[tor-commits] [metrics-tasks/master] Add code for bridge churn graph (#2794).
karsten at torproject.org
karsten at torproject.org
Thu Mar 24 15:06:16 UTC 2011
commit 8dde11c31d385d92d3dc41032438b62b693f6daf
Author: Karsten Loesing <karsten.loesing at gmx.net>
Date: Thu Mar 24 16:05:35 2011 +0100
Add code for bridge churn graph (#2794).
---
task-2794/.gitignore | 4 +
task-2794/README | 12 +++
task-2794/StillRunning.java | 139 +++++++++++++++++++++++++++++++++++++
task-2794/still-running-bridges.R | 15 ++++
4 files changed, 170 insertions(+), 0 deletions(-)
diff --git a/task-2794/.gitignore b/task-2794/.gitignore
new file mode 100644
index 0000000..939f812
--- /dev/null
+++ b/task-2794/.gitignore
@@ -0,0 +1,4 @@
+*.csv
+*.class
+*.png
+
diff --git a/task-2794/README b/task-2794/README
new file mode 100644
index 0000000..311c5b5
--- /dev/null
+++ b/task-2794/README
@@ -0,0 +1,12 @@
+Code to create a graph on "Uptimes of bridges that were running Jan 2,
+2011, 00:00:00 UTC":
+
+ - Generate assignments.csv and statuses.csv using the #2680 code and put
+ them in this directory.
+
+ - Compile and run the Java class:
+ $ javac StillRunning.java && java StillRunning
+
+ - Run the R code (note that this may take a few minutes!):
+ $ R --slave -f still-running-bridges.R
+
diff --git a/task-2794/StillRunning.java b/task-2794/StillRunning.java
new file mode 100644
index 0000000..111c085
--- /dev/null
+++ b/task-2794/StillRunning.java
@@ -0,0 +1,139 @@
+import java.io.*;
+import java.util.*;
+public class StillRunning {
+ public static void main(String[] args) throws Exception {
+
+ /* Parse bridge pool assignments. */
+ Map<String, String> assignments = new HashMap<String, String>();
+ BufferedReader br = new BufferedReader(new FileReader(
+ "assignments.csv"));
+ String line = br.readLine();
+ while ((line = br.readLine()) != null) {
+ String[] parts = line.split(",");
+ String fingerprint = parts[1];
+ String type = parts[2];
+ assignments.put(fingerprint, type);
+ }
+ br.close();
+
+ /* Parse running bridges in first status of the second day in the data
+ * set. */
+ br = new BufferedReader(new FileReader("statuses.csv"));
+ line = br.readLine();
+ if (!line.split(",")[15].equals("running")) {
+ System.out.println("Column 16 should be 'running'");
+ System.exit(1);
+ }
+ String dayOne = null, lastStatus = null;
+ List<String> fingerprints = new ArrayList<String>();
+ Map<String, String> addresses = new HashMap<String, String>();
+ while ((line = br.readLine()) != null) {
+ String[] parts = line.split(",");
+ String status = parts[0];
+ if (dayOne == null) {
+ dayOne = status.substring(0, "yyyy-mm-dd".length());
+ } else if (status.startsWith(dayOne)) {
+ continue;
+ }
+ String running = parts[15];
+ if (!running.equals("TRUE")) {
+ continue;
+ }
+ if (lastStatus != null && !status.equals(lastStatus)) {
+ break;
+ }
+ lastStatus = status;
+ String fingerprint = parts[1];
+ fingerprints.add(fingerprint);
+ String address = parts[4];
+ addresses.put(fingerprint, address);
+ }
+
+ /* Parse subsequent statuses and count how often these bridges
+ * occur. */
+ Map<String, Integer>
+ fingerprintAnyCount = new HashMap<String, Integer>(),
+ fingerprintSameCount = new HashMap<String, Integer>();
+ for (String fingerprint : fingerprints) {
+ fingerprintAnyCount.put(fingerprint, 1);
+ fingerprintSameCount.put(fingerprint, 1);
+ }
+ do {
+ String[] parts = line.split(",");
+ String status = parts[0];
+ String running = parts[15];
+ if (!running.equals("TRUE")) {
+ continue;
+ }
+ String fingerprint = parts[1];
+ if (!fingerprints.contains(fingerprint)) {
+ continue;
+ }
+ fingerprintAnyCount.put(fingerprint,
+ fingerprintAnyCount.get(fingerprint) + 1);
+ String address = parts[4];
+ if (addresses.get(fingerprint).equals(address)) {
+ fingerprintSameCount.put(fingerprint,
+ fingerprintSameCount.get(fingerprint) + 1);
+ }
+ } while ((line = br.readLine()) != null);
+
+ /* Create two lists of fingerprints, ordered by the number of
+ * occurrences. */
+ SortedMap<String, String>
+ sortAnyFingerprints = new TreeMap<String, String>(),
+ sortSameFingerprints = new TreeMap<String, String>();
+ for (Map.Entry<String, Integer> e : fingerprintAnyCount.entrySet()) {
+ sortAnyFingerprints.put(String.format("%05d %s", e.getValue(),
+ e.getKey()), e.getKey());
+ }
+ List<String> sortedAnyFingerprints = new ArrayList<String>(
+ sortAnyFingerprints.values());
+ for (Map.Entry<String, Integer> e : fingerprintSameCount.entrySet()) {
+ sortSameFingerprints.put(String.format("%05d %s", e.getValue(),
+ e.getKey()), e.getKey());
+ }
+ List<String> sortedSameFingerprints = new ArrayList<String>(
+ sortSameFingerprints.values());
+
+ /* Write bridges of first status to disk. */
+ BufferedWriter bw = new BufferedWriter(new FileWriter(
+ "still-running-bridges.csv"));
+ bw.write("status,anyid,sameid,type,addresschange\n");
+ for (String fingerprint : sortedAnyFingerprints) {
+ bw.write(lastStatus + ","
+ + sortedAnyFingerprints.indexOf(fingerprint) + ","
+ + sortedSameFingerprints.indexOf(fingerprint) + ","
+ + assignments.get(fingerprint) + ",FALSE\n");
+ }
+
+ /* Parse statuses once again and write bridges to disk. */
+ br = new BufferedReader(new FileReader("statuses.csv"));
+ line = br.readLine();
+ while ((line = br.readLine()) != null) {
+ String[] parts = line.split(",");
+ String status = parts[0];
+ if (status.startsWith(dayOne)) {
+ continue;
+ }
+ String running = parts[15];
+ if (!running.equals("TRUE")) {
+ continue;
+ }
+ String fingerprint = parts[1];
+ if (!fingerprints.contains(fingerprint)) {
+ continue;
+ }
+ String address = parts[4];
+ boolean addressChange = !addresses.get(fingerprint).equals(address);
+ bw.write(status + ","
+ + sortedAnyFingerprints.indexOf(fingerprint) + ","
+ + sortedSameFingerprints.indexOf(fingerprint) + ","
+ + assignments.get(fingerprint) + ","
+ + (addressChange ? "TRUE" : "FALSE") + "\n");
+ }
+ bw.close();
+ br.close();
+ }
+}
+
diff --git a/task-2794/still-running-bridges.R b/task-2794/still-running-bridges.R
new file mode 100644
index 0000000..8752b07
--- /dev/null
+++ b/task-2794/still-running-bridges.R
@@ -0,0 +1,15 @@
+library(ggplot2)
+anyip <- read.csv("still-running-bridges.csv", stringsAsFactors = FALSE)
+sameip <- anyip[anyip$addresschange == FALSE, ]
+data <- rbind(
+ data.frame(status = anyip$status, bridgeid = anyip$anyid, address = "any IP address"),
+ data.frame(status = sameip$status, bridgeid = sameip$sameid, address = "same IP address"))
+ggplot(data, aes(x = as.POSIXct(status),
+ y = (max(data$bridgeid) - bridgeid) / max(data$bridgeid))) +
+facet_grid(address ~ .) +
+geom_point(size = 0.2, colour = "springgreen3") +
+scale_x_datetime(name = "") +
+scale_y_continuous(name = "", formatter = "percent") +
+opts(title = "Uptimes of bridges that were running Jan 2, 2011, 00:00:00 UTC\n")
+ggsave(filename = "still-running-bridges.png", width = 8, height = 5, dpi = 72)
+
More information about the tor-commits
mailing list