[bug] Preprocessor considers all curly braces, whether legal tokens or not

[kleinesfilmroellchen]

Describe the bug

The Preprocessor tries to ensure that all code block braces match up. However, it primitively only checks the characters themselves, not whether they occur as a separate token. Therefore, curly braces which are part of a string (common, critical) or invalid token are also counted

To Reproduce

Run the following code through the preprocessor:

{ 4 5 "some tokens" writeln} .

Notice the missing space after writeln. The preprocessor will accept this code as legal, causing the interpreter to fail undeterministically.

Expected behavior

The above code should be rejected. Curly braces in non-token positions should not be counted.

Platform information

• Output of java -version: openjdk version "14.0.1" 2020-04-14 OpenJDK Runtime Environment (build 14.0.1+7) OpenJDK 64-Bit Server VM (build 14.0.1+7, mixed mode, sharing)
• Output of sof -v: sof version 0.1 (built 04.07.20, 20:33)
• Operating System, possibly relevant hardware information: Win10

[kleinesfilmroellchen]

Preprocessor.java line 88f is the culprit.

[kleinesfilmroellchen]

Fixed via 1d5e0b659ea839caa91adac08c69d5d1aadc39e3.