fix: make json slicing more robust (#611)

fixes #585
closes #599

* better slice json, do not use javascript constants after it

* remove console.log

* fix typo in module name

* removed slice, use accumulated string instead (because slice pos was calculated wrong sometime)

* some codestyle fixes

* Update util.js

* Update util.js

* remove unused variable

* Rely on closing all brackets instead of finding a semi-colon

Plus some changes to the error messages

* Clean up tests for cutting JSON

* remove unnessecary lines

Co-authored-by: Aleksei Zbiniakov <alexey@coccoc.com>
Co-authored-by: Alexey <zbinyakov@gmail.com>
Co-authored-by: TimeForANinja <t.kutscha@yahoo.de>

lib/info.js

@@ -74,7 +74,7 @@ exports.getBasicInfo = async(id, options) => {
   const jsonStr = util.between(body, 'ytplayer.config = ', '</script>');
   let config;
   if (jsonStr) {
-    config = jsonStr.slice(0, jsonStr.lastIndexOf(';ytplayer.load'));
+    config = util.cutAfterJSON(jsonStr);
     return gotConfig(id, options, additional, config, false);
   } else {
     // If the video page doesn't work, maybe because it has mature content.

lib/util.js

@@ -374,3 +374,55 @@ exports.parseAbbreviatedNumber = string => {
   }
   return null;
 };
+
+/**
+ * Match begin and end braces of input JSON, return only json
+ *
+ * @param {String} mixedJson
+ * @return {String}
+*/
+exports.cutAfterJSON = mixedJson => {
+  let open, close;
+  if (mixedJson[0] === '[') {
+    open = '[';
+    close = ']';
+  } else if (mixedJson[0] === '{') {
+    open = '{';
+    close = '}';
+  }
+
+  if (!open) {
+    throw new Error(`Can't cut unsupported JSON (need to begin with [ or { ) but got: ${mixedJson[0]}`);
+  }
+
+  // States if the loop is currently in a string
+  let isString = false;
+
+  // Current open brackets to be closed
+  let counter = 0;
+
+  let i;
+  for (i = 0; i < mixedJson.length; i++) {
+    // Toggle the isString boolean when leaving/entering string
+    if (mixedJson[i] === '"' && mixedJson[i - 1] !== '\\') {
+      isString = !isString;
+      continue;
+    }
+    if (isString) continue;
+
+    if (mixedJson[i] === open) {
+      counter++;
+    } else if (mixedJson[i] === close) {
+      counter--;
+    }
+
+    // All brackets have been closed, thus end of JSON is reached
+    if (counter === 0) {
+      // Return the cut JSON
+      return mixedJson.substr(0, i + 1);
+    }
+  }
+
+  // We ran through the whole string and ended up with an unclosed bracket
+  throw Error("Can't cut unsupported JSON (no matching closing bracket found)");
+};

test/util-test.js

@@ -434,4 +434,52 @@ describe('util.addFormatMeta()', () => {
       });
     });
   });
+  describe('util.cutAfterJSON()', () => {
+    it('Works with simple JSON', () => {
+      assert.equal(util.cutAfterJSON('{"a": 1, "b": 1}'), '{"a": 1, "b": 1}');
+    });
+    it('Cut extra characters after JSON', () => {
+      assert.equal(util.cutAfterJSON('{"a": 1, "b": 1}abcd'), '{"a": 1, "b": 1}');
+    });
+    it('Tolerant to string constants', () => {
+      assert.equal(util.cutAfterJSON('{"a": "}1", "b": 1}abcd'), '{"a": "}1", "b": 1}');
+    });
+    it('Tolerant to string with escaped quoting', () => {
+      assert.equal(util.cutAfterJSON('{"a": "\\"}1", "b": 1}abcd'), '{"a": "\\"}1", "b": 1}');
+    });
+    it('works with nested', () => {
+      assert.equal(
+        util.cutAfterJSON('{"a": "\\"1", "b": 1, "c": {"test": 1}}abcd'),
+        '{"a": "\\"1", "b": 1, "c": {"test": 1}}',
+      );
+    });
+    it('Works with utf', () => {
+      assert.equal(
+        util.cutAfterJSON('{"a": "\\"фыва", "b": 1, "c": {"test": 1}}abcd'),
+        '{"a": "\\"фыва", "b": 1, "c": {"test": 1}}',
+      );
+    });
+    it('Works with \\\\ in string', () => {
+      assert.equal(
+        util.cutAfterJSON('{"a": "\\\\фыва", "b": 1, "c": {"test": 1}}abcd'),
+        '{"a": "\\\\фыва", "b": 1, "c": {"test": 1}}',
+      );
+    });
+    it('Works with [ as start', () => {
+      assert.equal(
+        util.cutAfterJSON('[{"a": 1}, {"b": 2}]abcd'),
+        '[{"a": 1}, {"b": 2}]',
+      );
+    });
+    it('Returns an error when not beginning with [ or {', () => {
+      assert.throws(() => {
+        util.cutAfterJSON('abcd]}');
+      }, /Can't cut unsupported JSON \(need to begin with \[ or { \) but got: ./);
+    });
+    it('Returns an error when missing closing bracket', () => {
+      assert.throws(() => {
+        util.cutAfterJSON('{"a": 1,{ "b": 1}');
+      }, /Can't cut unsupported JSON \(no matching closing bracket found\)/);
+    });
+  });
 });