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

fix bugs in implementation for 'Nfa.getMatchesInRange'

Browse Source
This commit is contained in:
Humza Shahid
2025-09-29 22:57:19 +01:00
parent d37e510b24
commit b35d045a09
2 changed files with 19 additions and 170 deletions
Download Patch File Download Diff File Expand all files Collapse all files

187
fcore/search-list/nfa.sml
View File

@@ -169,194 +169,43 @@ struct
end end
local local
fun backtrackRange fun loop (pos, buffer, nfa, origNfa, startPos, acc, lastIdx) =
(hd, tl, prevStrings, origNfa, acc, absIdx, startIdx, finishIdx) = if pos = #textLength buffer then
case prevStrings of acc
prevHd :: prevTl => else if pos > lastIdx then
let
val prevIdx = absIdx - String.size prevHd
val tl = hd :: tl
in
if startIdx < prevIdx 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
)
| [] => acc
else if absIdx > finishIdx then
acc acc
else else
let let
val chr = String.sub (hd, strIdx) val chr = LineGap.substring (pos, 1, buffer)
val (nfa, state) = rebuild (nfa, chr, absIdx) val chr = String.sub (chr, 0)
val (nfa, state) = rebuild (nfa, chr, pos)
in in
case state of case state of
UNTESTED => VALID finishIdx =>
loop
( strIdx + 1
, hd
, tl
, prevStrings
, nfa
, origNfa
, acc
, absIdx + 1
, startIdx
, finishIdx
)
| VALID foundIdx =>
let let
val acc = PersistentVector.append (startIdx, foundIdx, acc) val acc = PersistentVector.append (startPos, finishIdx, acc)
in in
loop loop
( strIdx + 1 ( finishIdx + 1
, hd , buffer
, tl
, prevStrings
, origNfa , origNfa
, origNfa , origNfa
, finishIdx + 1
, acc , acc
, foundIdx + 1 , lastIdx
, foundIdx + 1
, finishIdx
) )
end end
| INVALID => | INVALID =>
let let val pos = startPos + 1
val prevIdx = absIdx - strIdx in loop (pos, buffer, origNfa, origNfa, pos, acc, lastIdx)
in
if startIdx < prevIdx then
backtrackRange
( hd
, tl
, prevStrings
, origNfa
, acc
, prevIdx
, startIdx
, finishIdx
)
else
let
val strIdx = startIdx - prevIdx + 1
val absIdx = prevIdx + strIdx
in
loop
( strIdx
, hd
, tl
, prevStrings
, origNfa
, origNfa
, acc
, absIdx
, startIdx + 1
, finishIdx
)
end
end end
| UNTESTED =>
loop (pos + 1, buffer, nfa, origNfa, startPos, acc, lastIdx)
end end
in in
(* Prerequisite: move buffer to 'start' parameter before calling *) (* Prerequisite: move buffer to 'start' parameter before calling *)
fun getMatchesInRange (startIdx, finishIdx, buffer: LineGap.t, nfa) = fun getMatchesInRange (startIdx, finishIdx, buffer: LineGap.t, nfa) =
let loop (startIdx, buffer, nfa, nfa, 0, PersistentVector.empty, finishIdx)
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
| [] => PersistentVector.empty)
| [] => PersistentVector.empty
end
end end
end end

2
lib/brolib-sml

Submodule lib/brolib-sml updated: d6e119a161...14bb447289