From fdbd4fe197f4e7ac7e1d2a00df4414c40e765f60 Mon Sep 17 00:00:00 2001
From: Dmitry Gozman <dgozman@gmail.com>
Date: Thu, 28 May 2020 14:49:39 -0700
Subject: [PATCH] fix(selectors): fix selector parsing for css attributes and
 quotes (#2389)

- css attribute selector may contain spaces;
- >> escaping inside strings was sometimes incorrect.

See added test cases for more details.
---
 src/injected/cssSelectorEngine.ts | 13 ++++++++++---
 src/selectors.ts                  |  3 +++
 test/queryselector.spec.js        | 32 +++++++++++++++++++++++++++++++
 3 files changed, 45 insertions(+), 3 deletions(-)

diff --git a/src/injected/cssSelectorEngine.ts b/src/injected/cssSelectorEngine.ts
index c469fd0009..de173bfcf1 100644
--- a/src/injected/cssSelectorEngine.ts
+++ b/src/injected/cssSelectorEngine.ts
@@ -219,6 +219,7 @@ function parentElementOrShadowHost(element: Element): Element | undefined {
 function split(selector: string): string[][] {
   let index = 0;
   let quote: string | undefined;
+  let insideAttr = false;
   let start = 0;
   let space: 'none' | 'before' | 'after' = 'none';
   const result: string[][] = [];
@@ -235,7 +236,7 @@ function split(selector: string): string[][] {
   };
   while (index < selector.length) {
     const c = selector[index];
-    if (!quote && c === ' ') {
+    if (!quote && !insideAttr && c === ' ') {
       if (space === 'none' || space === 'before')
         space = 'before';
       index++;
@@ -256,10 +257,16 @@ function split(selector: string): string[][] {
       } else if (c === quote) {
         quote = undefined;
         index++;
-      } else if (c === '\'' || c === '"') {
+      } else if (!quote && (c === '\'' || c === '"')) {
         quote = c;
         index++;
-      } else if (!quote && c === ',') {
+      } else if (!quote && c === '[') {
+        insideAttr = true;
+        index++;
+      } else if (!quote && insideAttr && c === ']') {
+        insideAttr = false;
+        index++;
+      } else if (!quote && !insideAttr && c === ',') {
         appendToResult();
         index++;
         start = index;
diff --git a/src/selectors.ts b/src/selectors.ts
index 97a99239d7..92891e9022 100644
--- a/src/selectors.ts
+++ b/src/selectors.ts
@@ -219,6 +219,9 @@ export function parseSelector(selector: string): types.ParsedSelector {
     } else if (c === quote) {
       quote = undefined;
       index++;
+    } else if (!quote && (c === '"' || c === '\'' || c === '`')) {
+      quote = c;
+      index++;
     } else if (!quote && c === '>' && selector[index + 1] === '>') {
       append();
       index += 2;
diff --git a/test/queryselector.spec.js b/test/queryselector.spec.js
index b1fc7b04b2..29a96b0078 100644
--- a/test/queryselector.spec.js
+++ b/test/queryselector.spec.js
@@ -494,6 +494,14 @@ describe('text selector', () => {
     expect(await page.$eval(`text="`, e => e.outerHTML)).toBe('<div> " </div>');
     expect(await page.$eval(`text='`, e => e.outerHTML)).toBe('<div> \' </div>');
 
+    await page.setContent(`<div>Hi''&gt;&gt;foo=bar</div>`);
+    expect(await page.$eval(`text="Hi''>>foo=bar"`, e => e.outerHTML)).toBe(`<div>Hi''&gt;&gt;foo=bar</div>`);
+    await page.setContent(`<div>Hi'"&gt;&gt;foo=bar</div>`);
+    expect(await page.$eval(`text="Hi'\\">>foo=bar"`, e => e.outerHTML)).toBe(`<div>Hi'"&gt;&gt;foo=bar</div>`);
+
+    await page.setContent(`<div>Hi&gt;&gt;<span></span></div>`);
+    expect(await page.$eval(`text="Hi>>">>span`, e => e.outerHTML)).toBe(`<span></span>`);
+
     await page.setContent(`<div>a<br>b</div><div>a</div>`);
     expect(await page.$eval(`text=a`, e => e.outerHTML)).toBe('<div>a<br>b</div>');
     expect(await page.$eval(`text=b`, e => e.outerHTML)).toBe('<div>a<br>b</div>');
@@ -622,6 +630,30 @@ describe('css selector', () => {
     expect(await page.$eval(`css=[attr='hello,world!']`, e => e.outerHTML)).toBe('<div attr="hello,world!"></div>');
     expect(await page.$eval(`css=div[attr="hello,world!"],span`, e => e.outerHTML)).toBe('<span></span>');
   });
+
+  it('should work with attribute selectors', async({page}) => {
+    await page.setContent(`<div attr="hello world" attr2="hello-''>>foo=bar[]" attr3="] span"><span></span></div>`);
+    await page.evaluate(() => window.div = document.querySelector('div'));
+    const selectors = [
+      `[attr="hello world"]`,
+      `[attr = "hello world"]`,
+      `[attr ~= world]`,
+      `[attr ^=hello ]`,
+      `[attr $= world ]`,
+      `[attr *= "llo wor" ]`,
+      `[attr2 |= hello]`,
+      `[attr = "Hello World" i ]`,
+      `[attr *= "llo WOR"i]`,
+      `[attr $= woRLD i]`,
+      `[attr2 = "hello-''>>foo=bar[]"]`,
+      `[attr2 $="foo=bar[]"]`,
+    ];
+    for (const selector of selectors)
+      expect(await page.$eval(selector, e => e === div)).toBe(true);
+    expect(await page.$eval(`[attr*=hello] span`, e => e.parentNode === div)).toBe(true);
+    expect(await page.$eval(`[attr*=hello] >> span`, e => e.parentNode === div)).toBe(true);
+    expect(await page.$eval(`[attr3="] span"] >> span`, e => e.parentNode === div)).toBe(true);
+  });
 });
 
 describe('attribute selector', () => {