@@ -6,9 +6,11 @@ import { monoStringToString, utf16ToString } from "../strings";
66import { MonoObject , MonoObjectRef , MonoString , MonoStringRef } from "../types/internal" ;
77import { Int32Ptr } from "../types/emscripten" ;
88import { wrap_error_root , wrap_no_error_root } from "../invoke-js" ;
9+ import { GraphemeSegmenter } from "./grapheme-segmenter" ;
910
1011const COMPARISON_ERROR = - 2 ;
1112const INDEXING_ERROR = - 1 ;
13+ let graphemeSegmenterCached : GraphemeSegmenter | null ;
1214
1315export function mono_wasm_compare_string ( culture : MonoStringRef , str1 : number , str1Length : number , str2 : number , str2Length : number , options : number , is_exception : Int32Ptr , ex_address : MonoObjectRef ) : number {
1416 const cultureRoot = mono_wasm_new_external_root < MonoString > ( culture ) ,
@@ -20,7 +22,7 @@ export function mono_wasm_compare_string(culture: MonoStringRef, str1: number, s
2022 const casePicker = ( options & 0x1f ) ;
2123 const locale = cultureName ? cultureName : undefined ;
2224 wrap_no_error_root ( is_exception , exceptionRoot ) ;
23- return compare_strings ( string1 , string2 , locale , casePicker ) ;
25+ return compareStrings ( string1 , string2 , locale , casePicker ) ;
2426 }
2527 catch ( ex : any ) {
2628 wrap_error_root ( is_exception , ex , exceptionRoot ) ;
@@ -37,19 +39,19 @@ export function mono_wasm_starts_with(culture: MonoStringRef, str1: number, str1
3739 exceptionRoot = mono_wasm_new_external_root < MonoObject > ( ex_address ) ;
3840 try {
3941 const cultureName = monoStringToString ( cultureRoot ) ;
40- const prefix = decode_to_clean_string ( str2 , str2Length ) ;
42+ const prefix = decodeToCleanString ( str2 , str2Length ) ;
4143 // no need to look for an empty string
4244 if ( prefix . length == 0 )
4345 return 1 ; // true
4446
45- const source = decode_to_clean_string ( str1 , str1Length ) ;
47+ const source = decodeToCleanString ( str1 , str1Length ) ;
4648 if ( source . length < prefix . length )
4749 return 0 ; //false
4850 const sourceOfPrefixLength = source . slice ( 0 , prefix . length ) ;
4951
5052 const casePicker = ( options & 0x1f ) ;
5153 const locale = cultureName ? cultureName : undefined ;
52- const result = compare_strings ( sourceOfPrefixLength , prefix , locale , casePicker ) ;
54+ const result = compareStrings ( sourceOfPrefixLength , prefix , locale , casePicker ) ;
5355 wrap_no_error_root ( is_exception , exceptionRoot ) ;
5456 return result === 0 ? 1 : 0 ; // equals ? true : false
5557 }
@@ -68,19 +70,19 @@ export function mono_wasm_ends_with(culture: MonoStringRef, str1: number, str1Le
6870 exceptionRoot = mono_wasm_new_external_root < MonoObject > ( ex_address ) ;
6971 try {
7072 const cultureName = monoStringToString ( cultureRoot ) ;
71- const suffix = decode_to_clean_string ( str2 , str2Length ) ;
73+ const suffix = decodeToCleanString ( str2 , str2Length ) ;
7274 if ( suffix . length == 0 )
7375 return 1 ; // true
7476
75- const source = decode_to_clean_string ( str1 , str1Length ) ;
77+ const source = decodeToCleanString ( str1 , str1Length ) ;
7678 const diff = source . length - suffix . length ;
7779 if ( diff < 0 )
7880 return 0 ; //false
7981 const sourceOfSuffixLength = source . slice ( diff , source . length ) ;
8082
8183 const casePicker = ( options & 0x1f ) ;
8284 const locale = cultureName ? cultureName : undefined ;
83- const result = compare_strings ( sourceOfSuffixLength , suffix , locale , casePicker ) ;
85+ const result = compareStrings ( sourceOfSuffixLength , suffix , locale , casePicker ) ;
8486 wrap_no_error_root ( is_exception , exceptionRoot ) ;
8587 return result === 0 ? 1 : 0 ; // equals ? true : false
8688 }
@@ -100,68 +102,57 @@ export function mono_wasm_index_of(culture: MonoStringRef, needlePtr: number, ne
100102 try {
101103 const needle = utf16ToString ( < any > needlePtr , < any > ( needlePtr + 2 * needleLength ) ) ;
102104 // no need to look for an empty string
103- if ( clean_string ( needle ) . length == 0 ) {
105+ if ( cleanString ( needle ) . length == 0 ) {
104106 wrap_no_error_root ( is_exception , exceptionRoot ) ;
105107 return fromBeginning ? 0 : srcLength ;
106108 }
107109
108110 const source = utf16ToString ( < any > srcPtr , < any > ( srcPtr + 2 * srcLength ) ) ;
109111 // no need to look in an empty string
110- if ( clean_string ( source ) . length == 0 ) {
112+ if ( cleanString ( source ) . length == 0 ) {
111113 wrap_no_error_root ( is_exception , exceptionRoot ) ;
112114 return fromBeginning ? 0 : srcLength ;
113115 }
114116 const cultureName = monoStringToString ( cultureRoot ) ;
115117 const locale = cultureName ? cultureName : undefined ;
116118 const casePicker = ( options & 0x1f ) ;
117-
118- const segmenter = new Intl . Segmenter ( locale , { granularity : "grapheme" } ) ;
119- const needleSegments = Array . from ( segmenter . segment ( needle ) ) . map ( s => s . segment ) ;
120- let i = 0 ;
121- let stop = false ;
122119 let result = - 1 ;
123- let segmentWidth = 0 ;
124- let index = 0 ;
125- let nextIndex = 0 ;
126- while ( ! stop ) {
127- // we need to restart the iterator in this outer loop because we have shifted it in the inner loop
128- const iteratorSrc = segmenter . segment ( source . slice ( i , source . length ) ) [ Symbol . iterator ] ( ) ;
129- let srcNext = iteratorSrc . next ( ) ;
130120
131- if ( srcNext . done )
132- break ;
121+ const graphemeSegmenter = graphemeSegmenterCached || ( graphemeSegmenterCached = new GraphemeSegmenter ( ) ) ;
122+ const needleSegments = [ ] ;
123+ let needleIdx = 0 ;
124+
125+ // Grapheme segmentation of needle string
126+ while ( needleIdx < needle . length ) {
127+ const needleGrapheme = graphemeSegmenter . nextGrapheme ( needle , needleIdx ) ;
128+ needleSegments . push ( needleGrapheme ) ;
129+ needleIdx += needleGrapheme . length ;
130+ }
131+
132+ let srcIdx = 0 ;
133+ while ( srcIdx < source . length ) {
134+ const srcGrapheme = graphemeSegmenter . nextGrapheme ( source , srcIdx ) ;
135+ srcIdx += srcGrapheme . length ;
133136
134- let matchFound = check_match_found ( srcNext . value . segment , needleSegments [ 0 ] , locale , casePicker ) ;
135- index = nextIndex ;
136- srcNext = iteratorSrc . next ( ) ;
137- if ( srcNext . done ) {
138- result = matchFound ? index : result ;
139- break ;
137+ if ( ! checkMatchFound ( srcGrapheme , needleSegments [ 0 ] , locale , casePicker ) ) {
138+ continue ;
140139 }
141- segmentWidth = srcNext . value . index ;
142- nextIndex = index + segmentWidth ;
143- if ( matchFound ) {
144- for ( let j = 1 ; j < needleSegments . length ; j ++ ) {
145- if ( srcNext . done ) {
146- stop = true ;
147- break ;
148- }
149- matchFound = check_match_found ( srcNext . value . segment , needleSegments [ j ] , locale , casePicker ) ;
150- if ( ! matchFound )
151- break ;
152140
153- srcNext = iteratorSrc . next ( ) ;
154- }
155- if ( stop )
141+ let j ;
142+ let srcNextIdx = srcIdx ;
143+ for ( j = 1 ; j < needleSegments . length ; j ++ ) {
144+ const srcGrapheme = graphemeSegmenter . nextGrapheme ( source , srcNextIdx ) ;
145+
146+ if ( ! checkMatchFound ( srcGrapheme , needleSegments [ j ] , locale , casePicker ) ) {
156147 break ;
148+ }
149+ srcNextIdx += srcGrapheme . length ;
157150 }
158-
159- if ( matchFound ) {
160- result = index ;
151+ if ( j === needleSegments . length ) {
152+ result = srcIdx - srcGrapheme . length ;
161153 if ( fromBeginning )
162154 break ;
163155 }
164- i = nextIndex ;
165156 }
166157 wrap_no_error_root ( is_exception , exceptionRoot ) ;
167158 return result ;
@@ -175,12 +166,12 @@ export function mono_wasm_index_of(culture: MonoStringRef, needlePtr: number, ne
175166 exceptionRoot . release ( ) ;
176167 }
177168
178- function check_match_found ( str1 : string , str2 : string , locale : string | undefined , casePicker : number ) : boolean {
179- return compare_strings ( str1 , str2 , locale , casePicker ) === 0 ;
169+ function checkMatchFound ( str1 : string , str2 : string , locale : string | undefined , casePicker : number ) : boolean {
170+ return compareStrings ( str1 , str2 , locale , casePicker ) === 0 ;
180171 }
181172}
182173
183- function compare_strings ( string1 : string , string2 : string , locale : string | undefined , casePicker : number ) : number {
174+ function compareStrings ( string1 : string , string2 : string , locale : string | undefined , casePicker : number ) : number {
184175 switch ( casePicker ) {
185176 case 0 :
186177 // 0: None - default algorithm for the platform OR
@@ -272,12 +263,12 @@ function compare_strings(string1: string, string2: string, locale: string | unde
272263 }
273264}
274265
275- function decode_to_clean_string ( strPtr : number , strLen : number ) {
266+ function decodeToCleanString ( strPtr : number , strLen : number ) {
276267 const str = utf16ToString ( < any > strPtr , < any > ( strPtr + 2 * strLen ) ) ;
277- return clean_string ( str ) ;
268+ return cleanString ( str ) ;
278269}
279270
280- function clean_string ( str : string ) {
271+ function cleanString ( str : string ) {
281272 const nStr = str . normalize ( ) ;
282273 return nStr . replace ( / [ \u200B - \u200D \uFEFF \0 ] / g, "" ) ;
283274}
0 commit comments