Note: Python functions and C++ extension functions (_bacon) are merged into one module.
-
Python 3. I am using 3.5, should probably work for other minor versions as well.
- Python 2 seems to work as well, at least for the core functionality (sessions/strategies), but I do not guarentee this
-
A modern enough C++ compiler supporting at least C++11
- These should be available with the OS typically
- On Windows I recommend using MSYS2, or you can use Visual Studio
- You may also need to install
libpython3,libpython3-dev, I am not sure
git clone https://github.com/sxyu/bacon2
cd bacon2
pip3 install .To use CMake, do as usual:
mkdir build && cd build
cmake .. -DCMAKE_BUILD_TYPE=Release
make -j4By default test building is enabled. To run the tests, use: ./bacon_test inside the build directory.
CMake configuration to setup the Python package is included but not required.
-
Rename
setup.pyto something else, then renamesetup.cmake.pytosetup.py -
pip3 install .
First time:
import bacon
sess = bacon.Session('fall_20XX')
oauth = bacon.ok.OAuthSession(session = sess)
# will open browser to authorize the first time
oauth.sync(assignment='cal/cs61a/faXX/proj01contest')
sess.run()
bacon.html.render(session=sess, template_path='hog.template.html', output_path='index.html')
# Have some script copy index.html to leaderboard serverLater, after reopening Python, everything is saved, and some of the options may be omitted:
import bacon
sess = bacon.Session('fall_20XX')
oauth = bacon.ok.OAuthSession(session = sess)
# should auto-refresh the token unless it has been > 8 hours
oauth.sync()
sess.run()
bacon.html.render(session=sess)Bacon will try to reused cached results and skip converting strategies whenever possible.
Launch Python 3 and import bacon
sess = bacon.Session() create a transient session
This type of session will be
deleted when you exit Python.
sess = bacon.Session('x') create a persistent session.
This session is automatically saved and
if you use the same session name after
reopening Python, everything including
strategies and results will continue
to exist.
bacon.sessions() list all persistent session names.
sess.add('id'[,'name']) add a strategy with id and display name
(display name is shown on scoreboard,
id is usually student's email)
if name is not specified then the id is used
as the name.
sess.add('id', 'name', x) (overflow from above) add a strategy
with id and display name
and initialize all rolls to x (default
is 0)
sess.add_random() add a strategy with each roll number
chosen uniformly at random
sess.add(Strategy) add a strategy object directly
sess.remove('id') remove a strategy with id
sess.remove(Strategy) remove a strategy object
sess.clear() clear all strategies
sess.clear_results() clear results (cache), forcing total re-evaluation next time run() is called
sess.unlink() 'unlinks' a session, deleting persistence
files and converting to transient session
which will be deleted when you exit Python
sess.ids() list all strategy ids
sess.names() list all strategy names
len(sess) number of strategies
Strategy(id[, name]) create an empty strategy with id and name
sess.strategies() get a list of Strategy objects in session
strat = sess['id'] get Strategy object by id from session
xgfzg = sess.by_name('x') get Strategy object by name from session;
if multiple strategies with name exist,
returns one with lexographically least id.
Very inefficient compared to get by id.
strat[our, opponent] get roll number in strategy
strat[our, opponent] = x set roll number in strategy
strat.set_const(i) set all rolls to a constant
strat.set_random() set all rolls randomly
strat.array() get numpy array of roll numbers (int8)
strat.set_array() set roll numbers from numpy array (must be int8)
strat.name get/set strategy name
strat.id get strategy id (immutable)
strat.win_rate(oppo) alt method to compute win rates
strat.draw() draw a strategy diagram identical to that in
the original Bacon
strat.win_rate_by_sampling(oppo[, num_samples = 10000])
compute win rate by sampling num_samples*2 games
(num_samples as player0/1 each)
sess.run([n_threads]) runs contest (tries to reuse old results).
By default uses n_threads=# cores.
Returns a Results object (which if not
stored has a repr that will print
out the rankings table).
res = sess.results() get the latest results objects. If you
have not ran the contest yet, errors.
sess.has_results() checks whether you have results.
res prints contest rankings incl. name, wins
str(res) gives more concise version of above
res.rankings returns a rankings list, which is a list
of tuples: (strategy index in contest, wins)
res.ids() get list of strategy ids in contest
res.names() get list of strategy names in contest
res[i] get the internal 'half-row' of win rates
for player i (representing the triangular
matrix) this row will have i entries and
allows you to use res[i][j] for i > j
res[i, j] get win rate of ith player in contest
playing against jth player (integers,
order is same as in res.list())
res.array() get numpy array of win rates:
arr[first_player, second_player]
To access session config:
sess.config()
This mostly behaves like a dict but only takes string keys and values. It is serialized with the rest of the session's state. This is used to store data about e.g. OK.
The bacon.config module contains game constants and functions from config.hpp and config.cpp. These are immutable in Python.
bacon.config.DICE_SIDES sides of dice, usally 6
bacon.config.GOAL goal score, usually 100
bacon.config.MIN_ROLLS min number of dice rolls in a turn
bacon.config.MAX_ROLLS max number of dice rolls in a turn
bacon.config.MOD_TROT time trot modulus
bacon.config.FERAL_HOGS_ABSDIFF feral hogs absolute difference
bacon.config.ENABLE_TIME_TROT whether time trot is enabled
bacon.config.ENABLE_SWINE_SWAP whether swine swap is enabled
bacon.config.ENABLE_FERAL_HOGS whether feral hogs is enabled
bacon.config.WIN_EPSILON min win rate rel. to 0.5 to consider a matchup a win
bacon.config.swine_swap(a, b) checks if two scores should result in
a swine swap
bacon.config.free_bacon(a) gets score obtainable through free bacon
- To render the HTML leaderboard (pretty hacky)
bacon.html.render(res, 'hog.template.html', 'output.html')
OR
bacon.html.render(session=sess, template_path='hog.template.html', output_path='output.html')
The second form allows you to use just
bacon.html.render(session=sess)
after the first call.
- To recursively converts all hog_contest.py in some directories returning a list of Strategy objects:
bacon.io.convert(paths...)
(kwargs: source_name_suffix: only converts file names with given suffix, verbose: whether to print verbosely)
- To sync a directory, i.e. convert only changed files (previously converted files are saved in the session config and compared)
bacon.io.sync_dir(session, dir_path[, source_name_suffix[, verbose]])
- Minor IO utils
bacon.io.write_py(Strategy, path) create valid hog_contest.py script
returning roll numbers for the
strategy (similar to hogmat)
bacon.io.write_legacy(Strategy, path) write legacy Bacon .strat format
strat = bacon.io.read_legacy(path) read legacy Bacon .strat format
- OK integration
oauth = bacon.ok.OAuthSession() # temporary session
oauth = bacon.ok.OAuthSession(session = bacon_session) # persists with Session
oauth.auth() # returns token, refreshes automaticallyTo download submissions:
oauth.download("hctest", "cal/cs61a/fa18/proj01contest"[, "hog_contest.py"])
(Arguments are: output directory, assignment name, name of file to download from the assignment, default "hog_contest.py")
To sync submissions directly from OK to session in one go (must create session in the second way: bacon.ok.OAuthSession(session)):
oauth.sync(assignment = "cal/cs61a/fa18/proj01contest")
assignment is optional after the first sync.
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent