Re #6372 Depend on tar-0.6.2.0

tar-0.6.2.0 supports Unicode in file names (encoded as UTF-8). Consequently, Stack no longer needs to UTF-8 encode file names. Also adds test of `stack sdist .` with Unicode in file names.
commercialhaskell · Apr 5, 2024 · d08d799 · d08d799
1 parent 7cba4de
commit d08d799
Show file tree

Hide file tree

Showing 14 changed files with 107 additions and 197 deletions.
diff --git a/package.yaml b/package.yaml
@@ -114,7 +114,7 @@ dependencies:
 - rio-prettyprint >= 0.1.8.0
 - split
 - stm
-- tar >= 0.5.1.1 && < 0.6.0.0
+- tar >= 0.6.2.0
 - template-haskell
 - text
 - time
@@ -157,7 +157,6 @@ library:
   - Build_stack
   - Paths_stack
   exposed-modules:
-  - Codec.Archive.Tar.Utf8
   - Control.Concurrent.Execute
   - Data.Attoparsec.Args
   - Data.Attoparsec.Combinators

diff --git a/src/Codec/Archive/Tar/Utf8.hs b/src/Codec/Archive/Tar/Utf8.hs
diff --git a/src/Stack/SDist.hs b/src/Stack/SDist.hs
@@ -15,15 +15,14 @@ module Stack.SDist
   , readLocalPackage
   ) where
 
-import qualified Codec.Archive.Tar.Utf8 as Tar
+import qualified Codec.Archive.Tar as Tar
 import qualified Codec.Archive.Tar.Entry as Tar
 import qualified Codec.Compression.GZip as GZip
 import           Conduit ( runConduitRes, sourceLazy, sinkFileCautious )
 import           Control.Concurrent.Execute
                    ( ActionContext (..), Concurrency (..) )
 import           Control.Monad.Extra ( whenJust )
 import qualified Data.ByteString as S
-import qualified Data.ByteString.Char8 as S8
 import qualified Data.ByteString.Lazy as L
 import           Data.Char ( toLower )
 import           Data.Data ( cast )
@@ -252,13 +251,9 @@ getSDistTarball mpvpBounds pkgDir = do
   -- prone and more predictable to read everything in at once, so that's what
   -- we're doing for now:
   let tarPath isDir fp =
-        case Tar.toTarPath isDir (forceUtf8Enc (pkgIdName FP.</> fp)) of
+        case Tar.toTarPath isDir (pkgIdName FP.</> fp) of
           Left e -> prettyThrowIO $ ToTarPathException e
           Right tp -> pure tp
-      -- convert a String of proper characters to a String of bytes in UTF8
-      -- encoding masquerading as characters. This is necessary for tricking the
-      -- tar package into proper character encoding.
-      forceUtf8Enc = S8.unpack . T.encodeUtf8 . T.pack
       packWith f isDir fp = liftIO $ f (pkgFp FP.</> fp) =<< tarPath isDir fp
       packDir = packWith Tar.packDirectoryEntry True
       packFile fp

diff --git a/stack-ghc-9.8.2.yaml b/stack-ghc-9.8.2.yaml
@@ -3,6 +3,13 @@
 
 snapshot: nightly-2024-04-04 # GHC 9.8.2
 
+extra-deps:
+# Dependency of tar >= 0.6.2.0:
+- os-string-2.0.2@sha256:32fa47f8345a2c0662fb602fc42e4b674e41ec48079b68bdecb4b6f68032c24e,3259
+# nightly-2024-04-04 provides tar-0.5.1.1, which does not support Unicode
+# filenames:
+- tar-0.6.2.0@sha256:619828cae098a7b6deeb0316e12f55011101d88f756787ed024ceedb81cf1eba,4576
+
 docker:
   enable: false
   repo: quay.io/benz0li/ghc-musl:9.8.1

diff --git a/stack-ghc-9.8.2.yaml.lock b/stack-ghc-9.8.2.yaml.lock
@@ -3,7 +3,21 @@
 # For more information, please see the documentation at:
 #   https://docs.haskellstack.org/en/stable/lock_files
 
-packages: []
+packages:
+- completed:
+    hackage: os-string-2.0.2@sha256:32fa47f8345a2c0662fb602fc42e4b674e41ec48079b68bdecb4b6f68032c24e,3259
+    pantry-tree:
+      sha256: 5b1dbaf1663005907462032997264b7942befb9fc108491c1fdd5e488ac0817e
+      size: 2217
+  original:
+    hackage: os-string-2.0.2@sha256:32fa47f8345a2c0662fb602fc42e4b674e41ec48079b68bdecb4b6f68032c24e,3259
+- completed:
+    hackage: tar-0.6.2.0@sha256:619828cae098a7b6deeb0316e12f55011101d88f756787ed024ceedb81cf1eba,4576
+    pantry-tree:
+      sha256: 1dde50961e9d1a6e6f820d918c0edbbd9673c83afdd300e0aae40a691e8151df
+      size: 2168
+  original:
+    hackage: tar-0.6.2.0@sha256:619828cae098a7b6deeb0316e12f55011101d88f756787ed024ceedb81cf1eba,4576
 snapshots:
 - completed:
     sha256: 31177dd01e2b7801b9347776d892d1e808396c21ab3b60f7f8c96856bdbe5f52

diff --git a/stack.cabal b/stack.cabal
@@ -152,7 +152,6 @@ flag supported-build
 
 library
   exposed-modules:
-      Codec.Archive.Tar.Utf8
       Control.Concurrent.Execute
       Data.Attoparsec.Args
       Data.Attoparsec.Combinators
@@ -409,7 +408,7 @@ library
     , rio-prettyprint >=0.1.8.0
     , split
     , stm
-    , tar >=0.5.1.1 && <0.6.0.0
+    , tar >=0.6.2.0
     , template-haskell
     , text
     , time
@@ -532,7 +531,7 @@ executable stack
     , split
     , stack
     , stm
-    , tar >=0.5.1.1 && <0.6.0.0
+    , tar >=0.6.2.0
     , template-haskell
     , text
     , time
@@ -634,7 +633,7 @@ executable stack-integration-test
     , rio-prettyprint >=0.1.8.0
     , split
     , stm
-    , tar >=0.5.1.1 && <0.6.0.0
+    , tar >=0.6.2.0
     , template-haskell
     , text
     , time
@@ -752,7 +751,7 @@ test-suite stack-unit-test
     , split
     , stack
     , stm
-    , tar >=0.5.1.1 && <0.6.0.0
+    , tar >=0.6.2.0
     , template-haskell
     , text
     , time

diff --git a/stack.yaml b/stack.yaml
@@ -7,9 +7,13 @@ extra-deps:
 # Cabal is pruned because process is a GHC boot package, and has to be specified
 # again.
 - Cabal-3.10.1.0@sha256:6d11adf7847d9734e7b02785ff831b5a0d11536bfbcefd6634b2b08411c63c94,12316
+# Dependency of tar >= 0.6.2.0:
+- os-string-2.0.2@sha256:32fa47f8345a2c0662fb602fc42e4b674e41ec48079b68bdecb4b6f68032c24e,3259
 - pantry-0.10.0@sha256:6f99ee8d7cfeeb0e2513638618acf80c72d018e7f10120048fa120a409b9dcd2,7864
 # GHC 9.6.4 comes with process-1.6.17.0, which can segfault on macOS.
 - process-1.6.18.0@sha256:cd0a3e0376b5a8525983d3131a31e52f9ffefc278ce635eec45a9d3987b8be3e,3025
+# lts-22.7 provides tar-0.5.1.1, which does not support Unicode filenames:
+- tar-0.6.2.0@sha256:619828cae098a7b6deeb0316e12f55011101d88f756787ed024ceedb81cf1eba,4576
 
 docker:
   enable: false

diff --git a/stack.yaml.lock b/stack.yaml.lock
@@ -11,6 +11,13 @@ packages:
       size: 9223
   original:
     hackage: Cabal-3.10.1.0@sha256:6d11adf7847d9734e7b02785ff831b5a0d11536bfbcefd6634b2b08411c63c94,12316
+- completed:
+    hackage: os-string-2.0.2@sha256:32fa47f8345a2c0662fb602fc42e4b674e41ec48079b68bdecb4b6f68032c24e,3259
+    pantry-tree:
+      sha256: 5b1dbaf1663005907462032997264b7942befb9fc108491c1fdd5e488ac0817e
+      size: 2217
+  original:
+    hackage: os-string-2.0.2@sha256:32fa47f8345a2c0662fb602fc42e4b674e41ec48079b68bdecb4b6f68032c24e,3259
 - completed:
     hackage: pantry-0.10.0@sha256:6f99ee8d7cfeeb0e2513638618acf80c72d018e7f10120048fa120a409b9dcd2,7864
     pantry-tree:
@@ -25,6 +32,13 @@ packages:
       size: 1675
   original:
     hackage: process-1.6.18.0@sha256:cd0a3e0376b5a8525983d3131a31e52f9ffefc278ce635eec45a9d3987b8be3e,3025
+- completed:
+    hackage: tar-0.6.2.0@sha256:619828cae098a7b6deeb0316e12f55011101d88f756787ed024ceedb81cf1eba,4576
+    pantry-tree:
+      sha256: 1dde50961e9d1a6e6f820d918c0edbbd9673c83afdd300e0aae40a691e8151df
+      size: 2168
+  original:
+    hackage: tar-0.6.2.0@sha256:619828cae098a7b6deeb0316e12f55011101d88f756787ed024ceedb81cf1eba,4576
 snapshots:
 - completed:
     sha256: 7b975b104cb3dbf0c297dfd01f936a4d2ee523241dd0b1ae960522b833fe3027

diff --git a/tests/integration/tests/6372-sdist-unicode-test/.gitignore b/tests/integration/tests/6372-sdist-unicode-test/.gitignore
@@ -0,0 +1 @@
+stack.yaml.lock
diff --git a/tests/integration/tests/6372-sdist-unicode-test/Main.hs b/tests/integration/tests/6372-sdist-unicode-test/Main.hs
@@ -0,0 +1,20 @@
+import StackTest
+
+-- | The test fails at runtime on the Windows Server 2022 GitHub-hosted runner
+-- only, at the point of outputting a Unicode character, with:
+--
+-- <stderr>: commitAndReleaseBuffer: invalid argument (cannot encode character '\1633')
+--
+-- That appears to be similar to
+-- https://gitlab.haskell.org/ghc/ghc/-/issues/8118, however: (1) the locale is
+-- set to C.UTF-8 and the active code page is 65001; and
+-- (2) `GHC.IO.Encoding.setLocaleEncoding utf8` has no effect.
+--
+-- Until the origin of the problem is identified, we disable the test on
+-- Windows.
+
+main :: IO ()
+main = unless isWindows $ do
+  stack ["clean"]
+  stack ["build", "--dry-run"]
+  stack ["sdist", "."]
diff --git a/tests/integration/tests/6372-sdist-unicode-test/files/package.yaml b/tests/integration/tests/6372-sdist-unicode-test/files/package.yaml
@@ -0,0 +1,8 @@
+name: test١٢٣
+version: 0.1.0.0
+license: BSD-3-Clause
+description: Test of Unicode characters in file names
+dependencies:
+- base < 5
+library:
+  source-dirs: src
diff --git a/tests/integration/tests/6372-sdist-unicode-test/files/src/Libαβγ.hs b/tests/integration/tests/6372-sdist-unicode-test/files/src/Libαβγ.hs
@@ -0,0 +1,6 @@
+module Libαβγ
+    ( someFunc
+    ) where
+
+someFunc :: IO ()
+someFunc = putStrLn "someFunc"
diff --git a/tests/integration/tests/6372-sdist-unicode-test/files/stack.yaml b/tests/integration/tests/6372-sdist-unicode-test/files/stack.yaml
@@ -0,0 +1 @@
+snapshot: lts-22.7