fix: regex continuation issue

marko-js · Jul 19, 2022 · 4e13e74 · 4e13e74
1 parent c91aa55
commit 4e13e74
Show file tree

Hide file tree

Showing 5 changed files with 29 additions and 11 deletions.
diff --git a/.changeset/gorgeous-laws-grin.md b/.changeset/gorgeous-laws-grin.md
@@ -0,0 +1,5 @@
+---
+"htmljs-parser": patch
+---
+
+Fix regression that causes incorrect expression continuations after regexps.
diff --git a/src/__tests__/fixtures/attr-regexp/__snapshots__/attr-regexp.expected.txt b/src/__tests__/fixtures/attr-regexp/__snapshots__/attr-regexp.expected.txt
@@ -0,0 +1,10 @@
+1╭─ <input pattern=/\s*?\S+\s*?/ name="test">
+ │  ││     │      ││             │   ││     ╰─ openTagEnd
+ │  ││     │      ││             │   │╰─ attrValue.value "\"test\""
+ │  ││     │      ││             │   ╰─ attrValue "=\"test\""
+ │  ││     │      ││             ╰─ attrName "name"
+ │  ││     │      │╰─ attrValue.value "/\\s*?\\S+\\s*?/"
+ │  ││     │      ╰─ attrValue "=/\\s*?\\S+\\s*?/"
+ │  ││     ╰─ attrName "pattern"
+ │  │╰─ tagName "input"
+ ╰─ ╰─ openTagStart
diff --git a/src/__tests__/fixtures/attr-regexp/input.marko b/src/__tests__/fixtures/attr-regexp/input.marko
@@ -0,0 +1 @@
+<input pattern=/\s*?\S+\s*?/ name="test">
diff --git a/...s__/fixtures/comments-within-open-tag/__snapshots__/comments-within-open-tag.expected.txt b/...s__/fixtures/comments-within-open-tag/__snapshots__/comments-within-open-tag.expected.txt
@@ -3,23 +3,23 @@
  │  ││   │  ││                           │    ││                                     ││ ╰─ closeTagName "div"
  │  ││   │  ││                           │    ││                                     │╰─ closeTagStart "</"
  │  ││   │  ││                           │    ││                                     ╰─ openTagEnd
- │  ││   │  ││                           │    │╰─ attrValue.value "\"world\""
- │  ││   │  ││                           │    ╰─ attrValue "=\"world\""
+ │  ││   │  ││                           │    │╰─ attrValue.value "\"world\" /* this is another comment */"
+ │  ││   │  ││                           │    ╰─ attrValue "=\"world\" /* this is another comment */"
  │  ││   │  ││                           ╰─ attrName "hello"
- │  ││   │  │╰─ attrValue.value "\"bar\""
- │  ││   │  ╰─ attrValue "=\"bar\""
+ │  ││   │  │╰─ attrValue.value "\"bar\" /*this is a comment*/"
+ │  ││   │  ╰─ attrValue "=\"bar\" /*this is a comment*/"
  │  ││   ╰─ attrName "foo"
  │  │╰─ tagName "div"
  ╰─ ╰─ openTagStart
 2╭─ <div foo="bar" // this is a test
- │  ││   │  │╰─ attrValue.value "\"bar\""
- │  ││   │  ╰─ attrValue "=\"bar\""
+ │  ││   │  │╰─ attrValue.value "\"bar\" // this is a test"
+ │  ││   │  ╰─ attrValue "=\"bar\" // this is a test"
  │  ││   ╰─ attrName "foo"
  │  │╰─ tagName "div"
  ╰─ ╰─ openTagStart
 3╭─      hello="world" // this is another test
- │       │    │╰─ attrValue.value "\"world\""
- │       │    ╰─ attrValue "=\"world\""
+ │       │    │╰─ attrValue.value "\"world\" // this is another test"
+ │       │    ╰─ attrValue "=\"world\" // this is another test"
  ╰─      ╰─ attrName "hello"
 4╭─ ></div>
  │  ││ │  ╰─ closeTagEnd(div)

diff --git a/src/states/EXPRESSION.ts b/src/states/EXPRESSION.ts
@@ -87,10 +87,13 @@ export const EXPRESSION: StateDefinition<ExpressionMeta> = {
             break;
           default: {
             if (
-              !canCharCodeBeFollowedByDivision(
+              canCharCodeBeFollowedByDivision(
                 this.getPreviousNonWhitespaceCharCode()
               )
             ) {
+              this.pos++;
+              this.consumeWhitespace();
+            } else {
               this.enterState(STATE.REGULAR_EXPRESSION);
             }
             break;
@@ -219,7 +222,6 @@ function buildPattern(type: PatternType) {
     `|(?<!-)-${
       type === PatternType.CONCISE_ATTRS ? "" : "(?:\\s*-\\s*-)*\\s*"
     }(?!-)` + // In concise mode we can't match multiple hyphens otherwise we can match an even number of hyphens
-    `|(?<![/*])/(?![/*${type === PatternType.HTML_ATTRS ? ">" : ""}])` + // We only continue after a forward slash if it isn't //, /* (or /> in html mode)
     `|(?<!\\.)\\.(?!\\.)` + // We only continue after a dot if it isn't a ...
     `|>${type === PatternType.HTML_ATTRS ? "{2,}" : "+"}` + // in html mode only consume closing angle brackets if it is >>
     "|[ \\t]+(?:in(?:stanceof)?|as|extends)(?=[ \\t]+[^=/,;:>])"; // We only continue after word operators (instanceof/in) when they are not followed by a terminator
@@ -233,7 +235,7 @@ function buildPattern(type: PatternType) {
     "|typeof" +
     "|void" +
     ")\\b";
-  const lookAheadPattern = `${space}*(?:${binary})\\s*|${space}+(?=[{(])`; // if we have spaces followed by an opening bracket, we'll consume the spaces and let the expression state handle the brackets
+  const lookAheadPattern = `${space}*(?:${binary})\\s*|${space}+(?=[{(]|/[^>])`; // if we have spaces followed by an opening bracket, we'll consume the spaces and let the expression state handle the brackets
   const lookBehindPattern = `(?<=${unary}|${binary})`;
   return new RegExp(`${lookAheadPattern}|${lookBehindPattern}`, "ym");
 }