Add --resume from XML output. Closes #316. See #243

CHANGELOG

@@ -1,5 +1,9 @@
 # Nmap Changelog ($Id$); -*-text-*-
 
+o [GH#316] Added scan resume from Nmap's XML output. Now you can --resume a
+  canceled scan from all 3 major output formats: -oN, -oG, and -oX.
+  [Tudor Emil Coman]
+
 o [Ndiff][GH#591] Fix a bug where hosts with the same IP but different
   hostnames were shown as changing hostnames between scans. Made sort stable
   with regard to hostnames. [Daniel Miller]

NmapOps.cc

@@ -389,6 +389,7 @@ void NmapOps::Initialize() {
   portlist = NULL;
   exclude_portlist = NULL;
   proxy_chain = NULL;
+  resuming = false;
 }
 
 bool NmapOps::SCTPScan() {

NmapOps.h

@@ -181,6 +181,7 @@ class NmapOps {
   /* Whether we have pcap functions (can be false on Windows). */
   bool have_pcap;
   int debugging;
+  bool resuming;
 
 #define PACKET_SEND_NOPREF 1
 #define PACKET_SEND_ETH_WEAK 2

main.cc

@@ -222,6 +222,7 @@ int main(int argc, char *argv[]) {
     if (gather_logfile_resumption_state(argv[2], &myargc, &myargv) == -1) {
       fatal("Cannot resume from (supposed) log file %s", argv[2]);
     }
+    o.resuming = true;
     return nmap_main(myargc, myargv);
   }
 

nmap.cc

@@ -1855,18 +1855,48 @@ int nmap_main(int argc, char *argv[]) {
   fflush(stderr);
 
   timep = time(NULL);
-
-  /* Brief info in case they forget what was scanned */
   Strncpy(mytime, ctime(&timep), sizeof(mytime));
   chomp(mytime);
-  char *xslfname = o.XSLStyleSheet();
+
-  xml_start_document("nmaprun");
+  if (!o.resuming) {
-  if (xslfname) {
+    /* Brief info in case they forget what was scanned */
-    xml_open_pi("xml-stylesheet");
+    char *xslfname = o.XSLStyleSheet();
-    xml_attribute("href", "%s", xslfname);
+    xml_start_document("nmaprun");
-    xml_attribute("type", "text/xsl");
+    if (xslfname) {
-    xml_close_pi();
+      xml_open_pi("xml-stylesheet");
+      xml_attribute("href", "%s", xslfname);
+      xml_attribute("type", "text/xsl");
+      xml_close_pi();
+      xml_newline();
+    }
+
+    xml_start_comment();
+    xml_write_escaped(" %s %s scan initiated %s as: %s ", NMAP_NAME, NMAP_VERSION, mytime, join_quoted(argv, argc).c_str());
+    xml_end_comment();
     xml_newline();
+
+    xml_open_start_tag("nmaprun");
+    xml_attribute("scanner", "nmap");
+    xml_attribute("args", "%s", join_quoted(argv, argc).c_str());
+    xml_attribute("start", "%lu", (unsigned long) timep);
+    xml_attribute("startstr", "%s", mytime);
+    xml_attribute("version", "%s", NMAP_VERSION);
+    xml_attribute("xmloutputversion", NMAP_XMLOUTPUTVERSION);
+    xml_close_start_tag();
+    xml_newline();
+
+    output_xml_scaninfo_records(&ports);
+
+    xml_open_start_tag("verbose");
+    xml_attribute("level", "%d", o.verbose);
+    xml_close_empty_tag();
+    xml_newline();
+    xml_open_start_tag("debugging");
+    xml_attribute("level", "%d", o.debugging);
+    xml_close_empty_tag();
+    xml_newline();
+  } else {
+    xml_start_tag("nmaprun", false);
   }
 
   std::string command;
@@ -1877,37 +1907,11 @@ int nmap_main(int argc, char *argv[]) {
     command += argv[i];
   }
 
-  xml_start_comment();
-  xml_write_escaped(" %s %s scan initiated %s as: %s ", NMAP_NAME, NMAP_VERSION, mytime, join_quoted(argv, argc).c_str());
-  xml_end_comment();
-  xml_newline();
-
   log_write(LOG_NORMAL | LOG_MACHINE, "# ");
   log_write(LOG_NORMAL | LOG_MACHINE, "%s %s scan initiated %s as: ", NMAP_NAME, NMAP_VERSION, mytime);
   log_write(LOG_NORMAL | LOG_MACHINE, "%s", command.c_str());
   log_write(LOG_NORMAL | LOG_MACHINE, "\n");
 
-  xml_open_start_tag("nmaprun");
-  xml_attribute("scanner", "nmap");
-  xml_attribute("args", "%s", join_quoted(argv, argc).c_str());
-  xml_attribute("start", "%lu", (unsigned long) timep);
-  xml_attribute("startstr", "%s", mytime);
-  xml_attribute("version", "%s", NMAP_VERSION);
-  xml_attribute("xmloutputversion", NMAP_XMLOUTPUTVERSION);
-  xml_close_start_tag();
-  xml_newline();
-
-  output_xml_scaninfo_records(&ports);
-
-  xml_open_start_tag("verbose");
-  xml_attribute("level", "%d", o.verbose);
-  xml_close_empty_tag();
-  xml_newline();
-  xml_open_start_tag("debugging");
-  xml_attribute("level", "%d", o.debugging);
-  xml_close_empty_tag();
-  xml_newline();
-
   /* Before we randomize the ports scanned, lets output them to machine
      parseable output */
   if (o.verbose)
@@ -2328,13 +2332,22 @@ int gather_logfile_resumption_state(char *fname, int *myargc, char ***myargv) {
   if (!q || ((unsigned int) (q - p) >= sizeof(nmap_arg_buffer) - 32))
     fatal("Unable to parse supposed log file %s.  Perhaps the Nmap execution had not finished at least one host?  In that case there is no use \"resuming\"", fname);
 
-
   strncpy(nmap_arg_buffer, "nmap --append-output ", sizeof(nmap_arg_buffer));
   if ((q - p) + 21 + 1 >= (int) sizeof(nmap_arg_buffer))
     fatal("0verfl0w");
   memcpy(nmap_arg_buffer + 21, p, q - p);
   nmap_arg_buffer[21 + q - p] = '\0';
 
+  q = strstr(nmap_arg_buffer, "-->");
+  if (q) {
+    *q = '\0';
+     char *unescaped = xml_unescape(nmap_arg_buffer);
+     if (sizeof(nmap_arg_buffer) < strlen(unescaped) + 1)
+       fatal("0verfl0w");
+     memcpy(nmap_arg_buffer, unescaped, strlen(unescaped) + 1);
+     free(unescaped);
+  }
+
   if (strstr(nmap_arg_buffer, "--randomize-hosts") != NULL) {
     error("WARNING: You are attempting to resume a scan which used --randomize-hosts.  Some hosts in the last randomized batch may be missed and others may be repeated once");
   }
@@ -2360,37 +2373,52 @@ int gather_logfile_resumption_state(char *fname, int *myargc, char ***myargv) {
       fatal("Unable to parse supposed log file %s.  Sorry", fname);
     *q = ' ';
   } else {
-    /* OK, I guess (hope) it is a normal log then (-oN) */
+    /* Let's see if it's an XML log (-oX) */
     q = p;
     found = NULL;
-    while ((q = strstr(q, "\nNmap scan report for ")))
+    while ((q = strstr(q, "\n<address addr=\"")))
-      found = q = q + 22;
+      found = q = q + 16;
-
-    /*  There may be some later IPs of the form :
-        "Nmap scan report for florence (x.x.7.10)" (dns reverse lookup)
-        or "Nmap scan report for x.x.7.10".
-    */
     if (found) {
-      q = strchr(found, '\n');
+      q = strchr(found, '"');
       if (!q)
         fatal("Unable to parse supposed log file %s.  Sorry", fname);
       *q = '\0';
-      p = strchr(found, '(');
+      if (inet_pton(AF_INET, found, &lastip) == 0)
-      if (!p) { /* No DNS reverse lookup, found should already contain IP */
+        fatal("Unable to parse supposed log file %s.  Sorry", fname);
-        lastipstr = strdup(found);
+      *q = '"';
-      } else { /* DNS reverse lookup, IP is between parentheses */
-        *q = '\n';
-        q--;
-        *q = '\0';
-        lastipstr = strdup(p + 1);
-      }
-      *q = p ? ')' : '\n'; /* recover changed chars */
-      if (inet_pton(AF_INET, lastipstr, &lastip) == 0)
-        fatal("Unable to parse ip (%s) in supposed log file %s.  Sorry", lastipstr, fname);
-      free(lastipstr);
     } else {
-      error("Warning: You asked for --resume but it doesn't look like any hosts in the log file were successfully scanned.  Starting from the beginning.");
+      /* OK, I guess (hope) it is a normal log then (-oN) */
-      lastip.s_addr = 0;
+      q = p;
+      found = NULL;
+      while ((q = strstr(q, "\nNmap scan report for ")))
+        found = q = q + 22;
+
+      /*  There may be some later IPs of the form :
+          "Nmap scan report for florence (x.x.7.10)" (dns reverse lookup)
+          or "Nmap scan report for x.x.7.10".
+      */
+      if (found) {
+        q = strchr(found, '\n');
+        if (!q)
+          fatal("Unable to parse supposed log file %s.  Sorry", fname);
+        *q = '\0';
+        p = strchr(found, '(');
+        if (!p) { /* No DNS reverse lookup, found should already contain IP */
+          lastipstr = strdup(found);
+        } else { /* DNS reverse lookup, IP is between parentheses */
+          *q = '\n';
+          q--;
+          *q = '\0';
+          lastipstr = strdup(p + 1);
+        }
+        *q = p ? ')' : '\n'; /* recover changed chars */
+        if (inet_pton(AF_INET, lastipstr, &lastip) == 0)
+          fatal("Unable to parse ip (%s) in supposed log file %s.  Sorry", lastipstr, fname);
+        free(lastipstr);
+      } else {
+        error("Warning: You asked for --resume but it doesn't look like any hosts in the log file were successfully scanned.  Starting from the beginning.");
+        lastip.s_addr = 0;
+      }
     }
   }
   o.resume_ip = lastip;

xml.cc

@@ -201,6 +201,68 @@ struct xml_writer {
 
 static struct xml_writer xml;
 
+char *xml_unescape(const char *str) {
+  char *result = NULL;
+  size_t n = 0, len;
+  const char *p;
+  int i;
+
+  i = 0;
+  for (p = str; *p != '\0'; p++) {
+    const char *repl;
+    char buf[32];
+
+    if (*p != '&') {
+      /* Based on the asumption that ampersand is only used for escaping. */
+      buf[0] = *p;
+      buf[1] = '\0';
+      repl = buf;
+    } else if (strncmp(p, "<", 4) == 0) {
+      repl = "<";
+      p += 3;
+    } else if (strncmp(p, "&gt;", 4) == 0) {
+      repl = ">";
+      p += 3;
+    } else if (strncmp(p, "&amp;", 5) == 0) {
+      repl = "&";
+      p += 4;
+    } else if (strncmp(p, "&quot;", 6) == 0) {
+      repl = "\"";
+      p += 5;
+    } else if (strncmp(p, "&apos;", 6) == 0) {
+      repl = "\'";
+      p += 5;
+    } else if (strncmp(p, "&#45;", 5) == 0) {
+      repl = "-";
+      p += 4;
+    } else {
+      /* Escaped control characters and anything outside of ASCII. */
+      Strncpy(buf, p + 3, sizeof(buf));
+      char *q;
+      q = strchr(buf, ';');
+      if(!q)
+        buf[0] = '\0';
+      else
+        *q = '\0';
+      repl = buf;
+    }
+
+    len = strlen(repl);
+    /* Double the size of the result buffer if necessary. */
+    if (i == 0 || i + len > n) {
+      n = (i + len) * 2;
+      result = (char *) safe_realloc(result, n + 1);
+    }
+    memcpy(result + i, repl, len);
+    i += len;
+  }
+  /* Trim to length. (Also does initial allocation when str is empty.) */
+  result = (char *) safe_realloc(result, i + 1);
+  result[i] = '\0';
+
+  return result;
+}
+
 /* Escape a string for inclusion in XML. This gets <>&, "' for attribute
    values, -- for inside comments, and characters with value > 0x7F. It
    also gets control characters with value < 0x20 to avoid parser
@@ -362,9 +424,10 @@ int xml_close_pi() {
 /* Open a start tag, like "<name". The tag must be later closed with
    xml_close_start_tag or xml_close_empty_tag. Usually the tag is closed
    after writing some attributes. */
-int xml_open_start_tag(const char *name) {
+int xml_open_start_tag(const char *name, const bool write) {
   assert(!xml.tag_open);
-  log_write(LOG_XML, "<%s", name);
+  if (write)
+    log_write(LOG_XML, "<%s", name);
   xml.element_stack.push_back(name);
   xml.tag_open = true;
   xml.root_written = true;
@@ -372,9 +435,10 @@ int xml_open_start_tag(const char *name) {
   return 0;
 }
 
-int xml_close_start_tag() {
+int xml_close_start_tag(const bool write) {
   assert(xml.tag_open);
-  log_write(LOG_XML, ">");
+  if(write)
+    log_write(LOG_XML, ">");
   xml.tag_open = false;
 
   return 0;
@@ -392,10 +456,10 @@ int xml_close_empty_tag() {
   return 0;
 }
 
-int xml_start_tag(const char *name) {
+int xml_start_tag(const char *name, const bool write) {
-  if (xml_open_start_tag(name) < 0)
+  if (xml_open_start_tag(name, write) < 0)
     return -1;
-  if (xml_close_start_tag() < 0)
+  if (xml_close_start_tag(write) < 0)
     return -1;
 
   return 0;

xml.h

@@ -137,10 +137,10 @@ int xml_end_comment();
 int xml_open_pi(const char *name);
 int xml_close_pi();
 
-int xml_open_start_tag(const char *name);
+int xml_open_start_tag(const char *name, const bool write = true);
-int xml_close_start_tag();
+int xml_close_start_tag(const bool write = true);
 int xml_close_empty_tag();
-int xml_start_tag(const char *name);
+int xml_start_tag(const char *name, const bool write = true);
 int xml_end_tag();
 
 int xml_attribute(const char *name, const char *fmt, ...) __attribute__ ((format (printf, 2, 3)));
@@ -151,5 +151,8 @@ int xml_depth();
 bool xml_tag_open();
 bool xml_root_written();
 
+
+char *xml_unescape(const char *str);
+
 #endif