Revert "Bug 1853082 - Don't rely on file extension to guess archive type in fetch-content. r=taskgraph-reviewers,bhearsum" (#543)

This reverts commit 02f110d.

That change is not backwards-compatible, since it causes files that
would previously not be extracted to now be detected as an archive (e.g.
python wheel as a zip archive).  Reverting for now, we can re-apply it
with a major version bump later.

Author: jcristau

src/taskgraph/run-task/fetch-content

@@ -10,7 +10,6 @@ import contextlib
 import datetime
 import gzip
 import hashlib
-import io
 import json
 import lzma
 import multiprocessing
@@ -333,41 +332,21 @@ def gpg_verify_path(path: pathlib.Path, public_key_data: bytes, signature_data:
             subprocess.run(["gpgconf", "--kill", "gpg-agent"], env=env)
 
 
-class ArchiveTypeNotSupported(Exception):
-    def __init__(self, path: pathlib.Path):
-        super(Exception, self).__init__("Archive type not supported for %s" % path)
-
-
-def open_stream(path: pathlib.Path):
-    """Attempt to identify a path as an extractable archive by looking at its
-    content."""
-    fh = path.open(mode="rb")
-    magic = fh.read(6)
-    fh.seek(0)
-    if magic[:2] == b"PK":
-        return "zip", fh
-    if magic[:2] == b"\x1f\x8b":
-        fh = gzip.GzipFile(fileobj=fh)
-    elif magic[:3] == b"BZh":
-        fh = bz2.BZ2File(fh)
-    elif magic == b"\xfd7zXZ\x00":
-        fh = lzma.LZMAFile(fh)
-    elif magic[:4] == b"\x28\xb5\x2f\xfd":
-        fh = ZstdDecompressor().stream_reader(fh)
-    fh = io.BufferedReader(fh)
-    try:
-        # A full tar info header is 512 bytes.
-        headers = fh.peek(512)
-        # 257 is the offset of the ustar magic.
-        magic = headers[257 : 257 + 8]
-        # For older unix tar, rely on TarInfo.frombuf's checksum check
-        if magic in (b"ustar\x0000", b"ustar  \x00") or tarfile.TarInfo.frombuf(
-            headers[:512], tarfile.ENCODING, "surrogateescape"
-        ):
-            return "tar", fh
-    except Exception as e:
-        pass
-    raise ArchiveTypeNotSupported(path)
+def open_tar_stream(path: pathlib.Path):
+    """"""
+    if path.suffix == ".bz2":
+        return bz2.open(str(path), "rb")
+    elif path.suffix in (".gz", ".tgz") :
+        return gzip.open(str(path), "rb")
+    elif path.suffix == ".xz":
+        return lzma.open(str(path), "rb")
+    elif path.suffix == ".zst":
+        dctx = ZstdDecompressor()
+        return dctx.stream_reader(path.open("rb"))
+    elif path.suffix == ".tar":
+        return path.open("rb")
+    else:
+        raise ValueError("unknown archive format for tar file: %s" % path)
 
 
 def archive_type(path: pathlib.Path):
@@ -380,7 +359,7 @@ def archive_type(path: pathlib.Path):
         return None
 
 
-def extract_archive(path, dest_dir):
+def extract_archive(path, dest_dir, typ):
     """Extract an archive to a destination directory."""
 
     # Resolve paths to absolute variants.
@@ -392,8 +371,8 @@ def extract_archive(path, dest_dir):
 
     # We pipe input to the decompressor program so that we can apply
     # custom decompressors that the program may not know about.
-    typ, ifh = open_stream(path)
     if typ == "tar":
+        ifh = open_tar_stream(path)
         # On Windows, the tar program doesn't support things like symbolic
         # links, while Windows actually support them. The tarfile module in
         # python does. So use that. But since it's significantly slower than
@@ -440,8 +419,10 @@ def repack_archive(
 ):
     assert orig != dest
     log("Repacking as %s" % dest)
-    orig_typ, ifh = open_stream(orig)
+    orig_typ = archive_type(orig)
     typ = archive_type(dest)
+    if not orig_typ:
+        raise Exception("Archive type not supported for %s" % orig.name)
     if not typ:
         raise Exception("Archive type not supported for %s" % dest.name)
 
@@ -467,7 +448,7 @@ def repack_archive(
         ctx = ZstdCompressor()
         if orig_typ == "zip":
             assert typ == "tar"
-            zip = zipfile.ZipFile(ifh)
+            zip = zipfile.ZipFile(orig)
             # Convert the zip stream to a tar on the fly.
             with ctx.stream_writer(fh) as compressor, tarfile.open(
                 fileobj=compressor, mode="w:"
@@ -509,6 +490,7 @@ def repack_archive(
                 raise Exception("Repacking a tar to zip is not supported")
             assert typ == "tar"
 
+            ifh = open_tar_stream(orig)
             if filter:
                 # To apply the filter, we need to open the tar stream and
                 # tweak it.
@@ -551,12 +533,11 @@ def fetch_and_extract(url, dest_dir, extract=True, sha256=None, size=None):
     if not extract:
         return
 
-    try:
-        extract_archive(dest_path, dest_dir)
+    typ = archive_type(dest_path)
+    if typ:
+        extract_archive(dest_path, dest_dir, typ)
         log("Removing %s" % dest_path)
         dest_path.unlink()
-    except ArchiveTypeNotSupported:
-        pass
 
 
 def fetch_urls(downloads):