perf: optimize expression state to use charcodes

src/states/ATTRIBUTE.ts

@@ -24,6 +24,36 @@ export interface AttrMeta extends Range {
   bound: boolean;
 }
 
+const HTML_VALUE_TERMINATORS = [
+  CODE.CLOSE_ANGLE_BRACKET,
+  CODE.COMMA,
+  [CODE.FORWARD_SLASH, CODE.CLOSE_ANGLE_BRACKET],
+];
+
+const CONCISE_VALUE_TERMINATORS = [
+  CODE.CLOSE_SQUARE_BRACKET,
+  CODE.SEMICOLON,
+  CODE.COMMA,
+];
+
+const HTML_NAME_TERMINATORS = [
+  CODE.CLOSE_ANGLE_BRACKET,
+  CODE.COMMA,
+  CODE.OPEN_PAREN,
+  CODE.EQUAL,
+  [CODE.COLON, CODE.EQUAL],
+  [CODE.FORWARD_SLASH, CODE.CLOSE_ANGLE_BRACKET],
+];
+
+const CONCISE_NAME_TERMINATORS = [
+  CODE.CLOSE_SQUARE_BRACKET,
+  CODE.SEMICOLON,
+  CODE.EQUAL,
+  CODE.COMMA,
+  CODE.OPEN_PAREN,
+  [CODE.COLON, CODE.EQUAL],
+];
+
 // We enter STATE.ATTRIBUTE when we see a non-whitespace
 // character after reading the tag name
 export const ATTRIBUTE: StateDefinition<AttrMeta> = {
@@ -73,40 +103,33 @@ export const ATTRIBUTE: StateDefinition<AttrMeta> = {
       attr.state = ATTR_STATE.VALUE;
       const expr = this.enterState(STATE.EXPRESSION);
       expr.terminatedByWhitespace = true;
-      expr.terminator = [
-        this.isConcise ? "]" : "/>",
-        this.isConcise ? ";" : ">",
-        ",",
-      ];
+      expr.terminator = this.isConcise
+        ? CONCISE_VALUE_TERMINATORS
+        : HTML_VALUE_TERMINATORS;
 
       this.rewind(1);
     } else if (code === CODE.OPEN_PAREN) {
       ensureAttrName(this, attr);
       attr.state = ATTR_STATE.ARGUMENT;
       this.skip(1); // skip (
-      this.enterState(STATE.EXPRESSION).terminator = ")";
+      this.enterState(STATE.EXPRESSION).terminator = CODE.CLOSE_PAREN;
       this.rewind(1);
     } else if (code === CODE.OPEN_CURLY_BRACE && attr.args) {
       ensureAttrName(this, attr);
       attr.state = ATTR_STATE.BLOCK;
       this.skip(1); // skip {
       const expr = this.enterState(STATE.EXPRESSION);
       expr.terminatedByWhitespace = false;
-      expr.terminator = "}";
+      expr.terminator = CODE.CLOSE_CURLY_BRACE;
       this.rewind(1);
     } else if (attr.state === undefined) {
       attr.state = ATTR_STATE.NAME;
       const expr = this.enterState(STATE.EXPRESSION);
       expr.terminatedByWhitespace = true;
       expr.skipOperators = true;
-      expr.terminator = [
-        this.isConcise ? "]" : "/>",
-        this.isConcise ? ";" : ">",
-        ":=",
-        "=",
-        ",",
-        "(",
-      ];
+      expr.terminator = this.isConcise
+        ? CONCISE_NAME_TERMINATORS
+        : HTML_NAME_TERMINATORS;
       this.rewind(1);
     } else {
       this.exitState();

src/states/EXPRESSION.ts

@@ -9,7 +9,7 @@ import {
 
 export interface ExpressionMeta extends Range {
   groupStack: number[];
-  terminator?: string | string[];
+  terminator?: number | (number | number[])[];
   skipOperators: boolean;
   terminatedByEOL: boolean;
   terminatedByWhitespace: boolean;
@@ -66,9 +66,13 @@ export const EXPRESSION: StateDefinition<ExpressionMeta> = {
         return;
       }
 
+      const { terminator } = expression;
+
       if (
-        expression.terminator &&
-        checkForTerminator(this, expression.terminator)
+        terminator &&
+        (typeof terminator === "number"
+          ? terminator === code
+          : checkForTerminators(this, code, terminator))
       ) {
         this.exitState();
         return;
@@ -279,21 +283,21 @@ function buildOperatorPattern(isConcise: boolean) {
   return new RegExp(`${lookAheadPattern}|${lookBehindPattern}`, "y");
 }
 
-function checkForTerminator(parser: Parser, terminator: string | string[]) {
-  if (typeof terminator === "string") {
-    if (parser.data[parser.pos] === terminator) {
-      return true;
-    } else if (terminator.length > 1) {
-      for (let i = 0; i < terminator.length; i++) {
-        if (parser.data[parser.pos + i] !== terminator[i]) {
-          return false;
+function checkForTerminators(
+  parser: Parser,
+  code: number,
+  terminators: (number | number[])[]
+) {
+  outer: for (const terminator of terminators) {
+    if (typeof terminator === "number") {
+      if (code === terminator) return true;
+    } else {
+      if (terminator[0] === code) {
+        for (let i = terminator.length; i-- > 1; ) {
+          if (parser.data.charCodeAt(parser.pos + i) !== terminator[i])
+            continue outer;
         }
-      }
-      return true;
-    }
-  } else {
-    for (let i = 0; i < terminator.length; i++) {
-      if (checkForTerminator(parser, terminator[i])) {
+
         return true;
       }
     }

src/states/INLINE_SCRIPT.ts

@@ -40,7 +40,7 @@ export const INLINE_SCRIPT: StateDefinition<ScriptletMeta> = {
       inlineScript.block = true;
       this.skip(1); // skip {
       const expr = this.enterState(STATE.EXPRESSION);
-      expr.terminator = "}";
+      expr.terminator = CODE.CLOSE_CURLY_BRACE;
       expr.skipOperators = true;
       this.rewind(1);
     } else {

src/states/OPEN_TAG.ts

@@ -31,6 +31,22 @@ export interface OpenTagMeta extends Range {
   parentTag: OpenTagMeta | undefined;
 }
 const PARSED_TEXT_TAGS = ["script", "style", "textarea", "html-comment"];
+const CONCISE_TAG_VAR_TERMINATORS = [
+  CODE.SEMICOLON,
+  CODE.OPEN_PAREN,
+  CODE.PIPE,
+  CODE.EQUAL,
+  [CODE.COLON, CODE.EQUAL],
+];
+
+const HTML_TAG_VAR_TERMINATORS = [
+  CODE.CLOSE_ANGLE_BRACKET,
+  CODE.OPEN_PAREN,
+  CODE.PIPE,
+  CODE.EQUAL,
+  [CODE.COLON, CODE.EQUAL],
+  [CODE.FORWARD_SLASH, CODE.CLOSE_ANGLE_BRACKET],
+];
 
 export const OPEN_TAG: StateDefinition<OpenTagMeta> = {
   name: "OPEN_TAG",
@@ -275,8 +291,8 @@ export const OPEN_TAG: StateDefinition<OpenTagMeta> = {
       expr.skipOperators = true;
       expr.terminatedByWhitespace = true;
       expr.terminator = this.isConcise
-        ? [";", "(", "|", "=", ":="]
-        : [">", "/>", "(", "|", "=", ":="];
+        ? CONCISE_TAG_VAR_TERMINATORS
+        : HTML_TAG_VAR_TERMINATORS;
       this.rewind(1);
     } else if (code === CODE.OPEN_PAREN && !tag.hasAttrs) {
       if (tag.hasArgs) {
@@ -291,14 +307,14 @@ export const OPEN_TAG: StateDefinition<OpenTagMeta> = {
       this.skip(1); // skip (
       const expr = this.enterState(STATE.EXPRESSION);
       expr.skipOperators = true;
-      expr.terminator = ")";
+      expr.terminator = CODE.CLOSE_PAREN;
       this.rewind(1);
     } else if (code === CODE.PIPE && !tag.hasAttrs) {
       tag.state = TAG_STATE.PARAMS;
       this.skip(1); // skip |
       const expr = this.enterState(STATE.EXPRESSION);
       expr.skipOperators = true;
-      expr.terminator = "|";
+      expr.terminator = CODE.PIPE;
       this.rewind(1);
     } else {
       if (tag.tagName) {

src/states/PLACEHOLDER.ts

@@ -84,7 +84,7 @@ export function checkForPlaceholder(parser: Parser, code: number) {
       parser.endText();
       parser.enterState(PLACEHOLDER).escape = escape;
       parser.skip(escape ? 2 : 3); // skip ${ or $!{
-      parser.enterState(STATE.EXPRESSION).terminator = "}";
+      parser.enterState(STATE.EXPRESSION).terminator = CODE.CLOSE_CURLY_BRACE;
       parser.rewind(1);
       return true;
     }

src/states/TAG_NAME.ts

@@ -127,7 +127,7 @@ export const TAG_NAME: StateDefinition<TagNameMeta> = {
       this.lookAtCharCodeAhead(1) === CODE.OPEN_CURLY_BRACE
     ) {
       this.skip(2); // skip ${
-      this.enterState(STATE.EXPRESSION).terminator = "}";
+      this.enterState(STATE.EXPRESSION).terminator = CODE.CLOSE_CURLY_BRACE;
       this.rewind(1);
     } else if (
       isWhitespaceCode(code) ||

src/states/TEMPLATE_STRING.ts

@@ -20,7 +20,7 @@ export const TEMPLATE_STRING: StateDefinition = {
       this.skip(1); // skip {
       const expr = this.enterState(STATE.EXPRESSION);
       expr.skipOperators = true;
-      expr.terminator = "}";
+      expr.terminator = CODE.CLOSE_CURLY_BRACE;
     } else {
       if (code === CODE.BACK_SLASH) {
         this.skip(1); // skip \

src/util/util.ts

@@ -21,15 +21,15 @@ export function getLoc(lines: number[], range: Range): Location {
 export function getPos(
   lines: number[],
   startLine: number,
-  pos: number
+  index: number
 ): Position {
   let max = lines.length - 1;
   let line = startLine;
 
   while (line < max) {
     const mid = (1 + line + max) >>> 1;
 
-    if (lines[mid] <= pos) {
+    if (lines[mid] <= index) {
       line = mid;
     } else {
       max = mid - 1;
@@ -38,12 +38,12 @@ export function getPos(
 
   return {
     line: line + 1,
-    column: line === 0 ? pos + 1 : pos - lines[line],
+    column: index - lines[line],
   };
 }
 
 export function getLines(src: string) {
-  const lines = [0];
+  const lines = [-1];
   for (let i = 0; i < src.length; i++) {
     if (src.charCodeAt(i) === CODE.NEWLINE) {
       lines.push(i);