OpenLanguage is a C# library providing lexers, parsers, and other processing tools for Open Office XML DSLs.
Features
WordprocessingML
Field Instructions: Parse field instructions into strongly-typed Abstract Syntax Trees (AST)
Grammar-Based: Uses GPLEX/GPPG for robust parsing of field instructions, expressions, and merge fields
Comprehensive Field Types: Support for all field instructions - both standard ECMA variations and those specified in the ISO docs for legacy compatibility
AST Manipulation: Programmatically access and modify parsed structures
Field Reconstruction: Convert ASTs back to valid field instruction strings with round-trip fidelity
SpreadsheetML
Formula Parsing: Parse SpreadsheetML formulas into Abstract Syntax Trees (AST)
Grammar-Based: Uses GPLEX lexer (.lex) and GPPG yacc parser (.y) for concise and efficient grammar specification and parsing logic
AST Manipulation: Access and modify parsed formula structures programmatically
Formula Reconstruction: Convert modified ASTs back to valid Excel formula strings
using OpenLanguage.SpreadsheetML.Formula;
// Parse an Excel formula
Ast.Node formula = FormulaParser.Parse("=SUM(A1:A10) * 2");
// Access the AST
Console.WriteLine($"Reconstructed: {formula.ToString()}");
// Try parsing with error handling
Ast.Node? maybeFormula = FormulaParser.TryParse("=INVALID_SYNTAX(");
if (maybeFormula == null)
{
Console.WriteLine("Parse failed - invalid syntax");
}
Field Instruction Processing
using OpenLanguage.WordprocessingML.FieldInstruction;
using OpenLanguage.WordprocessingML.FieldInstruction.Ast;
using OpenLanguage.WordprocessingML.Ast;
// Parse a field instruction into a strongly-typed AST node
var ast = FieldInstructionParser.Parse("MERGEFIELD FirstName \\* Upper");
// Check the type and use specific properties
if (ast is MergeFieldFieldInstruction mergeField)
{
Console.WriteLine($"Field Name: {mergeField.FieldName}");
if (mergeField.GeneralFormat?.Argument is StringLiteralNode format)
{
Console.WriteLine($"General Format: {format.Value}");
}
}
// Reconstruct field instruction
Console.WriteLine($"Field instruction: {ast.ToString()}");
# Format entire solution
dotnet csharpier format .
# Check formatting
dotnet csharpier check .
Git Hooks
Install git hooks to ensure code quality:
cmake --build build --target install-hooks
This installs a pre-commit hook that:
Runs code formatting
Executes all tests
Prevents commits if tests fail
Testing
The project uses xUnit for testing:
# Run all tests
dotnet test
# Run tests with coverage
dotnet test --collect:"XPlat Code Coverage"
# Run specific test project
dotnet test OpenLanguage.Test/
Performance
OpenLanguage is built with performance as a primary concern:
Native AOT Ready: Full compatibility with .NET Native AOT
Optimized Grammar: LALR yacc parser used for SpreadsheetML formula parsing with highly optimized, minimal lex grammar.
Compared against the ABNF grammar specification at each step of implementation.
Compatibility
Multi-framework: Supports both .NET 8.0 and .NET 9.0
Native AOT: Full support for ahead-of-time compilation
TODO
WordprocessingML
Field Instruction
Evaluation
Implement per-class evaluation of strongly typed field instructions.
Implement execution of parsed merge field.
Ensure evaluation is implemented with respect to configured decimalSymbol
Misc
Allow configuration option for decimalSymbol used for parsing floating point numbers
Complete exhaustive enumeration of CountryRegion enumerations.
LevelText
Define and implement paragraph numbering level text placeholder syntax grammar
Implement AST parsing of paragraph numbering level text
Implement placeholder value evaluator
Add API for final, evaluated numbering level text construction
SpreadsheetML
Test coverage is quite comprehensive, as are grammar and parser rule
specifications - there should not be anything left to complete here as far as
implementation of parsing, parsing dependencies, nor AST. However, optimization leaves a bit to be desired, and evaluation is unimplemented.
Optimization
Runtime memory consumption of FormulaParser and AST node classes, as
well as size of generated parser code, jumped by ~10x on adding the
builtin_function_call_head_raw rule. Investigate the cause and resolve.
Many Shift, Reduce, Shift/Reduce, and Reduce/Reduce conflicts are
automatically resolved on code generation of parser. Investigate the cause and
resolve.
Evaluation
Implement evaluation of formula, with a per-AST node class method called
Evaluate, on the abstract ExpressionNode class and overriden by derived
classes.
Implement all builtin SpreadsheetML functions.
Implement cell, sheet, table, and range reference resolution for read
and write of underlying value.
Implement arithmetic expression evaluation.
Implement function reference resolution and call evaluation -
regardless of whether user-defined, _xlpm.-prefixed function references, or
LAMBDA functions.
Bonus: implement a shared SpreadsheetContext to abstract common data
reading and writing operations
Bonus if completed: in SpreadsheetContext, use generic underlying data representation which
is derived from a common Spreadsheet class, allowing any underlying
matrix-like data representation to be manipulated by formulas.
Example Usage
Write a toy formula interpreter when evaluation is implemented.
Bonus: Realtime display and update of spreadsheet cell values in TUI
with matrix display and formula input prompt as well as vim-style selection of
corresponding cells. Simulaneously update underlying OPC package on modifying
values.
