nxu/test-regex.tal

%sp { #2018 DEO }
%nl { #0a18 DEO }

|0100
    ;expr1 compile #010e DEO nl
    emit-stack nl
    emit-arena nl

    LIT "= emit sp
    #01 emit-bool sp
    #01 emit-bool sp
    #00 emit-bool sp
    #01 emit-bool sp
    #01 emit-bool sp
    #00 emit-bool sp
    #00 emit-bool sp
    #00 emit-bool nl

    LIT "A emit sp
    ;test1 OVR2k rx-match emit-bool sp
    ;test2 OVR2k rx-match emit-bool sp
    ;test3 OVR2k rx-match emit-bool sp
    ;test4 OVR2k rx-match emit-bool sp
    ;test5 OVR2k rx-match emit-bool sp
    ;test6 OVR2k rx-match emit-bool sp
    ;test7 OVR2k rx-match emit-bool sp
    ;test8 OVR2k rx-match emit-bool nl

    LIT "B emit sp
    ;test1 ;graph1 rx-match emit-bool sp
    ;test2 ;graph1 rx-match emit-bool sp
    ;test3 ;graph1 rx-match emit-bool sp
    ;test4 ;graph1 rx-match emit-bool sp
    ;test5 ;graph1 rx-match emit-bool sp
    ;test6 ;graph1 rx-match emit-bool sp
    ;test7 ;graph1 rx-match emit-bool sp
    ;test8 ;graph1 rx-match emit-bool nl

    reset-arena
    #010f DEO BRK

( corresponds to regex: a(b|c)d* )
@expr1 "a(b|c)d* 00

( corresponds to regex: a(b|c)d* )
( accepts "ab" or "ac" followd by any number of d's )
@graph1
        03 "a =x1
    @x1 04 =x2 =x3
    @x2 03 "b =x4
    @x3 03 "c =x4
    @x4 05 =x5 0000
    @x5 03 "d =x4

( test case strings to try matching )
@test1 "ab 00 ( yes )
@test2 "acdd 00 ( yes )
@test3 "add 00 ( no )
@test4 "abd 00 ( yes )
@test5 "acddddddddddd 00 ( yes )
@test6 "bd 00 ( no )
@test7 "z 00 ( no )
@test8 00 ( no )

~regex.tal

@emit ( c^ -- )
    #18 DEO JMP2r

@emit-short ( short* -- )
    SWP emit-byte emit-byte JMP2r

@emit-byte ( byte^ -- )
    DUP #04 SFT ,&hex JSR #0f AND !&hex
    &hex #30 ADD DUP #39 GTH #27 MUL ADD emit
    JMP2r

@emit-bool ( byte^ -- )
    LIT "0 ADD emit JMP2r

( print stack size, followed by contents )
@emit-stack ( -> )
    sp LIT "n emit LIT "= emit ;stack-pos LDA2 ;stack-bot SUB2 #0004 DIV2 emit-short LIT ": emit
    ;stack-bot
    &loop
    DUP2 ;stack-pos LDA2 LTH2 ?&ok
    POP2 nl JMP2r
    &ok
    sp LDA2k emit-short
    #0002 ADD2 !&loop

( emit n bytes from the given address )
@emit-n ( addr* count^ -> addr2* )
    DUP #00 GTH ( addr count count>0? ) ?&ok ( addr count ) POP nl JMP2r
    &ok
    STH ( addr [count] ) sp LDAk emit-byte INC2 ( addr+1 [count] )
    STHr #01 SUB ( addr+1 count-1 )
    !emit-n

( emit the arena, with one line per node )
( parses node type, since node size is dynamic (3-5). )
@emit-arena ( -> )
    ;arena-bot
    &loop
    DUP2 ;arena-pos LDA2 LTH2 ?&ok POP2 JMP2r
    &ok
    DUP2 emit-short
    LIT ": emit sp
        LDAk #01 NEQ ?&c1 #03 emit-n !&loop
    &c1 LDAk #02 NEQ ?&c2 #03 emit-n !&loop
    &c2 LDAk #03 NEQ ?&c3 #04 emit-n !&loop
    &c3 LDAk #04 NEQ ?&c4 #05 emit-n !&loop
    &c4 LDAk #05 NEQ ?&c5 #05 emit-n !&loop
    &c5 ;unknown-node-type errorm