Normalize paths
Reject absolute paths and entries that escape the chosen destination through parent-directory components.
Family 03
Archives
Bundling files and compressing bytes are different operations. ZIP does both in one container; tar and gzip are often composed as two layers.
| Format | Model | Good fit | Watch for |
|---|---|---|---|
| ZIP .zipapplication/zip | A multi-entry archive. Each entry can be stored or compressed, with a central directory describing the collection. | Portable bundles where individual files should remain addressable. | Compression method and encryption support vary. A .zip extension does not guarantee that every reader supports every feature. |
| gzip .gzapplication/gzip | A compressed data stream with a small header and trailer; conventionally wraps one input stream rather than a directory tree. | Compressing a single file or the output of another archival format. | Original filename and timestamp are optional header fields, not a complete filesystem model. |
| tar .tarapplication/x-tar | A sequential archive of filesystem entries and metadata. Classic tar itself does not imply compression. | Preserving a directory tree before applying a stream compressor. | There are multiple tar variants and extensions. Metadata portability depends on producer and reader. |
Extraction safety
An archive is input to a filesystem operation. Validate entries before writing them, especially when the archive is untrusted.
Inspect links
Symbolic links and hard links can redirect later writes outside the expected extraction tree.
Apply limits
Bound entry count, expanded size, nesting depth, and processing time before extraction.