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

  plugin_version = "2.0.4"

  info.1.manufacturer = "Juniper"
  info.1.device = "Media Flow Controller (Access Logs) (NCSA)"
  info.1.version.1 = ""

  # 2011/10/27 - 1.0 - GMF - cloned apache_combined
  # 2011/11/18 - 1.0.1 - GMF - improved autodetect to look for the final numerical field
  # 2011-11-24 - 1.0.2 - GMF - Modified autodetection to look for this specific #Fields line, as requested ThreadID:1206750.
  # 2011-12-03 - 1.0.3 - GMF - Changed name to "Media Flow Controller" at Juniper's suggestion.
  # 2011-12-03 - 1.0.4 - GMF - Changed name to "Media Flow Controller Access Log" at Juniper's suggestion.
  # 2012-07-30 - 2.0 - GMF - Enhanced plug-in to handle any #Fields line, and to set up log fields, database fields, parsing, and filters appropriately.
  # 2012-08-01 - 2.0.1 - GMF - Added support for quotes around field values
  # 2012-08-10 - 2.0.2 - GMF - Changed name to have "Access Logs" in parentheses
  # 2012-09-18 - 2.0.3 - GMF - Added NCSA to label
  # 2012-10-19 - 2.0.4 - GMF - Added plugin_description

  # The name of the log format
  log.format.format_label = "Juniper Media Flow Controller Access Log Format (NCSA)"
  log.miscellaneous.log_data_type = "http_access"
  log.miscellaneous.log_format_type = "web_server"

  # Description contributed by Juniper
  create_profile_wizard_options.plugin_description = `
<p>This plugin detects and analyzes Media Flow Controller Access Logs that are in the NCSA Combined Log Format followed by many popular web servers including Apache. Juniper Media Flow Controller version 11.A and earlier releases generated access logs that conformed to the NCSA standard. All access logs recorded the following fields that collectively referred to as the  as Combined Log Format (CLF):</p>
<ul>
                %h %V %u %t "%r" %s %b "%{Referer}i" "%{User-Agent}i" %y
</ul>
<p>Please see Media Flow Controller CLI Reference Guide for a description of these fields.  Later releases of Juniper Media Flow Controller support the more flexible W3C access log format which is supported by a different Sawmill plug-in.</p>
<p>The reports generated by this plugin can be categorized as:</p>
<ul>
  <li>Date and Time based reports such as daily, weekly, monthly & yearly reports of user</li>
  <li>Content analysis reports based on content popularity, server domain(s), server response status and content referrers</li>
  <li>User demographic reports based on user's Country/State/City/Region, and ISP</li>
</ul>
</p>
`

  # The log is in this format if any of the first ten lines match this regular expression
#  log.format.autodetect_regular_expression = "^[^ ]* [^ ]* .* \\[../.../....[: ]..:..:...*\\] \"[A-Z]* [^ ]*(\"| HTTP.*\") [0-9]* [-0-9]* [^ ]* \"[^\"]*\" [0-9]+$"
#  log.format.autodetect_regular_expression = `#Fields: %h %V %u %t "%r" %s %b "%[{]Referer[}]i" "%[{]User-Agent[}]i" %y`
  log.format.autodetect_regular_expression = `^#Fields: .*%h`

  # Treat fields surrounded by square brackets (e.g. the date/time field) as a single quoted field.
#  log.format.treat_brackets_as_quotes = "true"
#  log.format.common_log_format = "true"
  log.format.ignore_format_lines = "true"

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

  # 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";

  log.format.parsing_regular_expression = '^';

  node log_fields_at_end = new_node();
  subroutine(create_log_field_at_end(node log_fields_at_end, string fieldname, string type), (
    log_fields_at_end{fieldname}{"type"} = type;
  ));

  # This subroutine creates a database field
  subroutine(create_database_field(string fieldname), (
#echo("create_database_field: " . fieldname);
    debug_message("create_database_field(" . fieldname . ")\n");
    string databasefieldpath = "profiles." . internal.profile_name . ".database.fields." . fieldname;
    (databasefieldpath . "") = "";
    node databasefield = databasefieldpath;
#    set_subnode_value(databasefield, "label", fieldname);
    databasefield;
  ));

  subroutine(create_log_field(string fieldname, string type), (
#    echo("create_log_field(" . fieldname . "; type=" . type . ")");
    string logfieldpath = "profiles." . internal.profile_name . ".log.fields." . fieldname;
    (logfieldpath . "") = "";
    node logfield = logfieldpath;
#    set_subnode_value(logfield, "label", fieldname);
    if (type ne '') then
      set_subnode_value(logfield, "type", type);
    logfield;
  ));

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

    string format_specifier = $1;
    fields = $2;

    bool quoted = false;
    if (matches_regular_expression(format_specifier, '^"([^"]+)"')) then (
      quoted = true;
      format_specifier = $1;
    );

#echo("format_specifier=" . format_specifier);

    string fieldname;

    # Handle special case hard-coded fields
    string field_regexp = '([^ ]+)';

    if (matches_regular_expression(format_specifier, '^R[0-9]+')) then
      fieldname = 'store_id';
    else if (matches_regular_expression(format_specifier, '^US|EMEA|JAPAN|CHINA|AUS$')) then
      fieldname = 'store_region';
    else if (matches_regular_expression(format_specifier, '^[A-Z][a-z]+-Profile$')) then
      fieldname = 'store_category';

    else if (format_specifier eq '%b') then
      fieldname = 'sc_bytes_content';
    else if (format_specifier eq '%c') then
      fieldname = 'x_cache_hit';
    else if (format_specifier eq '%d') then
      fieldname = 'date';
    else if (format_specifier eq '%f') then
      fieldname = 'cs_uri_stem';
    else if (format_specifier eq '%h') then
      fieldname = 'cs_host';
    else if (format_specifier eq '%m') then
      fieldname = 'cs_method';
    else if (format_specifier eq '%p') then
      fieldname = 'x_hotness';
    else if (format_specifier eq '%q') then
      fieldname = 'cs_uri_query';

    else if (format_specifier eq '%r') then (
      field_regexp = '([^ ]+) ([^ ]+) ([^ ]+)';
      create_log_field('operation', 'flat');
      create_log_field('cs_uri_stem', 'page');
      create_log_field_at_end(log_fields_at_end, 'file_type', 'flat');
      fieldname = 'protocol';
    ) # %r

    else if (format_specifier eq '%s') then
      fieldname = 'sc_status';

    else if (format_specifier eq '%t') then (
      field_regexp = '[[]([^:]+):([^ ]+) [^]]+[]]';
      create_log_field('date', 'date');
      fieldname = 'time';
    );

    else if (format_specifier eq '%u') then
      fieldname = 'user';
    else if (format_specifier eq '%v') then
      fieldname = 'x_server';
    else if (format_specifier eq '%y') then
      fieldname = 'sc_substatus';
    else if (format_specifier eq '%A') then
      fieldname = 'x-request-time';
    else if (format_specifier eq '%B') then
      fieldname = 'x-first-byte-out-time';
    else if (format_specifier eq '%C') then
      fieldname = 'cs(Cookie)';
    else if (format_specifier eq '%D') then
      fieldname = 'time-taken';
    else if (format_specifier eq '%E') then
      fieldname = 'x-time-used-ms';
    else if (format_specifier eq '%F') then
      fieldname = 'x-last-byte-out-time';
    else if (format_specifier eq '%H') then
      fieldname = 'cs-proto';
    else if (format_specifier eq '%I') then
      fieldname = 'cs-bytes';
    else if (format_specifier eq '%L') then
      fieldname = 'x-latency';
    else if (format_specifier eq '%M') then
      fieldname = 'x-data-len-ms';
    else if (format_specifier eq '%N') then
      fieldname = 'x-namespace';
    else if (format_specifier eq '%O') then
      fieldname = 'sc-bytes';
    else if (format_specifier eq '%R') then
      fieldname = 'x-revalidate-cache';
    else if (format_specifier eq '%U') then
      fieldname = 'cs-uri';
    else if (format_specifier eq '%V') then
      fieldname = 'server_domain';
    else if (format_specifier eq '%X') then
      fieldname = 'c-ip';
    else if (format_specifier eq '%Y') then
      fieldname = 's-ip';
    else if (format_specifier eq '%Z') then
      fieldname = 's-port';

    else if (matches_regular_expression(format_specifier, '^%[{]([^}]+)[}]([io]?)$')) then (
      fieldname = $1;
      if ($2 eq 'o') then
        fieldname .= '_out';
      else if ($2 eq 'i') then
        fieldname .= '_in';
    );

    else (
      error("Unknown format specified in #Fields header line: '" . format_specifier . "'");
    );

    if (log.format.parsing_regular_expression ne '^') then
      log.format.parsing_regular_expression .= ' ';
    if (quoted) then
      log.format.parsing_regular_expression .= '"';
    log.format.parsing_regular_expression .= field_regexp;
    if (quoted) then
      log.format.parsing_regular_expression .= '"';

    string unconverted_fieldname = fieldname;

    # 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_uri_stem') then (
      log_field_type = 'page';
      ("profiles." . internal.profile_name . ".log.fields.url.type") = 'flat';
    );
    if (fieldname eq 'cs_host') then (
      log_field_type = 'host';
      create_log_field_at_end(log_fields_at_end, 'location', '');
    );
    if (fieldname eq 'user_agent') then (
      log_field_type = 'agent';
      create_log_field_at_end(log_fields_at_end, 'web_browser', '');
      create_log_field_at_end(log_fields_at_end, 'operating_system', '');
    );
    if (fieldname eq 'cs_referer') then (
      log_field_type = 'url';
      create_log_field_at_end(log_fields_at_end, 'search_engine', '');
      create_log_field_at_end(log_fields_at_end, 'search_phrase', '');
    );

    # Create the log field
    create_log_field(fieldname, log_field_type);

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

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

      # Create derived field for agent
      else if (fieldname eq "cs_user_agent") then (
        create_database_field('operating_system');
        create_database_field('web_browser');
      );

      # Create derived field for IP
      else if (fieldname eq "cs_host") then (
        create_database_field('cs_host');
        create_database_field('location');
      );

      # Create derived fields for referrer
      else if (fieldname eq "cs_referer") then (
        create_database_field('search_engine');
        create_database_field('search_phrase');
      );

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

      # Create derived file type field, and other %r fields
      else if (fieldname eq "protocol") then (
        create_database_field('operation');
        create_database_field('cs_uri_stem');
        create_database_field('file_type');
      );
  
      # Don't add a database field for numerical fields
  #    else if (subnode_exists('database.fields', fieldname)) then (
      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);

    ); # if creating profile
 
  ); # while another field

  # Create any final log fields
  node lfae;
  foreach lfae log_fields_at_end (
#  echo("Final log field creation: " . node_name(lfae));
    create_log_field(node_name(lfae), @lfae{"type"});
  );

  # 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';
);
`

  # Get search engine and search phrase information from the referrer field (before it gets simplified).
  log.parsing_filters.compute_se_sp = {
value = `
if (get_search_engine_info(cs_referer)) then (
  search_engine = volatile.search_engine;
  search_phrase = volatile.search_phrase;
);
`
    requires_fields = {
      cs_referer = true
#      search_engine = true
#      search_phrase = true
    }
  }

  # Get web browser, operating system, web browser, and spider information from the user-agent field.
  log.parsing_filters.derive_from_user_agent = {
    value = `
get_user_agent_info(cs_user_agent);
web_browser = volatile.web_browser;
operating_system = volatile.operating_system;
#spider = volatile.spider;
`
    requires_fields = {
      cs_user_agent = true
    }
  }

  # Log Filters
  log.filters = {

    simplify_referrer = {
      label = "$lang_admin.log_filters.simplify_referrer_label"
      comment = "$lang_admin.log_filters.simplify_referrer_comment"
      value = "if (cs_referer eq '-') then cs_referer = '(no referrer)' else if (matches_regular_expression(cs_referer, '^([^:]+://[^/]+/)')) then cs_referer = $1 . '(omitted)'"
      requires_fields = {
        cs_referer = true
      }
    } # simplify_referrer

    internal_referrer = {
      label = "$lang_admin.log_filters.internal_referrer_label"
      comment = "$lang_admin.log_filters.internal_referrer_comment"
      value = "if (contains(referrer, 'mydomain.com/')) then referrer = '(internal referrer)';"
      requires_fields = {
        cs_referer = true
      }
      disabled = true
    } # internal_referrer

    not_authenticated = {
      label = "$lang_admin.log_filters.not_authenticated_label"
      comment = "$lang_admin.log_filters.not_authenticated_comment"
      value = "if (user eq '-') then user = '(not authenticated)';"
      requires_fields = {
        user = true
      }
    } # not_authenticated

    remove_query = {
      label = "$lang_admin.log_filters.remove_query_label"
      comment = "$lang_admin.log_filters.remove_query_comment"
      value = "if (contains(cs_uri_stem, '?')) then cs_uri_stem = substr(cs_uri_stem, 0, index(cs_uri_stem, '?') + 1) . '(parameters)';"
      requires_fields = {
        cs_uri_stem = true
      }
    } # remove_query

    categorize = {
      label = "$lang_admin.log_filters.categorize_hits_label"
      comment = "$lang_admin.log_filters.categorize_hits_comment"
      value = `if (starts_with(sc_status, '4')) then (
  errors = 1;
#  if (sc_status eq '404') then (
#    broken_links = 1;
#    hit_type = "broken link";
#  )
#  else (
#    hit_type = "error";
#  )
)
#else if (!starts_with(spider, '(')) then (
#  hit_type = "spider";
#  spiders = 1;
#)
#else if (!starts_with(worm, '(')) then (
#  hit_type = "worm";
#  worms = 1;
#)
#else if (!starts_with(screen_dimensions, '(')) then (
#  hit_type = "screen info";
#  screen_info_hits = 1;
#)
else if ((file_type eq 'JPEG') or (file_type eq 'JPG') or (file_type eq 'GIF') or (file_type eq 'ICO') or (file_type eq 'PNG') or (file_type eq 'CSS') or (file_type eq 'SWF') or (file_type eq 'JS')) then (
#  hit_type = "hit";
)
else (
#  hit_type = "page view";
  page_views = 1;
)`
      requires_fields = {
        sc_status = true
      }
    } # categorize

    mark_entry = {
      label = '$lang_admin.log_filters.mark_entry_label'
      comment = '$lang_admin.log_filters.mark_entry_comment'
      value = 'accesses = 1;'
    } # mark_entry

#    set_page_for_worm = {
#      label = "$lang_admin.log_filters.set_page_for_worm_label"
#      comment = "$lang_admin.log_filters.set_page_for_worm_comment"
#      value = "if (worms == 1) then page = '(worm)';"
#    } # set_page_for_worm

    strip_non_page_views = {
      label = '$lang_admin.log_filters.strip_non_page_views_label'
      comment = '$lang_admin.log_filters.strip_non_page_views_comment'
      value = "if (page_views == 0) then cs_uri_stem = substr(cs_uri_stem, 0, last_index(cs_uri_stem, '/') + 1) . '(nonpage)';"
      requires_fields = {
        cs_uri_stem = true
      }
    } # strip_non_page_views

  } # log.filters

  log.field_options = {

    sessions_page_field = "cs_uri_stem"
    sessions_visitor_id_field = "cs_host"
    sessions_event_field = "page_views"

  } # log.field_options


  database.numerical_fields = {

    accesses = {
      default = false
      requires_log_field = false
      entries_field = true
    } # accesses

    page_views = {
      default = true
      requires_log_field = false
    } # page_views

    errors = {
      requires_log_field = false
    } # errors      

#    broken_links = {
#      label = "$lang_stats.field_labels.broken_links"
#      default = true
#      requires_log_field = false
#      type = "int"
#      display_format_type = "integer"
#    } # broken_links

#    screen_info_hits = {
#      label = "$lang_stats.field_labels.screen_info_hits"
#      default = false
#      requires_log_field = false
#      type = "int"
#      display_format_type = "integer"
#    } # screen_info_hits

    unique_client_ips = {
      log_field = "cs_host"
      type = "unique"
    } # unique_client_ips

    sc_bytes_content = {
      type = "int"
      integer_bits = 64
      display_format_type = "bandwidth"
    } # sc_bytes_content

  } # database.numerical_fields


  create_profile_wizard_options = {
  

    # How the reports should be grouped in the report menu
    report_groups = {
      date_time_group = ""
      hit_type = ""
      content_group = {
        cs_uri_stem = true
        file_type = true
        x_namespace = true
        cs_host = true
      }
      users_group = {
        user = true
      }
      visitor_demographics_group = {
        hostname = true
        domain_description = true
        location = true
        organization = true
        isp = true
        domain = true
        authenticated_user = true
      }
      visitor_systems_group = {
        screen_dimensions = true
        screen_depth = true
        web_browser = true
        operating_system = true
      }
      referrer_group = {
        referrer = true
        referrer_description = true
        search_engine = true
        search_phrase = true
#        search_phrase_by_search_engine = true
      }
      other_group = {
        store_id = true
        store_region = true
        store_category = true
        sc_status = true
        worm = true
        spider = true
        server_domain = true
        server_response = true
        pragma_in = true
        pragma_out = true
        cache_control_in = true
        cache_control_out = true
        vary_out = true
        x_cache_hit = true
        protocol = true
        operation = true
      }
    } # report_groups

    snapons = {

      # Attach a top_level_domain snapon
      top_level_domain = {
	snapon = "top_level_domain"
	name = "top_level_domain"
	label = "$lang_admin.snapons.top_level_domain.label"
        parameters = {
          url_field.parameter_value = "cs_uri_stem"
          field_name = {
            parameter_value = "$lang_admin.field_labels.top_level_domain"
            final_node_name = "top_level_domain"
          }
        } # parameters
        requires_log_fields = {
          cs_uri_stem = true
        }
      } # top_level_domain

      # Attach a gateway_reports snapon
      gateway_reports = {
	snapon = "gateway_reports"
	name = "gateway_reports"
	label = "$lang_admin.snapons.gateway_reports.label"
        parameters = {
          user_field.parameter_value = "cs_host"
          have_category_field.parameter_value = false
#          category_field.parameter_value = "category"
          host_field.parameter_value = "top_level_domain"
          page_views_field.parameter_value = "page_views"
          bytes_in_field.parameter_value = "sc_bytes_content"
          sort_by_field.parameter_value = "page_views"
        } # parameters
  
      } # gateway_reports

# 2013-02-06 - GMF - Now added in gateway_reports
#      # Add the standard reports
#      add_standard_reports = {
#        name = "add_standard_reports"
#	label = "add_standard_reports"
#	snapon = "add_standard_reports"
#      } # add_standard_reports

    } # snapons

  } # create_profile_wizard_options

} # juniper_mfc