repo-buildinfo: add option to filter based on the DLL imports

For example:

	--imports-dll libstdc++-6.dll --built-with-package "mingw-w64-ucrt-x86_64-gcc<16.1.0"

to get all packages linking against an older libstdc++

msys2-repo-buildinfo

@@ -4,11 +4,13 @@ import argparse
 import fnmatch
 import os
 import sys
+from contextlib import contextmanager, closing
 
 from msys2_devtools.exttarfile import tarfile
 from msys2_devtools.utils import vercmp
 from fastprogress.fastprogress import progress_bar
 from concurrent.futures import ThreadPoolExecutor
+import pefile
 
 
 def find_dbs(target_dir: str) -> list[str]:
@@ -53,20 +55,57 @@ def parse_buildinfo(buildinfo: str) -> dict[str, list[str]]:
     return res
 
 
-def get_buildinfo(path: str) -> dict:
+@contextmanager
+def pefile_set_max_import_symbols(max_symbols: int):
+    old = pefile.MAX_IMPORT_SYMBOLS
+    pefile.MAX_IMPORT_SYMBOLS = max_symbols
+    try:
+        yield
+    finally:
+        pefile.MAX_IMPORT_SYMBOLS = old
+
+
+# python uses .pyd for extensions, ruby uses .so, and octave uses .oct
+PE_FILE_EXTENSIONS = (".dll", ".exe", ".pyd", ".so", ".oct")
+
+
+def get_imported_dlls(data: bytes, name: str) -> set[str]:
+    try:
+        with pefile_set_max_import_symbols(0x10000), \
+            closing(pefile.PE(data=data, fast_load=True)) as pe:
+            pe.parse_data_directories(directories=[
+                pefile.DIRECTORY_ENTRY['IMAGE_DIRECTORY_ENTRY_IMPORT']
+            ])
+
+            if len(pe.get_warnings()) > 0:
+                print(f"Warnings for PE ({name}):")
+                pe.show_warnings()
+
+            if not hasattr(pe, 'DIRECTORY_ENTRY_IMPORT'):
+                return set()
+            return {entry.dll.decode().lower() for entry in pe.DIRECTORY_ENTRY_IMPORT}
+    except pefile.PEFormatError:
+        print(f"PEFormatError while parsing PE file ({name}), skipping imports")
+        return set()
+
+
+def get_buildinfo(path: str, needs_files: bool, needs_imports: bool) -> tuple[dict, set[str], set[str]]:
     buildinfo = None
     files = set()
+    imports = set()
     with tarfile.TarFile.open(path, mode="r") as tar:
         for info in tar.getmembers():
             if info.name == ".BUILDINFO":
                 buildinfo = tar.extractfile(info).read().decode()
-            if info.isfile() and not info.name.startswith("."):
+            if needs_files and info.isfile() and not info.name.startswith("."):
                 files.add(info.name)
+            if needs_imports and info.isfile() and info.name.lower().endswith(PE_FILE_EXTENSIONS):
+                data = tar.extractfile(info).read()
+                imports.update(get_imported_dlls(data, info.name))
 
     if buildinfo is None:
         raise RuntimeError(f"Cannot find .BUILDINFO in {path}")
-
-    return parse_buildinfo(buildinfo), files
+    return parse_buildinfo(buildinfo), files, imports
 
 
 def parse_vercmp(comparison_expression) -> tuple[str, str | None, str | None]:
@@ -126,21 +165,28 @@ def main(argv):
                              "(optionally with a version constraint, e.g. 'foo>=1.0')")
     parser.add_argument("--contains-file",
                         help="filter packages that contain files matching this glob pattern (e.g. '*.a')")
+    parser.add_argument("--imports-dll",
+                        action="append",
+                        default=[],
+                        metavar="DLL",
+                        help="filter packages that import this DLL (e.g. 'libfoo.dll'); "
+                             "can be given multiple times")
 
     args = parser.parse_args(argv[1:])
 
     installed_filter = args.built_with_package
     file_pattern = args.contains_file
+    dll_filters = args.imports_dll
 
     found = set()
     paths = get_package_paths(args.root)
 
     def process_path(p):
-        buildinfo, files = get_buildinfo(p)
-        return buildinfo, files
+        buildinfo, files, imports = get_buildinfo(p, needs_files=file_pattern is not None, needs_imports=bool(dll_filters))
+        return buildinfo, files, imports
 
     with ThreadPoolExecutor(4) as executor:
-        for buildinfo, files in progress_bar(executor.map(process_path, paths), total=len(paths), leave=False):
+        for buildinfo, files, imports in progress_bar(executor.map(process_path, paths), total=len(paths), leave=False):
             if installed_filter is not None:
                 for installed_full_name in buildinfo["installed"]:
                     if package_matches(installed_full_name, installed_filter):
@@ -155,6 +201,10 @@ def main(argv):
                 else:
                     continue
 
+            if dll_filters:
+                if not any(dll in imports for dll in dll_filters):
+                    continue
+
             found.update(buildinfo["pkgbase"])
 
     for f in sorted(found):

pyproject.toml

@@ -34,6 +34,7 @@ all = [
     "pgpdump>=1.5,<2",
     "fastprogress>=1.0.3,<1.1",
     "requests-cache>=1.2.1,<2",
+    "pefile>=2024.8.26",
 ]
 
 [project.scripts]

uv.lock

@@ -3,7 +3,7 @@ revision = 3
 requires-python = ">=3.11.0, <4.0"
 
 [options]
-exclude-newer = "2026-04-24T11:04:34.845645838Z"
+exclude-newer = "2026-04-24T18:16:55.102883823Z"
 exclude-newer-span = "P7D"
 
 [[package]]
@@ -316,6 +316,7 @@ all = [
     { name = "fastprogress" },
     { name = "netaddr" },
     { name = "packageurl-python" },
+    { name = "pefile" },
     { name = "pgpdump" },
     { name = "requests-cache" },
 ]
@@ -354,6 +355,7 @@ requires-dist = [
     { name = "packageurl-python", marker = "extra == 'all'", specifier = ">=0.17.0,<0.18" },
     { name = "packageurl-python", marker = "extra == 'pypi-cache'", specifier = ">=0.17.0,<0.18" },
     { name = "packageurl-python", marker = "extra == 'sbom'", specifier = ">=0.17.0,<0.18" },
+    { name = "pefile", marker = "extra == 'all'", specifier = ">=2024.8.26" },
     { name = "pgpdump", marker = "extra == 'all'", specifier = ">=1.5,<2" },
     { name = "pgpdump", marker = "extra == 'sigstats'", specifier = ">=1.5,<2" },
     { name = "pydantic", specifier = ">=2.0,<3" },
@@ -398,6 +400,15 @@ wheels = [
     { url = "https://files.pythonhosted.org/packages/7a/c2/920ef838e2f0028c8262f16101ec09ebd5969864e5a64c4c05fad0617c56/packaging-26.1-py3-none-any.whl", hash = "sha256:5d9c0669c6285e491e0ced2eee587eaf67b670d94a19e94e3984a481aba6802f", size = 95831, upload-time = "2026-04-14T21:12:47.56Z" },
 ]
 
+[[package]]
+name = "pefile"
+version = "2024.8.26"
+source = { registry = "https://pypi.org/simple" }
+sdist = { url = "https://files.pythonhosted.org/packages/03/4f/2750f7f6f025a1507cd3b7218691671eecfd0bbebebe8b39aa0fe1d360b8/pefile-2024.8.26.tar.gz", hash = "sha256:3ff6c5d8b43e8c37bb6e6dd5085658d658a7a0bdcd20b6a07b1fcfc1c4e9d632", size = 76008, upload-time = "2024-08-26T20:58:38.155Z" }
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/54/16/12b82f791c7f50ddec566873d5bdd245baa1491bac11d15ffb98aecc8f8b/pefile-2024.8.26-py3-none-any.whl", hash = "sha256:76f8b485dcd3b1bb8166f1128d395fa3d87af26360c2358fb75b80019b957c6f", size = 74766, upload-time = "2024-08-26T21:01:02.632Z" },
+]
+
 [[package]]
 name = "pgpdump"
 version = "1.5"