Notes on using ANTLR4 to parse PL/SQL in C#.

antlr4-notes.md

Setting Up ANTLR

1. Download and set up ANTLR4
2. Download PlSqlLexer.g4 and PlSqlParser.g4 from https://github.com/antlr/grammars-v4/tree/master/sql/plsql

Building the Grammar

1. Build the lexer (https://stackoverflow.com/a/55379369/1958726):

   antlr4 PlSqlLexer.g4 -Dlanguage=CSharp
   

2. Build the parser:

   antlr4 PlSqlParser.g4 -Dlanguage=CSharp
   

Creating a Project

1. PowerShell:

   dotnet new console -n MyPlSqlProject
   cd .\MyPlSqlProject
   iwr https://github.com/antlr/grammars-v4/raw/master/sql/plsql/CSharp/PlSqlLexerBase.cs -OutFile PlSqlLexerBase.cs
   iwr https://github.com/antlr/grammars-v4/raw/master/sql/plsql/CSharp/PlSqlParserBase.cs -OutFile PlSqlParserBase.cs
   iwr https://raw.githubusercontent.com/antlr/antlr4/master/doc/resources/CaseChangingCharStream.cs -OutFile CaseChangingCharStream.cs
   

2. Copy the .cs files that ANTLR generated in the previous section into the project folder.

3. Fix namespaces; remove namespaces from PlSqlLexerBase.cs and PlSqlParserBase.cs, or add PlSqlParseTree usings to their subclasses, or do something else (maybe antlr4 takes a namespace argument when building the lexer and parser?).

4. Add missing constructor overload to PlSqlLexerBase.cs:

   public PlSqlLexerBase(ICharStream input, TextWriter output, TextWriter errorOutput) : base(input, output, errorOutput) { }
   

5. Add missing constructor overload to PlSqlParserBase.cs:

   protected PlSqlParserBase(ITokenStream input, TextWriter output, TextWriter errorOutput) : base(input, output, errorOutput) { }
   

6. In PlSqlLexerBase, change int la = _input.La(pos); to int la = InputStream.LA(pos); if it hasn't been fixed in the repo yet.

7. Insert into Program.cs:

   var charStream = CharStreams.fromstring("My PL/SQL"); // or .fromPath or whatever
   var caseChangingCharStream = new CaseChangingCharStream(charStream, true); // Required for PL/SQL; see https://github.com/antlr/grammars-v4/blob/master/sql/plsql/README.md
   
   var lexer = new PlSqlLexer(caseChangingCharStream);
   var commonTokenStream = new CommonTokenStream(lexer);
   var parser = new PlSqlParser(commonTokenStream);
   

8. For faster compilation, separate the ANTLR classes into their own class library project, and uncheck Build for it under Build > Configuration Manager.

Using XPath

If you just need to grab specific things, like all tableview names in a view DDL's from clause, you can use ANTLR's XPath class. Example:

var tableviews = XPath.FindAll(tree, "//create_view//from_clause//tableview_name", parser);

Getting a Context's Text, Including Whitespace

context.GetText() just returns the combined text of all child nodes; whitespace isn't included.

To get the full, original text, including whitespace, use:

caseChangingCharStream.GetText(new Interval(context.Start.StartIndex, context.Stop.StopIndex);

Thanks awfully.

Sure, glad it helped!

Thanks, it helped me.