def of_tab(self, tabfile):
"""
Initialize `stats_by_config` table from the spreadsheet `tabfile`
"""
util.fold_file(
tabfile, None, lambda acc, row: self.set_stats(row[0], util.stats_of_row([int(v) for v in row[1:]]))
)
def all_cells_matching(tabfile, config_pred):
"""
Return a list of all data cells in rows whose titles
match `config_pred`.
"""
def process_row(row):
title = row[0]
if config_pred(title):
return [int(x) for x in row[1::]]
else:
return []
return util.fold_file(tabfile, [], lambda acc,row: acc + process_row(row))
def bucketize(fname, get_bkt, num_buckets, pred=None):
"""
Group the rows in `fname` into buckets,
using `get_bkt` to assign a row of the file to a bucket.
Return the collected list of lists
(todo? optionally allow 'edit_row' function,
instead of mushing all values into a list)
"""
init = [[] for _ in range(1+num_buckets)]
def add_row(acc, row):
# TODO don't graph all points because ... (I had a reason at some point)
if (pred is None) or (pred(row[0])):
acc[get_bkt(row[0])].extend([int(x) for x in row[1::]])
return acc
return util.fold_file(fname, init, add_row)
def best_rows(tabfile, limit, metric, val_of_row, init=None):
""" (->* (Path-String (-> A A Boolean) (-> (Listof Nat) A)) (A) A)
Return the "value" of the best row, according to `metric`.
i.e. the row R such that for any other row r: metric(val_of_row(r), val_of_row(R)) = True
- `tabfile` filename, a tab-separated file
- `metric` a < judgment on rows
- `val_of_row` a pre-processing transformation on rows (used to filter, take mean, etc)
- `init` optionally, a first value to compare to
"""
cache = [None] * limit
cache[0] = init
def process_row(acc, row):
# Get value from row.
# Insert into the list `acc`
if config.is_untyped(row[0]) or config.is_typed(row[0]):
# Ignore fully typed / untyped configs
return acc
val = val_of_row(row)
return sorted_buffer.insert(acc, val, metric, 0)
return util.fold_file(tabfile, cache, process_row)
def results_of_config(self, config):
return util.fold_file(
self.source,
None,
lambda acc, row: acc or (util.stats_of_row([int(x) for x in row[1:]]) if row[0] == config else None),
)