From a395464aa6c0b9ecf030e1a9d07306d84faa19a8 Mon Sep 17 00:00:00 2001
From: "Daniel A. Wozniak" <daniel.wozniak@broadcom.com>
Date: Fri, 20 Sep 2024 14:50:22 -0700
Subject: [PATCH] Use tar instead of tarfile to create toolchain tarball

---
 .github/workflows/package-action.yml |  2 +-
 CHANGELOG.md                         |  5 ++++
 src/ppbt/__init__.py                 |  4 +--
 src/ppbt/build.py                    | 44 ++++++++++++----------------
 src/ppbt/common.py                   |  9 ++++--
 5 files changed, 33 insertions(+), 31 deletions(-)

diff --git a/.github/workflows/package-action.yml b/.github/workflows/package-action.yml
index 21d68e6..b3652a9 100644
--- a/.github/workflows/package-action.yml
+++ b/.github/workflows/package-action.yml
@@ -1,4 +1,4 @@
-name: Relenv Python Package
+name: Build ppbt python package
 
 on:
   workflow_call:
diff --git a/CHANGELOG.md b/CHANGELOG.md
index b9e47fb..4592e1b 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -1,3 +1,8 @@
+0.1.7
+=====
+- Use tar instead of tarfile to create toolchain tarballs
+
+
 0.1.6
 =====
 
diff --git a/src/ppbt/__init__.py b/src/ppbt/__init__.py
index ec6daea..e134309 100644
--- a/src/ppbt/__init__.py
+++ b/src/ppbt/__init__.py
@@ -3,8 +3,6 @@
 """
 Portable Python Build Toolchains.
 """
-from .common import environ, extract
-
-__version__ = "0.1.6"
+from .common import __version__, environ, extract
 
 ALL = ("environ", "extract")
diff --git a/src/ppbt/build.py b/src/ppbt/build.py
index 339b45a..abfe83f 100644
--- a/src/ppbt/build.py
+++ b/src/ppbt/build.py
@@ -24,7 +24,6 @@
 CT_NG_VER = "1.26.0"
 CT_URL = "http://crosstool-ng.org/download/crosstool-ng/crosstool-ng-{version}.tar.bz2"
 CT_GIT_REPO = "https://github.com/crosstool-ng/crosstool-ng.git"
-TC_URL = "https://{hostname}/relenv/{version}/toolchain/{host}/{triplet}.tar.xz"
 CICD = "CI" in os.environ
 PATCHELF_VERSION = "0.18.0"
 PATCHELF_SOURCE = (
@@ -308,37 +307,32 @@ def build_ppbt(branch=None, use_tempdir=True):
             subprocess.run(["make"])
 
         shutil.copy(source / "src" / "patchelf", build / triplet / "bin" / "patchelf")
-
         os.chdir(build)
         archive = f"{ triplet }.tar.xz"
         record = f"{ triplet }.tar.xz.record"
-        print(f"Archive is {archive}")
+        cmd = ["tar", "-C", f"{ build }", "-cJf", f"{ archive }", f"{ triplet }"]
+        subprocess.run(cmd)
+
+        # XXX: Figure out why things break when using the tarfile module.
         with open(record, "w") as rfp:
             rwriter = csv.writer(rfp)
-            with tarfile.open(archive, mode="w:xz", dereference=True) as fp:
-                for root, _dirs, files in os.walk(archdir):
-                    relroot = pathlib.Path(root).relative_to(build)
-                    for f in files:
-                        print(f"Archive {relroot} / {f}")
-                        relpath = relroot / f
-                        with open(relpath, "rb") as dfp:
-                            data = dfp.read()
-                            hsh = (
-                                base64.urlsafe_b64encode(hashlib.sha256(data).digest())
-                                .rstrip(b"=")
-                                .decode()
-                            )
-                            hashpath = str(
-                                pathlib.Path("ppbt") / "_toolchain" / relpath
-                            )
-                            rwriter.writerow([hashpath, f"sha256={hsh}", len(data)])
-                        try:
-                            fp.add(relpath, relpath, recursive=False)
-                        except FileNotFoundError:
-                            print(f"File not found while archiving: {relpath}")
+            for root, _dirs, files in os.walk(archdir):
+                relroot = pathlib.Path(root).relative_to(build)
+                for f in files:
+                    print(f"Archive {relroot} / {f}")
+                    relpath = relroot / f
+                    with open(relpath, "rb") as fp:
+                        data = fp.read()
+                        hsh = (
+                            base64.urlsafe_b64encode(hashlib.sha256(data).digest())
+                            .rstrip(b"=")
+                            .decode()
+                        )
+                        hashpath = str(pathlib.Path("ppbt") / "_toolchain" / relpath)
+                        rwriter.writerow([hashpath, f"sha256={hsh}", len(data)])
         print(f"Copying {archive} to {toolchain}")
-        print(f"Copying {record} to {toolchain}")
         shutil.copy(archive, toolchain)
+        print(f"Copying {record} to {toolchain}")
         shutil.copy(record, toolchain)
     finally:
         # if archdir and archdir.exists():
diff --git a/src/ppbt/common.py b/src/ppbt/common.py
index a544ba6..b377ef2 100644
--- a/src/ppbt/common.py
+++ b/src/ppbt/common.py
@@ -11,14 +11,18 @@
 
 from .build import build_arch, extract_archive, get_triplet
 
+__version__ = "0.1.7"
+
 triplet = get_triplet(build_arch())
 
 archive = pathlib.Path(__file__).parent / "_toolchain" / f"{triplet}.tar.xz"
 toolchain = pathlib.Path(__file__).parent / "_toolchain" / triplet
 toolchain_root = pathlib.Path(__file__).parent / "_toolchain"
 
-
-distinfo = pathlib.Path(__file__).resolve().parent.parent / "ppbt-0.1.0.dist-info"
+# This is not reliable, the version can be modified by setuptools at build time.
+distinfo = (
+    pathlib.Path(__file__).resolve().parent.parent / f"ppbt-{__version__}.dist-info"
+)
 
 log = logging.getLogger(__name__)
 
@@ -35,6 +39,7 @@ def extract(overwrite=False):
         record = distinfo / "RECORD"
         if record.exists():
             records = []
+            log.info("Update pkg metadata")
             with open(record, "r") as fp:
                 for row in csv.reader(fp):
                     records.append(row)