DGrady

% \labeledrule[width]{Top}{Bottom}
% 
% Useful for creating forms. The text above the rule is aligned to the
% surrounding baseline. A zero-width box contains the top label. The second
% command takes the bottom label and lowers it by \baselineskip, then packs it
% into a box with width zero, height of 1 em, and depth of \baselineskip plus
% the depth of the bottom label text. The height of the rule (0.4pt) is TeX's
% hard-coded height for \hrule.

\newcommand{\labeledrule}[3][\linewidth]{%

DGrady

// Automatically hide the toolbar and header in Jupyter Notebook 4.1.0
// This should go in ~/.jupyter/custom/custom.js
require(
    ['base/js/namespace', 'base/js/events'],
    function(Jupyter, events) {
        events.on("notebook_loaded.Notebook", function () {
            Jupyter.toolbar.actions.call('jupyter-notebook:toggle-toolbar')
            Jupyter.toolbar.actions.call('jupyter-notebook:toggle-header')
        })
    }

DGrady

function abspath() {
    # generate absolute path from relative path
    # $1     : relative filename
    # return : absolute path
    # From http://stackoverflow.com/a/23002317/514210
    if [[ -d "$1" ]]; then
        # dir
        (cd "$1"; pwd)
    elif [[ -f "$1" ]]; then
        # file

DGrady

# from operator import (
from cytoolz.curried.operator import (
    add,
    and_,
    attrgetter,
    # concat,
    contains,
    countOf,
    delitem,
    eq,

DGrady

def extent(collection):
    """
    Return the minimum and maximum values from an iterable in one pass
    """
    
    itr = iter(collection)
    
    first_value = next(itr)
    minimum = first_value
    maximum = first_value

DGrady

import pandas as pd

def describe_population(df: pd.DataFrame) -> pd.DataFrame:
    """
    Report the populated and uniqueness counts for each column of the input.
    
    The ratio columns are given as percents.
    """

    N = len(df)

DGrady

# See also https://github.com/minrk/findspark

import os
from pathlib import Path
import sys

def configure_spark(spark_home: str = None, python_path: str = None):
    
    if not spark_home:
        spark_home = os.environ['SPARK_HOME']

DGrady

"""
Problem: provide two-way communication with a subprocess in Python.

See also: 
- https://kevinmccarthy.org/2016/07/25/streaming-subprocess-stdin-and-stdout-with-asyncio-in-python/
- http://eli.thegreenplace.net/2017/interacting-with-a-long-running-child-process-in-python/
"""

import asyncio
import sys

DGrady

"""
Most of the time, it’s pretty easy to take a Pandas data frame that contains
timestamps and plot it, for example using `df.plot(…)`. Nowadays many of
high-level Matplotlib functions are canny enough to handle Pandas timestamps on
their own as well.

But not all of them.

So how do you get nice handling of Pandas timestamps when you need to use
Matplotlib directly?

DGrady

Fonts for nerds

One of the things you end up with when you spend too much time reading Hacker News is a folder of very slick monospaced fonts designed for code editors. Are any of these fonts measurably better than whatever’s already installed on your system? Nope! Here’s my list.

Spark by After the Flood

This one is kind of a gimmick, but an incredibly clever one. It translates sequences of characters like 123{30,60,90}456 into spark lines, using some fancy features of the OTF format. See also their source code repository for the project. I haven’t used this nearly enough to tell if it works well in practice, but I will now be on the constant lookout for use cases.

Consolas by Luc(as) de Groot for Microsoft