[25ae32e] | 1 | divert(-1) |
---|
| 2 | #-------------------------------------------------------------------------- |
---|
| 3 | |
---|
| 4 | # Macros defined here may be used in pattern specifications |
---|
| 5 | # You can modify this file according to your needs. |
---|
| 6 | |
---|
| 7 | # ENDOFSEGMENT and MORFIELD are macros expanded to, respectively, |
---|
| 8 | # end of segment marker (dependes on the format: flattened or not) |
---|
| 9 | # and the name of the annotation field containing morphological |
---|
| 10 | # information (standard value is 'lem'). These values are controlled |
---|
| 11 | # by programs using this file to expand search patterns (ser, grp, ...). |
---|
| 12 | |
---|
| 13 | # seg(type,form,annotation) |
---|
| 14 | |
---|
| 15 | define(`seg',`(\s*((\d+\s+)(\d+\s+)?)?dnl |
---|
| 16 | ifelse($1, `',`(\S+)', `($1)')\s+dnl |
---|
| 17 | ifelse($2, `',`(\S+)', `($2)')dnl |
---|
| 18 | ifelse($3, `',`((\s+\S+)*)', `(\s+($3))')\s*ENDOFSEGMENT)') |
---|
| 19 | |
---|
| 20 | # form(f) - segment containing the form f |
---|
| 21 | |
---|
| 22 | define(`form', `seg(,$1)') |
---|
| 23 | |
---|
| 24 | # field(f) segment containing auxiliary field f |
---|
| 25 | |
---|
| 26 | define(`field', `seg(,,`(\S+\s+)*($1)(\s+\S+)*')') |
---|
| 27 | |
---|
| 28 | # word, space, punct, number segments (assuming W, S, P, N segment types) |
---|
| 29 | |
---|
| 30 | define(`space', `seg(`S',`$1')') |
---|
| 31 | define(`word', `seg(`W',`$1')') |
---|
| 32 | define(`punct', `seg(`P',`$1')') |
---|
| 33 | define(`number', `seg(`N',`$1')') |
---|
| 34 | |
---|
| 35 | # macros specific to PMDB format |
---|
| 36 | |
---|
| 37 | define(`lexeme', `field(`MORFIELD:(\S+;)?$1,\S+')') |
---|
| 38 | define(`cat', `field(`MORFIELD:\S+,$1([,;]\S+)?')') |
---|
| 39 | |
---|
| 40 | |
---|
| 41 | # Place here your macro definitions. |
---|
| 42 | |
---|
| 43 | |
---|
| 44 | |
---|
| 45 | |
---|
| 46 | |
---|
| 47 | |
---|
| 48 | |
---|
| 49 | |
---|
| 50 | |
---|
| 51 | #-------------------------------------------------------------------------- |
---|
| 52 | divert(0) |
---|