# (C) 2007-2011 J.F.Dockes
# License: GPL V2
#
# Field names configuration. This defines how one may search ie for:
#   author:Hemingway
#
# Important: 
#   - the field names MUST be all lowercase alphabetic ascii here. They can
#     be anycased in the documents.

[prefixes]

#####################################################
# This section defines what prefix the terms inside named fields will be
# indexed with (in addition to prefix-less indexing for general search)
# ALL prefixes MUST be all ASCII UPPERCASE (NO DIGITS)
# 
# The field names should be the canonic ones, not the aliases defined in
# the following section. Don't change those which are predefined here, 
# quite a few are hard-coded in the c++. But you can add more (for new
# fields emitted by filters).
#
# Fields can have two relevance boost factors defined, such as in:
# caption = S ; wdfinc=10
#  and/or
# caption = S ; boost = 10
# The first line would boost the xapian "within document frequency" of caption terms by a factor of
# 10 at indexing time. The second one (NOT CURRENTLY IMPLEMENTED) would automatically boost the
# weight of a caption-based field query (ie: caption:mytitle or title:mytitle) at query time.
#
# The 'pfxonly' attribute can also be set on entries to express that terms from the field should be
# indexed only with a prefix (in general, field terms are indexed both with and without a prefix).
#
# The 'noterms' attribute is used at query time to avoid adding the field values to the terms to be
# highlighted.
#
# The values defined here in the default "fields" file are mostly also hard-coded in the c code, and
# here just for reference, can't change at all.
#
# Also reserved: F(parentid), Q(uniqueid)  XE (file ext), XP (for path elements), XSFN, XSFS, XXST, XXND, XXPG
# Using XX was not a good idea. 
#
# Note that you can disable generating terms for the fields: "filename", "containerfilename",
# "rclUnsplitFN", "dir" and "mimetype" by setting their prefix to empty in the local "fields"
# file. If you're not interested in the corresponding searches, this can save a significant amount
# of space.
#
# There was initially some effort made to have some compatibility with Xapian Omega, but it probably
# is now irrelevant, given how differently the indexes are now structured. In any case see
# xapian/applications/omega/docs/termprefixes.rst in the Xapian source for Xapian/Omega prefix
# usage. In a nutshell Omega reserves all single letters except X which is the start for
# user-defined prefixes.
#
# Recoll uses a lot of X-starting prefixes, and also some XX ones so that locally defined prefixes
# should start with XY to avoid any possible collision with the predefined ones.
#
# I hereby commit to not using XY for Recoll.
# *** USE XY for beginning your local prefixes *** e.g.:
# myfield = XYMYFIELD

author = A
xapdate = D
keywords= K
xapyearmon = M
title = S ; wdfinc = 10
mtype = T
ext = XE; noterms = 1
rclmd5 = XM
dir = XP ; noterms = 1
abstract = XS
filename = XSFN ; noterms = 1
containerfilename = XCFN ; pfxonly = 1 ; noterms = 1
rclUnsplitFN = XSFS
xapyear = Y
recipient = XTO
rclbes = XB ; noterms = 1
annotation = XA

[values]
###########
## Fields which will be stored in Xapian values, authorizing range query
## processing.
# Entries are specified as 'fieldname = valueslot;[px=val1;py=val2...]'.
# Xapian value slots are 32 bits numbers. Numbers below 1000 are reserved
# by Recoll or Xapian. Numbers above are available for user configuration
# Values have types, which can be 'int' or 'string' at the moment. ints have
# an additional 'len' attributes, which specifies the padding size used for
# sorting (leading zeroes: all xapian sorting is text-based). 10 is fine
# for an unsigned 32 bits integer.
# myfield = 1001; type=int; len = 10
# mystrfield = 1002; type = string

[stored]
############################
# Some fields are stored in the document data record inside the index and
# can be displayed in result lists. There is no necessity that stored fields
# should be indexed (have a prefix in the preceding section). Example: "url"
#
# Some fields are stored by default, don't add them here:
#    caption, mimetype, url
# Only canonical names should be used here, not aliases.
# "rclaptg" is used for viewer specialization (depending on local config)
# "rclbes" defines the backend type (ie normal fs, firefox cache). Should
#   probably be hardcoded, don't remove it
abstract=
author=
filename=
keywords=
rclaptg=
rclbes=
recipient=
annotation=

[aliases]
##########################
# This section defines field names aliases or synonyms. Any right hand side
# value will be turned into the lhs canonic name before further treatment
#
# The left-hand values in the recoll distribution file are well known and
# must match names used in the c++ code, or even the index data
# record. They can't change! But you can add others.
#
# Filters should only add canonic names to the meta array when indexing,
# not aliases.
abstract = summary dc:summary description xesam:description
author = creator dc:creator xesam:author xesam:creator from
title = caption title dc:title subject
# catg = dc:type contentCategory
dbytes = size xesam:size
dmtime = dc:date dc:datemodified datemodified contentmodified \
       xesam:contentmodified
ext = fileextension xesam:fileextension
# Don't add subject to keywords aliases, it's better to keep it for email
keywords = keyword xesam:keyword tag tags dc:subject xesam:subject \
	 dc:description
mtype = mime mimetype xesam:mimetype contenttype xesam:contenttype dc:format
recipient = to xesam:recipient
url = dc:identifier xesam:url
annotation = pdfannot

##################
# The queryaliases section defines aliases which are used exclusively at
# query time: there is no risk to pick up a random field from a document
# (e.g. an HTML meta field) and index it. 
[queryaliases]
filename = fn
containerfilename = cfn
annotation = annot pa

[xattrtofields]
######################
# Processing for extended file attributes.
#
# By default, extended attributes are processed as document fields (after
# removing the 'user' prefix from the name on Linux).
#
# You can enter case-sensitive name translations as "xattrname = fieldname". 
# Entering an empty translation will disable use of the attribute.
# The values from the extended attributes will extend, not replace, the
# data found from equivalent fields inside the document.
# 
# Special case: an extended attribute named "modificationdate" will set the
# "dmtime" (document date) only if it is not set by an internal document
# field (e.g. email "Date:").
#
# As an example, the following would map a quite plausible "tags" extended
# attribute into the "keywords" field.
tags = keywords

# Proposed or standard freedesktop.org extended attributes
xdg.tags = keywords
xdg.comment = abstract

# MacOS Finder comments
com.apple.metadata:kMDItemFinderComment = abstract
# MacOS Finder tags. Not sure how useful it would be to index these
#com.apple.metadata:_kMDItemUserTags = keywords
com.apple.FinderInfo =

# Some standard fields are not to be used (by default).
mime_type =
charset =

########################
# Sections reserved for specific filters follow
#

##########################
# Mail filter section. You can specify mail headers to be indexed 
# in addition to the standard ones: (To:, Cc:, From:, Subject:, Date, 
# Message-Id), along with the field name to be used. For this to be useful, 
# the field name should also be listed in the [prefixes] and possibly the
# [stored] sections.
#
# [mail]
# x-my-tag = mymailtag