Unverified Commit 59a47833 authored by Russell Cohen's avatar Russell Cohen Committed by GitHub
Browse files

First pass at optimizing SplitOnWordBoundaries (#3140) 

## Motivation and Context
@Velfi observed this function consuming a significant amount of codegen
time.

## Description
Optimize split on word boundaries, primarily by avoiding recomputation
of `isStartOfWord` when we already know it isn't the start of a word.

## Testing
- Original: 3.4s. Updated: 688ms
- Existing tests are exhaustive

On a `:aws:sdk:assemble` for the smoke test services, this improved from
53 seconds to 42 seconds
----

_By submitting this pull request, I confirm that you can use, modify,
copy, and redistribute this contribution, under the terms of your
choice._
parent c8edefee

codegen-core/src/main/kotlin/software/amazon/smithy/rust/codegen/core/util/Strings.kt

+17 −10
Original line number Diff line number Diff line
@@ -19,13 +19,16 @@ fun String.doubleQuote(): String = 
fun String.dq(): String = this.doubleQuote()



private val completeWords: List<String> = listOf("ipv4", "ipv6", "sigv4", "mib", "gib", "kib", "ttl")



private fun String.splitOnWordBoundaries(): List<String> {

    val out = mutableListOf<String>()

    // These are whole words but cased differently, e.g. `IPv4`, `MiB`, `GiB`, `TtL`

    var currentWord = ""



    var completeWordInProgress = true

    // emit the current word and update from the next character

    val emit = { next: Char ->

        completeWordInProgress = true

        if (currentWord.isNotEmpty()) {

            out += currentWord.lowercase()

        }
@@ -37,13 +40,17 @@ private fun String.splitOnWordBoundaries(): List<String> { 
    }

    val allLowerCase = this.lowercase() == this

    this.forEachIndexed { index, nextCharacter ->

        val peek = this.getOrNull(index + 1)

        val doublePeek = this.getOrNull(index + 2)

        val completeWordInProgress = completeWords.any {

            (currentWord + this.substring(index)).lowercase().startsWith(

        val computeWordInProgress = {

            val result = completeWordInProgress && currentWord.isNotEmpty() && completeWords.any {

                it.startsWith(currentWord, ignoreCase = true) && (currentWord + this.substring(index)).startsWith(

                    it,

            )

        } && !completeWords.contains(currentWord.lowercase())

                    ignoreCase = true,

                ) && !it.equals(currentWord, ignoreCase = true)

            }



            completeWordInProgress = result

            result

        }

        when {

            // [C] in these docs indicates the value of nextCharacter

            // A[_]B
@@ -53,15 +60,15 @@ private fun String.splitOnWordBoundaries(): List<String> { 
            currentWord.isEmpty() -> currentWord += nextCharacter.toString()



            // Abc[D]ef or Ab2[D]ef

            !completeWordInProgress && loweredFollowedByUpper(currentWord, nextCharacter) -> emit(nextCharacter)

            !computeWordInProgress() && loweredFollowedByUpper(currentWord, nextCharacter) -> emit(nextCharacter)



            // s3[k]ey

            !completeWordInProgress && allLowerCase && digitFollowedByLower(currentWord, nextCharacter) -> emit(

            !computeWordInProgress() && allLowerCase && digitFollowedByLower(currentWord, nextCharacter) -> emit(

                nextCharacter,

            )



            // DB[P]roxy, or `IAM[U]ser` but not AC[L]s

            endOfAcronym(currentWord, nextCharacter, peek, doublePeek) -> emit(nextCharacter)

            endOfAcronym(currentWord, nextCharacter, this.getOrNull(index + 1), this.getOrNull(index + 2)) -> emit(nextCharacter)



            // If we haven't found a word boundary, push it and keep going

            else -> currentWord += nextCharacter.toString()