#! /usr/bin/env python
#
# stats-to-csv <stats.log> <meta.dat> <wwwdir>
#
# Reads information from stats directory and outputs csv files <wwwdir>/<node>.<type>.csv.
# If any of these files already exists, we append (without writing the header line again).

import os.path
import os
import sys

Workers = set()
Proxies = set()

def readNodes(meta):
    for line in open(meta):
        m = line.split()

        if m[0] == "node":
            if m[2] == "worker":
                Workers.add(m[1])

            if m[2] == "proxy":
                Workers.add(m[1])

def processNode(stats, wwwdir, node, iface):

    print node, "..."

    def openFile(tag, columns):

        name = os.path.join(wwwdir, "%s.%s.csv" % (node, tag))

        if os.path.exists(name):
            return open(name, "a")
        else:
            f = open(name, "w")
            print >>f, "time," + ",".join(columns)

        return f

    iface_mbps = openFile("mbps", ["MBits/sec"])
    iface_pkts = openFile("pkts", ["TCP", "UDP", "ICMP", "Other"])
    cpu = openFile("cpu", ["CPU"])
    mem = openFile("mem", ["Memory"])
    cflow = openFile("in", ["MBits/sec"])

    def printEntry(t, entry):

        try:
            val = int(entry["parent-cpu"]) + int(entry["child-cpu"])
            print >>cpu, "%s,%s" % (t, val)
        except KeyError:
            pass

        try:
            val = int(entry["parent-vsize"]) + int(entry["child-vsize"])
            print >>mem, "%s,%s" % (t, val)
        except KeyError:
            pass

        if iface:
            try:
                print >>iface_mbps, "%s,%s" % (t, entry["interface-mbps"])
            except KeyError:
                pass

            try:
                tc = float(entry["interface-t"])
                ud = float(entry["interface-u"])
                ic = float(entry["interface-i"])
                ot = float(entry["interface-o"])
                print >>iface_pkts, "%s,%s,%s,%s,%s" % (t, tc, ud, ic, ot)

            except KeyError:
                pass

        if "in-mbps" in entry:
            print >>cflow, "%s,%s" % (t, entry["in-mbps"])

    entry = {}
    first = -1

    for line in open(stats):
        m = line.split()

        if m[1] != node:
            continue

        t = m[0]

        if t != first and first >= 0:
            printEntry(t, entry)
            entry = {}

        first = t

        try:
            entry["%s-%s" % (m[2], m[3])] = m[4]
        except IndexError:
            pass

    if first >= 0:
        printEntry(t, entry)

    iface_mbps.close()
    iface_pkts.close()
    cpu.close()
    mem.close()
    cflow.close()

if len(sys.argv) != 4:
    print "usage: %s <stats.log> <meta.dat> <www-dir>" % sys.argv[0]
    sys.exit(1)

stats = sys.argv[1]
meta = sys.argv[2]
wwwdir = sys.argv[3]

try:
    os.mkdir(wwwdir)
except OSError:
    pass

readNodes(meta)

try:
    for w in Workers:
        processNode(stats, wwwdir, w, True)

    for p in Proxies:
        processNode(stats, wwwdir, p, False)

    processNode(stats, wwwdir, "manager", False)
    processNode(stats, wwwdir, "cflow", False)
except IOError, err:
    print "Error: %s" % err
    sys.exit(1)

