Dockerfile.build_db

FROM amazonlinux:2023

RUN yum update -y && \
    yum upgrade -y && \
    yum install -y \
        git \ 
        bzip2 \
        python3 \
        tar \
        less \
        sqlite \
        wget \
        which && \
        yum clean all

# download retrosheet files
RUN wget https://www.retrosheet.org/alldata.zip
    
# make directories, run scripts to extract data from retrosheet files
# into Sqlite3 DB
RUN mkdir /baseball/alldata && \
    mkdir /baseball && \
    mkdir /baseball/webapp && \
    mkdir /baseball/vite-baseball-stats && \
    unzip -d /baseball/alldata alldata.zip

# copy files
# don't use bulk copy to avoid copying over junk/temp files
# use extraneous ARG to ignore cache after this point (force recopy)
# change ARG at command line to forcd the rebuild
ARG UPDATE_BUILD
COPY ./*.py /baseball

RUN cd /baseball && \
    ./retrosheet_init_db.py && \
    ./retrosheet_data_to_sql.py && \
    ./retrosheet_boxscores_to_sql.py

RUN cd /baseball && \
    ./retrosheet_parse_event_outcomes.py 0 25000 && \
    ./retrosheet_parse_event_outcomes.py 25000 50000 && \
    ./retrosheet_parse_event_outcomes.py 50000 75000 && \    
    ./retrosheet_parse_event_outcomes.py 75000 100000 && \
    ./retrosheet_parse_event_outcomes.py 100000 120000 && \
    ./retrosheet_parse_event_outcomes.py 120000 140000 && \
    ./retrosheet_parse_event_outcomes.py 140000 150000 && \
    ./retrosheet_parse_event_outcomes.py 150000 160000 && \
    ./retrosheet_parse_event_outcomes.py 170000 180000 && \
    ./retrosheet_parse_event_outcomes.py 180000 190000 && \
    ./retrosheet_parse_event_outcomes.py 190000 200000 && \
    sqlite3 baseball.db < create_situation_indexes.sql && \
    bzip2 baseball.db && \
    rm -r alldata