[OCaml] Mobile-friendly clone of cgit.
refactor rename Diff to Line_diff and line_diff to of_contents
Three modules were named Diff: this one computing diffs, Resolvers.Diff reading them out of a Git store, and Ui.Diff rendering them. Each was correct in its own scope, but repo.ml converts Resolvers.Diff.line to Ui.Diff.line in adjacent lines, and nothing in the name distinguished the computation from the two layers around it. Line_diff says what this module does and what it does not: it matches whole lines as opaque units, with no word-level refinement. Within it, line_diff stuttered against the module name, so it becomes of_contents, which also names its arguments — two file contents rather than two paths or two hashes. Resolvers.Diff still includes it, so Resolvers.Diff.of_contents remains the access path used by the tests. Also removes a duplicated mode line that the documentation commit introduced at the top of this file.
lib/diff.ml
@@ -1,179 +0,0 @@
1
Removed:
(* -*- mode: tuareg; -*- *)
2
Removed:
3
Removed:
(* -*- mode: tuareg; -*- *)
4
Removed:
5
Removed:
(** Line diffs between two blobs.
6
Removed:
7
Removed:
Computes a longest-common-subsequence diff and groups the result into hunks
8
Removed:
with the surrounding context, which is the shape a reader expects from
9
Removed:
[git diff]. Large inputs fall back to a coarser result rather than spending
10
Removed:
unbounded time and memory on the LCS matrix.
11
Removed:
12
Removed:
This module produces data only. Rendering it is {!module:Ui.Diff}'s job.
13
Removed:
14
Removed:
TODO: document the exact size threshold and the shape of the fallback, once
15
Removed:
the limit is settled. *)
16
Removed:
17
Removed:
type line_kind = Context | Addition | Deletion
18
Removed:
19
Removed:
type line = {
20
Removed:
kind : line_kind;
21
Removed:
old_number : int option;
22
Removed:
new_number : int option;
23
Removed:
text : string;
24
Removed:
}
25
Removed:
26
Removed:
type hunk = {
27
Removed:
old_start : int;
28
Removed:
old_count : int;
29
Removed:
new_start : int;
30
Removed:
new_count : int;
31
Removed:
lines : line list;
32
Removed:
}
33
Removed:
34
Removed:
type file = {
35
Removed:
path : string;
36
Removed:
old_hash : string option;
37
Removed:
new_hash : string option;
38
Removed:
old_mode : int option;
39
Removed:
new_mode : int option;
40
Removed:
binary : bool;
41
Removed:
hunks : hunk list;
42
Removed:
}
43
Removed:
44
Removed:
let split_lines content =
45
Removed:
match String.split_on_char '\n' content with
46
Removed:
| [] -> []
47
Removed:
| lines ->
48
Removed:
if content = "" then []
49
Removed:
else if String.ends_with ~suffix:"\n" content then
50
Removed:
List.rev lines |> List.tl |> List.rev
51
Removed:
else lines
52
Removed:
53
Removed:
let line_diff old_content new_content =
54
Removed:
let old_lines = Array.of_list (split_lines old_content) in
55
Removed:
let new_lines = Array.of_list (split_lines new_content) in
56
Removed:
let old_length = Array.length old_lines in
57
Removed:
let new_length = Array.length new_lines in
58
Removed:
let matrix_size = old_length * new_length in
59
Removed:
let rec all_deletions index acc =
60
Removed:
if index = old_length then List.rev acc
61
Removed:
else
62
Removed:
all_deletions (index + 1)
63
Removed:
({
64
Removed:
kind = Deletion;
65
Removed:
old_number = Some (index + 1);
66
Removed:
new_number = None;
67
Removed:
text = old_lines.(index);
68
Removed:
}
69
Removed:
:: acc)
70
Removed:
in
71
Removed:
let rec all_additions index acc =
72
Removed:
if index = new_length then List.rev acc
73
Removed:
else
74
Removed:
all_additions (index + 1)
75
Removed:
({
76
Removed:
kind = Addition;
77
Removed:
old_number = None;
78
Removed:
new_number = Some (index + 1);
79
Removed:
text = new_lines.(index);
80
Removed:
}
81
Removed:
:: acc)
82
Removed:
in
83
Removed:
if matrix_size > 4_000_000 then all_deletions 0 [] @ all_additions 0 []
84
Removed:
else
85
Removed:
let lengths = Array.make_matrix (old_length + 1) (new_length + 1) 0 in
86
Removed:
for old_index = old_length - 1 downto 0 do
87
Removed:
for new_index = new_length - 1 downto 0 do
88
Removed:
lengths.(old_index).(new_index) <-
89
Removed:
(if old_lines.(old_index) = new_lines.(new_index) then
90
Removed:
lengths.(old_index + 1).(new_index + 1) + 1
91
Removed:
else
92
Removed:
max
93
Removed:
lengths.(old_index + 1).(new_index)
94
Removed:
lengths.(old_index).(new_index + 1))
95
Removed:
done
96
Removed:
done;
97
Removed:
let rec build old_index new_index acc =
98
Removed:
if old_index = old_length then List.rev acc @ all_additions new_index []
99
Removed:
else if new_index = new_length then
100
Removed:
List.rev acc @ all_deletions old_index []
101
Removed:
else if old_lines.(old_index) = new_lines.(new_index) then
102
Removed:
build (old_index + 1) (new_index + 1)
103
Removed:
({
104
Removed:
kind = Context;
105
Removed:
old_number = Some (old_index + 1);
106
Removed:
new_number = Some (new_index + 1);
107
Removed:
text = old_lines.(old_index);
108
Removed:
}
109
Removed:
:: acc)
110
Removed:
else if
111
Removed:
lengths.(old_index + 1).(new_index)
112
Removed:
>= lengths.(old_index).(new_index + 1)
113
Removed:
then
114
Removed:
build (old_index + 1) new_index
115
Removed:
({
116
Removed:
kind = Deletion;
117
Removed:
old_number = Some (old_index + 1);
118
Removed:
new_number = None;
119
Removed:
text = old_lines.(old_index);
120
Removed:
}
121
Removed:
:: acc)
122
Removed:
else
123
Removed:
build old_index (new_index + 1)
124
Removed:
({
125
Removed:
kind = Addition;
126
Removed:
old_number = None;
127
Removed:
new_number = Some (new_index + 1);
128
Removed:
text = new_lines.(new_index);
129
Removed:
}
130
Removed:
:: acc)
131
Removed:
in
132
Removed:
build 0 0 []
133
Removed:
134
Removed:
let hunks ?(context = 3) lines =
135
Removed:
let lines = Array.of_list lines in
136
Removed:
let length = Array.length lines in
137
Removed:
let changed =
138
Removed:
Array.to_list (Array.mapi (fun index line -> (index, line.kind)) lines)
139
Removed:
|> List.filter_map (function
140
Removed:
| index, (Addition | Deletion) -> Some index
141
Removed:
| _, Context -> None)
142
Removed:
in
143
Removed:
let ranges =
144
Removed:
let add_range ranges index =
145
Removed:
let first = max 0 (index - context) in
146
Removed:
let last = min (length - 1) (index + context) in
147
Removed:
match ranges with
148
Removed:
| (range_first, range_last) :: rest when first <= range_last + 1 ->
149
Removed:
(range_first, max range_last last) :: rest
150
Removed:
| _ -> (first, last) :: ranges
151
Removed:
in
152
Removed:
List.fold_left add_range [] changed |> List.rev
153
Removed:
in
154
Removed:
let number_or_zero get_number slice =
155
Removed:
List.find_map get_number slice |> Option.value ~default:0
156
Removed:
in
157
Removed:
let make_hunk (first, last) =
158
Removed:
let rec slice index acc =
159
Removed:
if index > last then List.rev acc
160
Removed:
else slice (index + 1) (lines.(index) :: acc)
161
Removed:
in
162
Removed:
let lines = slice first [] in
163
Removed:
{
164
Removed:
old_start = number_or_zero (fun line -> line.old_number) lines;
165
Removed:
old_count =
166
Removed:
List.fold_left
167
Removed:
(fun count line ->
168
Removed:
if Option.is_some line.old_number then count + 1 else count)
169
Removed:
0 lines;
170
Removed:
new_start = number_or_zero (fun line -> line.new_number) lines;
171
Removed:
new_count =
172
Removed:
List.fold_left
173
Removed:
(fun count line ->
174
Removed:
if Option.is_some line.new_number then count + 1 else count)
175
Removed:
0 lines;
176
Removed:
lines;
177
Removed:
}
178
Removed:
in
179
Removed:
List.map make_hunk ranges
lib/line_diff.ml
@@ -0,0 +1,180 @@
1
Added:
(* -*- mode: tuareg; -*- *)
2
Added:
3
Added:
(** Line-oriented diffs between two blobs.
4
Added:
5
Added:
Computes a longest-common-subsequence diff and groups the result into hunks
6
Added:
with the surrounding context, which is the shape a reader expects from
7
Added:
[git diff]. Large inputs fall back to a coarser result rather than spending
8
Added:
unbounded time and memory on the LCS matrix.
9
Added:
10
Added:
The module is named for the granularity it works at: whole lines, matched as
11
Added:
opaque units, with no word- or character-level refinement.
12
Added:
13
Added:
It produces data only. Rendering it is {!module:Ui.Diff}'s job.
14
Added:
15
Added:
TODO: document the exact size threshold and the shape of the fallback, once
16
Added:
the limit is settled. *)
17
Added:
18
Added:
type line_kind = Context | Addition | Deletion
19
Added:
20
Added:
type line = {
21
Added:
kind : line_kind;
22
Added:
old_number : int option;
23
Added:
new_number : int option;
24
Added:
text : string;
25
Added:
}
26
Added:
27
Added:
type hunk = {
28
Added:
old_start : int;
29
Added:
old_count : int;
30
Added:
new_start : int;
31
Added:
new_count : int;
32
Added:
lines : line list;
33
Added:
}
34
Added:
35
Added:
type file = {
36
Added:
path : string;
37
Added:
old_hash : string option;
38
Added:
new_hash : string option;
39
Added:
old_mode : int option;
40
Added:
new_mode : int option;
41
Added:
binary : bool;
42
Added:
hunks : hunk list;
43
Added:
}
44
Added:
45
Added:
let split_lines content =
46
Added:
match String.split_on_char '\n' content with
47
Added:
| [] -> []
48
Added:
| lines ->
49
Added:
if content = "" then []
50
Added:
else if String.ends_with ~suffix:"\n" content then
51
Added:
List.rev lines |> List.tl |> List.rev
52
Added:
else lines
53
Added:
54
Added:
let of_contents old_content new_content =
55
Added:
let old_lines = Array.of_list (split_lines old_content) in
56
Added:
let new_lines = Array.of_list (split_lines new_content) in
57
Added:
let old_length = Array.length old_lines in
58
Added:
let new_length = Array.length new_lines in
59
Added:
let matrix_size = old_length * new_length in
60
Added:
let rec all_deletions index acc =
61
Added:
if index = old_length then List.rev acc
62
Added:
else
63
Added:
all_deletions (index + 1)
64
Added:
({
65
Added:
kind = Deletion;
66
Added:
old_number = Some (index + 1);
67
Added:
new_number = None;
68
Added:
text = old_lines.(index);
69
Added:
}
70
Added:
:: acc)
71
Added:
in
72
Added:
let rec all_additions index acc =
73
Added:
if index = new_length then List.rev acc
74
Added:
else
75
Added:
all_additions (index + 1)
76
Added:
({
77
Added:
kind = Addition;
78
Added:
old_number = None;
79
Added:
new_number = Some (index + 1);
80
Added:
text = new_lines.(index);
81
Added:
}
82
Added:
:: acc)
83
Added:
in
84
Added:
if matrix_size > 4_000_000 then all_deletions 0 [] @ all_additions 0 []
85
Added:
else
86
Added:
let lengths = Array.make_matrix (old_length + 1) (new_length + 1) 0 in
87
Added:
for old_index = old_length - 1 downto 0 do
88
Added:
for new_index = new_length - 1 downto 0 do
89
Added:
lengths.(old_index).(new_index) <-
90
Added:
(if old_lines.(old_index) = new_lines.(new_index) then
91
Added:
lengths.(old_index + 1).(new_index + 1) + 1
92
Added:
else
93
Added:
max
94
Added:
lengths.(old_index + 1).(new_index)
95
Added:
lengths.(old_index).(new_index + 1))
96
Added:
done
97
Added:
done;
98
Added:
let rec build old_index new_index acc =
99
Added:
if old_index = old_length then List.rev acc @ all_additions new_index []
100
Added:
else if new_index = new_length then
101
Added:
List.rev acc @ all_deletions old_index []
102
Added:
else if old_lines.(old_index) = new_lines.(new_index) then
103
Added:
build (old_index + 1) (new_index + 1)
104
Added:
({
105
Added:
kind = Context;
106
Added:
old_number = Some (old_index + 1);
107
Added:
new_number = Some (new_index + 1);
108
Added:
text = old_lines.(old_index);
109
Added:
}
110
Added:
:: acc)
111
Added:
else if
112
Added:
lengths.(old_index + 1).(new_index)
113
Added:
>= lengths.(old_index).(new_index + 1)
114
Added:
then
115
Added:
build (old_index + 1) new_index
116
Added:
({
117
Added:
kind = Deletion;
118
Added:
old_number = Some (old_index + 1);
119
Added:
new_number = None;
120
Added:
text = old_lines.(old_index);
121
Added:
}
122
Added:
:: acc)
123
Added:
else
124
Added:
build old_index (new_index + 1)
125
Added:
({
126
Added:
kind = Addition;
127
Added:
old_number = None;
128
Added:
new_number = Some (new_index + 1);
129
Added:
text = new_lines.(new_index);
130
Added:
}
131
Added:
:: acc)
132
Added:
in
133
Added:
build 0 0 []
134
Added:
135
Added:
let hunks ?(context = 3) lines =
136
Added:
let lines = Array.of_list lines in
137
Added:
let length = Array.length lines in
138
Added:
let changed =
139
Added:
Array.to_list (Array.mapi (fun index line -> (index, line.kind)) lines)
140
Added:
|> List.filter_map (function
141
Added:
| index, (Addition | Deletion) -> Some index
142
Added:
| _, Context -> None)
143
Added:
in
144
Added:
let ranges =
145
Added:
let add_range ranges index =
146
Added:
let first = max 0 (index - context) in
147
Added:
let last = min (length - 1) (index + context) in
148
Added:
match ranges with
149
Added:
| (range_first, range_last) :: rest when first <= range_last + 1 ->
150
Added:
(range_first, max range_last last) :: rest
151
Added:
| _ -> (first, last) :: ranges
152
Added:
in
153
Added:
List.fold_left add_range [] changed |> List.rev
154
Added:
in
155
Added:
let number_or_zero get_number slice =
156
Added:
List.find_map get_number slice |> Option.value ~default:0
157
Added:
in
158
Added:
let make_hunk (first, last) =
159
Added:
let rec slice index acc =
160
Added:
if index > last then List.rev acc
161
Added:
else slice (index + 1) (lines.(index) :: acc)
162
Added:
in
163
Added:
let lines = slice first [] in
164
Added:
{
165
Added:
old_start = number_or_zero (fun line -> line.old_number) lines;
166
Added:
old_count =
167
Added:
List.fold_left
168
Added:
(fun count line ->
169
Added:
if Option.is_some line.old_number then count + 1 else count)
170
Added:
0 lines;
171
Added:
new_start = number_or_zero (fun line -> line.new_number) lines;
172
Added:
new_count =
173
Added:
List.fold_left
174
Added:
(fun count line ->
175
Added:
if Option.is_some line.new_number then count + 1 else count)
176
Added:
0 lines;
177
Added:
lines;
178
Added:
}
179
Added:
in
180
Added:
List.map make_hunk ranges
lib/resolvers.ml
@@ -601,7 +601,7 @@
601
601
602
602
module Diff = struct
603
603
module Path_map = Map.Make (String)
604
Removed:
include Diff
604
Added:
include Line_diff
605
605
606
606
type tree_file = { hash : string; perm : Git.Tree.perm }
607
607
@@ -687,7 +687,9 @@
687
687
binary;
688
688
hunks =
689
689
(if binary then []
690
Removed:
else Diff.line_diff old_content new_content |> Diff.hunks);
690
Added:
else
691
Added:
Line_diff.of_contents old_content new_content
692
Added:
|> Line_diff.hunks);
691
693
}
692
694
in
693
695
build (file :: files) rest
lib/resolvers.mli
@@ -120,7 +120,7 @@
120
120
(** {1 Diffs} *)
121
121
122
122
module Diff : sig
123
Removed:
include module type of Diff
123
Added:
include module type of Line_diff
124
124
125
125
val of_commit : repository -> Commit.t -> (file list, error) Lwt_result.t
126
126
end
test/test_diff.ml
@@ -4,7 +4,7 @@
4
4
5
5
let test_line_diff () =
6
6
let open Ogit.Resolvers.Diff in
7
Removed:
match line_diff "first\nold\nlast\n" "first\nnew\nlast\n" with
7
Added:
match of_contents "first\nold\nlast\n" "first\nnew\nlast\n" with
8
8
| [ ctx1; del; add; ctx2 ] ->
9
9
Alcotest.(check string) "ctx1 text" "first" ctx1.text;
10
10
Alcotest.(check string) "del text" "old" del.text;
@@ -14,7 +14,7 @@
14
14
15
15
let test_empty_vs_empty () =
16
16
let open Ogit.Resolvers.Diff in
17
Removed:
let lines = line_diff "" "" in
17
Added:
let lines = of_contents "" "" in
18
18
Alcotest.(check int) "no lines" 0 (List.length lines);
19
19
let hunks = hunks lines in
20
20
Alcotest.(check int) "no hunks" 0 (List.length hunks)
@@ -22,7 +22,7 @@
22
22
let test_identical_content () =
23
23
let open Ogit.Resolvers.Diff in
24
24
let content = "line1\nline2\nline3\n" in
25
Removed:
let lines = line_diff content content in
25
Added:
let lines = of_contents content content in
26
26
List.iter
27
27
(fun line -> Alcotest.(check bool) "all context" true (line.kind = Context))
28
28
lines;
@@ -31,7 +31,7 @@
31
31
32
32
let test_entirely_new_file () =
33
33
let open Ogit.Resolvers.Diff in
34
Removed:
let lines = line_diff "" "new1\nnew2\n" in
34
Added:
let lines = of_contents "" "new1\nnew2\n" in
35
35
Alcotest.(check int) "2 additions" 2 (List.length lines);
36
36
List.iter
37
37
(fun line ->
@@ -40,7 +40,7 @@
40
40
41
41
let test_entirely_deleted_file () =
42
42
let open Ogit.Resolvers.Diff in
43
Removed:
let lines = line_diff "old1\nold2\n" "" in
43
Added:
let lines = of_contents "old1\nold2\n" "" in
44
44
Alcotest.(check int) "2 deletions" 2 (List.length lines);
45
45
List.iter
46
46
(fun line ->
@@ -49,8 +49,8 @@
49
49
50
50
let test_trailing_newline_handling () =
51
51
let open Ogit.Resolvers.Diff in
52
Removed:
let with_newline = line_diff "a\n" "a\n" in
53
Removed:
let without_newline = line_diff "a" "a" in
52
Added:
let with_newline = of_contents "a\n" "a\n" in
53
Added:
let without_newline = of_contents "a" "a" in
54
54
Alcotest.(check int) "with newline: 1 line" 1 (List.length with_newline);
55
55
Alcotest.(check int) "without newline: 1 line" 1 (List.length without_newline);
56
56
Alcotest.(check string) "text matches" "a" (List.hd with_newline).text;
@@ -58,7 +58,7 @@
58
58
59
59
let test_no_trailing_newline_diff () =
60
60
let open Ogit.Resolvers.Diff in
61
Removed:
let lines = line_diff "a\nb" "a\nc" in
61
Added:
let lines = of_contents "a\nb" "a\nc" in
62
62
Alcotest.(check int) "3 lines" 3 (List.length lines);
63
63
let kinds = List.map (fun l -> l.kind) lines in
64
64
Alcotest.(check bool)
@@ -73,7 +73,7 @@
73
73
in
74
74
match
75
75
hunks
76
Removed:
(line_diff
76
Added:
(of_contents
77
77
(String.concat "\n" old_content)
78
78
(String.concat "\n" new_content))
79
79
with
@@ -92,7 +92,7 @@
92
92
in
93
93
let result =
94
94
hunks
95
Removed:
(line_diff
95
Added:
(of_contents
96
96
(String.concat "\n" old_content)
97
97
(String.concat "\n" new_content))
98
98
in
@@ -105,7 +105,7 @@
105
105
let new_content =
106
106
String.concat "\n" (List.init 2001 (fun i -> string_of_int (i + 1000)))
107
107
in
108
Removed:
let lines = line_diff old_content new_content in
108
Added:
let lines = of_contents old_content new_content in
109
109
let has_deletions = List.exists (fun l -> l.kind = Deletion) lines in
110
110
let has_additions = List.exists (fun l -> l.kind = Addition) lines in
111
111
Alcotest.(check bool) "has deletions" true has_deletions;