anil.recoil.org
+1
-1
bin/html5check/dune
+1
-1
bin/html5check/dune
+11
-11
bin/html5check/html5check.ml
+11
-11
bin/html5check/html5check.ml
···
34
34
(** Format output based on the requested format *)
35
35
let format_output format result =
36
36
match format with
37
-
| `Text -> Html5_checker.format_text result
38
-
| `Json -> Html5_checker.format_json result
39
-
| `Gnu -> Html5_checker.format_gnu result
37
+
| `Text -> Htmlrw_check.to_text result
38
+
| `Json -> Htmlrw_check.to_json result
39
+
| `Gnu -> Htmlrw_check.to_gnu result
40
40
41
41
(** Run the validation *)
42
42
let run format errors_only exit_zero quiet verbose file =
···
46
46
Exit_code.io_error
47
47
| Ok (reader, ic, system_id) ->
48
48
(* Run validation *)
49
-
let result = Html5_checker.check ~system_id reader in
49
+
let result = Htmlrw_check.check ~system_id reader in
50
50
51
51
(* Close input if it's not stdin *)
52
52
if file <> "-" then close_in ic;
53
53
54
54
(* Get messages based on filtering *)
55
55
let messages =
56
-
if errors_only then Html5_checker.errors result
57
-
else Html5_checker.messages result
56
+
if errors_only then Htmlrw_check.errors result
57
+
else Htmlrw_check.messages result
58
58
in
59
59
60
60
(* Output based on mode *)
61
61
if quiet then begin
62
62
(* Only show counts *)
63
-
let error_count = List.length (Html5_checker.errors result) in
64
-
let warning_count = List.length (Html5_checker.warnings result) in
63
+
let error_count = List.length (Htmlrw_check.errors result) in
64
+
let warning_count = List.length (Htmlrw_check.warnings result) in
65
65
if errors_only then
66
66
Printf.printf "%d error%s\n" error_count (if error_count = 1 then "" else "s")
67
67
else
···
75
75
76
76
(* Show summary if verbose *)
77
77
if verbose && messages <> [] then begin
78
-
let error_count = List.length (Html5_checker.errors result) in
79
-
let warning_count = List.length (Html5_checker.warnings result) in
78
+
let error_count = List.length (Htmlrw_check.errors result) in
79
+
let warning_count = List.length (Htmlrw_check.warnings result) in
80
80
Printf.eprintf "\nSummary: %d error%s, %d warning%s\n"
81
81
error_count (if error_count = 1 then "" else "s")
82
82
warning_count (if warning_count = 1 then "" else "s")
···
84
84
end;
85
85
86
86
(* Determine exit code *)
87
-
if exit_zero || not (Html5_checker.has_errors result) then
87
+
if exit_zero || not (Htmlrw_check.has_errors result) then
88
88
Exit_code.ok
89
89
else
90
90
Exit_code.validation_errors
+5
-5
test/analyze_failures.ml
+5
-5
test/analyze_failures.ml
···
36
36
close_in ic;
37
37
38
38
let reader = Bytesrw.Bytes.Reader.of_string content in
39
-
let result = Html5_checker.check ~collect_parse_errors:true reader in
40
-
let errors = Html5_checker.errors result in
41
-
let warnings = Html5_checker.warnings result in
39
+
let result = Htmlrw_check.check ~collect_parse_errors:true reader in
40
+
let errors = Htmlrw_check.errors result in
41
+
let warnings = Htmlrw_check.warnings result in
42
42
43
43
let should_print = match mode with
44
44
| "isvalid" -> outcome = Valid && (errors <> [] || warnings <> []) && !count < 60
···
49
49
if mode = "isvalid" then begin
50
50
if errors <> [] then begin
51
51
Printf.printf "ERRORS:\n";
52
-
List.iter (fun e -> Printf.printf " %s\n" e.Html5_checker.Message.message) errors
52
+
List.iter (fun e -> Printf.printf " %s\n" e.Htmlrw_check.text) errors
53
53
end;
54
54
if warnings <> [] then begin
55
55
Printf.printf "WARNINGS:\n";
56
-
List.iter (fun w -> Printf.printf " %s\n" w.Html5_checker.Message.message) warnings
56
+
List.iter (fun w -> Printf.printf " %s\n" w.Htmlrw_check.text) warnings
57
57
end
58
58
end;
59
59
print_endline content;
+5
-5
test/debug_check.ml
+5
-5
test/debug_check.ml
···
27
27
print_node "" root;
28
28
print_endline "\n=== Checking... ===";
29
29
let reader2 = Bytesrw.Bytes.Reader.of_string html in
30
-
let result = Html5_checker.check ~collect_parse_errors:true ~system_id:test_file reader2 in
31
-
let errors = Html5_checker.errors result in
32
-
let warnings = Html5_checker.warnings result in
30
+
let result = Htmlrw_check.check ~collect_parse_errors:true ~system_id:test_file reader2 in
31
+
let errors = Htmlrw_check.errors result in
32
+
let warnings = Htmlrw_check.warnings result in
33
33
print_endline "=== Errors ===";
34
-
List.iter (fun e -> print_endline e.Html5_checker.Message.message) errors;
34
+
List.iter (fun e -> print_endline e.Htmlrw_check.text) errors;
35
35
print_endline "\n=== Warnings ===";
36
-
List.iter (fun e -> print_endline e.Html5_checker.Message.message) warnings;
36
+
List.iter (fun e -> print_endline e.Htmlrw_check.text) warnings;
37
37
print_endline "\n=== Expected ===";
38
38
print_endline "Consider adding a \xe2\x80\x9clang\xe2\x80\x9d attribute to the \xe2\x80\x9chtml\xe2\x80\x9d start tag to declare the language of this document."
+8
-8
test/debug_validator.ml
+8
-8
test/debug_validator.ml
···
15
15
Printf.printf "Input (%d bytes):\n%s\n\n" (String.length content) content;
16
16
17
17
let reader = Bytesrw.Bytes.Reader.of_string content in
18
-
let result = Html5_checker.check ~collect_parse_errors:true ~system_id:path reader in
18
+
let result = Htmlrw_check.check ~collect_parse_errors:true ~system_id:path reader in
19
19
20
-
let errors = Html5_checker.errors result in
21
-
let warnings = Html5_checker.warnings result in
20
+
let errors = Htmlrw_check.errors result in
21
+
let warnings = Htmlrw_check.warnings result in
22
22
23
23
Printf.printf "=== Results ===\n";
24
24
Printf.printf "Errors: %d\n" (List.length errors);
25
25
List.iter (fun msg ->
26
-
Printf.printf " [ERROR] %s\n" msg.Html5_checker.Message.message;
27
-
(match msg.Html5_checker.Message.location with
26
+
Printf.printf " [ERROR] %s\n" msg.Htmlrw_check.text;
27
+
(match msg.Htmlrw_check.location with
28
28
| Some loc -> Printf.printf " at line %d, col %d\n" loc.line loc.column
29
29
| None -> ())
30
30
) errors;
31
31
32
32
Printf.printf "Warnings: %d\n" (List.length warnings);
33
33
List.iter (fun msg ->
34
-
Printf.printf " [WARN] %s\n" msg.Html5_checker.Message.message;
35
-
(match msg.Html5_checker.Message.location with
34
+
Printf.printf " [WARN] %s\n" msg.Htmlrw_check.text;
35
+
(match msg.Htmlrw_check.location with
36
36
| Some loc -> Printf.printf " at line %d, col %d\n" loc.line loc.column
37
37
| None -> ())
38
38
) warnings;
39
39
40
40
Printf.printf "\n=== Formatted Output ===\n";
41
-
Printf.printf "%s\n" (Html5_checker.format_text result)
41
+
Printf.printf "%s\n" (Htmlrw_check.to_text result)
+8
-8
test/dune
+8
-8
test/dune
···
58
58
(executable
59
59
(name test_nesting_checker)
60
60
(modules test_nesting_checker)
61
-
(libraries html5rw.checker))
61
+
(libraries html5rw.check))
62
62
63
63
(executable
64
64
(name test_html5_checker)
65
65
(modules test_html5_checker)
66
-
(libraries bytesrw html5rw html5rw.checker str))
66
+
(libraries bytesrw html5rw html5rw.check str))
67
67
68
68
(rule
69
69
(alias runtest)
···
78
78
(library
79
79
(name expected_message)
80
80
(modules expected_message)
81
-
(libraries html5rw.checker str jsont jsont.bytesrw))
81
+
(libraries html5rw.check str jsont jsont.bytesrw))
82
82
83
83
(executable
84
84
(name test_validator)
85
85
(modules test_validator)
86
-
(libraries bytesrw html5rw html5rw.checker str jsont jsont.bytesrw test_report validator_messages expected_message))
86
+
(libraries bytesrw html5rw html5rw.check str jsont jsont.bytesrw test_report validator_messages expected_message))
87
87
88
88
(executable
89
89
(name debug_validator)
90
90
(modules debug_validator)
91
-
(libraries bytesrw html5rw html5rw.checker))
91
+
(libraries bytesrw html5rw html5rw.check))
92
92
93
93
(executable
94
94
(name analyze_failures)
95
95
(modules analyze_failures)
96
-
(libraries bytesrw html5rw html5rw.checker))
96
+
(libraries bytesrw html5rw html5rw.check))
97
97
98
98
(executable
99
99
(name debug_check)
100
100
(modules debug_check)
101
-
(libraries html5rw.checker bytesrw))
101
+
(libraries html5rw.check bytesrw))
102
102
103
103
(executable
104
104
(name test_roundtrip)
105
105
(modules test_roundtrip)
106
-
(libraries bytesrw html5rw html5rw.checker astring test_report))
106
+
(libraries bytesrw html5rw html5rw.check astring test_report))
+26
-26
test/expected_message.ml
+26
-26
test/expected_message.ml
···
2
2
3
3
type t = {
4
4
message: string;
5
-
error_code: Html5_checker.Error_code.t option;
5
+
error_code: Htmlrw_check.Error_code.t option;
6
6
line: int option;
7
7
column: int option;
8
8
element: string option;
···
80
80
if Str.string_match re msg 0 then
81
81
let child = Str.matched_group 1 msg in
82
82
let parent = Str.matched_group 2 msg in
83
-
Some ((`Element (`Not_allowed_as_child (`Child child, `Parent parent)) : Html5_checker.Error_code.t),
83
+
Some ((`Element (`Not_allowed_as_child (`Child child, `Parent parent)) : Htmlrw_check.Error_code.t),
84
84
Some child, None)
85
85
else None
86
86
···
90
90
if Str.string_match re msg 0 then
91
91
let attr = Str.matched_group 1 msg in
92
92
let element = Str.matched_group 2 msg in
93
-
Some ((`Attr (`Not_allowed (`Attr attr, `Elem element)) : Html5_checker.Error_code.t),
93
+
Some ((`Attr (`Not_allowed (`Attr attr, `Elem element)) : Htmlrw_check.Error_code.t),
94
94
Some element, Some attr)
95
95
else None
96
96
···
99
99
let re = Str.regexp {|Attribute "\([^"]+\)" not allowed here|} in
100
100
if Str.string_match re msg 0 then
101
101
let attr = Str.matched_group 1 msg in
102
-
Some ((`Attr (`Not_allowed_here (`Attr attr)) : Html5_checker.Error_code.t),
102
+
Some ((`Attr (`Not_allowed_here (`Attr attr)) : Htmlrw_check.Error_code.t),
103
103
None, Some attr)
104
104
else None
105
105
···
109
109
if Str.string_match re msg 0 then
110
110
let element = Str.matched_group 1 msg in
111
111
let attr = Str.matched_group 2 msg in
112
-
Some ((`Attr (`Missing (`Elem element, `Attr attr)) : Html5_checker.Error_code.t),
112
+
Some ((`Attr (`Missing (`Elem element, `Attr attr)) : Htmlrw_check.Error_code.t),
113
113
Some element, Some attr)
114
114
else None
115
115
···
119
119
if Str.string_match re msg 0 then
120
120
let parent = Str.matched_group 1 msg in
121
121
let child = Str.matched_group 2 msg in
122
-
Some ((`Element (`Missing_child (`Parent parent, `Child child)) : Html5_checker.Error_code.t),
122
+
Some ((`Element (`Missing_child (`Parent parent, `Child child)) : Htmlrw_check.Error_code.t),
123
123
Some parent, None)
124
124
else None
125
125
···
128
128
let re = Str.regexp {|Duplicate ID "\([^"]+\)"|} in
129
129
if Str.string_match re msg 0 then
130
130
let id = Str.matched_group 1 msg in
131
-
Some ((`Attr (`Duplicate_id (`Id id)) : Html5_checker.Error_code.t),
131
+
Some ((`Attr (`Duplicate_id (`Id id)) : Htmlrw_check.Error_code.t),
132
132
None, None)
133
133
else None
134
134
···
137
137
let re = Str.regexp {|The "\([^"]+\)" element is obsolete|} in
138
138
if Str.string_match re msg 0 then
139
139
let element = Str.matched_group 1 msg in
140
-
Some ((`Element (`Obsolete (`Elem element, `Suggestion "")) : Html5_checker.Error_code.t),
140
+
Some ((`Element (`Obsolete (`Elem element, `Suggestion "")) : Htmlrw_check.Error_code.t),
141
141
Some element, None)
142
142
else None
143
143
···
147
147
if Str.string_match re msg 0 then
148
148
let attr = Str.matched_group 1 msg in
149
149
let element = Str.matched_group 2 msg in
150
-
Some ((`Element (`Obsolete_attr (`Elem element, `Attr attr, `Suggestion None)) : Html5_checker.Error_code.t),
150
+
Some ((`Element (`Obsolete_attr (`Elem element, `Attr attr, `Suggestion None)) : Htmlrw_check.Error_code.t),
151
151
Some element, Some attr)
152
152
else None
153
153
···
156
156
let re = Str.regexp {|Stray end tag "\([^"]+\)"|} in
157
157
if Str.string_match re msg 0 then
158
158
let tag = Str.matched_group 1 msg in
159
-
Some ((`Tag (`Stray_end (`Tag tag)) : Html5_checker.Error_code.t),
159
+
Some ((`Tag (`Stray_end (`Tag tag)) : Htmlrw_check.Error_code.t),
160
160
Some tag, None)
161
161
else None
162
162
···
165
165
let re = Str.regexp {|Stray start tag "\([^"]+\)"|} in
166
166
if Str.string_match re msg 0 then
167
167
let tag = Str.matched_group 1 msg in
168
-
Some ((`Tag (`Stray_start (`Tag tag)) : Html5_checker.Error_code.t),
168
+
Some ((`Tag (`Stray_start (`Tag tag)) : Htmlrw_check.Error_code.t),
169
169
Some tag, None)
170
170
else None
171
171
···
175
175
if Str.string_match re msg 0 then
176
176
let role = Str.matched_group 1 msg in
177
177
let reason = Str.matched_group 2 msg in
178
-
Some ((`Aria (`Unnecessary_role (`Role role, `Elem "", `Reason reason)) : Html5_checker.Error_code.t),
178
+
Some ((`Aria (`Unnecessary_role (`Role role, `Elem "", `Reason reason)) : Htmlrw_check.Error_code.t),
179
179
None, None)
180
180
else None
181
181
···
185
185
if Str.string_match re msg 0 then
186
186
let role = Str.matched_group 1 msg in
187
187
let element = Str.matched_group 2 msg in
188
-
Some ((`Aria (`Bad_role (`Elem element, `Role role)) : Html5_checker.Error_code.t),
188
+
Some ((`Aria (`Bad_role (`Elem element, `Role role)) : Htmlrw_check.Error_code.t),
189
189
Some element, Some "role")
190
190
else None
191
191
···
196
196
let attr = Str.matched_group 1 msg in
197
197
let element = Str.matched_group 2 msg in
198
198
let condition = Str.matched_group 3 msg in
199
-
Some ((`Aria (`Must_not_specify (`Attr attr, `Elem element, `Condition condition)) : Html5_checker.Error_code.t),
199
+
Some ((`Aria (`Must_not_specify (`Attr attr, `Elem element, `Condition condition)) : Htmlrw_check.Error_code.t),
200
200
Some element, Some attr)
201
201
else None
202
202
···
207
207
let attr = Str.matched_group 1 msg in
208
208
let element = Str.matched_group 2 msg in
209
209
let condition = Str.matched_group 3 msg in
210
-
Some ((`Aria (`Must_not_use (`Attr attr, `Elem element, `Condition condition)) : Html5_checker.Error_code.t),
210
+
Some ((`Aria (`Must_not_use (`Attr attr, `Elem element, `Condition condition)) : Htmlrw_check.Error_code.t),
211
211
Some element, Some attr)
212
212
else None
213
213
···
225
225
String.trim (String.sub msg (colon_pos + 1) (String.length msg - colon_pos - 1))
226
226
with Not_found -> ""
227
227
in
228
-
Some ((`Attr (`Bad_value (`Elem element, `Attr attr, `Value value, `Reason reason)) : Html5_checker.Error_code.t),
228
+
Some ((`Attr (`Bad_value (`Elem element, `Attr attr, `Value value, `Reason reason)) : Htmlrw_check.Error_code.t),
229
229
Some element, Some attr)
230
230
else None
231
231
···
234
234
let re = Str.regexp {|End tag "\([^"]+\)" implied, but there were open elements|} in
235
235
if Str.string_match re msg 0 then
236
236
let tag = Str.matched_group 1 msg in
237
-
Some ((`Tag (`End_implied_open (`Tag tag)) : Html5_checker.Error_code.t),
237
+
Some ((`Tag (`End_implied_open (`Tag tag)) : Htmlrw_check.Error_code.t),
238
238
Some tag, None)
239
239
else None
240
240
···
243
243
let re = Str.regexp {|No "\([^"]+\)" element in scope but a "\([^"]+\)" end tag seen|} in
244
244
if Str.string_match re msg 0 then
245
245
let tag = Str.matched_group 1 msg in
246
-
Some ((`Tag (`Not_in_scope (`Tag tag)) : Html5_checker.Error_code.t),
246
+
Some ((`Tag (`Not_in_scope (`Tag tag)) : Htmlrw_check.Error_code.t),
247
247
Some tag, None)
248
248
else None
249
249
···
252
252
let re = Str.regexp {|Start tag "\([^"]+\)" seen in "table"|} in
253
253
if Str.string_match re msg 0 then
254
254
let tag = Str.matched_group 1 msg in
255
-
Some ((`Tag (`Start_in_table (`Tag tag)) : Html5_checker.Error_code.t),
255
+
Some ((`Tag (`Start_in_table (`Tag tag)) : Htmlrw_check.Error_code.t),
256
256
Some tag, None)
257
257
else None
258
258
···
335
335
336
336
let matches ~strictness ~expected ~actual =
337
337
let expected_norm = normalize_quotes expected.message in
338
-
let actual_norm = normalize_quotes actual.Html5_checker.Message.message in
338
+
let actual_norm = normalize_quotes actual.Htmlrw_check.text in
339
339
340
340
(* Check severity match *)
341
341
let severity_matches =
342
-
match (expected.severity, actual.Html5_checker.Message.severity) with
342
+
match (expected.severity, actual.Htmlrw_check.severity) with
343
343
| (None, _) -> true
344
-
| (Some `Error, Html5_checker.Message.Error) -> true
345
-
| (Some `Warning, Html5_checker.Message.Warning) -> true
346
-
| (Some `Info, Html5_checker.Message.Info) -> true
344
+
| (Some `Error, Htmlrw_check.Error) -> true
345
+
| (Some `Warning, Htmlrw_check.Warning) -> true
346
+
| (Some `Info, Htmlrw_check.Info) -> true
347
347
| _ -> false
348
348
in
349
349
350
350
(* Check location match *)
351
351
let location_matches =
352
-
match (expected.line, expected.column, actual.Html5_checker.Message.location) with
352
+
match (expected.line, expected.column, actual.Htmlrw_check.location) with
353
353
| (None, None, _) -> true
354
354
| (Some el, Some ec, Some loc) -> loc.line = el && loc.column = ec
355
355
| (Some el, None, Some loc) -> loc.line = el
···
358
358
359
359
(* Check error code match *)
360
360
let code_matches =
361
-
match (expected.error_code, actual.Html5_checker.Message.error_code) with
361
+
match (expected.error_code, actual.Htmlrw_check.error_code) with
362
362
| (None, _) -> true (* No expected code to match *)
363
363
| (Some ec, Some ac) -> error_codes_match ec ac
364
364
| (Some _, None) -> false (* Expected typed but got untyped *)
+2
-2
test/expected_message.mli
+2
-2
test/expected_message.mli
···
6
6
(** Structured expected message *)
7
7
type t = {
8
8
message: string; (** Full message text *)
9
-
error_code: Html5_checker.Error_code.t option; (** Parsed typed code *)
9
+
error_code: Htmlrw_check.Error_code.t option; (** Parsed typed code *)
10
10
line: int option; (** Expected line number *)
11
11
column: int option; (** Expected column number *)
12
12
element: string option; (** Element context *)
···
60
60
61
61
(** Check if actual message matches expected.
62
62
Returns the quality of match achieved. *)
63
-
val matches : strictness:strictness -> expected:t -> actual:Html5_checker.Message.t -> match_quality
63
+
val matches : strictness:strictness -> expected:t -> actual:Htmlrw_check.message -> match_quality
64
64
65
65
(** Check if match quality is acceptable given strictness *)
66
66
val is_acceptable : strictness:strictness -> match_quality -> bool
+47
-47
test/test_html5_checker.ml
+47
-47
test/test_html5_checker.ml
···
5
5
6
6
(** Helper to check if a message contains a substring *)
7
7
let message_contains msg substring =
8
-
String.lowercase_ascii msg.Html5_checker.Message.message
8
+
String.lowercase_ascii msg.Htmlrw_check.text
9
9
|> fun s -> String.length s >= String.length substring &&
10
10
try
11
11
ignore (Str.search_forward (Str.regexp_case_fold (Str.quote substring)) s 0);
···
21
21
<body><p>Hello world</p></body>
22
22
</html>|} in
23
23
let reader = reader_of_string html in
24
-
let result = Html5_checker.check reader in
25
-
let errors = Html5_checker.errors result in
24
+
let result = Htmlrw_check.check reader in
25
+
let errors = Htmlrw_check.errors result in
26
26
Printf.printf " Found %d error(s)\n" (List.length errors);
27
27
if List.length errors > 0 then begin
28
28
List.iter (fun msg ->
29
-
Printf.printf " - %s\n" msg.Html5_checker.Message.message
29
+
Printf.printf " - %s\n" msg.Htmlrw_check.text
30
30
) errors;
31
31
end else
32
32
Printf.printf " OK: No errors as expected\n"
···
36
36
Printf.printf "\nTest 2: Missing DOCTYPE\n";
37
37
let html = "<html><body>Hello</body></html>" in
38
38
let reader = reader_of_string html in
39
-
let result = Html5_checker.check reader in
40
-
let errors = Html5_checker.errors result in
39
+
let result = Htmlrw_check.check reader in
40
+
let errors = Htmlrw_check.errors result in
41
41
Printf.printf " Found %d error(s)\n" (List.length errors);
42
42
if List.length errors = 0 then
43
43
Printf.printf " Warning: Expected parse errors for missing DOCTYPE\n"
44
44
else begin
45
45
List.iter (fun msg ->
46
-
Printf.printf " - %s\n" msg.Html5_checker.Message.message
46
+
Printf.printf " - %s\n" msg.Htmlrw_check.text
47
47
) errors;
48
48
end
49
49
···
52
52
Printf.printf "\nTest 3: Obsolete <center> element\n";
53
53
let html = "<!DOCTYPE html><html><body><center>Centered</center></body></html>" in
54
54
let reader = reader_of_string html in
55
-
let result = Html5_checker.check reader in
56
-
let all_msgs = Html5_checker.messages result in
55
+
let result = Htmlrw_check.check reader in
56
+
let all_msgs = Htmlrw_check.messages result in
57
57
Printf.printf " Found %d message(s)\n" (List.length all_msgs);
58
58
let obsolete_msgs = List.filter (fun m ->
59
59
message_contains m "obsolete" || message_contains m "center"
···
61
61
if List.length obsolete_msgs > 0 then begin
62
62
Printf.printf " Found obsolete-related messages:\n";
63
63
List.iter (fun msg ->
64
-
Printf.printf " - %s\n" msg.Html5_checker.Message.message
64
+
Printf.printf " - %s\n" msg.Htmlrw_check.text
65
65
) obsolete_msgs;
66
66
end else
67
67
Printf.printf " Note: No obsolete element warnings found (checker may not be enabled)\n"
···
74
74
<div id="foo">Second</div>
75
75
</body></html>|} in
76
76
let reader = reader_of_string html in
77
-
let result = Html5_checker.check reader in
78
-
let all_msgs = Html5_checker.messages result in
77
+
let result = Htmlrw_check.check reader in
78
+
let all_msgs = Htmlrw_check.messages result in
79
79
Printf.printf " Found %d message(s)\n" (List.length all_msgs);
80
80
let id_msgs = List.filter (fun m ->
81
81
message_contains m "duplicate" || message_contains m "id"
···
83
83
if List.length id_msgs > 0 then begin
84
84
Printf.printf " Found ID-related messages:\n";
85
85
List.iter (fun msg ->
86
-
Printf.printf " - %s\n" msg.Html5_checker.Message.message
86
+
Printf.printf " - %s\n" msg.Htmlrw_check.text
87
87
) id_msgs;
88
88
end else
89
89
Printf.printf " Note: No duplicate ID errors found (checker may not be enabled)\n"
···
96
96
<h3>Skipped h2</h3>
97
97
</body></html>|} in
98
98
let reader = reader_of_string html in
99
-
let result = Html5_checker.check reader in
100
-
let all_msgs = Html5_checker.messages result in
99
+
let result = Htmlrw_check.check reader in
100
+
let all_msgs = Htmlrw_check.messages result in
101
101
Printf.printf " Found %d message(s)\n" (List.length all_msgs);
102
102
let heading_msgs = List.filter (fun m ->
103
103
message_contains m "heading" || message_contains m "skip"
···
105
105
if List.length heading_msgs > 0 then begin
106
106
Printf.printf " Found heading-related messages:\n";
107
107
List.iter (fun msg ->
108
-
Printf.printf " - %s\n" msg.Html5_checker.Message.message
108
+
Printf.printf " - %s\n" msg.Htmlrw_check.text
109
109
) heading_msgs;
110
110
end else
111
111
Printf.printf " Note: No heading structure warnings found (checker may not be enabled)\n"
···
117
117
<img src="test.jpg">
118
118
</body></html>|} in
119
119
let reader = reader_of_string html in
120
-
let result = Html5_checker.check reader in
121
-
let all_msgs = Html5_checker.messages result in
120
+
let result = Htmlrw_check.check reader in
121
+
let all_msgs = Htmlrw_check.messages result in
122
122
Printf.printf " Found %d message(s)\n" (List.length all_msgs);
123
123
let img_msgs = List.filter (fun m ->
124
124
message_contains m "alt" || (message_contains m "img" && message_contains m "attribute")
···
126
126
if List.length img_msgs > 0 then begin
127
127
Printf.printf " Found img/alt-related messages:\n";
128
128
List.iter (fun msg ->
129
-
Printf.printf " - %s\n" msg.Html5_checker.Message.message
129
+
Printf.printf " - %s\n" msg.Htmlrw_check.text
130
130
) img_msgs;
131
131
end else
132
132
Printf.printf " Note: No missing alt attribute errors found (checker may not be enabled)\n"
···
138
138
<a href="#">Link <a href="#">Nested</a></a>
139
139
</body></html>|} in
140
140
let reader = reader_of_string html in
141
-
let result = Html5_checker.check reader in
142
-
let all_msgs = Html5_checker.messages result in
141
+
let result = Htmlrw_check.check reader in
142
+
let all_msgs = Htmlrw_check.messages result in
143
143
Printf.printf " Found %d message(s)\n" (List.length all_msgs);
144
144
let nesting_msgs = List.filter (fun m ->
145
145
message_contains m "nesting" || message_contains m "nested" || message_contains m "ancestor"
···
147
147
if List.length nesting_msgs > 0 then begin
148
148
Printf.printf " Found nesting-related messages:\n";
149
149
List.iter (fun msg ->
150
-
Printf.printf " - %s\n" msg.Html5_checker.Message.message
150
+
Printf.printf " - %s\n" msg.Htmlrw_check.text
151
151
) nesting_msgs;
152
152
end else
153
153
Printf.printf " Note: No nesting errors found (checker may not be enabled)\n"
···
159
159
<form><form></form></form>
160
160
</body></html>|} in
161
161
let reader = reader_of_string html in
162
-
let result = Html5_checker.check reader in
163
-
let all_msgs = Html5_checker.messages result in
162
+
let result = Htmlrw_check.check reader in
163
+
let all_msgs = Htmlrw_check.messages result in
164
164
Printf.printf " Found %d message(s)\n" (List.length all_msgs);
165
165
let form_msgs = List.filter (fun m ->
166
166
message_contains m "form"
···
168
168
if List.length form_msgs > 0 then begin
169
169
Printf.printf " Found form-related messages:\n";
170
170
List.iter (fun msg ->
171
-
Printf.printf " - %s\n" msg.Html5_checker.Message.message
171
+
Printf.printf " - %s\n" msg.Htmlrw_check.text
172
172
) form_msgs;
173
173
end else
174
174
Printf.printf " Note: No form nesting errors found (checker may not be enabled)\n"
···
178
178
Printf.printf "\nTest 9: Output format testing\n";
179
179
let html = {|<!DOCTYPE html><html><body><p>Test</p></body></html>|} in
180
180
let reader = reader_of_string html in
181
-
let result = Html5_checker.check reader in
181
+
let result = Htmlrw_check.check reader in
182
182
183
183
Printf.printf " Testing text format:\n";
184
-
let text_output = Html5_checker.format_text result in
184
+
let text_output = Htmlrw_check.to_text result in
185
185
Printf.printf " Length: %d chars\n" (String.length text_output);
186
186
187
187
Printf.printf " Testing JSON format:\n";
188
-
let json_output = Html5_checker.format_json result in
188
+
let json_output = Htmlrw_check.to_json result in
189
189
Printf.printf " Length: %d chars\n" (String.length json_output);
190
190
191
191
Printf.printf " Testing GNU format:\n";
192
-
let gnu_output = Html5_checker.format_gnu result in
192
+
let gnu_output = Htmlrw_check.to_gnu result in
193
193
Printf.printf " Length: %d chars\n" (String.length gnu_output)
194
194
195
195
(** Test has_errors function *)
···
198
198
199
199
(* Valid document should have no errors *)
200
200
let valid_html = "<!DOCTYPE html><html><body><p>Valid</p></body></html>" in
201
-
let result1 = Html5_checker.check (reader_of_string valid_html) in
202
-
Printf.printf " Valid document has_errors: %b\n" (Html5_checker.has_errors result1);
201
+
let result1 = Htmlrw_check.check (reader_of_string valid_html) in
202
+
Printf.printf " Valid document has_errors: %b\n" (Htmlrw_check.has_errors result1);
203
203
204
204
(* Document with likely parse errors *)
205
205
let invalid_html = "<html><body><p>Unclosed" in
206
-
let result2 = Html5_checker.check (reader_of_string invalid_html) in
207
-
Printf.printf " Invalid document has_errors: %b\n" (Html5_checker.has_errors result2)
206
+
let result2 = Htmlrw_check.check (reader_of_string invalid_html) in
207
+
Printf.printf " Invalid document has_errors: %b\n" (Htmlrw_check.has_errors result2)
208
208
209
209
(** Test check_dom with pre-parsed document *)
210
210
let test_check_dom () =
···
212
212
let html = "<!DOCTYPE html><html><body><p>Test</p></body></html>" in
213
213
let reader = reader_of_string html in
214
214
let parsed = Html5rw.parse reader in
215
-
let result = Html5_checker.check_dom parsed in
216
-
let all_msgs = Html5_checker.messages result in
215
+
let result = Htmlrw_check.check_parsed parsed in
216
+
let all_msgs = Htmlrw_check.messages result in
217
217
Printf.printf " check_dom found %d message(s)\n" (List.length all_msgs);
218
218
Printf.printf " OK: check_dom completed successfully\n"
219
219
···
222
222
Printf.printf "\nTest 12: system_id parameter\n";
223
223
let html = "<!DOCTYPE html><html><body><p>Test</p></body></html>" in
224
224
let reader = reader_of_string html in
225
-
let result = Html5_checker.check ~system_id:"test.html" reader in
226
-
match Html5_checker.system_id result with
225
+
let result = Htmlrw_check.check ~system_id:"test.html" reader in
226
+
match Htmlrw_check.system_id result with
227
227
| Some id -> Printf.printf " system_id: %s\n" id
228
228
| None -> Printf.printf " Warning: system_id not set\n"
229
229
···
232
232
Printf.printf "\nTest 13: collect_parse_errors flag\n";
233
233
let html = "<!DOCTYPE html><html><body><p>Test</p></body></html>" in
234
234
235
-
let result_with = Html5_checker.check ~collect_parse_errors:true (reader_of_string html) in
236
-
let msgs_with = Html5_checker.messages result_with in
235
+
let result_with = Htmlrw_check.check ~collect_parse_errors:true (reader_of_string html) in
236
+
let msgs_with = Htmlrw_check.messages result_with in
237
237
Printf.printf " With parse errors: %d message(s)\n" (List.length msgs_with);
238
238
239
-
let result_without = Html5_checker.check ~collect_parse_errors:false (reader_of_string html) in
240
-
let msgs_without = Html5_checker.messages result_without in
239
+
let result_without = Htmlrw_check.check ~collect_parse_errors:false (reader_of_string html) in
240
+
let msgs_without = Htmlrw_check.messages result_without in
241
241
Printf.printf " Without parse errors: %d message(s)\n" (List.length msgs_without)
242
242
243
243
(** Test document accessor *)
···
245
245
Printf.printf "\nTest 14: document accessor\n";
246
246
let html = "<!DOCTYPE html><html><body><p>Test</p></body></html>" in
247
247
let reader = reader_of_string html in
248
-
let result = Html5_checker.check reader in
249
-
let _doc = Html5_checker.document result in
248
+
let result = Htmlrw_check.check reader in
249
+
let _doc = Htmlrw_check.document result in
250
250
Printf.printf " OK: document accessor works\n"
251
251
252
252
(** Test message severity filtering *)
···
254
254
Printf.printf "\nTest 15: Message severity filtering\n";
255
255
let html = "<!DOCTYPE html><html><body><p>Test</p></body></html>" in
256
256
let reader = reader_of_string html in
257
-
let result = Html5_checker.check reader in
257
+
let result = Htmlrw_check.check reader in
258
258
259
-
let all_msgs = Html5_checker.messages result in
260
-
let errors = Html5_checker.errors result in
261
-
let warnings = Html5_checker.warnings result in
259
+
let all_msgs = Htmlrw_check.messages result in
260
+
let errors = Htmlrw_check.errors result in
261
+
let warnings = Htmlrw_check.warnings result in
262
262
263
263
Printf.printf " Total messages: %d\n" (List.length all_msgs);
264
264
Printf.printf " Errors: %d\n" (List.length errors);
+34
-92
test/test_nesting_checker.ml
+34
-92
test/test_nesting_checker.ml
···
1
-
(** Simple test for nesting_checker functionality *)
2
-
3
-
let () =
4
-
(* Create a message collector *)
5
-
let collector = Html5_checker.Message_collector.create () in
1
+
(** Test for nesting checker functionality via public API *)
6
2
7
-
(* Get the nesting checker *)
8
-
let module C = (val Html5_checker__Nesting_checker.checker : Html5_checker__Checker.S) in
9
-
let state = C.create () in
3
+
let check_html html =
4
+
let reader = Bytesrw.Bytes.Reader.of_string html in
5
+
Htmlrw_check.check reader
10
6
7
+
let () =
11
8
(* Test 1: <a> cannot contain another <a> *)
12
9
Printf.printf "Test 1: Checking <a href> inside <a href>\n";
13
-
C.start_element state ~name:"a" ~namespace:None ~attrs:[("href", "#")] collector;
14
-
C.start_element state ~name:"a" ~namespace:None ~attrs:[("href", "#")] collector;
15
-
16
-
let errors1 = Html5_checker.Message_collector.errors collector in
10
+
let result1 = check_html "<a href='#'><a href='#'>nested</a></a>" in
11
+
let errors1 = Htmlrw_check.errors result1 in
17
12
Printf.printf " Found %d error(s)\n" (List.length errors1);
18
13
List.iter (fun msg ->
19
-
Printf.printf " - %s\n" msg.Html5_checker.Message.message
14
+
Printf.printf " - %s\n" msg.Htmlrw_check.text
20
15
) errors1;
21
16
22
-
C.end_element state ~name:"a" ~namespace:None collector;
23
-
C.end_element state ~name:"a" ~namespace:None collector;
24
-
Html5_checker.Message_collector.clear collector;
25
-
26
17
(* Test 2: <button> inside <a> *)
27
18
Printf.printf "\nTest 2: Checking <button> inside <a href>\n";
28
-
C.start_element state ~name:"a" ~namespace:None ~attrs:[("href", "#")] collector;
29
-
C.start_element state ~name:"button" ~namespace:None ~attrs:[] collector;
30
-
31
-
let errors2 = Html5_checker.Message_collector.errors collector in
19
+
let result2 = check_html "<a href='#'><button>click</button></a>" in
20
+
let errors2 = Htmlrw_check.errors result2 in
32
21
Printf.printf " Found %d error(s)\n" (List.length errors2);
33
22
List.iter (fun msg ->
34
-
Printf.printf " - %s\n" msg.Html5_checker.Message.message
23
+
Printf.printf " - %s\n" msg.Htmlrw_check.text
35
24
) errors2;
36
25
37
-
C.end_element state ~name:"button" ~namespace:None collector;
38
-
C.end_element state ~name:"a" ~namespace:None collector;
39
-
Html5_checker.Message_collector.clear collector;
40
-
41
26
(* Test 3: form inside form *)
42
27
Printf.printf "\nTest 3: Checking <form> inside <form>\n";
43
-
C.start_element state ~name:"form" ~namespace:None ~attrs:[] collector;
44
-
C.start_element state ~name:"form" ~namespace:None ~attrs:[] collector;
45
-
46
-
let errors3 = Html5_checker.Message_collector.errors collector in
28
+
let result3 = check_html "<form><form>nested</form></form>" in
29
+
let errors3 = Htmlrw_check.errors result3 in
47
30
Printf.printf " Found %d error(s)\n" (List.length errors3);
48
31
List.iter (fun msg ->
49
-
Printf.printf " - %s\n" msg.Html5_checker.Message.message
32
+
Printf.printf " - %s\n" msg.Htmlrw_check.text
50
33
) errors3;
51
34
52
-
C.end_element state ~name:"form" ~namespace:None collector;
53
-
C.end_element state ~name:"form" ~namespace:None collector;
54
-
Html5_checker.Message_collector.clear collector;
55
-
56
-
(* Test 4: header inside footer *)
35
+
(* Test 4: header inside footer (should be allowed) *)
57
36
Printf.printf "\nTest 4: Checking <header> inside <footer>\n";
58
-
C.start_element state ~name:"footer" ~namespace:None ~attrs:[] collector;
59
-
C.start_element state ~name:"header" ~namespace:None ~attrs:[] collector;
60
-
61
-
let errors4 = Html5_checker.Message_collector.errors collector in
37
+
let result4 = check_html "<footer><header>test</header></footer>" in
38
+
let errors4 = Htmlrw_check.errors result4 in
62
39
Printf.printf " Found %d error(s)\n" (List.length errors4);
63
-
List.iter (fun msg ->
64
-
Printf.printf " - %s\n" msg.Html5_checker.Message.message
65
-
) errors4;
66
-
67
-
C.end_element state ~name:"header" ~namespace:None collector;
68
-
C.end_element state ~name:"footer" ~namespace:None collector;
69
-
Html5_checker.Message_collector.clear collector;
40
+
if List.length errors4 > 0 then
41
+
List.iter (fun msg ->
42
+
Printf.printf " - %s\n" msg.Htmlrw_check.text
43
+
) errors4
44
+
else
45
+
Printf.printf " OK: No errors (header inside footer is valid)\n";
70
46
71
-
(* Test 5: input (not hidden) inside button *)
47
+
(* Test 5: input inside button *)
72
48
Printf.printf "\nTest 5: Checking <input type=text> inside <button>\n";
73
-
C.start_element state ~name:"button" ~namespace:None ~attrs:[] collector;
74
-
C.start_element state ~name:"input" ~namespace:None ~attrs:[("type", "text")] collector;
75
-
76
-
let errors5 = Html5_checker.Message_collector.errors collector in
49
+
let result5 = check_html "<button><input type='text'></button>" in
50
+
let errors5 = Htmlrw_check.errors result5 in
77
51
Printf.printf " Found %d error(s)\n" (List.length errors5);
78
52
List.iter (fun msg ->
79
-
Printf.printf " - %s\n" msg.Html5_checker.Message.message
53
+
Printf.printf " - %s\n" msg.Htmlrw_check.text
80
54
) errors5;
81
55
82
-
C.end_element state ~name:"input" ~namespace:None collector;
83
-
C.end_element state ~name:"button" ~namespace:None collector;
84
-
Html5_checker.Message_collector.clear collector;
85
-
86
56
(* Test 6: valid nesting - should not error *)
87
57
Printf.printf "\nTest 6: Checking valid nesting: <div> inside <div>\n";
88
-
C.start_element state ~name:"div" ~namespace:None ~attrs:[] collector;
89
-
C.start_element state ~name:"div" ~namespace:None ~attrs:[] collector;
90
-
91
-
let errors6 = Html5_checker.Message_collector.errors collector in
58
+
let result6 = check_html "<div><div>nested</div></div>" in
59
+
let errors6 = Htmlrw_check.errors result6 in
92
60
Printf.printf " Found %d error(s)\n" (List.length errors6);
93
61
if List.length errors6 = 0 then
94
-
Printf.printf " OK: No errors as expected\n";
95
-
96
-
C.end_element state ~name:"div" ~namespace:None collector;
97
-
C.end_element state ~name:"div" ~namespace:None collector;
98
-
Html5_checker.Message_collector.clear collector;
99
-
100
-
(* Test 7: area without map ancestor *)
101
-
Printf.printf "\nTest 7: Checking <area> without <map> ancestor\n";
102
-
C.start_element state ~name:"area" ~namespace:None ~attrs:[] collector;
103
-
104
-
let errors7 = Html5_checker.Message_collector.errors collector in
105
-
Printf.printf " Found %d error(s)\n" (List.length errors7);
106
-
List.iter (fun msg ->
107
-
Printf.printf " - %s\n" msg.Html5_checker.Message.message
108
-
) errors7;
109
-
110
-
C.end_element state ~name:"area" ~namespace:None collector;
111
-
Html5_checker.Message_collector.clear collector;
112
-
113
-
(* Test 8: area with map ancestor (valid) *)
114
-
Printf.printf "\nTest 8: Checking <area> with <map> ancestor (valid)\n";
115
-
C.start_element state ~name:"map" ~namespace:None ~attrs:[] collector;
116
-
C.start_element state ~name:"area" ~namespace:None ~attrs:[] collector;
117
-
118
-
let errors8 = Html5_checker.Message_collector.errors collector in
119
-
Printf.printf " Found %d error(s)\n" (List.length errors8);
120
-
if List.length errors8 = 0 then
121
-
Printf.printf " OK: No errors as expected\n";
122
-
123
-
C.end_element state ~name:"area" ~namespace:None collector;
124
-
C.end_element state ~name:"map" ~namespace:None collector;
62
+
Printf.printf " OK: No errors as expected\n"
63
+
else
64
+
List.iter (fun msg ->
65
+
Printf.printf " - %s\n" msg.Htmlrw_check.text
66
+
) errors6;
125
67
126
68
Printf.printf "\nAll tests completed!\n"
+3
-3
test/test_nfc_debug.ml
+3
-3
test/test_nfc_debug.ml
···
3
3
In_channel.input_all ic
4
4
) in
5
5
let reader = Bytesrw.Bytes.Reader.of_string content in
6
-
let result = Html5_checker.check ~system_id:"test.html" reader in
7
-
let warnings = Html5_checker.warnings result in
6
+
let result = Htmlrw_check.check ~system_id:"test.html" reader in
7
+
let warnings = Htmlrw_check.warnings result in
8
8
Printf.printf "Total warnings: %d\n" (List.length warnings);
9
9
List.iter (fun msg ->
10
-
Printf.printf "WARNING: %s\n" (Html5_checker.Message.message msg)
10
+
Printf.printf "WARNING: %s\n" msg.Htmlrw_check.text
11
11
) warnings
+6
-6
test/test_roundtrip.ml
+6
-6
test/test_roundtrip.ml
···
37
37
38
38
(* Count errors in validation result *)
39
39
let count_errors messages =
40
-
List.length (List.filter (fun (m : Html5_checker.Message.t) ->
41
-
m.severity = Html5_checker.Message.Error
40
+
List.length (List.filter (fun (m : Htmlrw_check.message) ->
41
+
m.severity = Htmlrw_check.Error
42
42
) messages)
43
43
44
44
(* Serialize a document to HTML string *)
···
65
65
let original_doc = Html5rw.root original_result in
66
66
67
67
(* Validate original *)
68
-
let checker_result = Html5_checker.check_dom ~system_id:path original_result in
69
-
let original_messages = Html5_checker.messages checker_result in
68
+
let checker_result = Htmlrw_check.check_parsed ~system_id:path original_result in
69
+
let original_messages = Htmlrw_check.messages checker_result in
70
70
let original_errors = count_errors original_messages in
71
71
let original_valid = original_errors = 0 in
72
72
···
77
77
let roundtrip_result = Html5rw.parse_bytes (Bytes.of_string serialized) in
78
78
79
79
(* Validate roundtripped document *)
80
-
let roundtrip_checker = Html5_checker.check_dom ~system_id:path roundtrip_result in
81
-
let roundtrip_messages = Html5_checker.messages roundtrip_checker in
80
+
let roundtrip_checker = Htmlrw_check.check_parsed ~system_id:path roundtrip_result in
81
+
let roundtrip_messages = Htmlrw_check.messages roundtrip_checker in
82
82
let roundtrip_errors = count_errors roundtrip_messages in
83
83
let roundtrip_valid = roundtrip_errors = 0 in
84
84
+7
-7
test/test_validator.ml
+7
-7
test/test_validator.ml
···
105
105
close_in ic;
106
106
107
107
let reader = Bytesrw.Bytes.Reader.of_string content in
108
-
let result = Html5_checker.check ~collect_parse_errors:true ~system_id:test.relative_path reader in
108
+
let result = Htmlrw_check.check ~collect_parse_errors:true ~system_id:test.relative_path reader in
109
109
110
110
(* Keep full message objects for proper matching *)
111
-
let error_msgs = Html5_checker.errors result in
112
-
let warning_msgs = Html5_checker.warnings result in
113
-
let info_msgs = Html5_checker.infos result in
111
+
let error_msgs = Htmlrw_check.errors result in
112
+
let warning_msgs = Htmlrw_check.warnings result in
113
+
let info_msgs = Htmlrw_check.infos result in
114
114
115
115
(* Extract text for reporting *)
116
-
let errors = List.map (fun m -> m.Html5_checker.Message.message) error_msgs in
117
-
let warnings = List.map (fun m -> m.Html5_checker.Message.message) warning_msgs in
118
-
let infos = List.map (fun m -> m.Html5_checker.Message.message) info_msgs in
116
+
let errors = List.map (fun m -> m.Htmlrw_check.text) error_msgs in
117
+
let warnings = List.map (fun m -> m.Htmlrw_check.text) warning_msgs in
118
+
let infos = List.map (fun m -> m.Htmlrw_check.text) info_msgs in
119
119
let expected_msg = Validator_messages.get messages test.relative_path in
120
120
121
121
let (passed, match_quality, details) = match test.expected with