Rewrite to be more automated.

brianhigh · Jan 6, 2016 · 0dffd90 · 0dffd90
1 parent d56a87f
commit 0dffd90
Show file tree

Hide file tree

Showing 9 changed files with 103 additions and 122 deletions.
diff --git a/data/home_cern/ipinfo.csv b/data/home_cern/ipinfo.csv
@@ -0,0 +1,10 @@
+"ip","country_code","country_name","region_code","region_name","city","zip_code","time_zone","latitude","longitude","metro_code","mean_rtt"
+"128.95.230.102","US","United States","WA","Washington","Seattle","98105","America/Los_Angeles",47.6606,-122.2919,"819",0
+"209.124.181.134","US","United States","WA","Washington","Seattle","98105","America/Los_Angeles",47.6606,-122.2919,"819",0
+"64.57.28.53","US","United States","MI","Michigan","Ann Arbor","48104","America/Detroit",42.2734,-83.7133,"505",0
+"198.71.45.24","US","United States","MI","Michigan","Ann Arbor","48104","America/Detroit",42.2734,-83.7133,"505",18.886
+"198.71.45.18","US","United States","MI","Michigan","Ann Arbor","48104","America/Detroit",42.2734,-83.7133,"505",34.893
+"198.71.45.14","US","United States","MI","Michigan","Ann Arbor","48104","America/Detroit",42.2734,-83.7133,"505",46.863
+"192.91.246.126","CH","Switzerland","GE","Geneva","Geneva","1205","Europe/Zurich",46.1956,6.1481,"0",182.871
+"192.65.196.38","CH","Switzerland","GE","Geneva","Geneva","1205","Europe/Zurich",46.1956,6.1481,"0",166.646333333333
+"188.184.37.205","CH","Switzerland","GE","Geneva","Meyrin","1217","Europe/Zurich",46.2324,6.0502,"0",164.322
diff --git a/data/home_cern/route.csv b/data/home_cern/route.csv
@@ -0,0 +1,12 @@
+"addr","mean_rtt","rtt.1","rtt.2","rtt.3"
+"128.95.230.102",0,0,0,0
+"10.132.1.73",0,0,0,0
+"10.132.1.75",0,0,0,0
+"209.124.181.134",0,0,0,0
+"64.57.28.53",0,0,0,0
+"198.71.45.24",18.886,18.886,18.886,18.886
+"198.71.45.18",34.893,34.893,34.893,34.893
+"198.71.45.14",46.863,46.863,46.863,46.863
+"192.91.246.126",182.871,182.871,182.871,182.871
+"192.65.196.38",166.646333333333,163.985,167.977,167.977
+"188.184.37.205",164.322,163.993,163.993,164.98
diff --git a/data/home_cern/route.txt b/data/home_cern/route.txt
@@ -0,0 +1,16 @@
+traceroute to home.cern (188.184.37.205), 30 hops max, 60 byte packets
+ 1  128.95.230.102  0.000 ms  0.000 ms  0.000 ms
+ 2  10.132.1.73  0.000 ms  0.000 ms  0.000 ms
+ 3  10.132.1.75  0.000 ms  0.000 ms  0.000 ms
+ 4  209.124.181.134  0.000 ms  0.000 ms  0.000 ms
+ 5  64.57.28.53  0.000 ms  0.000 ms  0.000 ms
+ 6  198.71.45.24  18.886 ms  18.886 ms  18.886 ms
+ 7  198.71.45.18  34.893 ms  34.893 ms  34.893 ms
+ 8  198.71.45.14  46.863 ms  46.863 ms  46.863 ms
+ 9  192.91.246.126  182.871 ms  182.871 ms  182.871 ms
+10  * * *
+11  * * *
+12  192.65.196.38  163.985 ms  167.977 ms  167.977 ms
+13  * * *
+14  * * *
+15  188.184.37.205  163.993 ms  163.993 ms  164.980 ms
diff --git a/data/udrive_uw_edu/ipinfo.csv b/data/udrive_uw_edu/ipinfo.csv
diff --git a/data/udrive_uw_edu/route.csv b/data/udrive_uw_edu/route.csv
diff --git a/data/udrive_uw_edu/route_4225roo_linux.txt b/data/udrive_uw_edu/route_4225roo_linux.txt
diff --git a/data/udrive_uw_edu/route_4225roo_windows.txt b/data/udrive_uw_edu/route_4225roo_windows.txt
diff --git a/data/udrive_uw_edu/route_mhsf_osx.txt b/data/udrive_uw_edu/route_mhsf_osx.txt
diff --git a/plot_route.R b/plot_route.R
@@ -1,96 +1,66 @@
-# Plot network hops to reach hosts by round-trip time (rtt) with error bars.
-
-# Data files were created with these commands from the Terminal (console):
-#
-#  DOS> tracert -d example.com > "route.txt"
-#
-#  bash-linux$ sudo traceroute -I \
-#          -n example.com 2>&1 | sed 's/\*/* ms/g' | tr -s " " "|" | \
-#          tail -n +2 > "route.txt"
-#          
-#  bash-osx$ traceroute -I \
-#          -n example.com 2>&1 | sed 's/\*/* ms/g' | tr -s " " "|" | \
-#          tail -n +3 > "route.txt"
-
-# Function to install packages as needed then load them into R.
-load_packages <- function(pkgs) {
-  # Install packages (if necessary) and load them into memory.
-
-  for (pkg in pkgs) {
-    if (! suppressWarnings(suppressPackageStartupMessages(require(
-      pkg, character.only=TRUE, quietly=TRUE))) ) {
-      install.packages(pkg, repos="http://cran.fhcrc.org", 
-                       dependencies=TRUE)
-      if (! suppressWarnings(suppressPackageStartupMessages(require(
-        pkg, character.only=TRUE, quietly=TRUE))) ) {
-        stop(paste0(c("Can't load package: ", pkg, "!"), 
-                    collapse = ""))
-      }
-    }
-  }
-}
-
-# Function to import route hop data from text files.
-import.route <- function(fname, troute.type, host) {
-    # We have two different formats, so parse differently for each type.
-    if (troute.type == "tracert") {
-        route <- read.fwf(fname, 
-                          widths=c(3, 6, 3, 6, 3, 6, 3, 16), 
-                          header=F, skip=4, strip.white=TRUE, 
-                          stringsAsFactors=F)
-        route <- route[,c(1, 2, 4, 6, 8)]
-        route <- route[complete.cases(route),]
-        names(route) <- c("hop", "query.1", "query.2", "query.3", "ip.addr")
-        route[,2:4] <- as.numeric(gsub("[<*]", "", as.matrix(route[2:4])))
-    } else {
-        route <- read.table(fname, sep="|", header=F, stringsAsFactors=F)
-        route <- route[,c(2, 3, 4, 6, 8)]
-        route <- route[complete.cases(route),]
-        names(route) <- c("hop", "ip.addr", "query.1", "query.2", "query.3")
-        route[,3:5] <- as.numeric(gsub("<", "", as.matrix(route[3:5])))
-        route <- route[, c(1, 3:5, 2)]
-    }
-
-    route$hop <- sprintf("%02d", as.numeric(route$hop))
-    route$hop <- as.factor(route$hop)
-    route.long <- melt(route, id=c("hop","ip.addr"))
-    names(route.long) <- c("hop", "ip.addr", "query", "rtt")
-    route.long$host <- rep(host, nrow(route.long))
-    route.long$query <- as.factor(gsub("query.", "", route.long$query, 
-                                       fixed = TRUE))
-    return(na.omit(as.data.table(route.long)))
-}
-
-# Load packages needed for this script.
-load_packages(c("reshape", "Rmisc", "ggplot2", "data.table"))
-
-# Define information about the data files to be used as route data.
-files <- c("data/www_gov_za/route.txt", 
-           "data/www_gov_bb/route.txt",
-           "data/www_gov_ro/route.txt",
-           "data/www_cubagob_cu/route.txt")
-types <- c("tracert", "tracert", "tracert", "tracert")
-hosts <- c("www.gov.za", "www.gov.bb", "www.gov.ro", "www.cubagob.cu")
-route.info <- cbind(file=files, type=types, host=hosts)
-
-# Process the routes using the route info.
-routes <- rbindlist(lapply(1:nrow(route.info), 
-                        function(n) import.route(files[n], types[n], hosts[n])))
-
-# Combine all route data into a singe data table.
-route.summary <- summarySE(routes, measurevar="rtt", groupvars=c("host", "hop"))
-
-
-# Make a basic line and point graph.
-
-# Plotting code (below) based on this example:
-# http://www.cookbook-r.com/Graphs/Plotting_means_and_error_bars_(ggplot2)
-
-# Use 95% confidence interval for error bars.
-g <- ggplot(route.summary, aes(x=hop, y=rtt, colour=host)) + 
-    geom_errorbar(aes(ymin=rtt-ci, ymax=rtt+ci), width=.1) +
-    geom_line(aes(group=host)) +
-    geom_point() + theme_bw() + 
-    xlab("network hop") + ylab("round-trip time (ms)") +
-    ggtitle("Round-trip time (ms) per network hop")
+# Plot network hops to reach hosts by round-trip time (rtt) with error bars.
+
+# Function to install packages as needed then load them into R.
+load_packages <- function(pkgs) {
+  # Install packages (if necessary) and load them into memory.
+
+  for (pkg in pkgs) {
+    if (! suppressWarnings(suppressPackageStartupMessages(require(
+      pkg, character.only=TRUE, quietly=TRUE))) ) {
+      install.packages(pkg, repos="http://cran.fhcrc.org", 
+                       dependencies=TRUE)
+      if (! suppressWarnings(suppressPackageStartupMessages(require(
+        pkg, character.only=TRUE, quietly=TRUE))) ) {
+        stop(paste0(c("Can't load package: ", pkg, "!"), 
+                    collapse = ""))
+      }
+    }
+  }
+}
+
+# Function to import and tidy route hop data from CSV file.
+import.route <- function(fname, host) {
+    if (file.exists(fname) == TRUE) {
+        route <- read.table(fname, sep=",", header=T, stringsAsFactors=F)
+        route$hop <- row.names(route)
+        route$mean_rtt <- NULL
+        route$hop <- sprintf("%02d", as.numeric(route$hop))
+        route$hop <- as.factor(route$hop)
+        route.long <- melt(route, id=c("hop","addr"))
+        names(route.long) <- c("hop", "addr", "query", "rtt")
+        route.long$host <- rep(host, nrow(route.long))
+        route.long$query <- as.factor(gsub("rtt.", "", route.long$query, 
+                                           fixed = TRUE))
+        return(na.omit(as.data.table(route.long)))
+    } else {
+        return(as.data.table(NULL))
+    }
+}
+
+# Load packages needed for this script.
+load_packages(c("reshape", "Rmisc", "ggplot2", "data.table"))
+
+# Search the "data" folder for "route.csv" files.
+hosts <- gsub("_", ".", list.dirs('data', recursive=FALSE, full.names=FALSE))
+files <- gsub("$", "/route.csv", list.dirs('data', recursive=FALSE))
+
+# Process the routes using the route info.
+routes <- rbindlist(lapply(1:length(files), 
+                        function(n) import.route(files[n], hosts[n])))
+
+# Combine all route data into a singe data table.
+route.summary <- summarySE(routes, measurevar="rtt", groupvars=c("host", "hop"))
+
+# Make a basic line and point graph with error bars.
+
+# Plotting code (below) based on this example:
+# http://www.cookbook-r.com/Graphs/Plotting_means_and_error_bars_(ggplot2)
+
+# Use 95% confidence interval for error bars.
+g <- ggplot(route.summary, aes(x=hop, y=rtt, colour=host)) + 
+    geom_errorbar(aes(ymin=rtt-ci, ymax=rtt+ci), width=.1, size=2) +
+    geom_line(aes(group=host), size=2) +
+    theme_bw() + 
+    xlab("network hop") + ylab("round-trip time (ms)") +
+    ggtitle("Round-trip time (ms) per network hop")
 plot(g)