X-Git-Url: http://git.annexia.org/?a=blobdiff_plain;f=goaljobs.ml;h=b5981ad78a6988c9f7c701ac411512ea39ee5f4a;hb=90eaa31d3acea2c640d662323142776a3eab517c;hp=1aa8c0d9fd9562ba52aa04a3ea7d540c15248ccf;hpb=7967057ada774e1b953de38a8d73f6009ce602ef;p=goaljobs.git

diff --git a/goaljobs.ml b/goaljobs.ml
index 1aa8c0d..b5981ad 100644
--- a/goaljobs.ml
+++ b/goaljobs.ml
@@ -16,6 +16,8 @@
  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
  *)
 
+open CalendarLib
+
 open Unix
 open Printf
 
@@ -35,8 +37,122 @@ let target_all vs = target (List.fold_left (&&) true vs)
 let target_exists vs = target (List.fold_left (||) false vs)
 let require () = ()
 
+type period_t = Seconds | Days | Months | Years
+let seconds = (1, Seconds)
+let sec = seconds and secs = seconds and second = seconds
+let minutes = (60, Seconds)
+let min = minutes and mins = minutes and minute = minutes
+let hours = (3600, Seconds)
+let hour = hours
+let days = (1, Days)
+let day = days
+let weeks = (7, Days)
+let week = weeks
+let months = (1, Months)
+let month = months
+let years = (1, Years)
+let year = years
+
+let periodic_jobs = ref []
+
+(* Register a periodic job. *)
+let every ?name i (j, t) f =
+  let period = i*j, t in (* 5 minutes -> ((5 * 60), Seconds) *)
+  periodic_jobs := (period, (name, f)) :: !periodic_jobs
+
+(* [next_time t period] returns the earliest event of [period]
+   strictly after time [t].
+
+   Visualising periods as repeated events on a timeline, this
+   returns [t']:
+
+   {v
+   events:  ---+---------+---------+---------+---------+---------+-----
+   times:          t     t'
+   }
+
+   Note that [period_t] events are not necessarily regular.
+   eg. The start of a month is not a fixed number of seconds
+   after the start of the previous month.  'Epoch' refers
+   to the Unix Epoch (ie. 1970-01-01 00:00:00 UTC).
+
+   If [period = i, Seconds i] then events are when
+   [t' mod i == 0] when t' is the number of seconds since
+   the Epoch.  This returns the next t' > t.
+
+   If [period = i, Days] then events happen at
+   midnight UTC every [i] days since the Epoch.
+   This returns the next midnight > t.
+
+   If [period = i, Months] then events happen at
+   midnight UTC on the 1st day of the month every [i] months
+   since the Epoch.  This returns midnight on the
+   1st day of the next month > t.
+
+   If [period = i, Years] then events happen at
+   midnight UTC on the 1st day of the year when
+   [(y - 1970) mod i == 0].  This returns midnight on the
+   1st day of the next year > t. *)
+
+let next_time =
+  (* Round up 'a' to the next multiple of 'i'. *)
+  let round_up_float a i =
+    let r = mod_float a i in
+    if r = 0. then a +. i else a +. (i -. r)
+  and round_up a i =
+    let r = a mod i in
+    if r = 0 then a + i else a + (i - r)
+  in
+
+  fun t -> function
+  | (i, Seconds) ->
+    let i = float_of_int i in
+    round_up_float t i
+
+  | (i, Years) ->
+    let tm = gmtime t in
+
+    (* Round 'tm' up to the first day of the next year. *)
+    let year = round_up tm.tm_year i in
+    let tm = { tm with tm_sec = 0; tm_min = 0; tm_hour = 0;
+                       tm_mday = 1; tm_mon = 0; tm_year = year } in
+    fst (mktime tm)
+
+  | (i, Days) ->
+    let t = Date.from_unixfloat t in
+    let t0 = Date.make 1970 1 1 in
+
+    (* Number of whole days since Unix Epoch. *)
+    let nb_days = Date.Period.safe_nb_days (Date.sub t t0) in
+
+    let nb_days = round_up nb_days i in
+    let t' = Date.add t0 (Date.Period.day nb_days) in
+    Date.to_unixfloat t'
+
+  | (i, Months) ->
+    (* Calculate number of whole months since Unix Epoch. *)
+    let tm = gmtime t in
+    let months = 12 * (tm.tm_year - 70) + tm.tm_mon in
+
+    let months = round_up months i in
+    let t0 = Date.make 1970 1 1 in
+    let t' = Date.add t0 (Date.Period.month months) in
+    Date.to_unixfloat t'
+
 let file_exists = Sys.file_exists
 
+let directory_exists path =
+  let s =
+    try Some (stat path)
+    with
+    | Unix_error (ENOENT, _, _) -> None
+    | Unix_error (err, _, _) ->
+      let msg = sprintf "directory_exists: %s: %s" path (error_message err) in
+      goal_failed msg in
+  match s with
+  | Some s -> s.st_kind = S_DIR
+  | None -> false
+
 let file_newer_than f1 f2 =
   let stat f =
     try Some (stat f)
@@ -103,11 +219,47 @@ let url_contains_string url str =
   unlink tmp;
   r
 
+(* Create a temporary directory.  It is *not* deleted on exit. *)
+let tmpdir () =
+  let chan = open_in "/dev/urandom" in
+  let data = String.create 16 in
+  really_input chan data 0 (String.length data);
+  close_in chan;
+  let data = Digest.to_hex (Digest.string data) in
+  let dir = Filename.temp_dir_name // sprintf "goaljobstmp%s" data in
+  mkdir dir 0o700;
+  dir
+
+(* Recursively remove directory. *)
+let rm_rf dir =
+  let cmd = sprintf "rm -rf %s" (quote dir) in
+  ignore (Sys.command cmd)
+
+let shell = ref "/bin/sh"
+
+(* Used by sh, shout etc.  Create a temporary directory and a
+ * 'script.sh' inside it containing the script to run.  Returns the
+ * temporary directory and command to run.
+ *)
+let create_script script =
+  let dir = tmpdir () in
+  let script_file = dir // "script.sh" in
+  let chan = open_out script_file in
+  fprintf chan "#!%s\n" !shell;
+  fprintf chan "set -e\n"; (* so that job exits on error *)
+  fprintf chan "set -x\n"; (* echo commands (must be last) *)
+  fprintf chan "\n";
+  output_string chan script;
+  close_out chan;
+  chmod script_file 0o700;
+  let cmd = sprintf "cd %s && exec %s" (quote dir) (quote script_file) in
+  dir, cmd
 
 let sh fs =
-  let do_sh cmd =
-    let cmd = "set -e\nset -x\n\n" ^ cmd in
+  let do_sh script =
+    let dir, cmd = create_script script in
     let r = Sys.command cmd in
+    rm_rf dir;
     if r <> 0 then (
       let msg = sprintf "external command failed with code %d" r in
       goal_failed msg
@@ -115,8 +267,8 @@ let sh fs =
   in
   ksprintf do_sh fs
 
-let do_shlines cmd =
-  let cmd = "set -e\nset -x\n\n" ^ cmd in
+let do_shlines script =
+  let dir, cmd = create_script script in
   let chan = open_process_in cmd in
   let lines = ref [] in
   let rec loop () =
@@ -126,6 +278,7 @@ let do_shlines cmd =
   in
   (try loop () with End_of_file -> ());
   let r = close_process_in chan in
+  rm_rf dir;
   match r with
   | WEXITED 0 -> List.rev !lines
   | WEXITED i ->
@@ -139,16 +292,12 @@ let do_shlines cmd =
     goal_failed msg
 let shlines fs = ksprintf do_shlines fs
 
-let do_shout cmd =
-  let lines = do_shlines cmd in
+let do_shout script =
+  let lines = do_shlines script in
   String.concat "\n" lines
 let shout fs = ksprintf do_shout fs
 
 (*
-val shell : string ref
-*)
-
-(*
 val replace_substring : string -> string -> string -> string
 *)
 
@@ -162,13 +311,75 @@ let change_file_extension ext filename =
 val filter_file_extension : string -> string list -> string
 *)
 
-(* XXX The Memory is not actually persistent yet. *)
-let memory = Hashtbl.create 13
+(* Persistent memory is stored in $HOME/.goaljobs-memory.  We have to
+ * lock this file each time we read or write because multiple concurrent
+ * jobs may access it at the same time.
+ *
+ * XXX Replace this with a more efficient and less fragile implementation.
+ *)
+
+type ('a, 'b) alternative = Either of 'a | Or of 'b
+let with_memory_locked ?(write = false) f =
+  let filename = getenv "HOME" // ".goaljobs-memory" in
+  let fd = openfile filename [O_RDWR; O_CREAT] 0o644 in
+  lockf fd (if write then F_LOCK else F_RLOCK) 0;
+  (* If the file is newly created with zero size, write an
+   * empty hash table.
+   *)
+  if (fstat fd).st_size = 0 then (
+    let empty : (string, string) Hashtbl.t = Hashtbl.create 13 in
+    let chan = out_channel_of_descr fd in
+    output_value chan empty;
+    Pervasives.flush chan;
+    ignore (lseek fd 0 SEEK_SET)
+  );
+
+  (* Run the function. *)
+  let r = try Either (f fd) with exn -> Or (exn) in
+  lockf fd F_ULOCK 0;
+  match r with
+  | Either x -> x
+  | Or exn -> raise exn
+
+let memory_exists key =
+  with_memory_locked (
+    fun fd ->
+      let chan = in_channel_of_descr fd in
+      let memory : (string, string) Hashtbl.t = input_value chan in
+      Hashtbl.mem memory key
+  )
+
+let memory_get key =
+  with_memory_locked (
+    fun fd ->
+      let chan = in_channel_of_descr fd in
+      let memory : (string, string) Hashtbl.t = input_value chan in
+      try Some (Hashtbl.find memory key) with Not_found -> None
+  )
+
+let memory_set key value =
+  with_memory_locked ~write:true (
+    fun fd ->
+      let chan = in_channel_of_descr fd in
+      let memory : (string, string) Hashtbl.t = input_value chan in
+      Hashtbl.replace memory key value;
+      let chan = out_channel_of_descr fd in
+      seek_out chan 0;
+      output_value chan memory;
+      Pervasives.flush chan;
+  )
 
-let memory_exists = Hashtbl.mem memory
-let memory_set = Hashtbl.replace memory
-let memory_get k = try Some (Hashtbl.find memory k) with Not_found -> None
-let memory_delete = Hashtbl.remove memory
+let memory_delete key =
+  with_memory_locked ~write:true (
+    fun fd ->
+      let chan = in_channel_of_descr fd in
+      let memory : (string, string) Hashtbl.t = input_value chan in
+      Hashtbl.remove memory key;
+      let chan = out_channel_of_descr fd in
+      seek_out chan 0;
+      output_value chan memory;
+      Pervasives.flush chan;
+  )
 
 let published_goals = ref []
 let publish name fn = published_goals := (name, fn) :: !published_goals
@@ -180,6 +391,11 @@ let goal_file_exists filename =
     let msg = sprintf "file '%s' required but not found" filename in
     goal_failed msg
   )
+let goal_directory_exists path =
+  if not (directory_exists path) then (
+    let msg = sprintf "directory '%s' required but not found" path in
+    goal_failed msg
+  )
 let goal_file_newer_than f1 f2 =
   if not (file_newer_than f1 f2) then (
     let msg = sprintf "file %s is required to be newer than %s" f1 f2 in
@@ -212,11 +428,24 @@ let goal_memory_exists k =
     goal_failed msg
   )
 
+let guard fn arg =
+  try fn arg; true
+  with
+  | Goal_result (Goal_failed msg) ->
+    prerr_endline ("error: " ^ msg);
+    false
+  | exn ->
+    prerr_endline (Printexc.to_string exn);
+    false
+
 (* Run the program. *)
-let init () =
+let rec init () =
   let prog = Sys.executable_name in
   let prog = Filename.basename prog in
 
+  (* Save the current working directory when the program started. *)
+  putenv "builddir" (getcwd ());
+
   let args = ref [] in
 
   let display_version () =
@@ -228,7 +457,8 @@ let init () =
     let names = !published_goals in
     let names = List.map fst names in
     let names = List.sort compare names in
-    List.iter print_endline names
+    List.iter print_endline names;
+    exit 0
   in
 
   let argspec = Arg.align [
@@ -254,21 +484,65 @@ Options:
   let args = List.rev !args in
 
   (* Was a goal named on the command line? *)
-  match args with
+  (match args with
   | name :: args ->
     (match get_goal name with
-    | Some fn -> fn args
+    | Some fn ->
+      exit (if guard fn args then 0 else 1)
     | None ->
       eprintf "error: no goal called '%s' was found.\n" name;
       eprintf "Use %s -l to list all published goals in this script.\n" name;
       exit 1
     )
   | [] ->
-    (* Does a published 'all' goal exist? *)
-    match get_goal "all" with
-    | Some fn -> fn []
-    | None ->
-      (* No published 'all' goal.  This is only a warning, because
-       * other top-level code may exist in the script.
-       *)
-      eprintf "warning: no 'all' goal found.\n"
+    (* If periodic jobs exist, fall through. *)
+    if !periodic_jobs = [] then (
+      (* Does a published 'all' goal exist? *)
+      match get_goal "all" with
+      | Some fn ->
+        exit (if guard fn [] then 0 else 1)
+      | None ->
+        (* No published 'all' goal. *)
+        eprintf "error: no goal called 'all' was found.\n";
+        exit 1
+    )
+  );
+
+  assert (!periodic_jobs <> []);
+
+  (* Run the periodic jobs.  Note these run forever, or until killed. *)
+  while true do
+    (* Find the next job to run. *)
+    let now = time () in
+    let jobs = List.map (
+      fun (period, (_, _ as name_f)) ->
+        next_time now period, name_f
+    ) !periodic_jobs in
+    let jobs = List.sort (fun (t1,_) (t2,_) -> compare t1 t2) jobs in
+
+    (* Find all jobs that have the same next time.
+     * XXX When we can handle parallel jobs we can do better here,
+     * but until them run all the ones which have the same time
+     * in series.
+     *)
+    let next_t = int_of_float (fst (List.hd jobs)) in
+    let jobs = List.filter (fun (t, _) -> int_of_float t = next_t) jobs in
+
+    (* Run next job(s) after waiting for the appropriate amount of time. *)
+    let seconds = next_t - int_of_float now in
+    eprintf "next job will run in %s\n%!" (printable_seconds seconds);
+    sleep seconds;
+
+    List.iter (
+      fun (_, (name, f)) ->
+        eprintf "running job: %s\n%!"
+          (match name with Some name -> name | None -> "[unnamed]");
+        ignore (guard f ())
+    ) jobs
+  done
+
+and printable_seconds s =
+  if s < 60 then sprintf "%d seconds" s
+  else if s < 6000 then sprintf "%d minutes, %d seconds" (s/60) (s mod 60)
+  else if s < 86400 then sprintf "%d hours, %d minutes" (s/3600) (s/60)
+  else sprintf "about %d days" (s/86400)