Nested Rules (#34)

[mr-martian]

This PR adds WITH rules, which are followed by a list of rules in {}, where the latter rules are only run if the WITH matches. These rules will be run with the same values of _TARGET_ and _MARK_ as FIND.

WITH (det def) IF (1X (n)) {
  MAP (@det) (*) ;
  SETPARENT (*) TO (1 _MARK_) ;
  SETCHILD (*) TO (-1 (predet)) ;
} ;

closes #18
closes #34
relevant to #45 and #103

[mr-martian]

note to self: I think in the refactoring process I lost the lines that let A change the rule target and that's probably behind a lot of the test failures.

[mr-martian]

I've changed the syntax, so the example above is

FIND (det def) IF (1X (n)) {
  MAP (@det) (*) ;
  SETPARENT (*) TO (1 _MARK_) ;
  SETCHILD (*) TO (-1 (predet)) ;
} ;

[mr-martian]

helps with #103
helps with #45
closes #34
closes #18 (probably)

[mr-martian]

FIND definitely could share unifications and regexes with its subrules, but I don't feel like wading into the semantics of that atm.

[TinoDidriksen]

Any attachment to FIND? I prefer WITH, because it reads as English. "With <these cohorts> do { <these rules> }".

[TinoDidriksen]

Segfault where context_stack.back().attach_to.cohort is null in.

Input:

"<LU>"
	"LU" <DAB:og> <DA:og> <2> <der> KC @CO #1->3
"<jeg>"
	"jeg" <DA:jeg> <insert> PERS 1S @SUBJ> #2->3
"<ikititerpakka>"
	"ikitippaa" <DAB:hente=OBJ=ind> <DA:henter> <7> <changed> <TIP+vv> <vder> <FST:iki·TIP·vv·TIR·vv·V·Ind·1Sg·3PlO> <B:ikititerpaa> <TV> <ex-iiv> <i> <mv> <btv> <hv> <nosubj> <tr> <subjH> V 3PO PR IND 1S @PRED #3->0
"<OBJ>"
	"OBJ" <DA:OBJ> <inc> PART @<OBJ #4->3
"<OBJ=ind>"
	"OBJ=ind" <DA:ind> <inc> PART @MV< #5->3
"<dem>"
	"de" <DA:dem> <insert> PERS 3P @<OBJ #6->3
"<TIR>"
	"TIR+vv" <DAB:efterhånden> <DA:efterhånden> <right> <8> <adv> <der> <TV> <nosubj> <subjH> ADV IND @<ADVL #7->3
"<da>"
	"da" <DA:da> <insert> KS @SUB #8->10
"<jeg>"
	"jeg" <DA:jeg> <insert> PERS 1S @SUBJ> #9->10
"<nassaaramalu>"
	"nassaarpoq" <DAB:finde=noget> <DA:finder> <1> <fn:discover> <aa:t0> <t0> <vt> <HTR> <changed> <FST:nassaar·V·Cau·1Sg·LU> <B:nassaarpoq> <longda> <HV> <©HV> <IV> <Conj-encl> <first-word> <i> <pseudo-i> <biv> <tam> <vOn> <mv> <hv> <left> <nosubj> <itr> <subjH> <vH> V PR CAU 1S @CAU #10->3
"<noget>"
	"noget" <DA:noget> <inc> PART @MV< #11->10
"<efter=først=at=have>"
	"efter=først=at=have" <DA:efter=først=at=have> <insert> X @X #12->13
"<ilioraariarlunga>"
	"ilioraavoq" <DAB:anbringe> <DA:anbragt> <left> <5> <fn:add> <aa:t1> <fn:institutionalize> <aa:t0> <vH> <vt> <vta+LOC> <fn:put_spatial> <HTR> <changed> <HTR+vv> <vder> <FST:iliorar·HTR·vv·RIAR·vv·V·Cont·1Sg> <B:ilioraariarpoq> <IV> <ex-iiv> <i> <vOh> <vOn> <mv> <biv> <hv> <nosubj> <subjH> <subjH> <vH> V PCP2 AKT 1S @CL-ADVL> #13->3
"<RIAR>"
	"RIAR+vv" <DAB:nil> <DA:nil> <right> <6> <adv> <der> <IV> <tam> <nosubj> <subjH> <ex-v> ADV CONT 1S @<ADVL #14->13
"<sumut>"
	"suna" <DAB:nil> <DA:nil> <mwe-prepart> <3> <interr> <FST:suna·Pron·Trm·Sg> <Int> PRON S @X #15->16
"<tamaanga>"
	"ma" <DAB:alle=vegne=hen> <DA:alle=vegne=hen> <left> <4> <aloc> <dei> <adir> <FST:ma·Adv·Trm> <TA> <dem> <mwe-core> ADV TRM @ADVL> #16->13
"<$.>"
	"." <DAB:.> <DA:.> <9> <clb> PU @PU #17->0

Grammar:

DELIMITERS = "<$.>" ;

MOVE WITHCHILD (*) (<left> @CAU) (NEGATE p (IMP)) BEFORE WITHCHILD (*) - (CONJ @CO) (pl (<mv>) OR (<hv>) LINK *Sp (<hv>)) ;

[mr-martian]

the segfault, as with so many other errors I had to fix, was overly-aggressive search-and-replace at the beginning of the process.