# Copyright (c) 2010 Flowerfire, Inc.  All Rights Reserved.
iisweb = {

  plugin_version = "2.0"

  info.1.manufacturer = "Microsoft"
  info.1.device = "IIS Extended (W3C)"
  info.1.version = ""

  # 2007-07-06 - 1.1 - KBB - Added broken_link field and log filter to set it
  # 2010-10-01 - 1.2 - MSG - Edited info lines.
  # 2012-07-10 - 1.3 - GMF - Added web server package snapon
  # 2012-08-03 - 1.4 - GMF - Added page parameter to web server package snapon
  # 2012-08-06 - 1.5 - GMF - Switched web_server_package to implement page_views and referrer analysis and agent analysis; turned off hit_type (it can later be implemented in web_server_package, if desired).
  # 2012-08-06 - 1.6 - GMF - Implemented file_type and worm with web_server_package
  # 2013-02-15 - 2.0 - GMF - Switched to log.filter_preprocessor for header parsing, because built-in parsing makes c-ip hierarchical, and that breaks web_server_package geo analysis

  # The name of the log format
  log.format.format_label = "IIS Extended (W3C) Web Server Log Format"
  log.miscellaneous.log_data_type = "generic_w3c"  
  log.miscellaneous.log_format_type = "web_server"

  # The log is in this format if any of the first ten lines match this regular expression
  log.format.autodetect_regular_expression = "^#Software: Microsoft Internet Information (Server|Services) [4-9]"

  # The format of dates and times in this log
  log.format.date_format = "yyyy-mm-dd"
  log.format.time_format = "hh:mm:ss"

  log.format.ignore_format_lines = "true"

  # Don't track these fields as discrete database fields.  Don't add sc_bytes as a database field here, because it will be added by the web_server_package snapon.
  auto_setup.omit_database_fields = "cs_uri_query,cs_cookie,sc_bytes,cs_bytes"

  # Look through a lot of lines to make sure we get past any process accounting information
  log.format.autodetect_lines = "200"

  # IIS does not use quotes to quote fields, so treat them as actual field values if we see them
  log.format.ignore_quotes = "true"

  # 2013-02-15 - GMF - Transplanted this from generic_w3c
  # This handles #Fields lines, and creates log and database fields from them
  log.filter_preprocessor = `
if (matches_regular_expression(current_log_line(), '^#Fields: +([^ ].*)$')) then (
  string fields = $1;
  string fieldname;
  v.logfieldindex = 1;
  string numerical_fields = "profiles." . internal.profile_name . ".database.numerical_fields";

  # This subroutine creates a database field
  subroutine(create_database_field(string fieldname, int top, int bottom), (
    #echo("create_database_field: " . fieldname); # debug
    debug_message("create_database_field(" . fieldname . ")\n");
    string databasefieldpath = "profiles." . internal.profile_name . ".database.fields." . fieldname;
    (databasefieldpath . "") = "";
    node databasefield = databasefieldpath;
    if (top ne 0) then
      set_subnode_value(databasefield, "suppress_top", top);
    if (bottom ne 0) then
      set_subnode_value(databasefield, "suppress_bottom", bottom);
    databasefield;
  ));

  subroutine(create_log_field(string fieldname, string type, bool withindex), (
    debug_message("create_log_field(" . fieldname . "; type=" . type . ")\n");
    string logfieldpath = "profiles." . internal.profile_name . ".log.fields." . fieldname;
    (logfieldpath . "") = "";
    node logfield = logfieldpath;
    if (withindex) then (
      set_subnode_value(logfield, "index", v.logfieldindex);
      v.logfieldindex++;
    );
    set_subnode_value(logfield, "subindex", 0);
    if (type ne '') then
      set_subnode_value(logfield, "type", type);
    logfield;
  ));

  # Assume there isn't a localtime field until we see one.
  v.parse_localtime = false;

  # Get the log format separator. Assume at least two fields.
  # 2010-02-01 - GMF - Changed the default from " " to "" (any whitespace).  With " " as the default, data which
  #                    has a spaces in the header, but tabs in the data, doesn't parse, because this assumes space
  #                    separators.  Truly generic W3C header parsing appears to be impossible, which is why we can do
  #                    per-plugin parsing now; but that's just moved the problem here, where we'll still have to
  #                    fight it forever.  Hopefully, "" gives better coverage to the universe of W3C variants than " ".
#  log.format.field_separator = " ";
  log.format.field_separator = "";
  if (matches_regular_expression(fields, '^[^ 	]+	[^ 	]+')) then (
    log.format.field_separator = "	";
  );

  # Extract the fields on at a time
  while (matches_regular_expression(fields, '^([^ 	]+)[ 	](.*)$') or matches_regular_expression(fields, '^([^ 	]+)$')) (

    string unconverted_fieldname = $1;
    fields = $2;

    # Clean up the field name
    fieldname = '';
    for (int i = 0; i < length(unconverted_fieldname); i++) (
      string c = lowercase(substr(unconverted_fieldname, i, 1));
      if (!matches_regular_expression(c, '^[a-z0-9]$')) then
        c = '_';
      fieldname .= c;
    );
    while (matches_regular_expression(fieldname, '^(.*)_$'))
      fieldname = $1;

    # Get the log field type
    string log_field_type = '';
    if (fieldname eq 'cs_referrer') then (
      fieldname = 'cs_referer';
    );
    if (fieldname eq 'cs_uri') then (
      fieldname = 'cs_uri_stem';
    );
    if (fieldname eq 'cs_referer') then (
      log_field_type = 'url';
    );
    else if (fieldname eq 'cs_uri_stem') then (
      log_field_type = 'page';
    );
    else if (fieldname eq 'cs_user_agent') then (
      log_field_type = 'agent';
    );
    else if (fieldname eq 'c_ip') then (
      log_field_type = 'host';
    );

    # Create the log field
    if ((fieldname ne "sc_bytes") and (fieldname ne "cs_bytes")) then
      create_log_field(fieldname, log_field_type, true);

    if (fieldname eq "localtime") then
      v.parse_localtime = true;

    # If we're creating a profile, create the database fields too.
    if (node_exists("volatile.creating_profile")) then (

      # Handle localtime by creating date_time and derived database fields
      if (fieldname eq "localtime") then (
        create_log_field('date', '', false);
        create_log_field('time', '', false);
        create_database_field('date_time', 0,  0);
        create_database_field('day_of_week', 0,  0);
        create_database_field('hour_of_day', 0,  0);
      ); # if localtime

      if ((fieldname eq "sc_bytes") or (fieldname eq "cs_bytes")) then
      ( );

      # Handle date by creating date_time and derived database fields
      else if (fieldname eq "date") then (
        create_log_field('localtime', '', false); # placeholder - 7/Nov/2006 - KBB
        create_database_field('date_time', 0,  0);
        create_database_field('day_of_week', 0,  0);
        create_database_field('hour_of_day', 0,  0);
      ); # if date

      else if (fieldname eq "time") then (
        create_database_field('date_time', 0,  0);
        create_database_field('day_of_week', 0,  0);
        create_database_field('hour_of_day', 0,  0);
      ); # if time

      # Create derived fields for agent
      else if (fieldname eq "cs_user_agent") then (
        create_database_field('operating_system', 0,  0);
        create_database_field('web_browser', 0,  0);
        create_database_field('spider', 0,  0);
      );

      # Create database field cs_ip and derived field for client IP 
      else if (fieldname eq "c_ip") then (
        create_database_field('c_ip', 0,  0);
        create_database_field('location', 0,  0);
      );

      # Create database field cs_referer and derived fields for referrer
      else if (fieldname eq "cs_referer") then (
        create_database_field('cs_referer', 1,  9);
        create_database_field('search_engine', 0,  0);
        create_database_field('search_phrase', 0,  0);
      );

      # Create derived file type field
      else if (fieldname eq "cs_uri_stem") then (
        create_database_field('cs_uri_stem', 0,  9);
        create_database_field('file_type', 0,  0);
      );

      # Don't add a database field for numerical fields
      else if (subnode_exists(numerical_fields, fieldname)) then (
        debug_message("Not adding numerical field: " . fieldname . "\n");
      );

      # Create a normal database field
      else
        create_database_field(fieldname, 0,  0);

    ); # if creating profile

  ); # while another field

  # Don't parse the #Fields line as a data line
  'reject';

); # if #Fields

# Don't parse any other # lines as data lines
else if (starts_with(current_log_line(), '#')) then (
  'reject';
);
`

  # Log Filters
  log.filters = {

    empty_uri_query = {
      label = "$lang_admin.log_filters.empty_uri_query_label"
      comment = "$lang_admin.log_filters.empty_uri_query_comment"
      value = "if (cs_uri_query eq '-') then cs_uri_query = '(empty)';"
      disabled = true
      requires_fields = {
        cs_uri_query = true
      }
    } # empty_uri_query

    add_cs_uri_query = {
      label = "$lang_admin.log_filters.add_cs_uri_query_label"
      comment = "$lang_admin.log_filters.add_cs_uri_query_comment"
      value = "cs_uri_stem = cs_uri_stem . '?' . cs_uri_query"
      disabled = true
      requires_fields = {
        cs_uri_query = true
        cs_uri_stem = true
      }
    } # add_cs_uri_query

  } # log.filters

  database.numerical_fields = {

    time_taken = {
      type = "int"
      integer_bits = 64
      display_format_type = duration_milliseconds
    } # time_taken

#    time_taken_avg = {
#      label = "$lang_stats.field_labels.average $lang_stats.field_labels.time_taken"
#      default = false
#      log_field = "time_taken"
#      requires_log_field = true
#      type = "int"
#      integer_bits = 64
#      aggregation_method = "average"
#      average_denominator_field = "hits"
#      display_format_type = duration_milliseconds
#    } # time_taken

  } # database.numerical_fields


  create_profile_wizard_options = {

    # How the reports should be grouped in the report menu
    report_groups = {
      date_time_group = ""
      content_group = {
        cs_uri_stem = true
        cs_uri_query = true
      }
      visitor_demographics_group = {
        c_ip = true
        cs_username = true
      }
      referrer_group = {
        referrer = true
      }
      server_group = {
        s_sitename = true
        s_computername = true
        s_ip = true
        s_port = true
        cs_host = true
      }
      other_group = {
        cs_version = true
#        spider = true
        cs_method = true
        sc_status = true
        sc_substatus = true
        sc_win32_status = true
        cs_cookie = true
      }
    } # report_groups

    # This plug-in does its derived fields manually--don't automatically create operating_system, web_browser, etc.
    manual_derived_fields = true

    snapons = {

      # Attach a web_server_package snapon
      web_server_package = {
	snapon = "web_server_package"
	name = "web_server_package"
	label = "$lang_admin.snapons.web_server_package.label"
        parameters = {
          user_agent_field.parameter_value = "cs_user_agent"
          page_field.parameter_value = "cs_uri_stem"
          client_ip_field.parameter_value = "c_ip"
          server_response_field.parameter_value = "sc_status"
          referrer_field.parameter_value = "referrer"
          authenticated_user_field.parameter_value = "cs_username"
          hits_field = {
            parameter_value = "$lang_stats.field_labels.hits"
            final_node_name = "hits"
          } # hits
          visitors_field = {
            parameter_value = "$lang_stats.field_labels.unique_client_ips"
            final_node_name = "unique_client_ips"
          } # visitors_field
          sc_bytes_field = {
            parameter_value = "$lang_stats.field_labels.sc_bytes"
            final_node_name = "sc_bytes"
          } # sc_bytes_field
          cs_bytes_field = {
            parameter_value = "$lang_stats.field_labels.cs_bytes"
            final_node_name = "cs_bytes"
          } # cs_bytes_field
        } # parameters
      } # web_server_package

    } # snapons

#    final_step = `
#include "templates.admin.profiles.setup_reports_util";
#
#string profile = "profiles." . volatile.new_profile_name;
#
## Remove cs_uri_query and cs_cookie from the database fields
#delete_database_field(profile, 'cs_uri_query');
#delete_database_field(profile, 'cs_cookie');
#
## Create the standard reports
#add_standard_reports(profile);
#
## Delete referrer filters if there's no referrer field
##if (!node_exists(profile . ".database.fields.referrer")) then (
##  (profile . ".log.filters.no_referrer.disabled") = true;
##  (profile . ".log.filters.internal_referrer.disabled") = true;
##);
#
#    ` # end final_step

  } # create_profile_wizard_options

##
##  Uncommenting this section will provide enhanced granularity visitor tracking by over-riding
##  default visitor tracking and uniqueness using c_ip with c_ip + cs_user_agent in
##  all reports including sessions analysis reporting
##
##  log.fields.visitor_id = ""
##  database.fields.visitor_id = ""
##  database.numerical_fields.visitors.log_field = "visitor_id"
##
##  log.filters.visitor_id = {
##      label = 'Client IP + user Agent'
##      comment = 'Provides better tracking of unique visitors when multiple users are coming from the same client IP (ISP/NAT/FIREWALL/ROUTER)'
##      value = 'visitor_id = c_ip . "+" . cs_user_agent;'
##  } # log.filters.visitor_id
##
##  log.field_options.sessions_visitor_id_field = "visitor_id"

} # iisweb