fix: avoid expr continuations for period when not followed by identifier

marko-js · Mar 8, 2023 · 61e6966 · 61e6966
1 parent 33d280d
commit 61e6966
Show file tree

Hide file tree

Showing 4 changed files with 49 additions and 4 deletions.
diff --git a/.changeset/bright-clouds-yawn.md b/.changeset/bright-clouds-yawn.md
@@ -0,0 +1,5 @@
+---
+"htmljs-parser": patch
+---
+
+Avoid continuing expressions after a period if after the whitespace is something that could not be an identifier.
diff --git a/src/__tests__/fixtures/scriptlet-eof/__snapshots__/scriptlet-eof.expected.txt b/src/__tests__/fixtures/scriptlet-eof/__snapshots__/scriptlet-eof.expected.txt
@@ -0,0 +1,27 @@
+1╭─ $ $global.
+ │   │╰─ scriptlet.value "$global."
+ ╰─  ╰─ scriptlet " $global."
+2├─ 
+3╭─ <header>
+ │  ││     ╰─ openTagEnd
+ │  │╰─ tagName "header"
+ ╰─ ╰─ openTagStart
+4╭─   <img.logo src=logo alt="Marko"/>
+ │  │ ││  ││    │  ││    │  ││      ╰─ openTagEnd:selfClosed "/>"
+ │  │ ││  ││    │  ││    │  │╰─ attrValue.value "\"Marko\""
+ │  │ ││  ││    │  ││    │  ╰─ attrValue "=\"Marko\""
+ │  │ ││  ││    │  ││    ╰─ attrName "alt"
+ │  │ ││  ││    │  │╰─ attrValue.value "logo"
+ │  │ ││  ││    │  ╰─ attrValue "=logo"
+ │  │ ││  ││    ╰─ attrName "src"
+ │  │ ││  │╰─ tagShorthandClass.quasis[0] "logo"
+ │  │ ││  ╰─ tagShorthandClass ".logo"
+ │  │ │╰─ tagName "img"
+ │  │ ╰─ openTagStart
+ ╰─ ╰─ text "\n  "
+5╭─ </header>
+ │  │ │     ╰─ closeTagEnd(header)
+ │  │ ╰─ closeTagName "header"
+ │  ├─ text "\n"
+ ╰─ ╰─ closeTagStart "</"
+6╰─ 
diff --git a/src/__tests__/fixtures/scriptlet-eof/input.marko b/src/__tests__/fixtures/scriptlet-eof/input.marko
@@ -0,0 +1,5 @@
+$ $global.
+
+<header>
+  <img.logo src=logo alt="Marko"/>
+</header>
diff --git a/src/states/EXPRESSION.ts b/src/states/EXPRESSION.ts
@@ -266,12 +266,17 @@ function lookBehindForOperator(data: string, pos: number): number {
     case CODE.OPEN_ANGLE_BRACKET:
     case CODE.CLOSE_ANGLE_BRACKET:
     case CODE.PERCENT:
-    case CODE.PERIOD:
     case CODE.PIPE:
     case CODE.QUESTION:
     case CODE.TILDE:
       return curPos;
 
+    case CODE.PERIOD: {
+      // Only matches `.` followed by something that could be an identifier.
+      const nextPos = lookAheadWhile(isWhitespaceCode, data, pos);
+      return isWordCode(data.charCodeAt(nextPos)) ? nextPos : -1;
+    }
+
     // special case -- and ++
     case CODE.PLUS:
     case CODE.HYPHEN: {
@@ -324,9 +329,11 @@ function lookAheadForOperator(data: string, pos: number): number {
     case CODE.OPEN_PAREN:
       return pos; // defers to base expression state to track block groups.
 
-    case CODE.PERIOD:
-      // Only match a dot if its not ...
-      return data.charCodeAt(pos + 1) === CODE.PERIOD ? -1 : pos + 1;
+    case CODE.PERIOD: {
+      // Only matches `.` followed by something that could be an identifier.
+      const nextPos = lookAheadWhile(isWhitespaceCode, data, pos + 1);
+      return isWordCode(data.charCodeAt(nextPos)) ? nextPos : -1;
+    }
 
     default: {
       for (const keyword of binaryKeywords) {
@@ -383,6 +390,7 @@ function isWordCode(code: number) {
     (code >= CODE.UPPER_A && code <= CODE.UPPER_Z) ||
     (code >= CODE.LOWER_A && code <= CODE.LOWER_Z) ||
     (code >= CODE.NUMBER_0 && code <= CODE.NUMBER_9) ||
+    code == CODE.DOLLAR ||
     code === CODE.UNDERSCORE
   );
 }