tmog-bin: follow tmog.org's versioned downloads and update to 1.0.0

TMOG 1.0.0 moved the site under /rtm/ and publishes each Linux tarball
under a versioned name with a .sha256 sidecar. The versionless
/downloads/TMOG-Task-Manager-Linux-x86_64.tar.gz path the hook fetched
now returns 404, so every upstream sync since 1.0.0 has failed.

Point source=() at the versioned tarball and have the hook read the
version from /rtm/version.txt and the checksum from the sidecar, which
drops the 8 MB download and the directory-name check the mutable URL
needed.
This commit is contained in:
Ryan Hughes committed 2026-09-26 19:57:09 -04:00
1 parent 3a70d3279c
commit 0c91711a9e
3 files changed
+44 -63

No files matched your search

+18 -30
View File
@@ -1,19 +1,12 @@
#!/bin/bash
# TMOG publishes no manifest for its Linux builds -- release.json describes the
# macOS DMG only -- so the version comes from /version.txt and the checksum has
# to be computed from the artifact itself. That is 8 MB, and only when the
# version has actually moved, so the six-hourly check normally costs one tiny
# request.
#
# The download path carries no version, which makes it worth proving that what
# arrived is what was announced: the tarball's top-level directory is named for
# the release, and a mismatch means the object served is not the one
# /version.txt describes. Reporting no update leaves the checked-in package
# alone and lets the next run try again, which is the right answer whether the
# cause is a half-published release or a stale CDN object.
# macOS DMG only -- so the version comes from /rtm/version.txt. Each Linux
# artifact is published under a versioned name with a `<artifact>.sha256`
# sidecar beside it, and that sidecar is the checksum reported here, so the
# six-hourly check costs two tiny requests and never the tarball itself.
set -euo pipefail
BASE_URL="https://tmog.org"
BASE_URL="https://tmog.org/rtm"
current=$(grep -m1 '^pkgver=' PKGBUILD | cut -d= -f2- | tr -d "\"'")
@@ -28,27 +21,22 @@ if [[ $version == "$current" ]]; then
exit 0
fi
tarball=$(mktemp)
trap 'rm -f "$tarball"' EXIT
curl -fsSL -o "$tarball" \
"$BASE_URL/downloads/TMOG-Task-Manager-Linux-x86_64.tar.gz?v=${version}-free"
# Every entry is listed rather than just the first: `head -1` would close the
# pipe under `tar` and take the whole hook down with SIGPIPE, and reading them
# all also catches a tarball that unpacks more than one top-level directory.
expected_dir="TaskManagerOG-${version}-linux-x86_64"
served_dir=$(tar tzf "$tarball" | cut -d/ -f1 | sort -u)
if [[ $served_dir != "$expected_dir" ]]; then
echo "Download holds $served_dir, but /version.txt announced $version; skipping" >&2
echo '{}'
exit 0
# The sidecar names the file it describes; insisting on that name catches a
# sidecar left over from another release or architecture. A failed download or
# a file without a trailing newline leaves `read` short, which the check below
# reports rather than letting set -e exit silently.
artifact="TaskManagerOG-${version}-linux-x86_64.tar.gz"
sha256="" name=""
read -r sha256 name < <(curl -fsSL "$BASE_URL/downloads/$artifact.sha256") || true
if [[ ! $sha256 =~ ^[0-9a-f]{64}$ || ${name#\*} != "$artifact" ]]; then
echo "Unusable checksum for $artifact: '$sha256 $name'" >&2
exit 1
fi
# "any" is bin/sync-upstream's name for the unsuffixed sha256sums array, which
# is the one this package has: upstream publishes x86_64 alone, so there is a
# single plain source=() rather than per-architecture arrays.
# is the one this package has: it builds x86_64 alone, so there is a single
# plain source=() rather than per-architecture arrays.
jq -n \
--arg pkgver "$version" \
--arg sha256 "$(sha256sum "$tarball" | cut -d' ' -f1)" \
--arg sha256 "$sha256" \
'{pkgver: $pkgver, sha256sums: {any: [$sha256]}}'