Generate run history from materials project database

mg-run-history

## Generate graph of run history

Generating the graph requires 2 scripts

1) A Python script to extract the dates of each compound from the
database. Replace DB, USER and PASSWORD below with appropriate values:

    ./mg_run_history.py DB USER PASSWORD --port 27018 > runs.csv


2) An R program to draw the graph:

    ./mg_run_history.r runs.csv runs.pdf

The resulting output is in runs.pdf.
It will be an area chart of the runs, with a loess smooth line added.

mg_run_history.py

#!/usr/bin/env python
"""
Fetch and write run history
"""
__author__ = 'Dan Gunter <dkgunter@lbl.gov>'
__date__ = '3/18/13'

import argparse
from datetime import datetime
import sys
import pymongo
import time

def write_dates(coll, ofile):
    ofile.write('completed_at\n')
    for rec in coll.find({},{'completed_at':1}):
        if not 'completed_at' in rec:
            continue
        s = rec['completed_at']
        if '.' in s:
            s = s.split('.')[0]
        date = datetime.strptime(s, '%Y-%m-%d %H:%M:%S')
        sec = int(time.mktime(date.utctimetuple()))
        ofile.write('{:d}\n'.format(sec))

def main(cmdline=sys.argv[1:]):
    retcode = 0
    parser = argparse.ArgumentParser()
    parser.add_argument('--host', dest='host', default='localhost')
    parser.add_argument('--port', dest='port', type=int, default=27017)
    parser.add_argument('database')
    parser.add_argument('user')
    parser.add_argument('password')
    args = parser.parse_args(cmdline)
    try:
        client = pymongo.MongoClient(host=args.host, port=args.port)
        db = client[args.database]
        db.authenticate(args.user, args.password)
    except Exception, err:
        parser.error('Error connecting: {}'.format(err))
        retcode = -1
    if retcode == 0:
        write_dates(db.tasks_dbv2, sys.stdout)
    return retcode

if __name__ == '__main__':
    sys.exit(main())

mg_run_history.r

#! /usr/bin/env Rscript
args <- commandArgs(TRUE)
if (length(args) != 2) {
    cat("usage: mg_run_history.r DATAFILE PLOTFILE\n")
    q(status=1)
}
require(ggplot2)
infile <- args[1]
outfile <- args[2]
cat(sprintf("reading data from file %s\n", infile))
d <- read.csv(infile)
d$date <- as.POSIXct(d$completed_at, origin='1970-01-01')
d.o <- d[order(d$completed_at),]
d.o$n <- 1
d.o$daysec <- floor(d.o$completed_at / (60*60*24)) * (60*60*24)
d.o.day <- aggregate(d.o$n, by=list(d.o$daysec), sum)
d.o.day$tot <- cumsum(d.o.day$x)
d.o.day$date <- as.POSIXct(d.o.day$Group.1, origin='1970-01-01')
pdf(file=outfile)
xlims <- with(d.o.day, c(min(date), max(date)))
p <- ggplot(d.o.day, aes(x=date, y=tot)) +
  geom_area(fill='grey') +
  geom_smooth(stat="smooth", span=0.5) +
  ylab("Number of compounds calculated") +
  xlab("Date") +
  scale_x_datetime(limits=xlims, expand=c(0,0)) + 
  scale_y_continuous(limits=c(0,60000), expand=c(0,0))
print(p)
dev.off()
q(status=0)