Add 'shf/' from commit 'b6c5a95b664aeb861d7b33ffc9eefe447ba99dd7'
git-subtree-dir: shf git-subtree-mainline:401408448fgit-subtree-split:b6c5a95b66
This commit is contained in:
254
shf/fcore/search-list/search-list.sml
Normal file
254
shf/fcore/search-list/search-list.sml
Normal file
@@ -0,0 +1,254 @@
|
||||
structure SearchList =
|
||||
struct
|
||||
structure Dfa = CaseInsensitiveDfa
|
||||
|
||||
fun buildLoop (idx, buffer, dfa, acc, curState, startPos, prevFinalPos) =
|
||||
let
|
||||
val buffer = LineGap.goToIdx (idx, buffer)
|
||||
in
|
||||
if idx = #textLength buffer then
|
||||
let
|
||||
val acc =
|
||||
if prevFinalPos < 0 then acc
|
||||
else PersistentVector.append (startPos, prevFinalPos, acc)
|
||||
in
|
||||
(buffer, acc)
|
||||
end
|
||||
else
|
||||
let
|
||||
val chr = LineGap.sub (idx, buffer)
|
||||
val newState = Dfa.nextState (dfa, curState, chr)
|
||||
val prevFinalPos =
|
||||
if Dfa.isFinal (dfa, newState) then idx else prevFinalPos
|
||||
in
|
||||
if Dfa.isDead newState then
|
||||
if prevFinalPos = ~1 then
|
||||
(* no match found: restart search from `startPos + 1` *)
|
||||
buildLoop (startPos + 1, buffer, dfa, acc, 0, startPos + 1, ~1)
|
||||
else
|
||||
(* match found: append and continue *)
|
||||
let
|
||||
val acc = PersistentVector.append (startPos, prevFinalPos, acc)
|
||||
|
||||
(* we start 1 idx after the final position we found *)
|
||||
val newStart = prevFinalPos + 1
|
||||
in
|
||||
buildLoop (newStart, buffer, dfa, acc, 0, newStart, ~1)
|
||||
end
|
||||
else
|
||||
buildLoop
|
||||
(idx + 1, buffer, dfa, acc, newState, startPos, prevFinalPos)
|
||||
end
|
||||
end
|
||||
|
||||
fun build (buffer, dfa) =
|
||||
if Vector.length dfa > 0 then
|
||||
let val buffer = LineGap.goToStart buffer
|
||||
in buildLoop (0, buffer, dfa, PersistentVector.empty, 0, 0, ~1)
|
||||
end
|
||||
else
|
||||
(buffer, PersistentVector.empty)
|
||||
|
||||
fun rangeLoop
|
||||
( dfa
|
||||
, bufferPos
|
||||
, buffer
|
||||
, finishIdx
|
||||
, searchList
|
||||
, curState
|
||||
, startPos
|
||||
, prevFinalPos
|
||||
) =
|
||||
if bufferPos = #textLength buffer orelse bufferPos > finishIdx then
|
||||
let
|
||||
val searchList =
|
||||
if prevFinalPos = ~1 then searchList
|
||||
else PersistentVector.append (startPos, prevFinalPos, searchList)
|
||||
in
|
||||
(buffer, searchList)
|
||||
end
|
||||
else
|
||||
let
|
||||
val buffer = LineGap.goToIdx (bufferPos, buffer)
|
||||
val chr = LineGap.sub (bufferPos, buffer)
|
||||
val newState = Dfa.nextState (dfa, curState, chr)
|
||||
val prevFinalPos =
|
||||
if Dfa.isFinal (dfa, newState) then bufferPos else prevFinalPos
|
||||
in
|
||||
if Dfa.isDead newState then
|
||||
if prevFinalPos = ~1 then
|
||||
(* no match found: restart search from `startPos + 1` *)
|
||||
rangeLoop
|
||||
( dfa
|
||||
, startPos + 1
|
||||
, buffer
|
||||
, finishIdx
|
||||
, searchList
|
||||
, 0
|
||||
, startPos + 1
|
||||
, ~1
|
||||
)
|
||||
else
|
||||
(* match found: append and continue *)
|
||||
let
|
||||
val searchList =
|
||||
PersistentVector.append (startPos, prevFinalPos, searchList)
|
||||
|
||||
(* we start 1 idx after the final position we found *)
|
||||
val newStart = prevFinalPos + 1
|
||||
in
|
||||
rangeLoop
|
||||
(dfa, newStart, buffer, finishIdx, searchList, 0, newStart, ~1)
|
||||
end
|
||||
else
|
||||
(* continue searching for match *)
|
||||
rangeLoop
|
||||
( dfa
|
||||
, bufferPos + 1
|
||||
, buffer
|
||||
, finishIdx
|
||||
, searchList
|
||||
, newState
|
||||
, startPos
|
||||
, prevFinalPos
|
||||
)
|
||||
end
|
||||
|
||||
fun buildRange (buffer, finishIdx, dfa) =
|
||||
if Vector.length dfa > 0 then
|
||||
rangeLoop
|
||||
( dfa
|
||||
, #idx buffer
|
||||
, buffer
|
||||
, finishIdx
|
||||
, PersistentVector.empty
|
||||
, 0
|
||||
, #idx buffer
|
||||
, ~1
|
||||
)
|
||||
else
|
||||
(buffer, PersistentVector.empty)
|
||||
|
||||
fun insertUntilMatch
|
||||
(idx, buffer, searchList, dfa, curState, startPos, prevFinalPos) =
|
||||
if idx = #textLength buffer then
|
||||
if prevFinalPos < 0 then
|
||||
(buffer, searchList)
|
||||
else if PersistentVector.isInRange (prevFinalPos, searchList) then
|
||||
(buffer, searchList)
|
||||
else
|
||||
let
|
||||
val searchList =
|
||||
PersistentVector.insertMatchKeepingAbsoluteInddices
|
||||
(startPos, prevFinalPos, searchList)
|
||||
in
|
||||
(buffer, searchList)
|
||||
end
|
||||
else if Dfa.isDead curState then
|
||||
if prevFinalPos = ~1 then
|
||||
(* no match found: restart search from `startPos + 1` *)
|
||||
insertUntilMatch
|
||||
(startPos + 1, buffer, searchList, dfa, 0, startPos + 1, ~1)
|
||||
else if PersistentVector.isInRange (prevFinalPos, searchList) then
|
||||
(buffer, searchList)
|
||||
else
|
||||
(* new match. Insert and continue *)
|
||||
let
|
||||
val searchList =
|
||||
PersistentVector.insertMatchKeepingAbsoluteInddices
|
||||
(startPos, prevFinalPos, searchList)
|
||||
val newStart = prevFinalPos + 1
|
||||
in
|
||||
insertUntilMatch (newStart, buffer, searchList, dfa, 0, newStart, ~1)
|
||||
end
|
||||
else
|
||||
let
|
||||
val buffer = LineGap.goToIdx (idx, buffer)
|
||||
val chr = LineGap.sub (idx, buffer)
|
||||
val newState = Dfa.nextState (dfa, curState, chr)
|
||||
val prevFinalPos =
|
||||
if Dfa.isFinal (dfa, newState) then idx else prevFinalPos
|
||||
in
|
||||
(* continue *)
|
||||
insertUntilMatch
|
||||
(idx + 1, buffer, searchList, dfa, newState, startPos, prevFinalPos)
|
||||
end
|
||||
|
||||
fun tryExtendingPrevMatch
|
||||
(idx, buffer, searchList, dfa, finalPos, curState, start) =
|
||||
if idx = #textLength buffer then
|
||||
let
|
||||
val searchList =
|
||||
PersistentVector.extendExistingMatch (start, finalPos, searchList)
|
||||
in
|
||||
(buffer, searchList)
|
||||
end
|
||||
else if Dfa.isDead curState then
|
||||
let
|
||||
val searchList =
|
||||
PersistentVector.extendExistingMatch (start, finalPos, searchList)
|
||||
in
|
||||
insertUntilMatch
|
||||
(finalPos + 1, buffer, searchList, dfa, 0, finalPos + 1, ~1)
|
||||
end
|
||||
else
|
||||
let
|
||||
val buffer = LineGap.goToIdx (idx, buffer)
|
||||
val chr = LineGap.sub (idx, buffer)
|
||||
val newState = Dfa.nextState (dfa, curState, chr)
|
||||
val finalPos = if Dfa.isFinal (dfa, newState) then idx else finalPos
|
||||
in
|
||||
(* continue *)
|
||||
tryExtendingPrevMatch
|
||||
(idx + 1, buffer, searchList, dfa, finalPos, newState, start)
|
||||
end
|
||||
|
||||
fun deleteBufferAndSearchList (start, length, buffer, searchList, dfa) =
|
||||
let
|
||||
val buffer = LineGap.delete (start, length, buffer)
|
||||
val searchList = PersistentVector.delete (start, length, searchList)
|
||||
val oldStart = PersistentVector.prevMatch (start, searchList, 1)
|
||||
in
|
||||
if Vector.length dfa = 0 then
|
||||
(buffer, searchList)
|
||||
else if oldStart >= start orelse oldStart = ~1 then
|
||||
(* no previous match, so try searching for a match from start of buffer *)
|
||||
insertUntilMatch (0, buffer, searchList, dfa, 0, 0, ~1)
|
||||
else
|
||||
tryExtendingPrevMatch
|
||||
(oldStart, buffer, searchList, dfa, ~1, 0, oldStart)
|
||||
end
|
||||
|
||||
(* inserts into buffer and searchList both *)
|
||||
fun insert (insIdx, insString, buffer, searchList, dfa) =
|
||||
let
|
||||
val buffer = LineGap.insert (insIdx, insString, buffer)
|
||||
|
||||
(* incremennt existing elements in the searchList after the insIdx
|
||||
* by the length of the string that was just inserted *)
|
||||
val searchList =
|
||||
let
|
||||
val searchListLeft = PersistentVector.splitLeft (insIdx, searchList)
|
||||
|
||||
val insLength = String.size insString
|
||||
val searchListRight =
|
||||
PersistentVector.splitRight (insIdx + insLength, searchList)
|
||||
val searchListRight = PersistentVector.empty
|
||||
in
|
||||
if PersistentVector.isEmpty searchListLeft then searchListRight
|
||||
else if PersistentVector.isEmpty searchListRight then searchListLeft
|
||||
else PersistentVector.merge (searchListLeft, searchListRight)
|
||||
end
|
||||
|
||||
val oldStart = PersistentVector.prevMatch (insIdx, searchList, 1)
|
||||
in
|
||||
if Vector.length dfa = 0 then
|
||||
(buffer, searchList)
|
||||
else if oldStart >= insIdx orelse oldStart = ~1 then
|
||||
(* no previous match, so try searching for a match from start of buffer *)
|
||||
insertUntilMatch (0, buffer, searchList, dfa, 0, 0, ~1)
|
||||
else
|
||||
tryExtendingPrevMatch
|
||||
(oldStart, buffer, searchList, dfa, ~1, 0, oldStart)
|
||||
end
|
||||
end
|
||||
Reference in New Issue
Block a user