utt/lib/terms.m4

52 lines
1.4 KiB
Plaintext
Raw Normal View History

divert(-1)
#--------------------------------------------------------------------------
# Macros defined here may be used in pattern specifications
# You can modify this file according to your needs.
# ENDOFSEGMENT and MORFIELD are macros expanded to, respectively,
# end of segment marker (dependes on the format: flattened or not)
# and the name of the annotation field containing morphological
# information (standard value is 'lem'). These values are controlled
# by programs using this file to expand search patterns (ser, grp, ...).
# seg(type,form,annotation)
define(`seg',`(\s*((\d+\s+)(\d+\s+)?)?dnl
ifelse($1, `',`(\S+)', `($1)')\s+dnl
ifelse($2, `',`(\S+)', `($2)')dnl
ifelse($3, `',`((\s+\S+)*)', `(\s+($3))')\s*ENDOFSEGMENT)')
# form(f) - segment containing the form f
define(`form', `seg(,$1)')
# field(f) segment containing auxiliary field f
define(`field', `seg(,,`(\S+\s+)*($1)(\s+\S+)*')')
# word, space, punct, number segments (assuming W, S, P, N segment types)
define(`space', `seg(`S',`$1')')
define(`word', `seg(`W',`$1')')
define(`punct', `seg(`P',`$1')')
define(`number', `seg(`N',`$1')')
# macros specific to PMDB format
define(`lexeme', `field(`MORFIELD:(\S+;)?$1,\S+')')
define(`cat', `field(`MORFIELD:\S+,$1([,;]\S+)?')')
# Place here your macro definitions.
#--------------------------------------------------------------------------
divert(0)