src/Pure/General/file.ML
author wenzelm
Thu, 10 Oct 2024 14:13:18 +0200
changeset 81146 87f173836d56
parent 77180 7af930cd0fce
permissions -rw-r--r--
tuned NEWS;
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
76884
a004c5322ea4 clarified modules;
wenzelm
parents: 75616
diff changeset
    14
  val symbolic_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
    15
  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
    16
  val full_path: Path.T -> Path.T -> Path.T
6182
4a07dfe3583f use Path.T;
wenzelm
parents: 6118
diff changeset
    17
  val tmp_path: Path.T -> Path.T
17826
afa2696eacce added assert;
wenzelm
parents: 16713
diff changeset
    18
  val exists: Path.T -> bool
afa2696eacce added assert;
wenzelm
parents: 16713
diff changeset
    19
  val rm: Path.T -> unit
40785
c755df0f7062 more permissive Isabelle_System.mkdir;
wenzelm
parents: 40746
diff changeset
    20
  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
    21
  val is_file: Path.T -> bool
42003
6e45dc518ebb replaced File.check by specific File.check_file, File.check_dir;
wenzelm
parents: 41944
diff changeset
    22
  val check_dir: Path.T -> Path.T
6e45dc518ebb replaced File.check by specific File.check_file, File.check_dir;
wenzelm
parents: 41944
diff changeset
    23
  val check_file: Path.T -> Path.T
43848
8f2bf02a0ccb some file and directory operations;
wenzelm
parents: 43845
diff changeset
    24
  val fold_dir: (string -> 'a -> 'a) -> Path.T -> 'a -> 'a
8f2bf02a0ccb some file and directory operations;
wenzelm
parents: 43845
diff changeset
    25
  val read_dir: Path.T -> string list
75616
986506233812 clarified signature: File.read_lines is based on scalable Bytes.T;
wenzelm
parents: 75615
diff changeset
    26
  val read: Path.T -> string
43848
8f2bf02a0ccb some file and directory operations;
wenzelm
parents: 43845
diff changeset
    27
  val read_lines: Path.T -> string list
6182
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
76884
a004c5322ea4 clarified modules;
wenzelm
parents: 75616
diff changeset
    49
(* symbolic path representation, e.g. "~~/src/Pure/ROOT.ML" *)
a004c5322ea4 clarified modules;
wenzelm
parents: 75616
diff changeset
    50
a004c5322ea4 clarified modules;
wenzelm
parents: 75616
diff changeset
    51
fun symbolic_path path =
a004c5322ea4 clarified modules;
wenzelm
parents: 75616
diff changeset
    52
  let
a004c5322ea4 clarified modules;
wenzelm
parents: 75616
diff changeset
    53
    val directories = rev (space_explode ":" (getenv "ISABELLE_DIRECTORIES"));
a004c5322ea4 clarified modules;
wenzelm
parents: 75616
diff changeset
    54
    val full_name = standard_path path;
a004c5322ea4 clarified modules;
wenzelm
parents: 75616
diff changeset
    55
    fun fold_path a =
a004c5322ea4 clarified modules;
wenzelm
parents: 75616
diff changeset
    56
      (case try (standard_path o Path.explode) a of
a004c5322ea4 clarified modules;
wenzelm
parents: 75616
diff changeset
    57
        SOME b =>
a004c5322ea4 clarified modules;
wenzelm
parents: 75616
diff changeset
    58
          if full_name = b then SOME a
a004c5322ea4 clarified modules;
wenzelm
parents: 75616
diff changeset
    59
          else
a004c5322ea4 clarified modules;
wenzelm
parents: 75616
diff changeset
    60
            (case try (unprefix (b ^ "/")) full_name of
a004c5322ea4 clarified modules;
wenzelm
parents: 75616
diff changeset
    61
              SOME name => SOME (a ^ "/" ^ name)
a004c5322ea4 clarified modules;
wenzelm
parents: 75616
diff changeset
    62
            | NONE => NONE)
a004c5322ea4 clarified modules;
wenzelm
parents: 75616
diff changeset
    63
      | NONE => NONE);
a004c5322ea4 clarified modules;
wenzelm
parents: 75616
diff changeset
    64
  in
a004c5322ea4 clarified modules;
wenzelm
parents: 75616
diff changeset
    65
    (case get_first fold_path directories of
a004c5322ea4 clarified modules;
wenzelm
parents: 75616
diff changeset
    66
      SOME name => name
a004c5322ea4 clarified modules;
wenzelm
parents: 75616
diff changeset
    67
    | NONE => Path.implode path)
a004c5322ea4 clarified modules;
wenzelm
parents: 75616
diff changeset
    68
  end;
a004c5322ea4 clarified modules;
wenzelm
parents: 75616
diff changeset
    69
a004c5322ea4 clarified modules;
wenzelm
parents: 75616
diff changeset
    70
42003
6e45dc518ebb replaced File.check by specific File.check_file, File.check_dir;
wenzelm
parents: 41944
diff changeset
    71
(* full_path *)
6e45dc518ebb replaced File.check by specific File.check_file, File.check_dir;
wenzelm
parents: 41944
diff changeset
    72
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
    73
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
    74
  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
    75
    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
    76
    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
    77
42003
6e45dc518ebb replaced File.check by specific File.check_file, File.check_dir;
wenzelm
parents: 41944
diff changeset
    78
fun full_path dir path =
6e45dc518ebb replaced File.check by specific File.check_file, File.check_dir;
wenzelm
parents: 41944
diff changeset
    79
  let
6e45dc518ebb replaced File.check by specific File.check_file, File.check_dir;
wenzelm
parents: 41944
diff changeset
    80
    val path' = Path.expand path;
6e45dc518ebb replaced File.check by specific File.check_file, File.check_dir;
wenzelm
parents: 41944
diff changeset
    81
    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
    82
  in absolute_path (dir + path') end;
5009
05b152a41922 moved Thy/file.ML to General/file.ML;
wenzelm
parents:
diff changeset
    83
6182
4a07dfe3583f use Path.T;
wenzelm
parents: 6118
diff changeset
    84
4a07dfe3583f use Path.T;
wenzelm
parents: 6118
diff changeset
    85
(* tmp_path *)
4a07dfe3583f use Path.T;
wenzelm
parents: 6118
diff changeset
    86
72511
460d743010bc clarified signature: overloaded "+" for Path.append;
wenzelm
parents: 72278
diff changeset
    87
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
    88
05b152a41922 moved Thy/file.ML to General/file.ML;
wenzelm
parents:
diff changeset
    89
26980
f7f48bb9a025 ident: naive caching prevents potentially slow external invocations;
wenzelm
parents: 26946
diff changeset
    90
(* directory entries *)
f7f48bb9a025 ident: naive caching prevents potentially slow external invocations;
wenzelm
parents: 26946
diff changeset
    91
77180
7af930cd0fce more robust on Windows, where C:\\ and \\SERVER\SHARE cause problems (line 920 of winbasicio.cpp);
wenzelm
parents: 76884
diff changeset
    92
val exists = can OS.FileSys.fileId o platform_path;
16261
28803c418b59 removed sysify_path, quote_sysity_path etc.;
wenzelm
parents: 16002
diff changeset
    93
28803c418b59 removed sysify_path, quote_sysity_path etc.;
wenzelm
parents: 16002
diff changeset
    94
val rm = OS.FileSys.remove o platform_path;
28803c418b59 removed sysify_path, quote_sysity_path etc.;
wenzelm
parents: 16002
diff changeset
    95
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
    96
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
    97
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
    98
fun is_file path = exists path andalso not (test_dir path);
40785
c755df0f7062 more permissive Isabelle_System.mkdir;
wenzelm
parents: 40746
diff changeset
    99
42003
6e45dc518ebb replaced File.check by specific File.check_file, File.check_dir;
wenzelm
parents: 41944
diff changeset
   100
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
   101
  if is_dir path then path
63668
5efaa884ac6c tuned error;
wenzelm
parents: 62854
diff changeset
   102
  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
   103
6e45dc518ebb replaced File.check by specific File.check_file, File.check_dir;
wenzelm
parents: 41944
diff changeset
   104
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
   105
  if is_file path then path
63668
5efaa884ac6c tuned error;
wenzelm
parents: 62854
diff changeset
   106
  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
   107
16261
28803c418b59 removed sysify_path, quote_sysity_path etc.;
wenzelm
parents: 16002
diff changeset
   108
43848
8f2bf02a0ccb some file and directory operations;
wenzelm
parents: 43845
diff changeset
   109
(* directory content *)
8f2bf02a0ccb some file and directory operations;
wenzelm
parents: 43845
diff changeset
   110
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
   111
fun fold_dir f path a =
75615
4494cd69f97f clarified modules;
wenzelm
parents: 75613
diff changeset
   112
  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
   113
    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
   114
      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
   115
        (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
   116
          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
   117
        | 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
   118
    in read a end);
43848
8f2bf02a0ccb some file and directory operations;
wenzelm
parents: 43845
diff changeset
   119
69427
ff2f39a221d4 clarified operations: uniform sorting of results;
wenzelm
parents: 69300
diff changeset
   120
fun read_dir path = sort_strings (fold_dir cons path []);
43848
8f2bf02a0ccb some file and directory operations;
wenzelm
parents: 43845
diff changeset
   121
8f2bf02a0ccb some file and directory operations;
wenzelm
parents: 43845
diff changeset
   122
75616
986506233812 clarified signature: File.read_lines is based on scalable Bytes.T;
wenzelm
parents: 75615
diff changeset
   123
(* read *)
43848
8f2bf02a0ccb some file and directory operations;
wenzelm
parents: 43845
diff changeset
   124
75615
4494cd69f97f clarified modules;
wenzelm
parents: 75613
diff changeset
   125
val read = File_Stream.open_input File_Stream.input_all;
5009
05b152a41922 moved Thy/file.ML to General/file.ML;
wenzelm
parents:
diff changeset
   126
75616
986506233812 clarified signature: File.read_lines is based on scalable Bytes.T;
wenzelm
parents: 75615
diff changeset
   127
val read_lines = Bytes.read #> Bytes.trim_split_lines;
986506233812 clarified signature: File.read_lines is based on scalable Bytes.T;
wenzelm
parents: 75615
diff changeset
   128
28028
c0f54a32491e renamed Buffer.write to File.write_buffer;
wenzelm
parents: 26980
diff changeset
   129
75615
4494cd69f97f clarified modules;
wenzelm
parents: 75613
diff changeset
   130
(* write *)
28028
c0f54a32491e renamed Buffer.write to File.write_buffer;
wenzelm
parents: 26980
diff changeset
   131
75615
4494cd69f97f clarified modules;
wenzelm
parents: 75613
diff changeset
   132
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
   133
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
   134
be5763901788 added write_list, append_list;
wenzelm
parents: 16603
diff changeset
   135
fun write path txt = write_list path [txt];
be5763901788 added write_list, append_list;
wenzelm
parents: 16603
diff changeset
   136
fun append path txt = append_list path [txt];
6182
4a07dfe3583f use Path.T;
wenzelm
parents: 6118
diff changeset
   137
28028
c0f54a32491e renamed Buffer.write to File.write_buffer;
wenzelm
parents: 26980
diff changeset
   138
56533
cd8b6d849b6a explicit 'document_files' in session ROOT specifications;
wenzelm
parents: 48911
diff changeset
   139
(* eq *)
5009
05b152a41922 moved Thy/file.ML to General/file.ML;
wenzelm
parents:
diff changeset
   140
16603
1776d276f848 added eq;
wenzelm
parents: 16304
diff changeset
   141
fun eq paths =
59058
a78612c67ec0 renamed "pairself" to "apply2", in accordance to @{apply 2};
wenzelm
parents: 56533
diff changeset
   142
  (case try (apply2 (OS.FileSys.fileId o platform_path)) paths of
26656
62fff5feb756 Library.is_equal;
wenzelm
parents: 26503
diff changeset
   143
    SOME ids => is_equal (OS.FileSys.compare ids)
16603
1776d276f848 added eq;
wenzelm
parents: 16304
diff changeset
   144
  | NONE => false);
1776d276f848 added eq;
wenzelm
parents: 16304
diff changeset
   145
5009
05b152a41922 moved Thy/file.ML to General/file.ML;
wenzelm
parents:
diff changeset
   146
end;