Javascript fix issues, unit tests, adding CI

[stephenberry]

Fix JavaScript library bugs and add test suite

Summary

This PR fixes two reported bugs in the JavaScript BEVE library and adds a comprehensive test suite with CI integration.

• Fixes does not parse correctly after compressing large string size #9: Strings >= 64 characters now encode/decode correctly
• Fixes TypeError: Cannot convert undefined or null to object #7: null and undefined values are now properly handled

Changes

Bug Fixes

Issue #9: String parsing fails for strings >= 64 characters

Root cause: Three bugs were identified:

1. String length encoded as character count instead of byte count (beve.js)

   • The code used string.length (character count) instead of UTF-8 byte count
   • This violated the BEVE spec which requires byte count for string SIZE
   • Fixed by encoding to UTF-8 first: new TextEncoder().encode(value).length

2. read_compressed() read extra bytes for multi-byte formats (beve.js)

   • After reading the header byte to determine config, the code read the full 2/4/8 bytes again
   • The header byte is part of the compressed value, so this read 1 extra byte
   • Fixed by reading from the current cursor position without pre-incrementing

3. Big-endian byte order in read_compressed() (beve_file.js)

   • BEVE specifies little-endian, but the code used big-endian bit shifting
   • Fixed by using proper little-endian reading

Issue #7: TypeError for null/undefined values

Root cause: The write_value function had no handling for null or undefined.

Fix: Added handling that matches JSON.stringify behavior:

• null → encodes as BEVE null (type 0, is_bool=0)
• undefined at top level → encodes as null
• undefined in arrays → encodes as null
• undefined in object values → property is omitted
• Empty objects {} → now supported (previously threw)

Also fixed typed array detection to check all elements are numbers, not just the first.

Test Suite

Added Jest test framework with 80 tests covering:

Category	Tests	Description
Primitive Types	11	Booleans, integers, floats
Null/Undefined (Issue #7)	10	null, undefined in various contexts
String Handling	5	Basic string operations
String Boundaries (Issue #9)	10	63, 64, 65, 127, 16383, 16384 char strings
UTF-8 Multi-byte	7	Emoji, Chinese characters, mixed content
Objects	7	Simple, nested, empty, deeply nested
Object Key Boundaries	5	Long keys, UTF-8 keys
Long String Values	4	Issue #9 exact scenarios
Typed Arrays	4	Integer and float arrays
Untyped Arrays	4	Mixed types, nested
Edge Cases	5	Empty strings, null chars, surrogate pairs
Compatibility	6	Reading existing .beve example files
Regression	3	Issue #9 specific regression tests

CI Integration

Added GitHub Actions workflow (.github/workflows/test.yml) that:

• Runs on push to main and on pull requests
• Sets up Node.js 20
• Runs the test suite

Files Changed

File	Change
javascript/beve.js	Bug fixes for issues #7 and #9
javascript/beve_file.js	Bug fix for issue #9 (byte order)
javascript/beve.test.js	New: comprehensive test suite
javascript/package.json	New: npm package config with Jest
javascript/package-lock.json	New: dependency lock file
.github/workflows/test.yml	New: CI workflow
.gitignore	Updated: ignore node_modules, allow package.json

Testing

cd javascript
npm install
npm test

All 80 tests pass.

Compatibility

• No breaking changes to the public API
• Encoded data from the fixed library can be read by other BEVE implementations
• Existing .beve files continue to work (verified by compatibility tests)