CentOS Advisory Parser (from September 2009, never got a chance to put this into production)

centos-advisory-parser.pl

#!/usr/bin/perl

$rc = 0;

while(<STDIN>) {
  chomp;

  ## Grab severity, CentOS version, architecture, and package name
  ##
  ## Sometimes, whitespace is missing between package name and "Update"
  ## e.g. CESA-2009:1335 had "opensslUpdate" instead of "openssl Update"
  ##
  ## Account for fact that mailman wraps Subject: line by not requiring "Update" at the end of line
  ## Also account for fact that Tru Huynh's e-mail subjects use "package - type update" format
  if ($_ =~ /^Subject:\s+\[CentOS-announce\]\s+\S+\s+(\w*)\s*CentOS\s+(\d+)\s+(x86_64|i386|i686)\s+(\S+)\s*(Update$|$|\-\s*$)/) {
    ## Severity may be missing, replace with "Unknown" if it is
    ## e.g. CESA-2009:1202, CESA-2009:1165, CESA-2009:1192, CESA-2009:1191, CESA-2009:1196, CESA-2009:1214
    ##      CESA-2009:1215, and CESA-2009:1212 had no severity in subject lines
    $severity = $1;
    if ($severity eq "") {
      $severity = "Unknown";
    }

    $version = $2;
    $arch = $3;
    $package = $4;
    if ($package =~ /(\S*)Update/) {
      $package = $1;
    }
  }

  ## Don't trust CentOS advisory number in subject, has had typos in the past
  ## e.g. "CESA-2009:1232" e-mailed out as "CESA-2009:123"
  ##      "CESA-2009:1165" e-mailed out as "CEBA-2009:1165"
  ##      "CESA-2009:1192" e-mailed out as "CEBA-2009:1192"
  ##      "CESA-2009:1191" e-mailed out as "CEBA-2009:1191"
  ##      "CESA-2009:1196" e-mailed out as "CEBA-2009:1196"
  ##      "CESA-2009:1202" e-mailed out as "CEBA-2009:1202"
  ##      "CESA-2009:1212" e-mailed out as "CEBA-2009:1212"
  ##      "CESA-2009:1214" e-mailed out as "CEEA-2009:1214"
  ##
  ## Do minimal matching of RHN URL, has had typos in the past
  ## e.g. CESA-2009:1212 had URI of "RHBA-2009-1212.html" instead of "RHSA-2009-1212.html"
  if ($_ =~ /(\d+)-(\d+)\.html/) {
    $advisory = sprintf("%04d-%d", $1, $2);
  }

  if ($_ =~ /\s*(.*)\.rpm/) {
    $rpm = $1;

    ## extra chopping: Karanbir Singh's e-mails carry MD5 hashes (which shouldn't be used)
    if ($rpm =~ /\s+(.*)/) {
      $rpm = $1;
    }

    ## extra chopping: Tru Huynh's e-mails carry relative paths, grab the last part of filename
    if ($rpm =~ /\//) {
      @rp = split("\/", $rpm);
      $rpm = $rp[$#rp];
    }

    $rpmfiles[$rc++] = sprintf("%s.rpm", $rpm);
  }
}

## Normalize our output and print
printf("CentOS %d/%s Advisory (CESA:%s)\n", $version, $arch, $advisory);
printf("Package: %s\n", $package);
printf("Severity: %s\n", $severity);

printf("\nFiles Updated in CentOS %d/%s Repository:\n", $version, $arch);
foreach $file (@rpmfiles) {
  printf("\t%s\n", $file);
}

printf("\nFor more information, please see https://rhn.redhat.com/errata/RHSA-%s.html\n", $advisory);