feat: add expr validator helpers

Author: DylanPiercey

.changeset/nine-cooks-sneeze.md

@@ -0,0 +1,5 @@
+---
+"htmljs-parser": minor
+---
+
+Add expression validator helpers.

src/__tests__/validate.test.ts

@@ -0,0 +1,72 @@
+import assert from "node:assert/strict";
+import { isValidAttrValue, isValidStatement } from "..";
+
+describe("validation helpers", () => {
+  describe("isValidStatement", () => {
+    it("accepts single-line expressions", () => {
+      assert.equal(isValidStatement("foo + bar"), true);
+    });
+
+    it("accepts indented continuation lines", () => {
+      assert.equal(isValidStatement("foo\n  + bar"), true);
+    });
+
+    it("rejects unindented continuation lines", () => {
+      assert.equal(isValidStatement("foo\nbar"), false);
+    });
+
+    it("accepts indented ternary continuation", () => {
+      assert.equal(isValidStatement("foo ?\n  bar : baz"), true);
+    });
+
+    it("rejects unterminated groups", () => {
+      assert.equal(isValidStatement("(foo"), false);
+    });
+
+    it("rejects mismatched closing groups", () => {
+      assert.equal(isValidStatement(")"), false);
+    });
+  });
+
+  describe("isValidAttrValue", () => {
+    it("accepts html attr values with operators", () => {
+      assert.equal(isValidAttrValue("foo + bar", false), true);
+    });
+
+    it("accepts html attr values containing =>", () => {
+      assert.equal(isValidAttrValue("foo=>bar", false), true);
+    });
+
+    it("rejects html attr values terminated by >", () => {
+      assert.equal(isValidAttrValue("foo >", false), false);
+    });
+
+    it("accepts concise attr values with >", () => {
+      assert.equal(isValidAttrValue("foo > bar", true), true);
+    });
+
+    it("rejects html attr values terminated by commas", () => {
+      assert.equal(isValidAttrValue("foo, bar", false), false);
+    });
+
+    it("accepts html attr values containing semicolons", () => {
+      assert.equal(isValidAttrValue("foo;", false), true);
+    });
+
+    it("rejects concise attr values terminated by semicolons", () => {
+      assert.equal(isValidAttrValue("foo;", true), false);
+    });
+
+    it("accepts html attr values with decrement operator", () => {
+      assert.equal(isValidAttrValue("foo --", false), true);
+    });
+
+    it("rejects concise attr values with decrement operator", () => {
+      assert.equal(isValidAttrValue("foo --", true), false);
+    });
+
+    it("rejects attr values separated only by whitespace", () => {
+      assert.equal(isValidAttrValue("foo bar", false), false);
+    });
+  });
+});

src/index.ts

@@ -12,6 +12,8 @@ export {
   type Range,
 } from "./internal";
 
+export { isValidStatement, isValidAttrValue } from "./util/validators";
+
 /**
  * Creates a new Marko parser.
  */

src/states/ATTRIBUTE.ts

@@ -323,7 +323,7 @@ function shouldTerminateHtmlAttrName(code: number, data: string, pos: number) {
   }
 }
 
-function shouldTerminateHtmlAttrValue(
+export function shouldTerminateHtmlAttrValue(
   this: STATE.ExpressionMeta,
   code: number,
   data: string,
@@ -368,7 +368,7 @@ function shouldTerminateConciseAttrName(
   }
 }
 
-function shouldTerminateConciseAttrValue(
+export function shouldTerminateConciseAttrValue(
   code: number,
   data: string,
   pos: number,

src/util/validators.ts

@@ -0,0 +1,102 @@
+import {
+  CODE,
+  STATE,
+  Parser,
+  type StateDefinition,
+  type Meta,
+} from "../internal";
+import {
+  shouldTerminateConciseAttrValue,
+  shouldTerminateHtmlAttrValue,
+} from "../states";
+
+const ROOT_STATE: StateDefinition = {
+  name: "ROOT",
+  enter() {
+    return ROOT_RANGE;
+  },
+  exit() {},
+  char() {},
+  eol() {},
+  eof() {},
+  return() {},
+};
+const ROOT_RANGE = {
+  state: ROOT_STATE,
+  parent: undefined as unknown as Meta,
+  start: 0,
+  end: 0,
+};
+
+export function isValidStatement(code: string): boolean {
+  return isValid(code, true, prepareStatement);
+}
+
+function prepareStatement(expr: STATE.ExpressionMeta) {
+  expr.operators = true;
+  expr.terminatedByEOL = true;
+  expr.consumeIndentedContent = true;
+}
+
+export function isValidAttrValue(code: string, concise: boolean): boolean {
+  return isValid(code, concise, prepareAttrValue);
+}
+
+function prepareAttrValue(expr: STATE.ExpressionMeta, concise: boolean) {
+  expr.operators = true;
+  expr.terminatedByWhitespace = true;
+  expr.shouldTerminate = concise
+    ? shouldTerminateConciseAttrValue
+    : shouldTerminateHtmlAttrValue;
+}
+
+function isValid(
+  data: string,
+  concise: boolean,
+  prepare: (expr: STATE.ExpressionMeta, concise: boolean) => void,
+) {
+  const parser = new Parser({});
+  const maxPos = (parser.maxPos = data.length);
+  parser.pos = 0;
+  parser.data = data;
+  parser.indent = "";
+  parser.forward = 1;
+  parser.textPos = -1;
+  parser.isConcise = concise;
+  parser.beginMixedMode = parser.endingMixedModeAtEOL = false;
+  parser.lines = parser.activeTag = parser.activeAttr = undefined;
+  parser.activeState = ROOT_STATE;
+  parser.activeRange = ROOT_RANGE;
+  const expr = parser.enterState(STATE.EXPRESSION);
+  prepare(expr, concise);
+
+  while (parser.pos < maxPos) {
+    const code = data.charCodeAt(parser.pos);
+
+    if (code === CODE.NEWLINE) {
+      parser.forward = 1;
+      parser.activeState.eol.call(parser, 1, parser.activeRange);
+    } else if (
+      code === CODE.CARRIAGE_RETURN &&
+      data.charCodeAt(parser.pos + 1) === CODE.NEWLINE
+    ) {
+      parser.forward = 2;
+      parser.activeState.eol.call(parser, 2, parser.activeRange);
+    } else {
+      parser.forward = 1;
+      parser.activeState.char.call(parser, code, parser.activeRange);
+    }
+
+    if (parser.activeRange === ROOT_RANGE) {
+      return false;
+    }
+
+    parser.pos += parser.forward;
+  }
+
+  return (
+    parser.pos === maxPos &&
+    parser.activeRange === expr &&
+    !expr.groupStack.length
+  );
+}