From 52bca474a6122d4130a15199afc37f622a71f466 Mon Sep 17 00:00:00 2001
From: Christoffer <christoffer@netravnen.de>
Date: Fri, 14 Dec 2018 13:05:20 +0000
Subject: [PATCH] Updated roagen script to be able to capture and use
 max-length value

---
 roagen.php | 357 +++++++++++++++++++++++++++++------------------------
 1 file changed, 196 insertions(+), 161 deletions(-)

diff --git a/roagen.php b/roagen.php
index 21e61e3..ad64699 100755
--- a/roagen.php
+++ b/roagen.php
@@ -5,196 +5,231 @@ echo shell_exec("/usr/bin/git -C ../registry/ pull origin master:master 2>&1");
 
 /*
  * Function:
- * startsWith( $string, 'word')
+ * startsWith ($string, "word", $length)
  * 
- * Find lines beginning with 'word'.
+ * Find lines beginning with "word". Optionally
+ * give the length of the string you are looking for.
  */
-function startsWith($haystack, $needle)
+function startsWith ($haystack, $needle, $length = "0")
 {
-    $length = strlen($needle);
-    return (substr($haystack, 0, $length) === $needle);
+  if ($length <= 0 || $length > (strlen ($needle)))
+    $length = strlen ($needle);
+
+  return (substr ($haystack, 0, $length) === $needle);
 }
 
 /*
  * Function:
- * endsWith( $string, 'word') 
+ * endsWith ($string, "word") 
  *   
- * Find lines ending with 'word'.
+ * Find lines ending with "word".
  */
-function endsWith($haystack, $needle)
+function endsWith ($haystack, $needle)
 {
-    $length = strlen($needle);
-    if ($length == 0) {
-        return true;
-    }
+  $length = strlen ($needle);
 
-    return (substr($haystack, -$length) === $needle);
+  if ($length == 0)
+    return true;
+
+  return (substr( $haystack, -$length) === $needle);
 }
 
-// Create empty array $roas
+/*
+ * Function:
+ * trim_special_chars ($string) 
+ *   
+ * Remove special characters.
+ */
+function trim_special_chars ($string)
+{
+  return (trim ($string, " \t\n\r\0\x0B"));
+}
+
+// Define array() we are going to populate with data.
 $roas = array();
 
+// Set folders we need to scan.
+$files6 = scandir("../registry/data/route6/");
+$files4 = scandir("../registry/data/route/");
+
 /*
  *
- * IPv6 route objects
+ * IPv6
  *
  */
 
-// Set directory for which all files are to be scanned
-$files = scandir('../registry/data/route6/');
-
-$h = 0; // count up when reading new file
-$i = 0; // used when loop through sub-arrays. I.e. when route objects have more than one listed origin asn
-
-foreach($files as $file)
+$i = 0; // Counter used with tmp $raw_array.
+$raw_array = array();  // tmp array() used for storing data to be processed
+foreach ($files6 as $file)
 {
-    $handle = fopen("../registry/data/route6/$file", "r");
-    while (($line = fgets($handle)) !== false) {
-        // extract var $ta
-        if (startsWith($line,'source')) {
-            $source = array();
-            preg_match('/([A-Z0-4]+)/',explode(':', $line)[1], $source);
-            $ta = $source[1];
-        }
-        // extract var $route
-        // extract var $maskLength
-        if (startsWith($line,'route6')) {
-            $prefix = array();
-            $line2 = explode('6: ', $line);
-            // validate v6 netmasks with following cmd
-            // cmd: ls -1 ../registry/data/route6/ | egrep -o "\b\_[0-9]+\b" | sed 's/\_//' | sort -n | uniq
-            preg_match('/([a-f0-9\:]{0,128})\/(32|40|4[4-9]|5[0-9]|6[0-4]|80)/', $line2[1], $prefix);
-            $route = $prefix[0];
-            $mask = $prefix[2];
-        }
-        // extract var $maxLength
-        /*if (startsWith($line,'max-length')) {
-            $maxLength = array();
-            preg_match('/([0-9]+)/',explode(':', $line)[1], $maxLength);
-            $maxLength = $maxLength[1];
-        } else {
-            $maxLength = '';
-        }*/
-        // extract var $asn
-        if (startsWith($line, 'origin')) {
-            $asn = array();
-            preg_match('/AS[0-9]+/', explode(':', $line)[1], $asn);
-            if (count($asn) > 1) {
-                foreach ($asn as $key => $value) {
-                    if (!empty($value)) {
-                        $roas['roas'][$i]['asn'] = $asn[0];
-                        $roas['roas'][$i]['prefix'] = $route;
-                        /*if (isset($ta)) {
-                            if ($ta == 'DN42') {
-                                $mask = ($mask <= 64 ? '64' : $mask);
-                                if ($maxLength != '') $mask = $maxLength;
-                            }
-                        } else {
-                            $ta = 'NULL';
-                        }*/
-                        $roas['roas'][$i]['maxLength'] = $mask;
-                        $roas['roas'][$i]['ta'] = (empty($ta) ? '' : $ta);
-                        $i++;
-                    }
-                }
-            } else {
-                $roas['roas'][$i]['asn'] = $asn[0];
-                $roas['roas'][$i]['prefix'] = $route;
-                /*if (isset($ta)) {
-                    if ($ta == 'DN42') {
-                        $mask = ($mask <= 64 ? '64' : $mask);
-                        if ($maxLength != '') $mask = $maxLength;
-                    }
-                } else {
-                    $ta = 'NULL';
-                }*/
-                $roas['roas'][$i]['maxLength'] = $mask;
-                $roas['roas'][$i]['ta'] = (empty($ta) ? '' : $ta);
-                $i++;
-            }
-        }
-    }
-    fclose($handle); // close $file when finished reading data
-    $h++;
+  $j = 0;
+
+  /*
+   * route6 with maxLength value set:
+   * - fd42:5d71:219::/48
+   * 
+   * $ cat ../registry/data/route6/fd42:5d71:219::_48
+   * route6:             fd42:5d71:219::/48
+   * origin:             AS4242420119
+   * max-length:         48
+   * mnt-by:             JRB0001-MNT
+   * source:             DN42
+   */
+  
+  $data = file("../registry/data/route6/$file");
+  
+  foreach ($data as $str)
+  {
+    $str = trim_special_chars ($str);
+    
+    if     (startsWith ($str, ("max"),    3))      $raw_array[$i]["max"]        = $str;
+    elseif (startsWith ($str, ("source"), 6))      $raw_array[$i]["source"]     = $str;
+    elseif (startsWith ($str, ("route"),  5))      $raw_array[$i]["route"]      = $str;
+    elseif (startsWith ($str, ("origin"), 6))      $raw_array[$i]["asn"][$j++]  = $str;
+    
+    // Catch max-length not set in route object.
+    if (empty ($raw_array[$i]["max"])) $raw_array[$i]["max"] = -1;
+  }
+  $i++;
+}
+
+$k = 0;
+
+foreach ($raw_array as $sub_array)
+{
+  // Extract prefix and subnet size
+  // Match prefix sizes 29-64, 80.
+  $prefix = array();
+  preg_match("/([a-f0-9\:]{0,128})\/(29|[3-5][0-9]|6[0-4]|80)/",
+   explode("6: ", $sub_array["route"])[1],
+   $prefix);
+  
+  // Extract ta information
+  $source = array();
+  preg_match ("/([A-Z0-4]+)/",
+   explode(":", $sub_array["source"])[1],
+   $source);
+  
+  // Try to extract max-length information
+  $maxlength = array();
+  if (($sub_array["max"]) != -1)
+    preg_match ("/([0-9]+)/",
+     explode(":", $sub_array["max"])[1],
+     $maxlength);
+
+  // Store extracted values
+  $_prefix    = $prefix[0];
+  $_ta        = (isset ($source[0]) ? $source[0] : "");
+  // We need to do conditional setting of maxLength to avoid errornous output.
+  if (($sub_array["max"]) != -1)
+    $_maxlength = (isset ($maxlength[0]) ? $maxlength[0] : "");
+  else
+    // Do fallback to default prefix size if max-length was not set.
+    $_maxlength = $prefix[2];
+  
+  // Loop through each asn in single route6 object and assign
+  // other values accordingly.
+  foreach ($sub_array["asn"] as $asn)
+  {
+    // Extract ASxxxxx from string.
+    preg_match ("/AS[0-9]+/", explode (":", $asn)[1], $_asn);
+    
+    $roas["roas"][$k]["asn"] = $_asn[0];
+    $roas["roas"][$k]["prefix"] = $_prefix;
+    $roas["roas"][$k]["maxLength"] = $_maxlength;
+    $roas["roas"][$k]["ta"] = $_ta;
+    
+    $k++;
+  }
 }
 
 /*
  *
- * IPv4 route objects
+ * IPv4
  *
  */
 
-// Set directory for which all files are to be scanned
-$files = scandir('../registry/data/route/');
-
-foreach($files as $file)
+$i = 0; // Counter used with tmp $raw_array.
+$raw_array = array();  // tmp array() used for storing data to be processed 
+foreach ($files4 as $file)
 {
-    $handle = fopen("../registry/data/route/$file", "r");
-    while (($line = fgets($handle)) !== false) {
-    // extract var $ta
-        if (startsWith($line,'source')) {
-            $source = array();
-            preg_match('/([A-Z0-4]+)/',explode(':', $line)[1], $source);
-            $ta = $source[1];
-        }
-        // extract var $route
-        // extract var $maskLength
-        if (startsWith($line,'route')) {
-            $prefix = array();
-            preg_match('/(25[0-5]|2[0-4][0-9]|[01]?[0-9][0-9]?)\.(25[0-5]|2[0-4][0-9]|[01]?[0-9][0-9]?)\.(25[0-5]|2[0-4][0-9]|[01]?[0-9][0-9]?)\.(25[0-5]|2[0-4][0-9]|[01]?[0-9][0-9]?)\/(3[0-2]|[0-2]?[0-9])/', explode(':', $line)[1], $prefix);
-            $route = $prefix[0];
-            $mask = $prefix[5];
-        }
-        // extract var $maxLength
-        /*$maxLength = ''; // ensure var is null when starting new loop
-        if (startsWith($line,'max-length')) {
-            $maxLength = array();
-            preg_match('/([0-9]+)/',explode(':', $line)[1], $maxLength);
-            $maxLength = $maxLength[1];
-        } else {
-            $maxLength = '';
-        }*/
-        // extract var $asn
-        if (startsWith($line, 'origin')) {
-            $asn = array();
-            preg_match('/AS[0-9]+/', explode(':', $line)[1], $asn);
-            if (count($asn) > 1) {
-                foreach ($asn as $key => $value) {
-                    if (!empty($value)) {
-                        $roas['roas'][$i]['asn'] = $asn[0];
-                        $roas['roas'][$i]['prefix'] = $route;
-                        /*if (isset($ta)) {
-                            if ($ta == 'DN42') {
-                                $mask = ($mask <= 28 ? '28' : $mask);
-                                if ($maxLength != '') $mask = $maxLength;
-                            }
-                        } else {
-                            $ta = 'NULL';
-                        }*/
-                        $roas['roas'][$i]['maxLength'] = $mask;
-                        $roas['roas'][$i]['ta'] = (empty($ta) ? '' : $ta);
-                        $i++;
-                    }
-                }
-            } else {
-                $roas['roas'][$i]['asn'] = $asn[0];
-                $roas['roas'][$i]['prefix'] = $route;
-                /*if (isset($ta)) {
-                    if ($ta == 'DN42') {
-                        $mask = ($mask <= 28 ? '28' : $mask);
-                        if ($maxLength != '') $mask = $maxLength;
-                    }
-                } else {
-                    $ta = 'NULL';
-                }*/
-                $roas['roas'][$i]['maxLength'] = $mask;
-                $roas['roas'][$i]['ta'] = (empty($ta) ? '' : $ta);
-                $i++;
-            }
-        }
-    }
-    fclose($handle); // close $file when finished reading data
-    $h++;
+  $j = 0;
+
+  /*
+   * route with maxLength value set:
+   * - 172.20.1.0/24
+   * 
+   * $ cat ../registry/data/route/172.20.1.0_24
+   * route:              172.20.1.0/24
+   * origin:             AS4242420119
+   * max-length:         24
+   * mnt-by:             JRB0001-MNT
+   * source:             DN42
+   */
+  
+  $data = file("../registry/data/route/$file");
+  
+  foreach ($data as $str)
+  {
+    $str = trim_special_chars ($str);
+    
+    if     (startsWith ($str, ("max"),    3))      $raw_array[$i]["max"]        = $str;
+    elseif (startsWith ($str, ("source"), 6))      $raw_array[$i]["source"]     = $str;
+    elseif (startsWith ($str, ("route"),  5))      $raw_array[$i]["route"]      = $str;
+    elseif (startsWith ($str, ("origin"), 6))      $raw_array[$i]["asn"][$j++]  = $str;
+    
+    // Catch max-length not set in route object.
+    if (empty ($raw_array[$i]["max"])) $raw_array[$i]["max"] = -1;
+  }
+  $i++;
+}
+
+foreach ($raw_array as $sub_array)
+{
+  // Extract prefix and subnet size
+  // Match prefix sizes 8-32.
+  $prefix = array();
+  preg_match("/([0-9\.]{7,15})\/([8-9]|[1-2][0-9]|3[0-2])/",
+   explode(":", $sub_array["route"])[1],
+   $prefix);
+  
+  // Extract ta information
+  $source = array();
+  preg_match ("/([A-Z0-4]+)/",
+   explode(":", $sub_array["source"])[1],
+   $source);
+  
+  // Try to extract max-length information
+  $maxlength = array();
+  if (($sub_array["max"]) != -1)
+    preg_match ("/([0-9]+)/",
+     explode(":", $sub_array["max"])[1],
+     $maxlength);
+
+  // Store extracted values
+  $_prefix    = $prefix[0];
+  $_ta        = (isset ($source[0]) ? $source[0] : "");
+  // We need to do conditional setting of maxLength to avoid errornous output.
+  if (($sub_array["max"]) != -1)
+    $_maxlength = (isset ($maxlength[0]) ? $maxlength[0] : "");
+  else
+    // Do fallback to default prefix size if max-length was not set.
+    $_maxlength = $prefix[2];
+  
+  // Loop through each asn in single route6 object and assign
+  // other values accordingly.
+  foreach ($sub_array["asn"] as $asn)
+  {
+    // Extract ASxxxxx from string.
+    preg_match ("/AS[0-9]+/", explode (":", $asn)[1], $_asn);
+    
+    $roas["roas"][$k]["asn"] = $_asn[0];
+    $roas["roas"][$k]["prefix"] = $_prefix;
+    $roas["roas"][$k]["maxLength"] = $_maxlength;
+    $roas["roas"][$k]["ta"] = $_ta;
+    
+    $k++;
+  }
 }
 
 // Do JSON encoding before writing result to file