[tor-commits] [metrics-tasks/master] Add graphing code for #4906.

[karsten at torproject.org]

commit 31f7b2a72901f7484f72e87e8013f15d89f28992
Author: Karsten Loesing <karsten.loesing at gmx.net>
Date:   Sun Jan 15 06:10:23 2012 +0100

    Add graphing code for #4906.
---
 task-4906/.gitignore |    7 +++++++
 task-4906/Eval.java  |   44 ++++++++++++++++++++++++++++++++++++++++++++
 task-4906/README     |   19 +++++++++++++++++++
 task-4906/eval.R     |   22 ++++++++++++++++++++++
 4 files changed, 92 insertions(+), 0 deletions(-)

diff --git a/task-4906/.gitignore b/task-4906/.gitignore
new file mode 100644
index 0000000..515a3f8
--- /dev/null
+++ b/task-4906/.gitignore
@@ -0,0 +1,7 @@
+*.class
+in/
+lib/
+Rplots.pdf
+out.csv
+*.png
+
diff --git a/task-4906/Eval.java b/task-4906/Eval.java
new file mode 100644
index 0000000..1d2d702
--- /dev/null
+++ b/task-4906/Eval.java
@@ -0,0 +1,44 @@
+import java.io.*;
+import java.util.*;
+import org.torproject.descriptor.*;
+
+public class Eval {
+  public static void main(String[] args) throws IOException {
+    SortedMap<String, String> lines = new TreeMap<String, String>();
+    File inputDirectory = new File("in");
+    RelayDescriptorReader reader =
+        DescriptorSourceFactory.createRelayDescriptorReader();
+    reader.addDirectory(inputDirectory);
+    Iterator<DescriptorFile> descriptorFiles = reader.readDescriptors();
+    while (descriptorFiles.hasNext()) {
+      DescriptorFile descriptorFile = descriptorFiles.next();
+      for (Descriptor descriptor : descriptorFile.getDescriptors()) {
+        if (!(descriptor instanceof ExtraInfoDescriptor)) {
+          continue;
+        }
+        ExtraInfoDescriptor extraInfoDescriptor =
+            (ExtraInfoDescriptor) descriptor;
+        BandwidthHistory dirreqWriteHistory =
+            extraInfoDescriptor.getDirreqWriteHistory();
+        if (dirreqWriteHistory != null) {
+          String nickname = extraInfoDescriptor.getNickname();
+          for (Map.Entry<Long, Long> e :
+              dirreqWriteHistory.getBandwidthValues().entrySet()) {
+            long intervalEnd = e.getKey();
+            long writtenDirreqBytes = e.getValue();
+            String key = nickname + ","
+                + String.valueOf(intervalEnd / 1000L);
+            String value = key + "," + String.valueOf(writtenDirreqBytes);
+            lines.put(key, value);
+          }
+        }
+      }
+    }
+    BufferedWriter bw = new BufferedWriter(new FileWriter("out.csv"));
+    for (String line : lines.values()) {
+      bw.write(line + "\n");
+    }
+    bw.close();
+  }
+}
+
diff --git a/task-4906/README b/task-4906/README
new file mode 100644
index 0000000..7b1f97f
--- /dev/null
+++ b/task-4906/README
@@ -0,0 +1,19 @@
+Download and extract extra-info descriptors published by the directory
+authorities.  Put them in an in/ subdirectory.
+
+  $ grep -REil "^extra-info gabelmoo|^extra-info moria1|^extra-info tor26|
+    ^extra-info ides|^extra-info dannenberg|^extra-info urras|
+    ^extra-info maatuska|^extra-info dizum" extra-infos-20??-?? |
+    xargs -I{} mv {} in/
+
+Put metrics-lib's descriptor.jar in lib/.
+
+Run the Java program.
+
+  $ javac -cp lib/descriptor.jar Eval.java
+  $ java -cp .:lib/descriptor.jar Eval
+
+Run R.
+
+  $ R --slave -f eval.R
+
diff --git a/task-4906/eval.R b/task-4906/eval.R
new file mode 100644
index 0000000..393fb7e
--- /dev/null
+++ b/task-4906/eval.R
@@ -0,0 +1,22 @@
+library(ggplot2)
+d <- read.csv("out.csv", header = FALSE,
+  col.names = c("dirauth", "intervalend", "writtendirreqbytes"))
+d <- data.frame(dirauth = d$dirauth,
+  date = as.Date(d$intervalend / 86400, origin = "1970-01-01 00:00:00"),
+  writtendirreqbytes = d$writtendirreqbytes, count = 1)
+c <- aggregate(x = list(count = d$count),
+  by = list(dirauth = d$dirauth, date = d$date), FUN = sum)
+d <- aggregate(x = list(writtendirreqbytes = d$writtendirreqbytes),
+  by = list(dirauth = d$dirauth, date = d$date), FUN = sum)
+e <- data.frame(dirauth = d$dirauth, date = d$date,
+  writtendirreqbytespersec = (d$writtendirreqbytes / c$count) / 900)
+ggplot(e, aes(x = date, y = writtendirreqbytespersec / 1024,
+  colour = dirauth)) +
+geom_line() +
+scale_x_date(name = "\nDate") +
+scale_y_continuous(name = "KiB/s\n") +
+scale_colour_hue(name = "Directory authority") +
+opts(title = "Written bytes on CONN_TYPE_DIR connections\n")
+ggsave("dirauth-written-dirreq-bytes.png", width = 8, height = 6,
+  dpi = 100)
+