fetch library
Downloads and archives: the "download and add" half of installing a
package, shared by just and its plugins.
- HttpFetcher: HTTPS only (every redirect too), polite retries, limits, every failure a FetchException, and CachedText that still reads offline.
- DownloadCache and checkPinned: downloads pinned by size and SHA-256, kept per user.
- GitHubRepo: where a repository's release files are, without the rate-limited GitHub API.
- ZipArchive reads a zip, refusing anything a tool should never unpack (paths outside the folder, links, ZIP64, bombs); ZipWriter writes one the same way every time. SafePaths holds the path rules both follow.
- CancelToken stops any of it.
Classes
- ByteRange
-
A part of a file to fetch: an HTTP
Rangeof bytes, so a tool can read a zip's directory, or one file in it, without downloading all of it. - CachedText
- Text fetched through an on-disk copy: fresh when the server answered, the cached copy when it said "unchanged" — or, offline, when it could not be reached at all or sent something unreadable.
- CancelToken
- A request to stop a download or an install, checked between steps.
- DownloadCache
- The per-user cache shared by every project on this machine: archives stored under their SHA-256, so a version downloaded once installs anywhere without the network, and thumbnails.
- FetchResponse
- What a server answered.
- GitHubRepo
-
A GitHub repository, as
https://github.com/<owner>/<repo>names it. - HashedFiles
- Files named by the SHA-256 of their bytes, in one folder: archives, catalog pieces. Every read checks the bytes again, so a damaged or replaced copy is never used — it is deleted, and reads as missing.
- HttpFetcher
- Fetches files over HTTPS.
- SafePaths
- Checks paths from untrusted archives so they can only ever land inside the folder they are extracted to, on every desktop platform.
- ZipArchive
- A ZIP archive read fully into memory and checked before any file in it is unpacked.
- ZipDirectory
- A ZIP archive's directory, read from the end of the file alone: every entry's path, sizes and flags, checked by every rule of ZipArchive.read that needs only the directory. What a tool reads to check an archive on a server — with HTTP ranges — before downloading all of it; readContent then fetches and unpacks one small file.
- ZipEntry
- One file or folder in a ZipArchive or a ZipDirectory.
- ZipLimits
- How much an archive may hold before ZipArchive.read refuses it.
- ZipWriter
- Builds a ZIP archive whose bytes depend only on its files' paths and contents.
Functions
-
checkPinned(
Uint8List bytes, {required int size, required String sha256, required String what}) → void -
Checks that
bytes, downloaded aswhat, are thesizebytes whose SHA-256 issha256. Throws DownloadException when they are not: changed after they were pinned, or broken on the way. -
isFetchAllowed(
Uri uri) → bool -
Whether
urimay be fetched:https:, or plainhttp:to this machine only.
Typedefs
- ByteProgress = void Function(int received, int? total)
-
Reports bytes received so far, out of
totalwhen known.
Exceptions / Errors
- CancelledException
- Work stopped because a CancelToken was cancelled.
- DownloadException
- A download whose bytes are not what was pinned.
- FetchException
- Something that could not be fetched: the server refused, the network failed, or the download broke a limit.
- UnsafePathException
- A path in an archive or a package that is not safe to write.
- ZipFormatException
- A ZIP archive that is damaged, uses something this reader refuses, or breaks one of its ZipLimits.
- ZipNeedsBytes
- Thrown by ZipDirectory when the bytes it was given end before what it must read: bytes start to end (end not included) of the archive are needed too. Fetch them and try again.