diff options
| -rw-r--r-- | compiler/standardClient.ey | 308 | ||||
| -rwxr-xr-x | examples/compiler-missing.sh | 4 | ||||
| l--------- | examples/working-compiler/regex.test | 1 |
3 files changed, 312 insertions, 1 deletions
diff --git a/compiler/standardClient.ey b/compiler/standardClient.ey index 575c2ab..3971bcc 100644 --- a/compiler/standardClient.ey +++ b/compiler/standardClient.ey @@ -272,6 +272,11 @@ # no types need to be abstracted, function can be called concreteArgs _ dump _ len dearray f "attempting to call function (w.o. abstraction)" dump + 0 concreteArgs len 1 sub range { ==i + i concreteArgs * sys .typed .type _ dump + i inputs * sys .typed .type _ dump + neq { "invalid input type at argument index " dump i dump "" die } rep + } each * } { [ ] ==argTypes # the type stack of the new function @@ -415,4 +420,307 @@ { 0 dumpIndented } > -- /dump deff +# regex support +# ideas taken from http://swtch.com/~rsc/regexp/regexp3.html +{ + [ "(" ")" "[" "]" "-" "|" "^" "*" "." ] { _ { 0 -01 * eq }_ "'" -102 cat deff }' each + + 0 ==currentCapture + + { ==b ==a + [ + [ /SPLIT 1 a len 1 add ] + a _ len dearray + [ /JUMP b len ] + b _ len dearray + ] + } /alternative deff + + { ==b ==a + a b cat + } /sequence deff + + { ==a + [ + a _ len dearray + [ /SPLIT a len neg 1 ] + ] + } /star deff + + { ==p + [ + [ /TERM p ] + ] + } /terminal deff + + { ==a + [ + a _ len dearray + [ /MATCH ] + ] + } /final deff + + { ==i ==a + [ + [ /SAVE i 2 mul ] + a _ len dearray + [ /SAVE i 2 mul 1 add ] + ] + } /capture deff + + { + [ ] + } /empty deff + + { ==str + str len 0 eq { + 1 neg + } { + 0 str * + } ? * + } /head deff + + { ==s + 1 s str .postfix + } /tail deff + + { ==re + # "(parse) re: " re cat dump + + re seq ==a =re + re len 0 neq { + re head '| { + re tail =re + re parse ==b =re + a b alternative =a + } rep + } rep + re a + } /parse deff + + { ==re + # "(seq) re: " re cat dump + + empty _ ==a + ==l + + { + # "(seq loop) re: " re cat dump + re head '| not + re head ') not and + re head 1 neg neq and + } { + re head '* { + l star =l + re tail =re + } { + a l sequence =a + re atom =l =re + } ? * + } loop + a l sequence =a + re a + } /seq deff + + { ==re + # "(atom) re: " re cat dump + empty ==a + + [ + re head '( + { + re tail parse currentCapture capture =a =re + currentCapture 1 add =currentCapture + re head ') not { ") expected" die } rep + re tail =re + } + re head '[ + { + re tail =re + re head '^ { + re tail chars ==nset =re + { nset not } ==set + re head '] not { "] expected" die } rep + re tail =re + } { + re tail chars ==set =re + re head '] not { "] expected" die } rep + re tail =re + } ? * + set terminal =a + } + re head '. + { + { -- 1 } terminal =a + re tail =re + } + 1 + { + re head ==t + { t eq } terminal =a + re tail =re + } + ] conds + + re a + # "(atom end) re: " re cat dump + } /atom deff + + { ==re + # "(chars) re: " re cat dump + + re head '] { + re tail chars2 ==set =re + set "']" | or =set + } { + chars2 ==set =re + } ? * + re set + } /chars deff + + { ==re + # "(chars2) re: " re cat dump + + re head '- { + re tail chars2 ==set =re + set "'-" | or =set + } { + charsR ==set =re + } ? * + re set + } /chars2 deff + + { ==re + # "(charsR) re: " re cat dump + + re charsN ==set =re + { re head '] not } { + re charsN set or =set =re + } loop + re set + } /charsR deff + + { ==re + # "(charsN) re: " re cat dump + + re head ==start + re tail =re + re head '- { + re tail =re + re head ==end + re tail =re + { "TODO" die } ==set + } { + { start eq } ==set + } ? * + re set + } /charsN deff + + { < + 0 ==pc + [ 10 { 0 0 } rep ] ==captures + > } /newThread deff + + |add /origadd deff + + # TODO think about implementation efficiency + { < ==maxSize + 0 ==size + [ maxSize { 0 } rep ] =*get + [ maxSize { 1 } rep ] =*pcFree + + { ==thread + thread .pc pcFree { + thread size |get =[] + 0 thread .pc |pcFree =[] + size 1 origadd =size + } rep + } /add deff + + { + 0 =size + [ maxSize { 1 } rep ] =pcFree + } /clear deff + > } /threadList deff + + { ==thread ==newpc < + newpc ==pc + [ thread .captures 20 dearray ] ==captures + > } /cloneThread deff + + { ==prog ==string + 0 ==position + string len ==maxPosition + 0 ==done + 0 ==matched + + prog len _ threadList ==clist + threadList ==nlist + + newThread _ ==thread clist .add + + { position maxPosition le done not and } { + 0 ==i + { i clist .size lt done not and } { + i clist .get _ =thread + .pc _ ==pc + prog * =*code + [ + 0 code /MATCH streq + { + 1 =matched + clist .clear + } + 0 code /TERM streq + { + position maxPosition lt { + position string * 1 code * { pc 1 add thread cloneThread nlist .add } rep + } rep + } + 0 code /JUMP streq + { pc 1 code add thread cloneThread clist .add } + 0 code /SPLIT streq + { + pc 1 code add thread cloneThread clist .add + pc 2 code add thread cloneThread clist .add + } + 0 code /SAVE streq + { + position 1 code thread .captures =[] + pc 1 add thread cloneThread clist .add + } + 1 + { |code dump "unsupported regex-VM instruction" die } + ] conds + i 1 add =i + } loop + + clist nlist =clist =nlist + nlist .clear + position 1 add =position + } loop + + matched { + currentCapture ==i + { i } { i 1 sub =i + string + i 2 mul thread .captures * _ ==start -01 str .postfix + i 2 mul 1 add thread .captures * start sub -01 str .inplacePrefix + } loop + } rep + matched + } /execute deff + + parse + [ + [ /SPLIT 3 1 ] + [ /TERM { -- 1 } ] + [ /JUMP 2 neg ] + ] -01 cat + final ==prog -- + { prog execute } +} /enregex deff + +{ enregex * } /regex deff + # vim: syn=elymas diff --git a/examples/compiler-missing.sh b/examples/compiler-missing.sh index 13474dd..6f93a96 100755 --- a/examples/compiler-missing.sh +++ b/examples/compiler-missing.sh @@ -1 +1,3 @@ -{( cd working-compiler; ls *.test ); (cd working; ls *.ey)} | sed -e 's/\..*//g' | sort | uniq -c | sort -nr +( cd working; ls *.ey ) | sed -e 's/.ey/.test/' | while read n; do + [ ! -r working-compiler/"$n" ] && echo $n +done diff --git a/examples/working-compiler/regex.test b/examples/working-compiler/regex.test new file mode 120000 index 0000000..1f308c9 --- /dev/null +++ b/examples/working-compiler/regex.test @@ -0,0 +1 @@ +../working/regex.ey
\ No newline at end of file |