Add 7z archive writer with LZMA/LZMA2 compression

[DanNsk]

Summary

Adds write support for 7z archives. Files can be compressed with LZMA or LZMA2 (default). Each file is compressed independently (non-solid mode). Archives produced pass 7z t validation and open correctly in 7-Zip GUI.

The writer plugs into the existing WriterFactory/SevenZipFactory infrastructure and supports both sync and async APIs. SevenZipWriterOptions controls compression type, header compression, and LZMA encoder settings.

Also adds LZMA2 to the CompressionType enum and a Lzma2EncoderStream that implements chunk-based LZMA2 framing on top of the existing LZMA encoder.

What's not included (but will/can add what i can if this PR will be accepted)

• Solid mode (all files in one compression block)
• Encryption
• IWritableArchive integration
• BCJ/BCJ2 filters
• Truly async compression - LZMA is inherently CPU-bound and the underlying encoder is synchronous, so the async writer methods delegate to sync internally. Don't see a good way to make this genuinely async without Task.Run in library code, which just hides threading from the caller. Open to ideas if there's a better approach.

Testing

Round-trip tests cover single/multiple files, directories, empty files, large data, both compression methods, compressed and uncompressed headers, and factory-based creation. Tests pass on net10.0 and net48. Archives also verified externally with 7-Zip.

[kilo-code-bot]

Code Review Summary

Status: 1 Active Issue | Recommendation: Address before merge

Overview

Severity	Count
CRITICAL	0
WARNING	1
SUGGESTION	0

Issue Details (click to expand)

WARNING

File	Line	Issue
src/SharpCompress/Writers/SevenZip/SevenZipWriter.cs	97	Empty non-seekable streams can leave orphaned packed data - FIX IN PROGRESS (rewind added, awaiting final verification)

Note: The issue has been partially addressed with a rewind fix at lines 106-107 that erases orphaned encoder bytes when empty streams are detected. The author indicated they will extend this with byte probe later.

Files Reviewed (14 files)

• src/SharpCompress/Common/CompressionType.cs - Added LZMA2 enum value
• src/SharpCompress/Common/SevenZip/ArchiveWriter.cs - New header writer
• src/SharpCompress/Common/SevenZip/SevenZipFilesInfo.cs - New files info writer
• src/SharpCompress/Common/SevenZip/SevenZipHeaderStructures.cs - New header structures
• src/SharpCompress/Common/SevenZip/SevenZipSignatureHeaderWriter.cs - New signature header
• src/SharpCompress/Common/SevenZip/SevenZipStreamsCompressor.cs - New streams compressor
• src/SharpCompress/Factories/SevenZipFactory.cs - Added IWriterFactory
• src/SharpCompress/Writers/SevenZip/SevenZipWriter.Async.cs - Async support
• src/SharpCompress/Writers/SevenZip/SevenZipWriter.Factory.cs - Factory methods
• src/SharpCompress/Writers/SevenZip/SevenZipWriter.cs - Main writer implementation
• src/SharpCompress/Writers/SevenZip/SevenZipWriterOptions.cs - Writer options
• src/SharpCompress/Writers/WriterOptionsExtensions.cs - LeaveStreamOpen support
• tests/SharpCompress.Test/SevenZip/SevenZipWriterTests.cs - Comprehensive tests

Assessment: The implementation is well-structured with solid test coverage. The existing warning at line 97 has been partially addressed with the rewind fix. Consider verifying the empty stream handling works correctly with non-seekable streams before merging.

[adamhathcock]

This is great, I'll have to inspect hits closely

I was close to merging in a new PR that changes the API just for 7z files because you can't really do random access of bytes (extract them) as it's likely there's multiple files in a single stream. I'm interested in how you handled it.

[adamhathcock]

Looks like it's just the writer and does a compresssion stream per file. I'll sleep on it but probably want to take it as-is if the tests pass!

[DanNsk]

Looks like it's just the writer and does a compresssion stream per file. I'll sleep on it but probably want to take it as-is if the tests pass!

Yes it doesn't do solid archives yet, I honestly didn't want to add everything at once - just to make PR easier to comprehend