Retail draft word counts

Fixed issues with some sets, added more set codes for draftable sets,
added CSV output file.

retail_set_word_counts.py

@@ -1,11 +1,9 @@
 import json
 import statistics
 
-
 # Counts words on each card from each set using MTGJSON data.
 # I downloaded "all set files" from https://mtgjson.com/downloads/all-files/
 
-
 def load_set(code):
     with open(f'mtgjson/sets/{code}.json', encoding='utf8') as f:
         data = json.load(f)
@@ -14,14 +12,15 @@ def load_set(code):
     cards = data['cards']
 
 
-def wc_fo(text):
-    # full oracle text word count
-    return len(text.split())
-
-
 def analyze_set(code):
     d = load_set(code)
+    if 'data' in d:
+        d = d['data']
+    if 'cards' in d:
         cards = d['cards']
+    else:
+        print(f'No card data found in {code}.json')
+        return
     faces = {}
     #list text for all cards
     for card in cards:
@@ -33,7 +32,11 @@ def analyze_set(code):
     for card in cards:
         if 'side' in card:
             if card['side'] == 'a':
+                if 'text' not in card:
+                    card['text'] = ""
+                elif card['otherFaceIds']:
                     card['text'] += " " + faces[card['otherFaceIds'][0]]
+
         # make card number an int
         if 'number' in card:
             try:
@@ -80,14 +83,22 @@ def analyze_set(code):
     return textline
 
 
-codes = ['LEA','3ED','4ED','5ED','6ED','7ED','8ED','9ED','10E','M10','M11','M12','M13','M14','M15','ORI','M19','M20','M21','ARN','ATQ','LEG','DRK','FEM','ICE','HML','ALL','MIR','VIS','WTH','TMP','STH','EXO','USG','ULG','UDS','MMQ','NEM','PCY','INV','PLS','APC','ODY','TOR','JUD','ONS','LGN','SCG','MRD','DST','5DN','CHK','BOK','SOK','RAV','GPT','DIS','CSP','TSP','PLC','FUT','LRW','MOR','SHM','EVE','ALA','CON','ARB','ZEN','WWK','ROE','SOM','MBS','NPH','ISD','DKA','AVR','RTR','GTC','DGM','THS','BNG','JOU','KTK','FRF','DTK','BFZ','OGW','SOI','EMN','KLD','AER','AKH','HOU','XLN','RIX','DOM','GRN','RNA','WAR','ELD','THB','IKO','ZNR']
+codes = []
-
+codes += ['LEA','3ED','4ED','5ED','6ED','7ED','8ED','9ED','10E','M10','M11','M12','M13','M14','M15','ORI','M19','M20',
+         'M21']
+codes += ['POR','P02','PTK','S99']
+codes += ['ARN','ATQ','LEG','DRK','FEM','HML']
+codes += ['ICE','ALL','MIR','VIS','WTH','TMP','STH','EXO','USG','ULG','UDS','MMQ','NEM','PCY','INV','PLS','APC',
+          'ODY','TOR','JUD','ONS','LGN','SCG','MRD','DST','5DN','CHK','BOK','SOK','RAV','GPT','DIS','CSP',
+          'TSP','PLC','FUT','LRW','MOR','SHM','EVE','ALA','CON_','ARB','ZEN','WWK','ROE','SOM','MBS','NPH',
+          'ISD','DKA','AVR','RTR','GTC','DGM','THS','BNG','JOU','KTK','FRF','DTK','BFZ','OGW','SOI','EMN','KLD','AER',
+          'AKH','HOU','XLN','RIX','DOM','GRN','RNA','WAR','ELD','THB','IKO','ZNR']
+codes += ['MMA','MM2','EMA','MM3','IMA','A25','UMA','2XM']
+codes += ['MH1','CMR','JMP']
 
 output = open('retail_set_words.csv', 'w+')
+output.write('Set,c (mean),c (med),u (mean),u (med),r (mean),r (med),m (mean),m (med)\n')
 for code in codes:
-    # These sets trigger errors due to meld, MDFC, and other issues. Fixes needed.
-    if code in ['CON', 'SOI', 'EMN', 'ELD', 'ZNR']:
-        continue
     line = analyze_set(code)
     output.write(line + '\n')
 output.close()

retail_set_words.csv

@@ -0,0 +1,120 @@
+Set,c (mean),c (med),u (mean),u (med),r (mean),r (med),m (mean),m (med)
+LEA,14.7,10.0,16.1,14,22.2,17.5,0,0
+3ED,14.9,10,15.4,14,21.3,17,0,0
+4ED,14.8,10,16.2,14,25.2,20.5,0,0
+5ED,15.5,14,17.8,15.5,23.8,20,0,0
+6ED,10.4,9.5,12.4,11.5,18.5,15.0,0,0
+7ED,10.2,9.0,11.4,10.5,14.3,14.0,0,0
+8ED,9.8,9,11.9,11,16.5,14,0,0
+9ED,10.5,9.0,12.5,11.0,18.5,15,0,0
+10E,13.6,12,17.2,14.0,20.5,18,0,0
+M10,10.0,9,14.8,12.0,21.1,19,32.3,33
+M11,11.7,10,14.5,12.5,23.9,24,26.9,26
+M12,11.8,11,17.5,18.0,25.0,22,32.9,32
+M13,11.6,11,21.6,18.0,22.2,19,34.1,36
+M14,12.2,11,16.9,17.0,25.9,25,37.5,37
+M15,16.8,15,22.2,19.0,30.5,32,41.9,40
+ORI,19.7,17,25.2,22.5,29.3,29,58.4,47.5
+M19,11.3,11,19.4,18.0,26.1,24,39.9,39.5
+M20,15.2,15.0,23.4,21,31.8,32,47.5,51
+M21,16.8,16,24.3,23.0,32.5,34,46.3,50.5
+POR,6.6,8.0,10.4,10,15.2,13,0,0
+P02,7.1,8.0,12.8,12,13.7,14,0,0
+PTK,8.0,9,14.0,12,18.0,17,0,0
+S99,5.9,4,9.4,10,12.5,11,0,0
+ARN,21.3,14.0,21.7,19,24.0,20.5,0,0
+ATQ,17.2,16.0,21.3,18,33.1,25.0,0,0
+LEG,16.9,13,21.5,17.5,25.4,20,0,0
+DRK,14.5,11,21.3,17.5,28.7,27,0,0
+FEM,23.1,19.5,29.8,27,28.9,23.5,0,0
+HML,17.6,12,18.6,15,25.2,22.0,0,0
+ICE,21.3,17.0,23.5,19,33.2,33,0,0
+ALL,20.7,19.0,31.0,29,39.9,41.0,0,0
+MIR,22.3,21,20.0,19.0,29.4,25.5,0,0
+VIS,19.7,18.5,25.7,22,28.8,26.0,0,0
+WTH,18.4,16.0,21.2,17,26.4,22.5,0,0
+TMP,14.3,13.0,17.2,14.0,22.6,19.0,0,0
+STH,14.1,11,21.9,20.5,23.5,22.0,0,0
+EXO,16.6,15,20.7,18.5,26.4,25.0,0,0
+USG,15.8,14.0,17.3,15.0,22.3,21.0,0,0
+ULG,16.6,14,20.5,15.5,22.9,22.0,0,0
+UDS,13.6,11,22.3,22.5,18.5,16.0,0,0
+MMQ,15.0,13.0,18.0,16.0,20.4,18.0,0,0
+NEM,19.0,15,23.5,22.5,29.1,23.0,0,0
+PCY,15.0,14,18.0,17.5,23.2,21.0,0,0
+INV,19.5,16.0,17.9,15.0,24.1,20.5,0,0
+PLS,20.3,17,24.9,22.5,27.3,25,0,0
+APC,17.8,17,27.8,28.0,26.3,17.0,0,0
+ODY,18.5,19.0,18.9,17.5,22.8,20.0,0,0
+TOR,23.0,23,25.9,26.5,26.2,26.5,0,0
+JUD,19.3,18,26.5,28.0,26.3,25.5,0,0
+ONS,20.9,18.5,23.6,21.0,23.7,23.0,0,0
+LGN,21.0,19,22.1,22,28.3,25,0,0
+SCG,23.5,23,23.2,22.0,30.0,31.0,0,0
+MRD,13.8,12.5,19.9,16.5,27.3,24.0,0,0
+DST,16.7,16,24.6,23,27.5,27,0,0
+5DN,21.6,20,18.1,17,25.4,23,0,0
+CHK,19.8,17.0,22.8,20,27.7,24.5,0,0
+BOK,23.9,18,30.4,30,30.1,28,0,0
+SOK,21.7,19,22.5,22,29.4,27,0,0
+RAV,19.9,17.0,22.4,19.0,27.9,24.0,0,0
+GPT,19.9,20,19.5,18,27.7,24,0,0
+DIS,20.2,20.5,25.7,23.0,31.2,29.0,0,0
+CSP,22.7,21.0,28.8,24,36.4,34.0,0,0
+TSP,26.5,23,29.8,30.0,35.5,33.5,0,0
+PLC,24.0,18.0,30.5,29,34.6,28.0,0,0
+FUT,29.5,26.0,32.1,28.0,31.8,27.5,0,0
+LRW,21.8,18,27.4,27.0,31.4,29.0,0,0
+MOR,21.8,18.0,35.0,36.0,30.7,28.0,0,0
+SHM,20.6,15,20.0,17.5,24.5,24,0,0
+EVE,22.6,20.0,25.4,21.5,27.0,28.0,0,0
+ALA,16.8,16,20.3,17.0,27.3,25,29.7,30
+CON,18.9,17.5,20.2,18.0,24.8,23,27.0,24.0
+ARB,24.4,24.5,23.0,18.5,26.4,21,26.6,24.0
+ZEN,20.2,19,25.0,23.5,31.6,28,31.6,28
+WWK,20.9,21.5,27.1,26.0,28.3,31,34.9,36.0
+ROE,17.2,16.5,22.3,22.5,28.1,27,33.1,39
+SOM,16.5,15,22.2,20.5,27.8,23,33.7,29
+MBS,15.6,14.0,21.5,20.5,25.6,25,30.8,26.5
+NPH,18.6,19.0,24.8,21.5,31.2,31,29.2,29.5
+ISD,16.7,15,20.7,18,26.6,23,35.1,34.5
+DKA,19.5,20.5,25.2,26.5,33.1,30.0,37.9,31.0
+AVR,18.0,15,20.4,17.0,26.9,24,31.3,33
+RTR,16.9,15,20.3,21.0,28.2,25,31.3,30
+GTC,16.8,15,22.0,20.0,30.4,29,38.3,35
+DGM,16.6,13.0,24.6,19.5,34.4,35,40.4,39
+THS,17.8,14,26.8,24.0,31.3,29,41.5,44
+BNG,19.0,16.5,27.2,25.0,34.4,35,42.6,43.5
+JOU,18.5,18.0,26.9,26.5,24.6,23,39.1,37.5
+KTK,18.5,18,22.7,21.0,27.7,25,38.7,38
+FRF,21.1,21.0,27.1,26.0,39.1,42,40.6,36.5
+DTK,20.6,18,27.8,23.5,35.1,37,39.1,40
+BFZ,19.7,20,24.9,23.5,30.7,32,37.3,36
+OGW,17.3,17.5,22.7,21.0,32.2,31.5,38.5,39.5
+SOI,19.4,19,27.6,24.5,34.4,31.5,42.9,42.0
+EMN,18.4,18,27.7,26.5,32.0,29.5,44.9,42
+KLD,18.2,18.0,20.4,20.0,32.2,31,44.1,46
+AER,18.8,19,25.4,24.0,29.6,28.5,38.0,37.0
+AKH,18.1,18,25.2,25.5,30.7,29,42.0,39
+HOU,17.7,17.5,26.8,25.5,31.9,30.5,43.1,43.5
+XLN,16.3,16,20.5,19.5,32.2,28,29.3,35
+RIX,14.5,13.5,23.5,20.0,35.2,33.5,41.1,39
+DOM,18.2,17,24.9,21.0,29.4,26,40.5,39
+GRN,17.6,13,26.1,24.0,30.3,30,44.7,41
+RNA,15.8,14,24.9,24.5,29.6,28,49.1,46
+WAR,17.1,16,27.7,28.0,41.5,45.0,48.3,49.0
+ELD,21.5,22,26.0,25.0,32.9,32,42.5,45
+THB,18.2,18,32.1,30.0,34.9,33,50.8,51
+IKO,18.5,18,29.7,27.0,37.0,39,55.9,55
+ZNR,22.4,21,28.9,28.0,35.3,36.0,54.4,50
+MMA,26.3,26,26.9,27.5,33.1,34,29.3,28
+MM2,21.2,20,23.7,24.0,27.6,27,31.9,30
+EMA,18.4,18,21.2,19.0,23.2,21,31.1,29
+MM3,18.4,14,18.3,14.5,27.9,25,32.1,34
+IMA,21.2,18,22.9,23.5,25.9,24,25.6,28
+A25,19.3,17,22.8,19.0,23.8,22,32.8,36
+UMA,22.3,21,24.4,24.0,28.2,24,31.6,33.0
+2XM,21.9,21,23.1,21.0,23.6,21,35.7,36.0
+MH1,25.0,26.0,24.8,22.0,31.0,27,45.7,50
+CMR,21.0,17.0,29.2,26.0,36.5,36,38.5,37.0
+JMP,15.6,14.5,18.9,16,25.6,24,32.1,35