LOGGING = LOGGING = False
+parser.LOGGING = LOGGING = True
 
 def parse_sheets(download_place):
     to_return = {}
@@ -120,6 +121,8 @@ def load_parsed_state(excel_filename):
     
 def save_parsed_state(excel_filename, obj):
     filepath = parsed_file_path(excel_filename)
+    if DEBUG_NO_SAVE_STATES:
+        print("Saved! (fake because DEBUG_NO_SAVE_STATES)")
     
     with open(filepath, "w", encoding="utf-8") as fp:
         json.dump(obj, fp=fp, ensure_ascii=False, sort_keys=True)
@@ -149,7 +152,12 @@ def run_session():
     print("main(); parse links starting...")
     EXCEL_LINKS = links_parser.parse_links(FACULTETS if DEBUG_ONE_FAC is None else [DEBUG_ONE_FAC])
     
+    if len(EXCEL_LINKS) < 5 and not DEBUG_ONE_FAC:
+        raise Exception("Safety exception: excel links count < 5; maybe in vstu.ru tech works")
+        
+    
     last_changeds = set()
+    states = []
     for excel_dict in EXCEL_LINKS:
         try:
             last_changeds.add(excel_dict['last_changed'])
@@ -179,6 +187,7 @@ def run_session():
                     try:
                         del state['excel']['different_in_this_session']
                     except: pass
+                    states.append(state)
                     save_parsed_state(excel_filename, state)
                     continue
             
@@ -195,6 +204,7 @@ def run_session():
             state['sheets'] = parse_sheets(download_place)
             
             save_parsed_state(excel_filename, state)
+            states.append(state)
             
         except Exception as e:
             faileds.append({
@@ -222,6 +232,20 @@ def run_session():
                 "all_files": EXCEL_LINKS,
                 "faileds": faileds
             }, fp=fp, ensure_ascii=False)
+        
+    with open("result_v2.json", 'w', encoding="utf-8") as fp:
+        all_files = states
+        json.dump({
+                "version": 2,
+                "notice": "ОТКАЗ ОТ ОТВЕТСТВЕННОСТИ: ПРЕДОСТАВЛЯЕТСЯ КАК-ЕСТЬ (AS-IS) БЕЗ КАКИХ ЛИБО ГАРАНТИЙ",
+                "contact": "https://fazziclay.com/ или fazziclay@gmail.com",
+                "api_notices": {
+                    "just_save_and_check_diffs": "просто сохраните и проверяйте разницу"
+                },
+                "actual_at": currt(),
+                "all_files": sorted(all_files, key=lambda d: d['excel']['url']),
+                "faileds": faileds
+            }, fp=fp, ensure_ascii=False)
 
     # Delete a non-empty directory and its contents
     try:
@@ -242,6 +266,11 @@ def main():
             print("BEGIN run_session();")
             run_session()
             print("END run_session();")
+            
+            if DEBUG_ONE_FAC:
+                print("DEBUG_ONE_FAC; break infinity-loop")
+                break
+            
         except Exception as e:
             print("Exception in run_session();")
             traceback.print_exception(e)

parser.py

@@ -98,7 +98,7 @@ class Parser:
         
         head_joined = " ||| ".join([v for v in head if isinstance(v, str) and v.strip()])
         print(head_joined)
-        if "1 неделя" in head_joined or "1 НЕДЕЛЯ" in head_joined or "2 неделя" in head_joined or "2 НЕДЕЛЯ" in head_joined or "ИЗМЕНЕНИЯ" in head_joined or "изменения" in head_joined or "vtf-vstu.ru" in head_joined:
+        if (len(head_joined) == 0) or "1 неделя" in head_joined or "1 НЕДЕЛЯ" in head_joined or "2 неделя" in head_joined or "2 НЕДЕЛЯ" in head_joined or "ИЗМЕНЕНИЯ" in head_joined or "изменения" in head_joined or "vtf-vstu.ru" in head_joined:
             head_rx -= 1
             self.raw_no_schedule.append(head_joined)
             head = self.reader.get_row_values(head_rx) # get all ROW (months, groups)