forked from
jonmsterling.com/ocaml-forester
ocaml
formatting
+73
-87
+5
-5
lib/compiler/URI_util.ml
+5
-5
lib/compiler/URI_util.ml
···
26
26
let config = forest.config in
27
27
let keys =
28
28
List.of_seq @@
29
-
let@ uri = Seq.filter_map @~ addrs in
30
-
if URI.host config.url = URI.host uri then
31
-
let@ prefix', key = Option.bind @@ URI_scheme.split_addr uri in
32
-
if prefix = prefix' then Some key else None
33
-
else None
29
+
let@ uri = Seq.filter_map @~ addrs in
30
+
if URI.host config.url = URI.host uri then
31
+
let@ prefix', key = Option.bind @@ URI_scheme.split_addr uri in
32
+
if prefix = prefix' then Some key else None
33
+
else None
34
34
in
35
35
let next =
36
36
match mode with
+10
-10
lib/compiler/test/Test_compiler.ml
+10
-10
lib/compiler/test/Test_compiler.ml
···
28
28
29
29
let test_batch_run ~env () =
30
30
let forest, history =
31
-
with_test_forest
32
-
~raw_trees
33
-
~env
34
-
~config
35
-
(fun path ->
36
-
Sys.chdir (Eio.Path.native_exn path);
37
-
let@ () = Reporter.easy_run in
38
-
let forest = State.make ~env ~config ~dev: false () in
39
-
Driver.run_with_history Load_all_configured_dirs forest
40
-
)
31
+
let@ path =
32
+
with_test_forest
33
+
~raw_trees
34
+
~env
35
+
~config
36
+
in
37
+
Sys.chdir (Eio.Path.native_exn path);
38
+
let@ () = Reporter.easy_run in
39
+
let forest = State.make ~env ~config ~dev: false () in
40
+
Driver.run_with_history Load_all_configured_dirs forest
41
41
in
42
42
Alcotest.(check @@ list action)
43
43
"all actions have run"
+1
lib/compiler/test/Test_diagnostic_store.ml
+1
lib/compiler/test/Test_diagnostic_store.ml
···
1
+
+3
-2
lib/compiler/test/Test_expansion.ml
+3
-2
lib/compiler/test/Test_expansion.ml
···
61
61
let@ () = Reporter.easy_run in
62
62
let forest = State.make ~env ~config ~dev: false () in
63
63
let expanded =
64
-
expand ~forest
64
+
expand
65
+
~forest
65
66
{|
66
67
\subtree[foo]{
67
68
\title{Hello}
···
79
80
evaluated
80
81
81
82
let test_visible ~env () =
82
-
let forest = State.make ~env ~config ~dev: false () in
83
+
let forest = State.make ~env ~config ~dev: false () in
83
84
let code =
84
85
Result.get_ok @@
85
86
parse_string
+1
-1
lib/compiler/test/Test_import_graph.ml
+1
-1
lib/compiler/test/Test_import_graph.ml
···
14
14
15
15
let config = {(Config.default ()) with trees = ["imports"]}
16
16
17
-
let mk_vertex v = T.Uri_vertex (URI_scheme.named_uri ~base:config.url v)
17
+
let mk_vertex v = T.Uri_vertex (URI_scheme.named_uri ~base: config.url v)
18
18
19
19
let has_edge g v w =
20
20
Forest_graph.mem_edge g (mk_vertex v) (mk_vertex w)
+13
-12
lib/frontend/Config_parser.ml
+13
-12
lib/frontend/Config_parser.ml
···
23
23
let keys (tbl : Toml.Types.value Toml.Types.Table.t) =
24
24
let rec go current keys tbl =
25
25
List.fold_left
26
-
(fun acc (key, value) ->
27
-
match value with
28
-
| Toml.Types.TBool _
29
-
| TInt _
30
-
| TFloat _
31
-
| TString _
32
-
| TDate _
33
-
| TArray _ ->
34
-
(key :: current) :: acc
35
-
| TTable tbl ->
36
-
go (key :: current) acc tbl
37
-
)
26
+
begin
27
+
fun acc (key, value) ->
28
+
match value with
29
+
| Toml.Types.TBool _
30
+
| TInt _
31
+
| TFloat _
32
+
| TString _
33
+
| TDate _
34
+
| TArray _ ->
35
+
(key :: current) :: acc
36
+
| TTable tbl ->
37
+
go (key :: current) acc tbl
38
+
end
38
39
keys
39
40
(Toml.Types.Table.to_list tbl)
40
41
in
+40
-56
lib/search/Index.ml
+40
-56
lib/search/Index.ml
···
4
4
* SPDX-License-Identifier: GPL-3.0-or-later
5
5
*)
6
6
7
+
open Forester_prelude
7
8
open Forester_core
8
9
open Spelll
9
10
···
21
22
number_of_docs: int;
22
23
}
23
24
24
-
let average_doc_length
25
-
: t -> float
26
-
= fun {number_of_tokens; number_of_docs; _} ->
25
+
let average_doc_length {number_of_tokens; number_of_docs; _} : float =
27
26
Float.of_int number_of_tokens /. Float.of_int number_of_docs
28
27
29
-
let add_one
30
-
: T.content T.article -> t -> t
31
-
= fun article ({index; number_of_tokens; number_of_docs;} as t) ->
28
+
let add_one (article : T.content T.article) ({index; number_of_tokens; number_of_docs;} as t) : t =
32
29
if Option.is_none T.(article.frontmatter.uri) then t
33
30
else
34
31
let tokens_in_article = Tokenizer.tokenize_article article in
···
36
33
let new_tokens = ref 0 in
37
34
let new_index =
38
35
List.fold_left
39
-
(fun index (ocurrences, token) ->
40
-
match Index.retrieve_l ~limit: 0 index token with
41
-
| [] ->
42
-
(* Unseen token*)
43
-
(* TODO: add to list of ocurrences*)
44
-
let ocurrence = Ocurrences.singleton ([ocurrences], uri) in
45
-
new_tokens := !new_tokens + 1;
46
-
Index.add index token ocurrence
47
-
| ids :: [] ->
48
-
Index.add index token (Ocurrences.add ([ocurrences], uri) ids)
49
-
| _ ->
50
-
(* We are using limit=0, so this shouldn't happen*)
51
-
assert false
52
-
)
36
+
begin
37
+
fun index (ocurrences, token) ->
38
+
match Index.retrieve_l ~limit: 0 index token with
39
+
| [] ->
40
+
(* Unseen token*)
41
+
(* TODO: add to list of ocurrences*)
42
+
let ocurrence = Ocurrences.singleton ([ocurrences], uri) in
43
+
new_tokens := !new_tokens + 1;
44
+
Index.add index token ocurrence
45
+
| ids :: [] ->
46
+
Index.add index token (Ocurrences.add ([ocurrences], uri) ids)
47
+
| _ ->
48
+
(* We are using limit=0, so this shouldn't happen*)
49
+
assert false
50
+
end
53
51
index
54
52
tokens_in_article
55
53
in
···
59
57
number_of_tokens = number_of_tokens + !new_tokens
60
58
}
61
59
62
-
let add
63
-
: T.content T.article list -> t -> t
64
-
=
60
+
let add : T.content T.article list -> t -> t =
65
61
List.fold_right add_one
66
62
67
-
let search
68
-
: ?fuzz: int -> t -> string -> (int list list * URI.t) list
69
-
= fun ?(fuzz = 0) index term ->
70
-
Tokenizer.tokenize term
71
-
|> List.concat_map
72
-
(fun str ->
73
-
List.concat_map Ocurrences.to_list @@
74
-
Index.retrieve_l ~limit: fuzz index.index str
75
-
)
63
+
let search ?(fuzz = 0) (index : t) (term : string) : (int list list * URI.t) list =
64
+
let@ str = List.concat_map @~ Tokenizer.tokenize term in
65
+
List.concat_map Ocurrences.to_list @@
66
+
Index.retrieve_l ~limit: fuzz index.index str
76
67
77
68
module BM_25 = struct
78
69
let sum = List.fold_left (+.) 0.
···
87
78
List.length @@
88
79
Tokenizer.tokenize_article d
89
80
90
-
let score
91
-
: T.content T.article -> string -> t -> float
92
-
= fun d q index ->
81
+
let score (d : T.content T.article) (q : string) (index : t) : float =
93
82
let tokens = Tokenizer.tokenize q in
94
83
assert (List.length tokens > 0);
95
84
let avg_len = average_doc_length index in
96
85
let k_1 = 1.5 in
97
86
let b = 0.75 in
98
87
sum @@
99
-
List.map
100
-
(fun q_i ->
101
-
let num_occurrences =
102
-
Float.of_int @@
103
-
List.length @@ search index q_i
104
-
in
105
-
(* Format.printf "num_occurrences: %f" num_occurrences; *)
106
-
idf q index *.
107
-
begin
108
-
(num_occurrences *. k_1 +. 1.) /.
109
-
(num_occurrences +. k_1 *. (1. -. b +. (b *. doc_length d /. avg_len))) +.
110
-
1.
111
-
end
112
-
)
113
-
tokens
88
+
let@ q_i = List.map @~ tokens in
89
+
let num_occurrences =
90
+
Float.of_int @@
91
+
List.length @@ search index q_i
92
+
in
93
+
(* Format.printf "num_occurrences: %f" num_occurrences; *)
94
+
idf q index *.
95
+
begin
96
+
(num_occurrences *. k_1 +. 1.) /.
97
+
(num_occurrences +. k_1 *. (1. -. b +. (b *. doc_length d /. avg_len))) +.
98
+
1.
99
+
end
114
100
end
115
101
116
102
let create articles =
···
124
110
125
111
let marshal (v : t) filename =
126
112
let oc = open_out_bin filename in
127
-
Fun.protect
128
-
~finally: (fun () -> close_out oc)
129
-
(fun () -> Marshal.to_channel oc v [])
113
+
let@ () = Fun.protect ~finally: (fun () -> close_out oc) in
114
+
Marshal.to_channel oc v []
130
115
131
116
let unmarshal filename : t =
132
117
let ic = open_in_bin filename in
133
-
Fun.protect
134
-
~finally: (fun () -> close_in ic)
135
-
(fun () -> Marshal.from_channel ic)
118
+
let@ () = Fun.protect ~finally: (fun () -> close_in ic) in
119
+
Marshal.from_channel ic