From: arjen <arjen>
Date: Sat, 29 Mar 2003 09:04:10 +0000 (+0000)
Subject: Extract the hostname out of the 'From:' or 'Message-Id:' line
X-Git-Tag: V0_0_6~10
X-Git-Url: http://www.andromeda.nl/gitweb/?p=gnucomo.git;a=commitdiff_plain;h=2d3b3b083f9be0ceff1010464d0c0eb81b681571

Extract the hostname out of the 'From:' or 'Message-Id:' line
of an email header.
---

diff --git a/src/gcm_input/message.cpp b/src/gcm_input/message.cpp
index 0e9dca9..55c1d1a 100644
--- a/src/gcm_input/message.cpp
+++ b/src/gcm_input/message.cpp
@@ -8,7 +8,7 @@
 ***********************
 **      FILE NAME      : message.cpp
 **      SYSTEM NAME    : Gnucomo - Gnu Computer Monitoring
-**      VERSION NUMBER : $Revision: 1.8 $
+**      VERSION NUMBER : $Revision: 1.9 $
 **
 **  DESCRIPTION      :  Implementation of the message handling classes
 **
@@ -20,13 +20,17 @@
 ********************************
 **      ORIGINAL AUTHOR : Arjen Baart - arjen@andromeda.nl
 **      CREATION DATE   : Sep 16, 2002
-**      LAST UPDATE     : Feb 28, 2003
+**      LAST UPDATE     : Mar 28, 2003
 **      MODIFICATIONS   : 
 **************************************************************************/
 
 /*****************************
    $Log: message.cpp,v $
-   Revision 1.8  2003-03-16 09:42:40  arjen
+   Revision 1.9  2003-03-29 09:04:10  arjen
+   Extract the hostname out of the 'From:' or 'Message-Id:' line
+   of an email header.
+
+   Revision 1.8  2003/03/16 09:42:40  arjen
    Read IRIX system logs.
 
    Revision 1.7  2003/02/21 08:08:05  arjen
@@ -57,7 +61,7 @@
 
 *****************************/
 
-static const char *RCSID = "$Id: message.cpp,v 1.8 2003-03-16 09:42:40 arjen Exp $";
+static const char *RCSID = "$Id: message.cpp,v 1.9 2003-03-29 09:04:10 arjen Exp $";
 
 #include <algorithm>
 #include "message.h"
@@ -127,6 +131,7 @@ client_message::client_message(std::istream *in, gnucomo_database db)
 static const String syslog_date_re("[[:alpha:]]{3} [ 123][0-9] [0-9]{2}:[0-9]{2}:[0-9]{2}");
 static const String mail_date_re("[[:alpha:]]{3}, [ 123]?[0-9] [[:alpha:]]{3} [0-9]{4} [0-9]{2}:[0-9]{2}:[0-9]{2} [+-][0-9]{4}");
 static const String unix_date_re("[[:alpha:]]{3} [[:alpha:]]{3} [ 123][0-9] [0-9]{2}:[0-9]{2}:[0-9]{2} [0-9]{4}");
+static const String email_address_re("[[:alnum:]_.-]+@[[:alnum:]_.-]+");
 
 static const regex re_syslog(syslog_date_re + " [[:alnum:]]+ [[:alpha:]]+.*:.+");
 static const regex re_syslog_irix(syslog_date_re + " [0-7][A-T]:[[:alnum:]]+ [[:alpha:]]+.*:.+");
@@ -137,9 +142,10 @@ static const regex re_errorlog("^\\[" + unix_date_re + "\\] \\[(error|notice)\\]
 static const regex re_rpm("[[:alnum:]+-]+-[0-9][[:alnum:].-]");
 
 static const regex re_syslog_date("[[:alpha:]]{3} [ 123][0-9] [0-9]{2}:[0-9]{2}:[0-9]{2}");
-static const regex re_uxmail_from("^From - " + unix_date_re);
+static const regex re_uxmail_from("^From [^ \t]+[ ]+" + unix_date_re);
 static const regex re_mail_From("^From:[[:blank:]]+");
 static const regex re_mail_Date("^Date:[[:blank:]]+" + mail_date_re);
+static const regex re_mail_MsId("^Message-Id:[[:blank:]]+");
 static const regex re_email_address("[[:alnum:]_.-]+@[[:alnum:]_.-]+");
 static const regex re_email_user("[[:alnum:]_.-]+@");
 
@@ -155,7 +161,7 @@ static const regex re_email_user("[[:alnum:]_.-]+@");
 **  VARS CHANGED   :
 **  FUNCTIONS USED :
 **  SEE ALSO       :
-**  LAST MODIFIED  : Nov 16, 2002
+**  LAST MODIFIED  : Mar 28, 2003
 **=========================================================================
 */
 
@@ -183,7 +189,19 @@ double client_message::classify(String host, UTC arriv, String serv)
          {
             from_address = line(re_email_address);
             from_address(re_email_user) = "";            //  Remove the user part;
-            hostname = from_address;
+            if (from_address != "")
+            {
+               hostname = from_address;
+            }
+         }
+         if (line == re_mail_MsId)
+         {
+            from_address = line(re_email_address);
+            from_address(re_email_user) = "";            //  Remove the user part;
+            if (from_address != "")
+            {
+              hostname = from_address;
+            }
          }
          if (line == re_mail_Date)
          {
@@ -195,6 +213,7 @@ double client_message::classify(String host, UTC arriv, String serv)
    {
       //  Push the first line back, we need to read it again.
       --input;
+
    }
 
    /*
@@ -301,7 +320,7 @@ double client_message::classify(String host, UTC arriv, String serv)
 **  VARS CHANGED   :
 **  FUNCTIONS USED :
 **  SEE ALSO       :
-**  LAST MODIFIED  : Feb 19, 2003
+**  LAST MODIFIED  : Mar 28, 2003
 **=========================================================================
 */
 
@@ -651,6 +670,10 @@ int client_message::enter()
                i++;
             }
             
+            if (!isdigit(line[version_start + 1]))
+            {
+               version_start = next_version_start;
+            }
             String package(line(0,version_start));
             String version(line(version_start + 1, ~line));
             String paramid;