netsec graphing

1 files changed, 223 insertions, 0 deletions

diff --git a/netsec.py b/netsec.py
new file mode 100755
index 0000000..b1f4f30
--- /dev/null
+++ b/netsec.py
@@ -0,0 +1,223 @@
+#!/usr/bin/env python
+import defapp
+
+from plot import decode
+from util import load_csv_file, write_csv_file
+
+from math import ceil
+
+from numpy import amin, amax, mean, median, std, histogram, zeros, arange
+
+from os.path  import splitext, basename
+from optparse import make_option as o
+
+from gnuplot  import gnuplot, FORMATS, Plot, label, curve
+
+options = [
+    # output options
+    o('-f', '--format', action='store', dest='format', type='choice',
+      choices=FORMATS, help='output format'),
+    o(None, '--save-script', action='store_true', dest='save_script'),
+    o('-p', '--prefix', action='store', dest='prefix'),
+
+    # formatting options
+    # These may or may not be supported by a particular experiment plotter.
+    o(None, '--smooth', action='store_true', dest='smooth'),
+    o(None, '--hist', action='store_true', dest='histogram'),
+    ]
+
+defaults = {
+    # output options
+    'format' : 'pdf',
+    'save_script' : False,
+    'prefix' : '',
+
+    # formatting options
+    'histogram' : False,
+    'smooth' : False,
+    }
+
+def get_stats_label(samples):
+    avg = mean(samples)
+    med = median(samples)
+    dev = std(samples)
+    max = amax(samples)
+    min = amin(samples)
+    return  "min=%.2f max=%.2f avg=%.2f  median=%.2f  std=%.2f" \
+        % (min, max, avg, med, dev)
+
+class NetsecPlotter(defapp.App):
+    def __init__(self):
+        defapp.App.__init__(self, options, defaults, no_std_opts=True)
+        self.tmpfiles = []
+
+    def make_plot(self, fname=None):
+        p = Plot()
+        p.output = "%s%s.%s" % (self.options.prefix, fname, self.options.format)
+        p.format = self.options.format
+        return p
+
+    def setup_png(self, plot):
+        # standard png options; usually correct; never tweaked for paper
+        if self.options.format == 'png':
+            plot.font_size = 'large'
+            plot.size   = (1024, 768)
+            plot.xticks = (0, 1)
+            plot.yticks = (0, 0.1)
+            plot.default_style = "linespoints"
+            return True
+        else:
+            return False
+
+    def write(self, data, name, ext='data'):
+        if self.options.save_script:
+            fname = "%s.%s" % (name, ext)
+            write_csv_file(fname, data)
+            return fname
+        else:
+            tmp = write_csv_file(None, data)
+            # keep a reference so that it isn't deleted
+            self.tmpfiles.append(tmp)
+            return tmp.name
+
+    def load(self, datafile):
+        data = load_csv_file(datafile)
+        print "loaded %d lines" % len(data)
+        return data
+
+    def write_histogram(self, samples, name):
+        max = amax(samples)
+        (hist, edges) = histogram(samples, bins=max,
+                range=(0.5,max+.5))
+        data = zeros((len(edges)-1, 3))
+        cumulative = 0
+        for i in xrange(len(hist)):
+            data[i, 0]  = (edges[i] + edges[i + 1]) / 2.0
+            data[i, 1]  = hist[i]
+            cumulative += hist[i]
+            data[i, 2]  = cumulative
+
+        if len(hist) > 20:
+            label_freq = 10
+        else:
+            label_freq = 1
+
+        for_file = []
+        for i, row in enumerate(data):
+            label = '%d' % row[0] if row[0] % label_freq == 0 else ''
+            for_file.append([row[0], row[1], row[2], label])
+
+        return (data, self.write(for_file, name, ext='hist'))
+
+    def render(self, p):
+        if self.options.save_script:
+            p.gnuplot_save(p.output + '.plot')
+        else:
+            p.gnuplot_exec()
+
+    def plot_hchisto(self, datafile, name, conf):
+        data  = self.load(datafile)
+
+        #max_val = amax(data[:,1])
+
+        if self.options.histogram:
+            name += '_hist'
+
+        p = self.make_plot(name)
+
+        # place a label on the graph
+        p.labels = [label(0.5, 0.9, get_stats_label(data[:,1]),
+                          coord=['graph', 'screen'], align='center')]
+
+        (data, fname) = self.write_histogram(data[:,1], name)
+        p.xlabel = "hop count"
+
+        p.ylabel = "number of sources"
+        p.setup_histogram(gap=1, boxwidth=1.0)
+        p.title = "hop counts;"
+
+        if 'per-host' in conf:
+            p.title += " HC per-host: %s" % conf['per-host']
+
+        if 'as-num' in conf:
+            p.title += " AS=%s, IP/mask=%s/%s MB=%s Unique Hosts=%s" % \
+                    (conf['as-num'], conf['as-str'], conf['as-mask'], \
+                    conf['megabytes'], conf['num-hosts'])
+
+#            p.xrange = (0, ceil(max_cost))
+        p.xticks = (0, 10)
+#            p.yticks = (0, 1)
+        #p.yrange = (0, (ceil(amax(data[:,1]) / 100) * 100))
+
+
+        ymax = amax(data[:,1]) 
+        p.curves = [curve(histogram=fname, col=2, labels_col=4)]
+        p.yrange = (1, amax(data[:,1]) + 2)
+
+        if ymax > 10000:
+            p.ylog = True
+
+            #print "not implemented yet..."
+            #return
+            ## plot raw samples
+            #p.title = "raw decoding cost; input=%s; host=%s" \
+            #    % (conf['file'], conf['host'])
+
+            #p.ylabel = "decoding cost (ms)"
+            #p.xlabel = "frame number"
+            #p.xrange = (0, len(data))
+            ##p.xticks = (0, 100)
+            #p.yticks = (0, 1)
+            #p.yrange = (1, ceil(max_cost) + 2)
+
+            #p.curves = [curve(fname=fname, xcol=2, ycol=1, title="decoding cost")]
+
+
+        #### Styling.
+
+        if not self.setup_png(p):
+            p.rounded_caps = True
+            p.font = 'Helvetica'
+
+            p.font_size = '10'
+            p.size = ('20cm', '10cm')
+            p.monochrome   = False
+            p.dashed_lines = False
+            p.key = 'off'
+            p.default_style = 'points lw 1'
+
+        if self.options.smooth:
+            p.default_style += " smooth bezier"
+
+        self.render(p)
+
+
+    def plot_file(self, datafile):
+        bname     = basename(datafile)
+        name, ext = splitext(bname)
+        conf      = decode(name)
+
+        if 'per-host' in conf or 'as-num' in conf:
+            self.plot_hchisto(datafile, name, conf)
+
+        #for plot_type in plotters:
+        #    if plot_type in conf:
+        #        try:
+        #            plotters[plot_type](datafile, name, conf)
+        #        except IOError as err:
+        #            self.err("Skipped '%s' (%s)." % err)
+        #        break
+        #else:
+        #    self.err("Skipped '%s'; unkown experiment type."
+        #             % bname)
+
+        # release all tmp files
+        self.tmpfiles = []
+
+    def default(self, _):
+        for i, datafile in enumerate(self.args):
+            self.out("[%d/%d] Processing %s ..." % (i + 1, len(self.args), datafile))
+            self.plot_file(datafile)
+
+if __name__ == "__main__":
+    NetsecPlotter().launch()