Skip to content

Add allMatches to Traverse function to improve performance #511

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Draft
jtmaxwell3 wants to merge 19 commits into
base: master
Choose a base branch
from
Draft

Add allMatches to Traverse function to improve performance #511

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
19 commits
Select commit Hold shift + click to select a range
7af8809
Add allMatches to Traverse function to improve performance
jtmaxwell3 Sep 17, 2026
33c9144
Fix format
jtmaxwell3 Sep 17, 2026
3f6ab43
Fix format
jtmaxwell3 Sep 17, 2026
1fcd3e7
Fix format
jtmaxwell3 Sep 17, 2026
b348f48
Fix bug in how matches are skipped
jtmaxwell3 Sep 18, 2026
99900ac
Fix format
jtmaxwell3 Sep 18, 2026
da352c7
Fix format
jtmaxwell3 Sep 18, 2026
e9e8d12
Add minimal failing repro tests for traversal-dedup match divergence …
johnml1135 Sep 21, 2026
091659e
Implement two pass traversal with lattice
jtmaxwell3 Sep 22, 2026
cd4c7b6
Fix format
jtmaxwell3 Sep 22, 2026
24d97e9
Fix format
jtmaxwell3 Sep 22, 2026
d5ad117
Fix format
jtmaxwell3 Sep 22, 2026
95f647c
Fix bugs
jtmaxwell3 Sep 23, 2026
04dd4a9
Fix format
jtmaxwell3 Sep 23, 2026
30a3877
Fix unit test
jtmaxwell3 Sep 23, 2026
225f308
Add VariableBindings to LatticeNodeKey; add lattice to Nondeterministic
jtmaxwell3 Sep 24, 2026
14a13a8
Cache lattice arc results
jtmaxwell3 Sep 24, 2026
2907bfb
Fix format
jtmaxwell3 Sep 24, 2026
8a51794
Fix format
jtmaxwell3 Sep 24, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
22 changes: 21 additions & 1 deletion src/SIL.Machine/FiniteState/DeterministicFsaTraversalMethod.cs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -22,7 +22,8 @@ public override IEnumerable<FstResult<TData, TOffset>> Traverse(
ref int annIndex,
Register<TOffset>[,] initRegisters,
IList<TagMapCommand> initCmds,
ISet<int> initAnns
ISet<int> initAnns,
bool allMatches
)
{
Stack<DeterministicFsaTraversalInstance<TData, TOffset>> instStack = InitializeStack(
Expand All @@ -33,15 +34,21 @@ ISet<int> initAnns
);

var curResults = new List<FstResult<TData, TOffset>>();
var lattice = !allMatches ? CreateFstLattice() : null;

while (instStack.Count != 0)
{
DeterministicFsaTraversalInstance<TData, TOffset> inst = instStack.Pop();
DeterministicFsaTraversalInstance<TData, TOffset> origInst = !allMatches
? CopyInstanceAndBindings(inst)
: null;

bool releaseInstance = true;
foreach (Arc<TData, TOffset> arc in inst.State.Arcs)
{
if (CheckInputMatch(arc, inst.AnnotationIndex, inst.VariableBindings))
{
int resultCount = !allMatches ? curResults.Count : 0;
foreach (
DeterministicFsaTraversalInstance<TData, TOffset> ni in Advance(
inst,
Expand All @@ -51,6 +58,13 @@ DeterministicFsaTraversalInstance<TData, TOffset> ni in Advance(
)
)
{
if (!allMatches)
{
if (curResults.Count > resultCount)
RecordFinalArc(lattice, origInst, arc);
if (RecordedInstance(lattice, ni, origInst, arc))
continue;
}
instStack.Push(ni);
}

Expand All @@ -63,6 +77,12 @@ DeterministicFsaTraversalInstance<TData, TOffset> ni in Advance(
ReleaseInstance(inst);
}

if (!allMatches)
{
var newResults = ExtractResults(lattice, allMatches);
curResults = newResults;
}

CheckAcceptingStartState(initAnns, initRegisters, curResults);

return curResults;
Expand Down
3 changes: 2 additions & 1 deletion src/SIL.Machine/FiniteState/DeterministicFstTraversalMethod.cs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -26,7 +26,8 @@ public override IEnumerable<FstResult<TData, TOffset>> Traverse(
ref int annIndex,
Register<TOffset>[,] initRegisters,
IList<TagMapCommand> initCmds,
ISet<int> initAnns
ISet<int> initAnns,
bool allMatches
)
{
Stack<DeterministicFstTraversalInstance<TData, TOffset>> instStack = InitializeStack(
Expand Down
4 changes: 2 additions & 2 deletions src/SIL.Machine/FiniteState/Fst.cs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -387,7 +387,7 @@ out IEnumerable<FstResult<TData, TOffset>> results
}

List<FstResult<TData, TOffset>> curResults = traversalMethod
.Traverse(ref annIndex, initRegisters, cmds, initAnns)
.Traverse(ref annIndex, initRegisters, cmds, initAnns, allMatches)
.ToList();
if (curResults.Count > 0)
{
Expand Down Expand Up @@ -424,7 +424,7 @@ private int ResultCompare(FstResult<TData, TOffset> x, FstResult<TData, TOffset>
compare = -compare;
if (IsDeterministic)
{
compare = x.IsLazy ? -compare : compare;
compare = (x.IsLazy || y.IsLazy) ? -compare : compare;
}
else if (compare == 0)
{
Expand Down
3 changes: 2 additions & 1 deletion src/SIL.Machine/FiniteState/ITraversalMethod.cs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -11,7 +11,8 @@ IEnumerable<FstResult<TData, TOffset>> Traverse(
ref int annIndex,
Register<TOffset>[,] initRegisters,
IList<TagMapCommand> initCmds,
ISet<int> initAnns
ISet<int> initAnns,
bool allMatches
);
}
}
55 changes: 43 additions & 12 deletions src/SIL.Machine/FiniteState/NondeterministicFsaTraversalMethod.cs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -24,7 +24,8 @@ public override IEnumerable<FstResult<TData, TOffset>> Traverse(
ref int annIndex,
Register<TOffset>[,] initRegisters,
IList<TagMapCommand> initCmds,
ISet<int> initAnns
ISet<int> initAnns,
bool allMatches
)
{
Stack<NondeterministicFsaTraversalInstance<TData, TOffset>> instStack = InitializeStack(
Expand All @@ -35,6 +36,7 @@ ISet<int> initAnns
);

var curResults = new List<FstResult<TData, TOffset>>();
var lattice = !allMatches ? CreateFstLattice() : null;
var traversed = new HashSet<Tuple<State<TData, TOffset>, int, Register<TOffset>[,]>>(
AnonymousEqualityComparer.Create<Tuple<State<TData, TOffset>, int, Register<TOffset>[,]>>(
KeyEquals,
Expand All @@ -44,6 +46,9 @@ ISet<int> initAnns
while (instStack.Count != 0)
{
NondeterministicFsaTraversalInstance<TData, TOffset> inst = instStack.Pop();
NondeterministicFsaTraversalInstance<TData, TOffset> origInst = !allMatches
? CopyInstanceAndBindings(inst)
: null;

bool releaseInstance = true;
VariableBindings varBindings = null;
Expand All @@ -69,24 +74,36 @@ ISet<int> initAnns
}

ti.Visited.Add(arc.Target);
int resultCount = !allMatches ? curResults.Count : 0;
NondeterministicFsaTraversalInstance<TData, TOffset> newInst = EpsilonAdvance(
ti,
arc,
curResults
);
Tuple<State<TData, TOffset>, int, Register<TOffset>[,]> key = Tuple.Create(
newInst.State,
newInst.AnnotationIndex,
newInst.Registers
);
if (!traversed.Contains(key))
bool skip = false;
if (!allMatches)
{
instStack.Push(newInst);
traversed.Add(key);
if (curResults.Count > resultCount)
RecordFinalArc(lattice, origInst, arc);
if (RecordedInstance(lattice, newInst, origInst, arc))
skip = true;
}
if (!skip)
{
Tuple<State<TData, TOffset>, int, Register<TOffset>[,]> key = Tuple.Create(
newInst.State,
newInst.AnnotationIndex,
newInst.Registers
);
if (!traversed.Contains(key))
{
instStack.Push(newInst);
traversed.Add(key);
}
if (isInstReusable)
releaseInstance = false;
varBindings = null;
}
if (isInstReusable)
releaseInstance = false;
varBindings = null;
}
}
else
Expand All @@ -99,6 +116,7 @@ ISet<int> initAnns
? inst
: CopyInstance(inst);

int resultCount = !allMatches ? curResults.Count : 0;
foreach (
NondeterministicFsaTraversalInstance<TData, TOffset> newInst in Advance(
ti,
Expand All @@ -109,6 +127,13 @@ NondeterministicFsaTraversalInstance<TData, TOffset> newInst in Advance(
)
{
newInst.Visited.Clear();
if (!allMatches)
{
if (curResults.Count > resultCount)
RecordFinalArc(lattice, origInst, arc);
if (RecordedInstance(lattice, newInst, origInst, arc))
continue;
}
Tuple<State<TData, TOffset>, int, Register<TOffset>[,]> key = Tuple.Create(
newInst.State,
newInst.AnnotationIndex,
Expand All @@ -132,6 +157,12 @@ NondeterministicFsaTraversalInstance<TData, TOffset> newInst in Advance(
ReleaseInstance(inst);
}

if (!allMatches)
{
var newResults = ExtractResults(lattice, allMatches);
curResults = newResults;
}

CheckAcceptingStartState(initAnns, initRegisters, curResults);

return curResults;
Expand Down
3 changes: 2 additions & 1 deletion src/SIL.Machine/FiniteState/NondeterministicFstTraversalMethod.cs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -27,7 +27,8 @@ public override IEnumerable<FstResult<TData, TOffset>> Traverse(
ref int annIndex,
Register<TOffset>[,] initRegisters,
IList<TagMapCommand> initCmds,
ISet<int> initAnns
ISet<int> initAnns,
bool allMatches
)
{
Stack<NondeterministicFstTraversalInstance<TData, TOffset>> instStack = InitializeStack(
Expand Down
Loading
Loading