11:57 a.m.
On Monday, 31 July 2017 17:40:57 CEST Richard W.M. Jones wrote:
+(* Parse a os-release file.
+ *
+ * Only few fields are parsed, falling back to the usual detection if we
+ * cannot read all of them.
+ *
+ * For the format of os-release, see also:
+ * http://www.freedesktop.org/software/systemd/man/os-release.html
+ *)
+let rec parse_os_release release_file data =
+ let chroot = Chroot.create ~name:"parse_os_release" () in
+ let lines =
+ Chroot.f chroot (
+ fun () ->
+ if not (is_small_file release_file) then (
+ eprintf "%s: not a regular file or too large\n" release_file;
+ None
+ )
+ else
+ Some (read_whole_file release_file)
+ ) () in
I see this bit (chroot + read_whole_file) is common among various
functions in this file -- could you please place it in an helper
function?
+ match lines with
+ | None -> false
+ | Some lines ->
+ let lines = String.nsplit "\n" lines in
+
+ List.iter (
+ fun line ->
+ let line = String.trim line in
+ if line = "" || line.[0] = '#' then
+ ()
+ else (
+ let key, value = String.split "=" line in
+ let value =
+ let n = String.length value in
+ if n >= 2 && value.[0] = '"' && value.[n-1]
= '"' then
+ String.sub value 1 (n-2)
+ else
+ value in
+ if key = "ID" then (
+ let distro = distro_of_os_release_id value in
+ match distro with
+ | Some _ as distro -> data.distro <- distro
+ | None -> ()
+ )
+ else if key = "PRETTY_NAME" then
+ data.product_name <- Some value
+ else if key = "VERSION_ID" then
+ parse_version_from_major_minor value data
+ )
+ ) lines;
+
+ (* If we haven't got all the fields, exit right away. *)
+ if data.distro = None || data.product_name = None then
+ false
+ else (
+ (* os-release in Debian and CentOS does not provide the full
+ * version number (VERSION_ID), just the major part of it. If
+ * we detect that situation then bail out and use the release
+ * files instead.
+ *)
+ match data with
+ | { distro = Some (DISTRO_DEBIAN|DISTRO_CENTOS);
+ version = Some (_, 0) } ->
+ false
+ | _ -> true
+ )
parse_os_release should also reset the version to Some (0, 0) (instead
of None) in case the distro is one of the "rolling" ones (only Void at
the moment) -- see also src/inspect-fs-unix.c, lines 240 and on.
+(* Ubuntu has /etc/lsb-release containing:
+ * DISTRIB_ID=Ubuntu # Distro
+ * DISTRIB_RELEASE=10.04 # Version
+ * DISTRIB_CODENAME=lucid
+ * DISTRIB_DESCRIPTION="Ubuntu 10.04.1 LTS" # Product name
+ *
+ * [Ubuntu-derived ...] Linux Mint was found to have this:
+ * DISTRIB_ID=LinuxMint
+ * DISTRIB_RELEASE=10
+ * DISTRIB_CODENAME=julia
+ * DISTRIB_DESCRIPTION="Linux Mint 10 Julia"
+ * Linux Mint also has /etc/linuxmint/info with more information,
+ * but we can use the LSB file.
+ *
+ * Mandriva has:
+ * LSB_VERSION=lsb-4.0-amd64:lsb-4.0-noarch
+ * DISTRIB_ID=MandrivaLinux
+ * DISTRIB_RELEASE=2010.1
+ * DISTRIB_CODENAME=Henry_Farman
+ * DISTRIB_DESCRIPTION="Mandriva Linux 2010.1"
+ * Mandriva also has a normal release file called /etc/mandriva-release.
+ *
+ * CoreOS has a /etc/lsb-release link to /usr/share/coreos/lsb-release containing:
+ * DISTRIB_ID=CoreOS
+ * DISTRIB_RELEASE=647.0.0
+ * DISTRIB_CODENAME="Red Dog"
+ * DISTRIB_DESCRIPTION="CoreOS 647.0.0"
+ *)
+and parse_lsb_release release_file data =
TBH, since both parse_os_release and parse_lsb_release are rewritten
in OCaml, I'd try a better approach for them: add an helper function
that read such kind of files (ignoring empty lines, and those starting
with '#'), split them in lines, and map the lines into (key, value)
pairs, also un-quoting the value. It would not map 1:1 the C
implementations, but in OCaml it would be so much easier than in C
(and that is basically what I'd have liked to do in the first place
when implementing parse_os_release, but it'd have been so much memory
waste).
+let rec check_linux_root mountable data =
+ let os_type = OS_TYPE_LINUX in
+ data.os_type <- Some os_type;
+
+ let rec loop = function
+ | (release_file, parse_fun) :: tests ->
+ if verbose () then
+ eprintf "check_linux_root: checking %s\n%!" release_file;
+ if Is.is_file ~followsymlinks:true release_file then (
+ if parse_fun release_file data then () (* true => finished *)
+ else loop tests
+ ) else loop tests
+ | [] -> ()
+ in
+ loop linux_root_tests;
Would it be possible to move the above loop in an own helper function,
so the check_$OS functions of other OSes can have their own counterpart
of linux_root_tests? It is true that we don't add new identifications
often, but since it's already implemented here...
+and check_fstab_entry md_map root_mountable os_type aug entry =
+ if verbose () then
+ eprintf "check_fstab_entry: augeas path: %s\n%!" entry;
+
+ let is_bsd =
+ os_type = OS_TYPE_FREEBSD ||
+ os_type = OS_TYPE_NETBSD ||
+ os_type = OS_TYPE_OPENBSD in
'match' here?
--
Pino Toscano