process_retrosheet.py

import csv
import json
from collections import defaultdict

datafiles = ['./GL{0}.TXT'.format(x) for x in range(1960,2014)]

counts = defaultdict(lambda: defaultdict(lambda: defaultdict(int)))

max_i = 0

def process_linescore(s):
    results = []
    while s:
        s0 = s[0]
        if s0 == "(":
            results.append(int(s[1:3]))
            s = s[4:]
        else:
            if s0 == 'x':
                results.append(0)
            else:
                results.append(int(s0))
            s = s[1:]
    return results

for df in datafiles:
    with open(df, 'rb') as datafile:
        reader = csv.reader(datafile, delimiter=',')
        for line in reader:
            visitor_score = int(line[9])
            home_score = int(line[10])
            if visitor_score == home_score:
                result = 't'
            elif visitor_score > home_score:
                result = 'v'
            else:
                result = 'h'

            visitor_linescore = process_linescore(line[19])
            home_linescore = process_linescore(line[20])
            if len(visitor_linescore) != len(home_linescore):
                continue  # sanity check
            score_diff = 0
            for idx in range(0, len(visitor_linescore)):
                for bottom in (False, True):
                    if bottom:
                        score_diff += home_linescore[idx]
                    else:
                        score_diff -= visitor_linescore[idx]
                    effective_diff = max(min(score_diff, 10), -10) + 10
                    inning_num = idx * 2 + int(bottom)
                    counts[inning_num][effective_diff][result] += 1
print(json.dumps(counts))