Gradle is used to build the project. gradle-intellij-plugin is used to pull
any given version of IntelliJ to add its JARs to the classpath, build forms,
publish to the marketplace etc.
GitHub Actions are used for continuous integration, see files in .github/workflows.
An ANTLR v4 grammar in src/main/antlr is used to generate a lexer/parser.
ANTLRv4ParserDefinition uses the antlr4-intellij-adaptor library to
delegate parsing/lexing to these generated classes.
Syntax highlighting is also using the generated lexer (by delegation) in
org.antlr.intellij.plugin.ANTLRv4SyntaxHighlighter.getHighlightingLexer.
The plugin embeds a complete version of ANTLR v4. To highlight warnings in the
editor, the file is parsed again in ANTLRv4ExternalAnnotator, but this time
using the "official" parser. The annotator constructs a org.antlr.v4.Tool
with all the flags configured in the Configure ANTLR... dialog.
GrammarIssuesCollector sets up an error listener, then processes all the issues
reported by the Tool to show them in the editor.
Contrary to ANTLRWorks, the IntelliJ plugin does not run the actual generated parser to test a grammar. Instead, it uses ANTLR interpreters. This is mainly because the IDE classpath and the project classpath are totally different, meaning generated parsers and their custom code and dependencies are not available from the IDE process.
The interpreter provides a convenient way to "run" grammars without needing
any generated code. A major drawback of this approach is that custom code will
not be executed: @members, actions, predicates...
The entry point for the preview window is PreviewPanel, which holds the main
Swing component displayed in the tool window. It uses ANTLRv4PluginController
to maintain a cache of parsed grammars that can be previewed. Everytime the
editor switches to another .g4 grammar, the preview is updated accordingly:
the input text and selected rule are restored to their previous state (if any),
and the interpreter results are updated. Interpreter results include profiling
data, a list of tokens and a tree of matched rules and terminal nodes (both
in graphical form and in JTree form).
PreviewPanel acts as an orchestrator between inputs (grammar changed event,
input text changed event) and outputs (profiler, token list, parse tree etc.).
Since the preview panel has full control over the interpreter, it can monitor the parsing phase to detect potential infinite loops caused by bad grammars, thus preventing IDE freezes or OutOfMemoryExceptions.