src/Pure/General/file.ML
author wenzelm
Fri, 24 Jun 2022 23:31:28 +0200
changeset 75615 4494cd69f97f
parent 75613 1b50bcd108b7
child 75616 986506233812
permissions -rw-r--r--
clarified modules;
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
6118
caa439435666 fixed titles;
wenzelm
parents: 5009
diff changeset
     1
(*  Title:      Pure/General/file.ML
64698
e022a69db531 tuned comments;
wenzelm
parents: 64304
diff changeset
     2
    Author:     Makarius
5009
05b152a41922 moved Thy/file.ML to General/file.ML;
wenzelm
parents:
diff changeset
     3
64698
e022a69db531 tuned comments;
wenzelm
parents: 64304
diff changeset
     4
File-system operations.
5009
05b152a41922 moved Thy/file.ML to General/file.ML;
wenzelm
parents:
diff changeset
     5
*)
05b152a41922 moved Thy/file.ML to General/file.ML;
wenzelm
parents:
diff changeset
     6
05b152a41922 moved Thy/file.ML to General/file.ML;
wenzelm
parents:
diff changeset
     7
signature FILE =
05b152a41922 moved Thy/file.ML to General/file.ML;
wenzelm
parents:
diff changeset
     8
sig
60970
e08d868ceca9 clarified File.standard_path vs. File.platform_path (like Isabelle/Scala operations);
wenzelm
parents: 59058
diff changeset
     9
  val standard_path: Path.T -> string
16261
28803c418b59 removed sysify_path, quote_sysity_path etc.;
wenzelm
parents: 16002
diff changeset
    10
  val platform_path: Path.T -> string
62549
9498623b27f0 File.bash_string operations in ML as in Scala -- exclusively for GNU bash, not perl and not user output;
wenzelm
parents: 62468
diff changeset
    11
  val bash_path: Path.T -> string
69223
44d68a00917c more operations;
wenzelm
parents: 64698
diff changeset
    12
  val bash_paths: Path.T list -> string
72278
199dc903131b clarified signature;
wenzelm
parents: 70998
diff changeset
    13
  val bash_platform_path: Path.T -> string
73314
87403fde8cc3 more direct make_directory in ML and Scala, but ssh still requires perl for Windows UNC paths (see a5dbad753552);
wenzelm
parents: 73264
diff changeset
    14
  val absolute_path: Path.T -> Path.T
42003
6e45dc518ebb replaced File.check by specific File.check_file, File.check_dir;
wenzelm
parents: 41944
diff changeset
    15
  val full_path: Path.T -> Path.T -> Path.T
6182
4a07dfe3583f use Path.T;
wenzelm
parents: 6118
diff changeset
    16
  val tmp_path: Path.T -> Path.T
17826
afa2696eacce added assert;
wenzelm
parents: 16713
diff changeset
    17
  val exists: Path.T -> bool
afa2696eacce added assert;
wenzelm
parents: 16713
diff changeset
    18
  val rm: Path.T -> unit
40785
c755df0f7062 more permissive Isabelle_System.mkdir;
wenzelm
parents: 40746
diff changeset
    19
  val is_dir: Path.T -> bool
69300
8b6ab9989bcd is_file/is_dir/read_dir: more uniform treatment of errors and boundary cases, notably for symlinks in ssh;
wenzelm
parents: 69223
diff changeset
    20
  val is_file: Path.T -> bool
42003
6e45dc518ebb replaced File.check by specific File.check_file, File.check_dir;
wenzelm
parents: 41944
diff changeset
    21
  val check_dir: Path.T -> Path.T
6e45dc518ebb replaced File.check by specific File.check_file, File.check_dir;
wenzelm
parents: 41944
diff changeset
    22
  val check_file: Path.T -> Path.T
43848
8f2bf02a0ccb some file and directory operations;
wenzelm
parents: 43845
diff changeset
    23
  val fold_dir: (string -> 'a -> 'a) -> Path.T -> 'a -> 'a
8f2bf02a0ccb some file and directory operations;
wenzelm
parents: 43845
diff changeset
    24
  val read_dir: Path.T -> string list
28028
c0f54a32491e renamed Buffer.write to File.write_buffer;
wenzelm
parents: 26980
diff changeset
    25
  val fold_lines: (string -> 'a -> 'a) -> Path.T -> 'a -> 'a
43848
8f2bf02a0ccb some file and directory operations;
wenzelm
parents: 43845
diff changeset
    26
  val read_lines: Path.T -> string list
6182
4a07dfe3583f use Path.T;
wenzelm
parents: 6118
diff changeset
    27
  val read: Path.T -> string
4a07dfe3583f use Path.T;
wenzelm
parents: 6118
diff changeset
    28
  val write: Path.T -> string -> unit
4a07dfe3583f use Path.T;
wenzelm
parents: 6118
diff changeset
    29
  val append: Path.T -> string -> unit
16713
be5763901788 added write_list, append_list;
wenzelm
parents: 16603
diff changeset
    30
  val write_list: Path.T -> string list -> unit
be5763901788 added write_list, append_list;
wenzelm
parents: 16603
diff changeset
    31
  val append_list: Path.T -> string list -> unit
16603
1776d276f848 added eq;
wenzelm
parents: 16304
diff changeset
    32
  val eq: Path.T * Path.T -> bool
5009
05b152a41922 moved Thy/file.ML to General/file.ML;
wenzelm
parents:
diff changeset
    33
end;
05b152a41922 moved Thy/file.ML to General/file.ML;
wenzelm
parents:
diff changeset
    34
05b152a41922 moved Thy/file.ML to General/file.ML;
wenzelm
parents:
diff changeset
    35
structure File: FILE =
05b152a41922 moved Thy/file.ML to General/file.ML;
wenzelm
parents:
diff changeset
    36
struct
05b152a41922 moved Thy/file.ML to General/file.ML;
wenzelm
parents:
diff changeset
    37
26503
4dec4460244f discontinued unused hooks explode_platform_path_fn, platform_path_fn, shell_path_fn;
wenzelm
parents: 26220
diff changeset
    38
(* system path representations *)
6224
0c08846be6f3 fail_safe close;
wenzelm
parents: 6218
diff changeset
    39
60970
e08d868ceca9 clarified File.standard_path vs. File.platform_path (like Isabelle/Scala operations);
wenzelm
parents: 59058
diff changeset
    40
val standard_path = Path.implode o Path.expand;
62468
d97e13e5ea5b clarified modules;
wenzelm
parents: 60982
diff changeset
    41
val platform_path = ML_System.platform_path o standard_path;
26980
f7f48bb9a025 ident: naive caching prevents potentially slow external invocations;
wenzelm
parents: 26946
diff changeset
    42
73264
440546ea20e6 clarified modules;
wenzelm
parents: 72511
diff changeset
    43
val bash_path = Bash.string o standard_path;
440546ea20e6 clarified modules;
wenzelm
parents: 72511
diff changeset
    44
val bash_paths = Bash.strings o map standard_path;
6224
0c08846be6f3 fail_safe close;
wenzelm
parents: 6218
diff changeset
    45
73264
440546ea20e6 clarified modules;
wenzelm
parents: 72511
diff changeset
    46
val bash_platform_path = Bash.string o platform_path;
72278
199dc903131b clarified signature;
wenzelm
parents: 70998
diff changeset
    47
70292
bc9d02f916c4 tuned whitespace;
wenzelm
parents: 69483
diff changeset
    48
42003
6e45dc518ebb replaced File.check by specific File.check_file, File.check_dir;
wenzelm
parents: 41944
diff changeset
    49
(* full_path *)
6e45dc518ebb replaced File.check by specific File.check_file, File.check_dir;
wenzelm
parents: 41944
diff changeset
    50
73314
87403fde8cc3 more direct make_directory in ML and Scala, but ssh still requires perl for Windows UNC paths (see a5dbad753552);
wenzelm
parents: 73264
diff changeset
    51
val absolute_path =
87403fde8cc3 more direct make_directory in ML and Scala, but ssh still requires perl for Windows UNC paths (see a5dbad753552);
wenzelm
parents: 73264
diff changeset
    52
  Path.expand #> (fn path =>
87403fde8cc3 more direct make_directory in ML and Scala, but ssh still requires perl for Windows UNC paths (see a5dbad753552);
wenzelm
parents: 73264
diff changeset
    53
    if Path.is_absolute path then path
87403fde8cc3 more direct make_directory in ML and Scala, but ssh still requires perl for Windows UNC paths (see a5dbad753552);
wenzelm
parents: 73264
diff changeset
    54
    else Path.explode (ML_System.standard_path (OS.FileSys.getDir ())) + path);
87403fde8cc3 more direct make_directory in ML and Scala, but ssh still requires perl for Windows UNC paths (see a5dbad753552);
wenzelm
parents: 73264
diff changeset
    55
42003
6e45dc518ebb replaced File.check by specific File.check_file, File.check_dir;
wenzelm
parents: 41944
diff changeset
    56
fun full_path dir path =
6e45dc518ebb replaced File.check by specific File.check_file, File.check_dir;
wenzelm
parents: 41944
diff changeset
    57
  let
6e45dc518ebb replaced File.check by specific File.check_file, File.check_dir;
wenzelm
parents: 41944
diff changeset
    58
    val path' = Path.expand path;
6e45dc518ebb replaced File.check by specific File.check_file, File.check_dir;
wenzelm
parents: 41944
diff changeset
    59
    val _ = Path.is_current path' andalso error "Bad file specification";
73314
87403fde8cc3 more direct make_directory in ML and Scala, but ssh still requires perl for Windows UNC paths (see a5dbad753552);
wenzelm
parents: 73264
diff changeset
    60
  in absolute_path (dir + path') end;
5009
05b152a41922 moved Thy/file.ML to General/file.ML;
wenzelm
parents:
diff changeset
    61
6182
4a07dfe3583f use Path.T;
wenzelm
parents: 6118
diff changeset
    62
4a07dfe3583f use Path.T;
wenzelm
parents: 6118
diff changeset
    63
(* tmp_path *)
4a07dfe3583f use Path.T;
wenzelm
parents: 6118
diff changeset
    64
72511
460d743010bc clarified signature: overloaded "+" for Path.append;
wenzelm
parents: 72278
diff changeset
    65
fun tmp_path path = Path.variable "ISABELLE_TMP" + Path.base path;
5009
05b152a41922 moved Thy/file.ML to General/file.ML;
wenzelm
parents:
diff changeset
    66
05b152a41922 moved Thy/file.ML to General/file.ML;
wenzelm
parents:
diff changeset
    67
26980
f7f48bb9a025 ident: naive caching prevents potentially slow external invocations;
wenzelm
parents: 26946
diff changeset
    68
(* directory entries *)
f7f48bb9a025 ident: naive caching prevents potentially slow external invocations;
wenzelm
parents: 26946
diff changeset
    69
23861
72bb3494746f replaced info by ident (for full identification, potentially content-based);
wenzelm
parents: 23826
diff changeset
    70
val exists = can OS.FileSys.modTime o platform_path;
16261
28803c418b59 removed sysify_path, quote_sysity_path etc.;
wenzelm
parents: 16002
diff changeset
    71
28803c418b59 removed sysify_path, quote_sysity_path etc.;
wenzelm
parents: 16002
diff changeset
    72
val rm = OS.FileSys.remove o platform_path;
28803c418b59 removed sysify_path, quote_sysity_path etc.;
wenzelm
parents: 16002
diff changeset
    73
69300
8b6ab9989bcd is_file/is_dir/read_dir: more uniform treatment of errors and boundary cases, notably for symlinks in ssh;
wenzelm
parents: 69223
diff changeset
    74
fun test_dir path = the_default false (try OS.FileSys.isDir (platform_path path));
8b6ab9989bcd is_file/is_dir/read_dir: more uniform treatment of errors and boundary cases, notably for symlinks in ssh;
wenzelm
parents: 69223
diff changeset
    75
fun is_dir path = exists path andalso test_dir path;
8b6ab9989bcd is_file/is_dir/read_dir: more uniform treatment of errors and boundary cases, notably for symlinks in ssh;
wenzelm
parents: 69223
diff changeset
    76
fun is_file path = exists path andalso not (test_dir path);
40785
c755df0f7062 more permissive Isabelle_System.mkdir;
wenzelm
parents: 40746
diff changeset
    77
42003
6e45dc518ebb replaced File.check by specific File.check_file, File.check_dir;
wenzelm
parents: 41944
diff changeset
    78
fun check_dir path =
69300
8b6ab9989bcd is_file/is_dir/read_dir: more uniform treatment of errors and boundary cases, notably for symlinks in ssh;
wenzelm
parents: 69223
diff changeset
    79
  if is_dir path then path
63668
5efaa884ac6c tuned error;
wenzelm
parents: 62854
diff changeset
    80
  else error ("No such directory: " ^ Path.print (Path.expand path));
42003
6e45dc518ebb replaced File.check by specific File.check_file, File.check_dir;
wenzelm
parents: 41944
diff changeset
    81
6e45dc518ebb replaced File.check by specific File.check_file, File.check_dir;
wenzelm
parents: 41944
diff changeset
    82
fun check_file path =
69300
8b6ab9989bcd is_file/is_dir/read_dir: more uniform treatment of errors and boundary cases, notably for symlinks in ssh;
wenzelm
parents: 69223
diff changeset
    83
  if is_file path then path
63668
5efaa884ac6c tuned error;
wenzelm
parents: 62854
diff changeset
    84
  else error ("No such file: " ^ Path.print (Path.expand path));
42003
6e45dc518ebb replaced File.check by specific File.check_file, File.check_dir;
wenzelm
parents: 41944
diff changeset
    85
16261
28803c418b59 removed sysify_path, quote_sysity_path etc.;
wenzelm
parents: 16002
diff changeset
    86
43848
8f2bf02a0ccb some file and directory operations;
wenzelm
parents: 43845
diff changeset
    87
(* directory content *)
8f2bf02a0ccb some file and directory operations;
wenzelm
parents: 43845
diff changeset
    88
69300
8b6ab9989bcd is_file/is_dir/read_dir: more uniform treatment of errors and boundary cases, notably for symlinks in ssh;
wenzelm
parents: 69223
diff changeset
    89
fun fold_dir f path a =
75615
4494cd69f97f clarified modules;
wenzelm
parents: 75613
diff changeset
    90
  check_dir path |> File_Stream.open_dir (fn stream =>
69300
8b6ab9989bcd is_file/is_dir/read_dir: more uniform treatment of errors and boundary cases, notably for symlinks in ssh;
wenzelm
parents: 69223
diff changeset
    91
    let
8b6ab9989bcd is_file/is_dir/read_dir: more uniform treatment of errors and boundary cases, notably for symlinks in ssh;
wenzelm
parents: 69223
diff changeset
    92
      fun read x =
8b6ab9989bcd is_file/is_dir/read_dir: more uniform treatment of errors and boundary cases, notably for symlinks in ssh;
wenzelm
parents: 69223
diff changeset
    93
        (case OS.FileSys.readDir stream of
8b6ab9989bcd is_file/is_dir/read_dir: more uniform treatment of errors and boundary cases, notably for symlinks in ssh;
wenzelm
parents: 69223
diff changeset
    94
          NONE => x
8b6ab9989bcd is_file/is_dir/read_dir: more uniform treatment of errors and boundary cases, notably for symlinks in ssh;
wenzelm
parents: 69223
diff changeset
    95
        | SOME entry => read (f entry x));
8b6ab9989bcd is_file/is_dir/read_dir: more uniform treatment of errors and boundary cases, notably for symlinks in ssh;
wenzelm
parents: 69223
diff changeset
    96
    in read a end);
43848
8f2bf02a0ccb some file and directory operations;
wenzelm
parents: 43845
diff changeset
    97
69427
ff2f39a221d4 clarified operations: uniform sorting of results;
wenzelm
parents: 69300
diff changeset
    98
fun read_dir path = sort_strings (fold_dir cons path []);
43848
8f2bf02a0ccb some file and directory operations;
wenzelm
parents: 43845
diff changeset
    99
8f2bf02a0ccb some file and directory operations;
wenzelm
parents: 43845
diff changeset
   100
44879
3b6613366dd7 more orthogonal signature;
wenzelm
parents: 43848
diff changeset
   101
(*
3b6613366dd7 more orthogonal signature;
wenzelm
parents: 43848
diff changeset
   102
  scalable iterator:
3b6613366dd7 more orthogonal signature;
wenzelm
parents: 43848
diff changeset
   103
  . avoid size limit of TextIO.inputAll and overhead of many TextIO.inputLine
48911
5debc3e4fa81 tuned messages: end-of-input rarely means physical end-of-file from the past;
wenzelm
parents: 48446
diff changeset
   104
  . optional terminator at end-of-input
44879
3b6613366dd7 more orthogonal signature;
wenzelm
parents: 43848
diff changeset
   105
*)
75615
4494cd69f97f clarified modules;
wenzelm
parents: 75613
diff changeset
   106
fun fold_lines f path a = File_Stream.open_input (fn file =>
28028
c0f54a32491e renamed Buffer.write to File.write_buffer;
wenzelm
parents: 26980
diff changeset
   107
  let
28510
66b95e857bde fold_lines: more tuning, avoiding extra split_last;
wenzelm
parents: 28500
diff changeset
   108
    fun read buf x =
75615
4494cd69f97f clarified modules;
wenzelm
parents: 75613
diff changeset
   109
      (case File_Stream.input file of
75595
ecbd0b38256b removed unused operations;
wenzelm
parents: 75593
diff changeset
   110
        "" => (case Buffer.content buf of "" => x | line => f line x)
28510
66b95e857bde fold_lines: more tuning, avoiding extra split_last;
wenzelm
parents: 28500
diff changeset
   111
      | input =>
75595
ecbd0b38256b removed unused operations;
wenzelm
parents: 75593
diff changeset
   112
          (case String.fields (fn c => c = #"\n") input of
28510
66b95e857bde fold_lines: more tuning, avoiding extra split_last;
wenzelm
parents: 28500
diff changeset
   113
            [rest] => read (Buffer.add rest buf) x
75595
ecbd0b38256b removed unused operations;
wenzelm
parents: 75593
diff changeset
   114
          | line :: more => read_more more (f (Buffer.content (Buffer.add line buf)) x)))
ecbd0b38256b removed unused operations;
wenzelm
parents: 75593
diff changeset
   115
    and read_more [rest] x = read (Buffer.add rest Buffer.empty) x
ecbd0b38256b removed unused operations;
wenzelm
parents: 75593
diff changeset
   116
      | read_more (line :: more) x = read_more more (f line x);
28028
c0f54a32491e renamed Buffer.write to File.write_buffer;
wenzelm
parents: 26980
diff changeset
   117
  in read Buffer.empty a end) path;
c0f54a32491e renamed Buffer.write to File.write_buffer;
wenzelm
parents: 26980
diff changeset
   118
43848
8f2bf02a0ccb some file and directory operations;
wenzelm
parents: 43845
diff changeset
   119
fun read_lines path = rev (fold_lines cons path []);
8f2bf02a0ccb some file and directory operations;
wenzelm
parents: 43845
diff changeset
   120
75615
4494cd69f97f clarified modules;
wenzelm
parents: 75613
diff changeset
   121
val read = File_Stream.open_input File_Stream.input_all;
5009
05b152a41922 moved Thy/file.ML to General/file.ML;
wenzelm
parents:
diff changeset
   122
28028
c0f54a32491e renamed Buffer.write to File.write_buffer;
wenzelm
parents: 26980
diff changeset
   123
75615
4494cd69f97f clarified modules;
wenzelm
parents: 75613
diff changeset
   124
(* write *)
28028
c0f54a32491e renamed Buffer.write to File.write_buffer;
wenzelm
parents: 26980
diff changeset
   125
75615
4494cd69f97f clarified modules;
wenzelm
parents: 75613
diff changeset
   126
fun write_list path ss = File_Stream.open_output (fn stream => File_Stream.outputs stream ss) path;
4494cd69f97f clarified modules;
wenzelm
parents: 75613
diff changeset
   127
fun append_list path ss = File_Stream.open_append (fn stream => File_Stream.outputs stream ss) path;
16713
be5763901788 added write_list, append_list;
wenzelm
parents: 16603
diff changeset
   128
be5763901788 added write_list, append_list;
wenzelm
parents: 16603
diff changeset
   129
fun write path txt = write_list path [txt];
be5763901788 added write_list, append_list;
wenzelm
parents: 16603
diff changeset
   130
fun append path txt = append_list path [txt];
6182
4a07dfe3583f use Path.T;
wenzelm
parents: 6118
diff changeset
   131
28028
c0f54a32491e renamed Buffer.write to File.write_buffer;
wenzelm
parents: 26980
diff changeset
   132
56533
cd8b6d849b6a explicit 'document_files' in session ROOT specifications;
wenzelm
parents: 48911
diff changeset
   133
(* eq *)
5009
05b152a41922 moved Thy/file.ML to General/file.ML;
wenzelm
parents:
diff changeset
   134
16603
1776d276f848 added eq;
wenzelm
parents: 16304
diff changeset
   135
fun eq paths =
59058
a78612c67ec0 renamed "pairself" to "apply2", in accordance to @{apply 2};
wenzelm
parents: 56533
diff changeset
   136
  (case try (apply2 (OS.FileSys.fileId o platform_path)) paths of
26656
62fff5feb756 Library.is_equal;
wenzelm
parents: 26503
diff changeset
   137
    SOME ids => is_equal (OS.FileSys.compare ids)
16603
1776d276f848 added eq;
wenzelm
parents: 16304
diff changeset
   138
  | NONE => false);
1776d276f848 added eq;
wenzelm
parents: 16304
diff changeset
   139
5009
05b152a41922 moved Thy/file.ML to General/file.ML;
wenzelm
parents:
diff changeset
   140
end;