blaseball-streak-finder

This repo contains a command line tool called streak-finder that helps find blaseball winning and losing streaks.

You can use command line flags or configuration options to filter on conditions, including:

• Season
• At home or away
• Streaks versus a team, division, or league
• Streaks versus a pitcher

This allows you to find not just winning and losing streaks, but also winning and losing streaks against a set of opponents.

Table of Contents

• Screenshots
• Installation
  • pip
  • source
• Quick Start
  • Command line flags
  • Configuration file
• Data
• Configuration Examples
• Scripts
• Software architecture
• Who is this tool for?
• Future work
  • Libraries used

Screenshots

The streak-finder tool can print plain text tables to the console, or output tables in HTML or Markdown format. Here are a few examples:

Look for winning streaks from the Tigers from Season 3 and 4,

Compare to winning streaks by the Pies at the same:

Show the top winning streaks in Season 1 for the Good League:

Compare to top winning streaks in Season 1 for the Evil League:

Show game-by-game summaries of the worst losing streaks in blaseball history:

This tool also works on antique hardware!

The tool also offers the ability to export tables in HTML format, in both short and long versions:

Installation

pip

pip install git+https://github.com/ch4zm/blaseball-core-game-data.git#egg=blaseball_core_game_data
pip install git+https://github.com/ch4zm/blaseball-streak-finder.git#egg=blaseball_streak_finder

source

Start by cloning the package:

git clone https://github.com/ch4zm/blaseball-streak-finder
cd blaseball-streak-finder

If installing from source, it is recommended you install the package into a virtual environment. For example:

virtualenv vp
source vp/bin/activate

Now build and install the package:

python setup.py build
python setup.py install

Now test that the tool is available on the command line, and try out a command to search for some streaks:

which streak-finder
streak-finder --min 8 --short

Quick Start

The way this tool works is, it creates a data frame object, applies some filters to it based on command line flags provided by the user, then runs the data frame through a data viewer (which makes the nice tables). All command line flags can also be specified in a config file.

Command line flags

Command line flags are grouped into data options and view options.

Data options:

• Winning or Losing Streaks: Use the --winning/--losing flags to specify winning/losing streaks

• Season: Set season for game data using --season. For multiple seasons, repeat the flag: --season 1 --season 2

• (Optional) Our Team: Specify only one of the following:

  • Team: use the --team flag to specify the short name of your team (use --help to see valid choices). For multiple teams, use multiple --team flags.
  • Division: use the --division flag to specify the name of a division. Surround division name in quotes, e.g., --division "Lawful Evil"
  • League: use the --league flag to specify the Good/Evil league

• (Optional) Versus Team: Specify only one of the following:

  • Versus Team: use the --versus-team flag to specify the short name of the opposing team (use --help to see valid choices). For multiple teams, use multiple --versus-team flags.
  • Versus Division: use the --versus-division flag to specify the name of the versus division. Surround division name in quotes, e.g., --versus-division "Lawful Evil"
  • Versus League: use the --versus-league flag to specify the versus Good/Evil league

(If neither flag is specified, it will include all games between all teams.)

View options:

• Minimum: Specify the minimum number of wins or losses to qualify as a streak with --min N

• HTML: Use --html to specify that the output should be in HTML table format. If no --output file is specified, it will print the HTML to stdout.

• Markdown: Use --markdown to specify that the output should be in Markdown table format. If no --output file is specified, it will print the Markdown to stdout.

• Output File: Use --output to specify the output file when using the --html or --markdown flags. This flag has no effect when --html or --markdown are not present.

• Use Short Output: Use --short to display streaks in short format (one line per streak; default option).

• Use Long Output: Use --long to display streaks in long format (one table per streak summarizing each game in the streak).

• Use Nicknames: Use --nickname flag to use team nicknames in table (e.g., Sunbeams)

• Use Full Names: Use --fullname flag to use full team name in table (e.g., Hellmouth Sunbeams)

Using a configuration file:

• Config file: use the -c or --config file to point to a configuration file (see next section).

Configuration file

(Note: several configuration file examples are provided in a section below.)

Every command line flag can be specified in a configuration file as well. To reproduce the following command line call,

streak-finder --season 1 --season 2 --team Tigers --versus-team Pies --min 5 --fullname 

we could create a configuration file named config.ini with the contents:

season = [1, 2]
team = Tigers
versus-team = Pies
min = 5
fullname = True

and run blaseball-streak-finder specifying that configuration file:

streak-finder --config config.ini
# or
streak-finder -c config.ini

This would produce identical output to the command with all the flags.

You can also use both a config file and command line flags; the command line flags will take precedence over config file options if a parameter is specified by both.

Data

The data set used by this tool comes from blaseball.com's /games API endpoint. The data set is imported from blaseball-core-game-data.

Configuration Examples

See config.example.ini in the repo for an example config file.

Show winning streaks from season 3 and season 4 for the Hades Tigers:

[data]
season = [3, 4]
team = Tigers
winning
min = 5

Include winning streaks by the Pies too:

[data]
season = [3, 4]
team = [Tigers, Pies]
winning
min = 5

Show the top winning streaks from season 1 for all teams in the Good League:

[data]
season = 1
league = Good
winning
min = 7

Compare to top winning streaks from season 1 for all teams in the Evil League:

[data]
season = 1
league = Evil
winning
min = 7

Look for losing streaks that were 12 games or more, and print a game-by-game summary of the streaks using the long format:

[data]
min = 12
losing
long

Look for winning streaks by the Firefighters that were 7 games or more, and output the summary table to the file short.html:

[data]
team = Firefighters
min = 7
winning
short
html
output = short.html

Repeat the above streak-finding search, but output a table for each streak with a game-by-game breakdown (long format) to the file long.html:

[data]
team = Firefighters
min = 7
winning
long
html
output = long.html

Repeat the above streak-finding search, but output a Markdown table for each streak with a game-by-game breakdown (long format) to the file long.md:

[data]
team = Firefighters
min = 7
winning
long
markdown
output = long.md

Python API

If you prefer to call this tool from Python directly, rather than from the command line, you can import and call the streak_finder function and pass it a list of strings containing the flags you would normally pass on the command line.

from streak_finder.command import streak_finder

flags = "--winning --team Millennials --versus-team Flowers --fullname"
output = streak_finder(flags.split(" "))
game_ids = [j.strip() for j in output.split("\n")]
print(game_ids)

If you prefer to store the JSON game object returned by game-finder as a native Python dictionary, you can use json.loads():

from game_finder.command import game_finder
import json

flags = "--json --season 1 --season 2 --team Tigers --versus-team Pies"
output = game_finder(flags.split(" "))
json_data = json.loads(output)
print(json_data)

Software architecture

This software consists of three parts:

• The command line flag and config file parser (uses configargparse library) - see cli/command.py
• The StreakData object that stores the game data in a Pandas data frame (uses pandas library) - see cli/streak_data.py
• The View object that provides a presentation layer on top of the Pandas data frame (uses panda's DataFrame.to_string() and DataFrame.to_html() methods to print the data) - see cli/view.py (there are two classes, one for plain text and one for HTML)

Who is this tool for?

This tool is for the blaseball community. It will be useful to people interested in exploring game data, people who are brainstorming about lore for their team, and people who are looking for a starting point for developing their own blaseball tool.

Future work

• Add pitcher filter

Libraries used

This command line tool uses the following libraries under the hood:

• blaseball-core-game-data
• pandas for organizing/filtering data
• configarparse for handling CLI arguments