From ad7fbcc364005f605f75fbb1149a0df71b0bc4b1 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Maximilian=20Ke=C3=9Fler?= <git@maximilian-kessler.de>
Date: Sun, 12 Nov 2023 00:01:22 +0100
Subject: [PATCH] improved parsing of endgame data

---
 endgames.py  | 41 ++++++++++++++++++++++++++++-------------
 get_sheet.py | 11 ++++++++---
 2 files changed, 36 insertions(+), 16 deletions(-)

diff --git a/endgames.py b/endgames.py
index f667517..25748f5 100644
--- a/endgames.py
+++ b/endgames.py
@@ -16,8 +16,6 @@ with open(DATA_FILE, 'r') as f:
 
 def analyze_game(game_id: int):
     max_draw_pile_size = 15
-    probabilities = {}
-    raw_output = None
     try:
         result = subprocess.run(['./endgame-analyzer', '-g', str(game_id), '-d', str(max_draw_pile_size), '--interactive', '0', '--quiet', '-r'], stdout=subprocess.PIPE, timeout=10)
         raw_output = result.stdout
@@ -27,13 +25,17 @@ def analyze_game(game_id: int):
 
     print(output)
 
-    # Check if the game was just over before reaching the specified draw pile size
-    if re.search(r'The given draw pile size \({}\) cannot be obtained with the specified replay.'.format(max_draw_pile_size), output):
-        print('detected empty output')
-        return {str(size): 0 for size in range(1, max_draw_pile_size + 1)}
-
     # Now, parse all results that we obtained (unclear how many depending on whether we ran into the timeout)
-    x =                  'Probability with 13       cards left in deck: 5855/6237 ~ 93.875%'
+    probabilities = {}
+
+    # Check if the game was just over before reaching the specified draw pile size
+    for m in re.finditer(r'The given draw pile size \(([0-9]+)\) cannot be obtained with the specified replay.', output):
+        if m.group(1) == str(max_draw_pile_size):
+            print('detected empty output')
+            return {str(size): 0 for size in range(1, max_draw_pile_size + 1)}
+        else:
+            probabilities[str(m.group(1))] = 0
+
     for m in re.finditer('Probability with ([0-9]+) cards left in deck: .*/.* ~ ([0-9.]+)', output):
         probabilities[str(m.group(1))] = m.group(2)
 
@@ -41,13 +43,26 @@ def analyze_game(game_id: int):
 
 
 def full_analyze_game(game_id: int):
-    probabilities = {}
+    max_draw_pile_size = 10
     try:
-        result = subprocess.run(['./endgame-analyzer', '-g', str(game_id), '-d', str(10), '-i', '0', '--all-clues', '-r', '--quiet'], stdout=subprocess.PIPE, timeout=180)
-    except subprocess.TimeoutExpired:
-        return probabilities
-    output = result.stdout.decode('utf-8')
+        result = subprocess.run(['./endgame-analyzer', '-g', str(game_id), '-d', str(max_draw_pile_size), '-i', '0', '--all-clues', '-r', '--quiet'], stdout=subprocess.PIPE, timeout=180)
+        raw_output = result.stdout
+    except subprocess.TimeoutExpired as time_err:
+        raw_output = time_err.stdout
+    output = raw_output.decode('utf-8')
+
     print(output)
+
+    probabilities = {}
+    zero_dict = {
+        (('+' if clue_modifier >= 0 else '') + str(clue_modifier)): 0 for clue_modifier in range(-8, 9)
+    }
+    for m in re.finditer(r'The given draw pile size \(([0-9]+)\) cannot be obtained with the specified replay.', output):
+        if m.group(1) == str(max_draw_pile_size):
+            return {str(size): zero_dict for size in range(1, max_draw_pile_size + 1)}
+        else:
+            probabilities[str(m.group(1))] = zero_dict
+
     for m in re.finditer('Probability with ([0-9]+) cards left in deck and [0-8] clues \((.[0-8])\).*: .*/.* ~ ([0-9.]*)', output):
         if m.group(1) not in probabilities.keys():
             probabilities[m.group(1)] = {}
diff --git a/get_sheet.py b/get_sheet.py
index ca95634..6c51136 100644
--- a/get_sheet.py
+++ b/get_sheet.py
@@ -14,7 +14,7 @@ from endgames import analyze_game_cached, full_analyze_game_cached
 global_db_connection_manager.read_config()
 global_db_connection_manager.connect()
 
-session = requests_cache.CachedSession('.hanab-live.cache',expire_after=300)
+session = requests_cache.CachedSession('.hanab-live.cache',expire_after=30000)
 
 player_mapping = {
         'RamaNoVarjan': 'Ramanujan',
@@ -133,7 +133,8 @@ def sort_players_by_num_games(games_dict):
     return sorted(player_cols, key = lambda col: -nums[col])
 
 
-def lookup_val(endgame_dict, clue_modifier):
+def lookup_val(endgame_dict, clue_modifier) -> str:
+    print('looking up val {} in {}'.format(clue_modifier, endgame_dict))
     if clue_modifier > 0:
         for lookup in range(clue_modifier, 0, -1):
             val = endgame_dict.get('+' + str(lookup), None)
@@ -144,7 +145,8 @@ def lookup_val(endgame_dict, clue_modifier):
             val = endgame_dict.get(str(lookup))
             if val is not None:
                 return val
-    return endgame_dict.get('+0', None)
+    retval = endgame_dict.get('+0', None)
+    return retval
 
 
 if __name__ == "__main__":
@@ -197,6 +199,7 @@ if __name__ == "__main__":
             writer.writerow(endgame)
 
     all_endgames = full_analyze_endgames(games)
+    print(all_endgames.keys())
     fieldnames = ['Game ID'] + [str(i) for i in range(1, 11)]
     for clue_modifier in range(-2, 3):
         filename = 'endgames{}.csv'.format(clue_modifier)
@@ -205,9 +208,11 @@ if __name__ == "__main__":
         with open(filename, 'a') as f:
             writer = csv.DictWriter(f, fieldnames=fieldnames)
             for game_id, endgame in sorted(all_endgames.items()):
+                print(endgame)
                 row = {'Game ID': game_id}
                 for deck_size in range(1, 11):
                     val = lookup_val(endgame.get(str(deck_size), {}), clue_modifier)
+                    print('looked up val {}'.format(val))
                     if val is not None:
                         row[str(deck_size)] = val
                     else: