logstats: add arm64 detection support

Parse the new ARM64 flag to detect if we are running on arm64 hardware.

Since all that wow64/xta/prism stuff gets confusing fast just derive the
cygwin and host arch from the UA instead and display that.
Imo easier to understand for people not familiar with the internals.

Author: lazka

msys2-logstats

@@ -33,17 +33,15 @@ class UserAgent:
     pacman_version: Tuple[int, int, int]
     windows_version: Tuple[int, int]
     build_number: int
-    is_wow64: bool
-    cpu_arch: str
+    cygwin_arch: str
+    host_arch: str
     libalpm_version: Tuple[int, int, int]
 
 
 @dataclass
 class ClientInfo:
     pacman_version: str
     windows_edition: str
-    cpu_arch: str
-    is_wow64: str
     user_agent: UserAgent
     ci: str
 
@@ -56,21 +54,28 @@ def parse_user_agent(ua: str) -> UserAgent:
     assert is_valid_user_agent(ua)
     m = re.match(r"pacman/([^\s]+) \(([^-]+)-([^-]+)-?([0-9]+|)-?(.+|) ([^)]+)\) libalpm/(.+)", ua)
     assert m is not None
-    pacman_version, _, windows_version, build_number, wow64, arch, alpm_version = m.groups()
+    pacman_version, _, windows_version, build_number, extra, arch, alpm_version = m.groups()
 
     pacman_version = tuple(map(int, pacman_version.split("-")[0].split(".")))
     windows_version = tuple(map(int, windows_version.split(".")))
     build_number = int(build_number or "-1")
-    is_wow64 = bool(wow64)
-    cpu_arch = arch
+    cygwin_arch = arch
+    if extra in ("WOW64", "WOW") and cygwin_arch == "i686":
+        host_arch = "x86_64"
+    elif extra == "ARM64":
+        host_arch = "aarch64"
+    elif extra:
+        host_arch = "???"
+    else:
+        host_arch = cygwin_arch
     alpm_version = tuple(map(int, alpm_version.split(".")))
 
     return UserAgent(
         pacman_version,
         windows_version,
         build_number,
-        is_wow64,
-        cpu_arch,
+        cygwin_arch,
+        host_arch,
         alpm_version
     )
 
@@ -79,22 +84,31 @@ def test_parse_user_agent():
     ua = "pacman/6.0.1 (MSYS_NT-10.0-19042 x86_64) libalpm/13.0.1"
     assert parse_user_agent(ua).pacman_version == (6, 0, 1)
     ua = "pacman/6.0.1 (MSYS_NT-10.0-19042-WOW64 i686) libalpm/13.0.1"
-    assert parse_user_agent(ua).is_wow64
+    assert parse_user_agent(ua).cygwin_arch == "i686"
+    assert parse_user_agent(ua).host_arch == "x86_64"
     assert parse_user_agent(ua).build_number == 19042
     ua = "pacman/5.2.1 (MSYS_NT-6.1-7601-WOW64 i686) libalpm/12.0.1"
-    assert parse_user_agent(ua).is_wow64
+    assert parse_user_agent(ua).cygwin_arch == "i686"
+    assert parse_user_agent(ua).host_arch == "x86_64"
     assert parse_user_agent(ua).build_number == 7601
     ua = "pacman/5.1.0 (MINGW64_NT-10.0 x86_64) libalpm/11.0.0"
     assert parse_user_agent(ua).build_number == -1
-    assert parse_user_agent(ua).cpu_arch == "x86_64"
+    assert parse_user_agent(ua).cygwin_arch == "x86_64"
+    assert parse_user_agent(ua).host_arch == "x86_64"
     ua = "pacman/5.0.1 (UCRT64_NT-10.0-WOW i686) libalpm/10.0.1"
     assert parse_user_agent(ua).build_number == -1
-    assert parse_user_agent(ua).is_wow64
-    assert parse_user_agent(ua).cpu_arch == "i686"
+    assert parse_user_agent(ua).host_arch == "x86_64"
+    assert parse_user_agent(ua).cygwin_arch == "i686"
     ua = "pacman/4.2.1-313-g5535-dirty (MINGW64_NT-6.1 x86_64) libalpm/9.0.1"
     assert parse_user_agent(ua).pacman_version == (4, 2, 1)
     ua = "pacman/6.0.1 (MSYS_NT-6.0-6002 x86_64) libalpm/13.0.1"
     assert parse_user_agent(ua).windows_version == (6, 0)
+    ua = "pacman/6.1.0 (MSYS_NT-10.0-22631-ARM64 x86_64) libalpm/14.0.0"
+    assert parse_user_agent(ua).host_arch == "aarch64"
+    assert parse_user_agent(ua).cygwin_arch == "x86_64"
+    ua = "pacman/6.1.0 (MSYS_NT-10.0-22631-ARM64 i686) libalpm/14.0.0"
+    assert parse_user_agent(ua).host_arch == "aarch64"
+    assert parse_user_agent(ua).cygwin_arch == "i686"
 
 
 def test_get_windows_edition():
@@ -282,13 +296,13 @@ def print_pacman(clients, show_ci):
 
 def print_system_arch(clients, show_ci):
     table = []
-    for (cpu_arch, is_wow64, ci), count in Counter([(u.cpu_arch, u.is_wow64, u.ci) for u in clients]).most_common():
+    for (cygwin_arch, host_arch, ci), count in Counter([(u.user_agent.cygwin_arch, u.user_agent.host_arch, u.ci) for u in clients]).most_common():
         pcnt = count / len(clients) * 100
-        line = [cpu_arch, is_wow64, ci, f"{pcnt:.2f}%", f"{count}"]
+        line = [cygwin_arch, host_arch, ci, f"{pcnt:.2f}%", f"{count}"]
         if not show_ci:
             line.pop(2)
         table.append(line)
-    headers = ["Arch", "WOW64", "CI", "% Clients", "Clients"]
+    headers = ["Cygwin Arch", "Host Arch", "CI", "% Clients", "Clients"]
     if not show_ci:
         headers.pop(2)
     print()
@@ -392,8 +406,6 @@ def main(argv):
         client_info = ClientInfo(
             ".".join(map(str, user_agent.pacman_version)),
             get_windows_edition(user_agent),
-            user_agent.cpu_arch,
-            str(user_agent.is_wow64),
             user_agent,
             ci
         )