humza/sml-projects
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

first pass of 'get matches in range from nfa' functionality

Browse Source
This commit is contained in:
Humza Shahid
2025-09-29 12:18:45 +01:00
parent 8d29bfab78
commit b6720ed5f1
2 changed files with 267 additions and 72 deletions
Download Patch File Download Diff File Expand all files Collapse all files

BIN
fcore/search-list/nfa
View File

Binary file not shown.

201
fcore/search-list/nfa.sml
View File

@@ -57,7 +57,6 @@ struct
| [] => UNTESTED | [] => UNTESTED
end end
local
fun rebuildConcat (lst, chr, idx) = fun rebuildConcat (lst, chr, idx) =
case lst of case lst of
[(hd, _)] => [(hd, _)] =>
@@ -132,9 +131,14 @@ struct
| ALTERNATION (lst, UNTESTED) => rebuildAlternation (lst, chr, idx, []) | ALTERNATION (lst, UNTESTED) => rebuildAlternation (lst, chr, idx, [])
| ALTERNATION (_, state) => (nfa, state) | ALTERNATION (_, state) => (nfa, state)
| _ => | _ => raise Fail "nfa.sml 69: not char literal or concat or alternation"
raise Fail "nfa.sml 69: not char literal or concat or alternation"
(* get all matches in string.
* Todo:
* - Append {start: int, finish: int} into PersistentVector instead
* - Search through gap buffer instead of string
* *)
local
fun loop (pos, str, nfa, origNfa, startPos, acc) = fun loop (pos, str, nfa, origNfa, startPos, acc) =
if pos = String.size str then if pos = String.size str then
PersistentVector.toVector acc PersistentVector.toVector acc
@@ -158,6 +162,197 @@ struct
fun getMatches (str, nfa) = fun getMatches (str, nfa) =
loop (0, str, nfa, nfa, 0, PersistentVector.empty) loop (0, str, nfa, nfa, 0, PersistentVector.empty)
end end
local
fun backtrackRange
(hd, tl, prevStrings, origNfa, acc, absIdx, startIdx, finishIdx) =
case prevStrings of
prevHd :: prevTl =>
let
val prevIdx = absIdx - String.size prevHd
val tl = hd :: tl
in
if prevIdx < startIdx then
(* keep backtracking *)
backtrackRange
( prevHd
, tl
, prevTl
, origNfa
, acc
, prevIdx
, startIdx
, finishIdx
)
else
let
val strIdx = startIdx - prevIdx + 1
val absIdx = absIdx + strIdx
in
loop
( strIdx
, prevHd
, tl
, prevTl
, origNfa
, origNfa
, acc
, absIdx
, absIdx
, finishIdx
)
end
end
| [] => raise Fail "nfa.sml 188: should not backtrack to empty list"
and loop
( strIdx
, hd
, tl
, prevStrings
, nfa
, origNfa
, acc
, absIdx
, startIdx
, finishIdx
) =
if strIdx = String.size hd then
case tl of
newHd :: newTl =>
loop
( 0
, newHd
, newTl
, hd :: prevStrings
, nfa
, origNfa
, acc
, absIdx
, startIdx
, finishIdx
)
| [] => PersistentVector.toVector acc
else if absIdx > finishIdx then
PersistentVector.toVector acc
else
let
val chr = String.sub (hd, strIdx)
val (nfa, state) = rebuild (nfa, chr, absIdx)
in
case state of
UNTESTED =>
loop
( startIdx + 1
, hd
, tl
, prevStrings
, nfa
, origNfa
, acc
, absIdx + 1
, startIdx
, finishIdx
)
| VALID _ =>
let
val acc = PersistentVector.append (startIdx, acc)
in
loop
( startIdx + 1
, hd
, tl
, prevStrings
, origNfa
, origNfa
, acc
, absIdx + 1
, absIdx + 1
, finishIdx
)
end
| INVALID =>
let
val prevIdx = absIdx - strIdx
in
if prevIdx < startIdx then
backtrackRange
( hd
, tl
, prevStrings
, origNfa
, acc
, prevIdx
, startIdx
, finishIdx
)
else
let
val strIdx = startIdx - prevIdx + 1
val absIdx = absIdx + strIdx
in
loop
( strIdx
, hd
, tl
, prevStrings
, origNfa
, origNfa
, acc
, absIdx
, absIdx
, finishIdx
)
end
end
end
in
(* Prerequisite: move buffer to 'start' parameter before calling *)
fun getMatchesInRange (startIdx, finishIdx, buffer: LineGap.t, nfa) =
let
val {rightStrings, idx = bufferIdx, ...} = buffer
val strIdx = startIdx - bufferIdx
in
case rightStrings of
hd :: tl =>
if strIdx < String.size hd then
(* strIdx is in this string *)
loop
( strIdx
, hd
, tl
, []
, nfa
, nfa
, PersistentVector.empty
, startIdx
, startIdx
, finishIdx
)
else
(* strIdx is in tl *)
(case tl of
stlhd :: stltl =>
let
val strIdx = strIdx - String.size hd
in
loop
( strIdx
, stlhd
, stltl
, []
, nfa
, nfa
, PersistentVector.empty
, startIdx
, startIdx
, finishIdx
)
end
| [] => Vector.fromList [])
| [] => Vector.fromList []
end
end
end end
structure ParseNfa = structure ParseNfa =