anil.recoil.org
OCaml HTML5 parser/serialiser based on Python's JustHTML
switch to a typed error variant
+1
-1
examples/error_handling.ml
+1
-1
examples/error_handling.ml
+4
-1
lib/html5rw/html5rw.ml
+4
-1
lib/html5rw/html5rw.ml
···
36
36
37
37
(** {1 Sub-modules} *)
38
38
39
+
(** Parse error code types *)
40
+
module Parse_error_code = Parse_error_code
41
+
39
42
(** DOM types and manipulation functions *)
40
43
module Dom = Dom
41
44
···
96
99
(** Get the namespace from a fragment context *)
97
100
let fragment_context_namespace = Parser.fragment_context_namespace
98
101
99
-
(** Get the error code string *)
102
+
(** Get the error code *)
100
103
let error_code = Parser.error_code
101
104
102
105
(** Get the line number of an error (1-indexed) *)
+23
-5
lib/html5rw/html5rw.mli
+23
-5
lib/html5rw/html5rw.mli
···
122
122
123
123
(** {1 Sub-modules} *)
124
124
125
+
(** Parse error code types.
126
+
127
+
This module provides the {!Parse_error_code.t} variant type that represents
128
+
all WHATWG-defined parse errors plus tree construction errors.
129
+
130
+
@see <https://html.spec.whatwg.org/multipage/parsing.html#parse-errors>
131
+
WHATWG: Parse errors *)
132
+
module Parse_error_code = Parse_error_code
133
+
125
134
(** DOM types and manipulation functions.
126
135
127
136
This module provides the core types for representing HTML documents as
···
334
343
WHATWG: Complete list of parse errors *)
335
344
type parse_error = Parser.parse_error
336
345
337
-
(** Get the error code string.
346
+
(** Get the error code.
347
+
348
+
Returns the {!Parse_error_code.t} variant representing this error.
349
+
This allows pattern matching on specific error types:
350
+
351
+
{[
352
+
match Html5rw.error_code err with
353
+
| Parse_error_code.Unexpected_null_character -> (* handle *)
354
+
| Parse_error_code.Eof_in_tag -> (* handle *)
355
+
| Parse_error_code.Tree_construction_error msg -> (* handle tree error *)
356
+
| _ -> (* other *)
357
+
]}
338
358
339
-
Error codes are lowercase with hyphens, matching the WHATWG specification
340
-
names. Examples: ["unexpected-null-character"], ["eof-in-tag"],
341
-
["missing-end-tag-name"].
359
+
Use {!Parse_error_code.to_string} to convert to a string representation.
342
360
343
361
@see <https://html.spec.whatwg.org/multipage/parsing.html#parse-errors>
344
362
WHATWG: Parse error codes *)
345
-
val error_code : parse_error -> string
363
+
val error_code : parse_error -> Parse_error_code.t
346
364
347
365
(** Get the line number where the error occurred (1-indexed).
348
366
+209
lib/html5rw/parse_error_code.ml
+209
lib/html5rw/parse_error_code.ml
···
1
+
(*---------------------------------------------------------------------------
2
+
Copyright (c) 2025 Anil Madhavapeddy <anil@recoil.org>. All rights reserved.
3
+
SPDX-License-Identifier: MIT
4
+
---------------------------------------------------------------------------*)
5
+
6
+
(** Parse error codes as defined by the WHATWG HTML5 specification.
7
+
8
+
@see <https://html.spec.whatwg.org/multipage/parsing.html#parse-errors>
9
+
*)
10
+
11
+
type t =
12
+
| Abrupt_closing_of_empty_comment
13
+
| Abrupt_doctype_public_identifier
14
+
| Abrupt_doctype_system_identifier
15
+
| Absence_of_digits_in_numeric_character_reference
16
+
| Cdata_in_html_content
17
+
| Character_reference_outside_unicode_range
18
+
| Control_character_in_input_stream
19
+
| Control_character_reference
20
+
| Duplicate_attribute
21
+
| End_tag_with_attributes
22
+
| End_tag_with_trailing_solidus
23
+
| Eof_before_tag_name
24
+
| Eof_in_cdata
25
+
| Eof_in_comment
26
+
| Eof_in_doctype
27
+
| Eof_in_script_html_comment_like_text
28
+
| Eof_in_tag
29
+
| Incorrectly_closed_comment
30
+
| Incorrectly_opened_comment
31
+
| Invalid_character_sequence_after_doctype_name
32
+
| Invalid_first_character_of_tag_name
33
+
| Missing_attribute_value
34
+
| Missing_doctype_name
35
+
| Missing_doctype_public_identifier
36
+
| Missing_doctype_system_identifier
37
+
| Missing_end_tag_name
38
+
| Missing_quote_before_doctype_public_identifier
39
+
| Missing_quote_before_doctype_system_identifier
40
+
| Missing_semicolon_after_character_reference
41
+
| Missing_whitespace_after_doctype_public_keyword
42
+
| Missing_whitespace_after_doctype_system_keyword
43
+
| Missing_whitespace_before_doctype_name
44
+
| Missing_whitespace_between_attributes
45
+
| Missing_whitespace_between_doctype_public_and_system_identifiers
46
+
| Nested_comment
47
+
| Noncharacter_character_reference
48
+
| Noncharacter_in_input_stream
49
+
| Non_void_html_element_start_tag_with_trailing_solidus
50
+
| Null_character_reference
51
+
| Surrogate_character_reference
52
+
| Surrogate_in_input_stream
53
+
| Unexpected_character_after_doctype_system_identifier
54
+
| Unexpected_character_in_attribute_name
55
+
| Unexpected_character_in_unquoted_attribute_value
56
+
| Unexpected_equals_sign_before_attribute_name
57
+
| Unexpected_null_character
58
+
| Unexpected_question_mark_instead_of_tag_name
59
+
| Unexpected_solidus_in_tag
60
+
| Unknown_named_character_reference
61
+
| Tree_construction_error of string
62
+
63
+
let to_string = function
64
+
| Abrupt_closing_of_empty_comment -> "abrupt-closing-of-empty-comment"
65
+
| Abrupt_doctype_public_identifier -> "abrupt-doctype-public-identifier"
66
+
| Abrupt_doctype_system_identifier -> "abrupt-doctype-system-identifier"
67
+
| Absence_of_digits_in_numeric_character_reference ->
68
+
"absence-of-digits-in-numeric-character-reference"
69
+
| Cdata_in_html_content -> "cdata-in-html-content"
70
+
| Character_reference_outside_unicode_range ->
71
+
"character-reference-outside-unicode-range"
72
+
| Control_character_in_input_stream -> "control-character-in-input-stream"
73
+
| Control_character_reference -> "control-character-reference"
74
+
| Duplicate_attribute -> "duplicate-attribute"
75
+
| End_tag_with_attributes -> "end-tag-with-attributes"
76
+
| End_tag_with_trailing_solidus -> "end-tag-with-trailing-solidus"
77
+
| Eof_before_tag_name -> "eof-before-tag-name"
78
+
| Eof_in_cdata -> "eof-in-cdata"
79
+
| Eof_in_comment -> "eof-in-comment"
80
+
| Eof_in_doctype -> "eof-in-doctype"
81
+
| Eof_in_script_html_comment_like_text ->
82
+
"eof-in-script-html-comment-like-text"
83
+
| Eof_in_tag -> "eof-in-tag"
84
+
| Incorrectly_closed_comment -> "incorrectly-closed-comment"
85
+
| Incorrectly_opened_comment -> "incorrectly-opened-comment"
86
+
| Invalid_character_sequence_after_doctype_name ->
87
+
"invalid-character-sequence-after-doctype-name"
88
+
| Invalid_first_character_of_tag_name ->
89
+
"invalid-first-character-of-tag-name"
90
+
| Missing_attribute_value -> "missing-attribute-value"
91
+
| Missing_doctype_name -> "missing-doctype-name"
92
+
| Missing_doctype_public_identifier -> "missing-doctype-public-identifier"
93
+
| Missing_doctype_system_identifier -> "missing-doctype-system-identifier"
94
+
| Missing_end_tag_name -> "missing-end-tag-name"
95
+
| Missing_quote_before_doctype_public_identifier ->
96
+
"missing-quote-before-doctype-public-identifier"
97
+
| Missing_quote_before_doctype_system_identifier ->
98
+
"missing-quote-before-doctype-system-identifier"
99
+
| Missing_semicolon_after_character_reference ->
100
+
"missing-semicolon-after-character-reference"
101
+
| Missing_whitespace_after_doctype_public_keyword ->
102
+
"missing-whitespace-after-doctype-public-keyword"
103
+
| Missing_whitespace_after_doctype_system_keyword ->
104
+
"missing-whitespace-after-doctype-system-keyword"
105
+
| Missing_whitespace_before_doctype_name ->
106
+
"missing-whitespace-before-doctype-name"
107
+
| Missing_whitespace_between_attributes ->
108
+
"missing-whitespace-between-attributes"
109
+
| Missing_whitespace_between_doctype_public_and_system_identifiers ->
110
+
"missing-whitespace-between-doctype-public-and-system-identifiers"
111
+
| Nested_comment -> "nested-comment"
112
+
| Noncharacter_character_reference -> "noncharacter-character-reference"
113
+
| Noncharacter_in_input_stream -> "noncharacter-in-input-stream"
114
+
| Non_void_html_element_start_tag_with_trailing_solidus ->
115
+
"non-void-html-element-start-tag-with-trailing-solidus"
116
+
| Null_character_reference -> "null-character-reference"
117
+
| Surrogate_character_reference -> "surrogate-character-reference"
118
+
| Surrogate_in_input_stream -> "surrogate-in-input-stream"
119
+
| Unexpected_character_after_doctype_system_identifier ->
120
+
"unexpected-character-after-doctype-system-identifier"
121
+
| Unexpected_character_in_attribute_name ->
122
+
"unexpected-character-in-attribute-name"
123
+
| Unexpected_character_in_unquoted_attribute_value ->
124
+
"unexpected-character-in-unquoted-attribute-value"
125
+
| Unexpected_equals_sign_before_attribute_name ->
126
+
"unexpected-equals-sign-before-attribute-name"
127
+
| Unexpected_null_character -> "unexpected-null-character"
128
+
| Unexpected_question_mark_instead_of_tag_name ->
129
+
"unexpected-question-mark-instead-of-tag-name"
130
+
| Unexpected_solidus_in_tag -> "unexpected-solidus-in-tag"
131
+
| Unknown_named_character_reference -> "unknown-named-character-reference"
132
+
| Tree_construction_error s -> s
133
+
134
+
let of_string = function
135
+
| "abrupt-closing-of-empty-comment" -> Abrupt_closing_of_empty_comment
136
+
| "abrupt-doctype-public-identifier" -> Abrupt_doctype_public_identifier
137
+
| "abrupt-doctype-system-identifier" -> Abrupt_doctype_system_identifier
138
+
| "absence-of-digits-in-numeric-character-reference" ->
139
+
Absence_of_digits_in_numeric_character_reference
140
+
| "cdata-in-html-content" -> Cdata_in_html_content
141
+
| "character-reference-outside-unicode-range" ->
142
+
Character_reference_outside_unicode_range
143
+
| "control-character-in-input-stream" -> Control_character_in_input_stream
144
+
| "control-character-reference" -> Control_character_reference
145
+
| "duplicate-attribute" -> Duplicate_attribute
146
+
| "end-tag-with-attributes" -> End_tag_with_attributes
147
+
| "end-tag-with-trailing-solidus" -> End_tag_with_trailing_solidus
148
+
| "eof-before-tag-name" -> Eof_before_tag_name
149
+
| "eof-in-cdata" -> Eof_in_cdata
150
+
| "eof-in-comment" -> Eof_in_comment
151
+
| "eof-in-doctype" -> Eof_in_doctype
152
+
| "eof-in-script-html-comment-like-text" ->
153
+
Eof_in_script_html_comment_like_text
154
+
| "eof-in-tag" -> Eof_in_tag
155
+
| "incorrectly-closed-comment" -> Incorrectly_closed_comment
156
+
| "incorrectly-opened-comment" -> Incorrectly_opened_comment
157
+
| "invalid-character-sequence-after-doctype-name" ->
158
+
Invalid_character_sequence_after_doctype_name
159
+
| "invalid-first-character-of-tag-name" ->
160
+
Invalid_first_character_of_tag_name
161
+
| "missing-attribute-value" -> Missing_attribute_value
162
+
| "missing-doctype-name" -> Missing_doctype_name
163
+
| "missing-doctype-public-identifier" -> Missing_doctype_public_identifier
164
+
| "missing-doctype-system-identifier" -> Missing_doctype_system_identifier
165
+
| "missing-end-tag-name" -> Missing_end_tag_name
166
+
| "missing-quote-before-doctype-public-identifier" ->
167
+
Missing_quote_before_doctype_public_identifier
168
+
| "missing-quote-before-doctype-system-identifier" ->
169
+
Missing_quote_before_doctype_system_identifier
170
+
| "missing-semicolon-after-character-reference" ->
171
+
Missing_semicolon_after_character_reference
172
+
| "missing-whitespace-after-doctype-public-keyword" ->
173
+
Missing_whitespace_after_doctype_public_keyword
174
+
| "missing-whitespace-after-doctype-system-keyword" ->
175
+
Missing_whitespace_after_doctype_system_keyword
176
+
| "missing-whitespace-before-doctype-name" ->
177
+
Missing_whitespace_before_doctype_name
178
+
| "missing-whitespace-between-attributes" ->
179
+
Missing_whitespace_between_attributes
180
+
| "missing-whitespace-between-doctype-public-and-system-identifiers" ->
181
+
Missing_whitespace_between_doctype_public_and_system_identifiers
182
+
| "nested-comment" -> Nested_comment
183
+
| "noncharacter-character-reference" -> Noncharacter_character_reference
184
+
| "noncharacter-in-input-stream" -> Noncharacter_in_input_stream
185
+
| "non-void-html-element-start-tag-with-trailing-solidus" ->
186
+
Non_void_html_element_start_tag_with_trailing_solidus
187
+
| "null-character-reference" -> Null_character_reference
188
+
| "surrogate-character-reference" -> Surrogate_character_reference
189
+
| "surrogate-in-input-stream" -> Surrogate_in_input_stream
190
+
| "unexpected-character-after-doctype-system-identifier" ->
191
+
Unexpected_character_after_doctype_system_identifier
192
+
| "unexpected-character-in-attribute-name" ->
193
+
Unexpected_character_in_attribute_name
194
+
| "unexpected-character-in-unquoted-attribute-value" ->
195
+
Unexpected_character_in_unquoted_attribute_value
196
+
| "unexpected-equals-sign-before-attribute-name" ->
197
+
Unexpected_equals_sign_before_attribute_name
198
+
| "unexpected-null-character" -> Unexpected_null_character
199
+
| "unexpected-question-mark-instead-of-tag-name" ->
200
+
Unexpected_question_mark_instead_of_tag_name
201
+
| "unexpected-solidus-in-tag" -> Unexpected_solidus_in_tag
202
+
| "unknown-named-character-reference" -> Unknown_named_character_reference
203
+
| s -> Tree_construction_error s
204
+
205
+
let of_string_opt s = Some (of_string s)
206
+
207
+
let is_whatwg_standard = function
208
+
| Tree_construction_error _ -> false
209
+
| _ -> true
+330
lib/html5rw/parse_error_code.mli
+330
lib/html5rw/parse_error_code.mli
···
1
+
(*---------------------------------------------------------------------------
2
+
Copyright (c) 2025 Anil Madhavapeddy <anil@recoil.org>. All rights reserved.
3
+
SPDX-License-Identifier: MIT
4
+
---------------------------------------------------------------------------*)
5
+
6
+
(** Parse error codes as defined by the WHATWG HTML5 specification.
7
+
8
+
The HTML5 parser never fails - it always produces a DOM tree. However,
9
+
the specification defines these error codes for conformance checkers to
10
+
report issues in HTML documents.
11
+
12
+
Each error code corresponds to a specific condition in the WHATWG
13
+
specification's parsing algorithm.
14
+
15
+
@see <https://html.spec.whatwg.org/multipage/parsing.html#parse-errors>
16
+
WHATWG: Parse errors *)
17
+
18
+
type t =
19
+
| Abrupt_closing_of_empty_comment
20
+
(** Parser encounters [<!-->] or [<!--->]; comment is treated as
21
+
correctly closed.
22
+
23
+
@see <https://html.spec.whatwg.org/multipage/parsing.html#parse-error-abrupt-closing-of-empty-comment> *)
24
+
25
+
| Abrupt_doctype_public_identifier
26
+
(** [>] found in DOCTYPE public identifier before closing quote;
27
+
sets document to quirks mode.
28
+
29
+
@see <https://html.spec.whatwg.org/multipage/parsing.html#parse-error-abrupt-doctype-public-identifier> *)
30
+
31
+
| Abrupt_doctype_system_identifier
32
+
(** [>] found in DOCTYPE system identifier before closing quote;
33
+
sets document to quirks mode.
34
+
35
+
@see <https://html.spec.whatwg.org/multipage/parsing.html#parse-error-abrupt-doctype-system-identifier> *)
36
+
37
+
| Absence_of_digits_in_numeric_character_reference
38
+
(** Numeric character reference has no digits (e.g., [&#qux;]);
39
+
the reference is not resolved.
40
+
41
+
@see <https://html.spec.whatwg.org/multipage/parsing.html#parse-error-absence-of-digits-in-numeric-character-reference> *)
42
+
43
+
| Cdata_in_html_content
44
+
(** CDATA section found outside SVG or MathML foreign content;
45
+
treated as a bogus comment.
46
+
47
+
@see <https://html.spec.whatwg.org/multipage/parsing.html#parse-error-cdata-in-html-content> *)
48
+
49
+
| Character_reference_outside_unicode_range
50
+
(** Numeric reference exceeds U+10FFFF; resolves to U+FFFD
51
+
REPLACEMENT CHARACTER.
52
+
53
+
@see <https://html.spec.whatwg.org/multipage/parsing.html#parse-error-character-reference-outside-unicode-range> *)
54
+
55
+
| Control_character_in_input_stream
56
+
(** Control code point (other than ASCII whitespace or NULL)
57
+
appears in the input; parsed as-is.
58
+
59
+
@see <https://html.spec.whatwg.org/multipage/parsing.html#parse-error-control-character-in-input-stream> *)
60
+
61
+
| Control_character_reference
62
+
(** Numeric reference to a control character; handled per
63
+
specification replacement rules.
64
+
65
+
@see <https://html.spec.whatwg.org/multipage/parsing.html#parse-error-control-character-reference> *)
66
+
67
+
| Duplicate_attribute
68
+
(** Tag contains duplicate attribute names; later duplicates
69
+
are removed.
70
+
71
+
@see <https://html.spec.whatwg.org/multipage/parsing.html#parse-error-duplicate-attribute> *)
72
+
73
+
| End_tag_with_attributes
74
+
(** End tag includes attributes; attributes are ignored.
75
+
76
+
@see <https://html.spec.whatwg.org/multipage/parsing.html#parse-error-end-tag-with-attributes> *)
77
+
78
+
| End_tag_with_trailing_solidus
79
+
(** End tag has [/] before [>] (like [</br/>]); treated as
80
+
regular end tag.
81
+
82
+
@see <https://html.spec.whatwg.org/multipage/parsing.html#parse-error-end-tag-with-trailing-solidus> *)
83
+
84
+
| Eof_before_tag_name
85
+
(** End of input where tag name expected; [<] or [</] is
86
+
treated as text.
87
+
88
+
@see <https://html.spec.whatwg.org/multipage/parsing.html#parse-error-eof-before-tag-name> *)
89
+
90
+
| Eof_in_cdata
91
+
(** End of input within CDATA section; treated as immediately closed.
92
+
93
+
@see <https://html.spec.whatwg.org/multipage/parsing.html#parse-error-eof-in-cdata> *)
94
+
95
+
| Eof_in_comment
96
+
(** End of input within comment; comment is treated as
97
+
immediately closed.
98
+
99
+
@see <https://html.spec.whatwg.org/multipage/parsing.html#parse-error-eof-in-comment> *)
100
+
101
+
| Eof_in_doctype
102
+
(** End of input within DOCTYPE; sets document to quirks mode.
103
+
104
+
@see <https://html.spec.whatwg.org/multipage/parsing.html#parse-error-eof-in-doctype> *)
105
+
106
+
| Eof_in_script_html_comment_like_text
107
+
(** End of input within HTML-like comment syntax inside a script element.
108
+
109
+
@see <https://html.spec.whatwg.org/multipage/parsing.html#parse-error-eof-in-script-html-comment-like-text> *)
110
+
111
+
| Eof_in_tag
112
+
(** End of input within a start or end tag; the tag is ignored.
113
+
114
+
@see <https://html.spec.whatwg.org/multipage/parsing.html#parse-error-eof-in-tag> *)
115
+
116
+
| Incorrectly_closed_comment
117
+
(** Comment closed by [--!>] instead of [-->]; treated as
118
+
correctly closed.
119
+
120
+
@see <https://html.spec.whatwg.org/multipage/parsing.html#parse-error-incorrectly-closed-comment> *)
121
+
122
+
| Incorrectly_opened_comment
123
+
(** [<!] not followed by [--] (e.g., [<!ELEMENT]); content is
124
+
treated as a bogus comment.
125
+
126
+
@see <https://html.spec.whatwg.org/multipage/parsing.html#parse-error-incorrectly-opened-comment> *)
127
+
128
+
| Invalid_character_sequence_after_doctype_name
129
+
(** Neither "PUBLIC" nor "SYSTEM" after DOCTYPE name; sets
130
+
document to quirks mode.
131
+
132
+
@see <https://html.spec.whatwg.org/multipage/parsing.html#parse-error-invalid-character-sequence-after-doctype-name> *)
133
+
134
+
| Invalid_first_character_of_tag_name
135
+
(** Non-ASCII-alpha character where tag name start expected;
136
+
[<] is treated as text.
137
+
138
+
@see <https://html.spec.whatwg.org/multipage/parsing.html#parse-error-invalid-first-character-of-tag-name> *)
139
+
140
+
| Missing_attribute_value
141
+
(** [>] where attribute value expected (e.g., [<div id=>]);
142
+
attribute gets empty string value.
143
+
144
+
@see <https://html.spec.whatwg.org/multipage/parsing.html#parse-error-missing-attribute-value> *)
145
+
146
+
| Missing_doctype_name
147
+
(** DOCTYPE has no name; sets document to quirks mode.
148
+
149
+
@see <https://html.spec.whatwg.org/multipage/parsing.html#parse-error-missing-doctype-name> *)
150
+
151
+
| Missing_doctype_public_identifier
152
+
(** [>] where public identifier expected; sets quirks mode.
153
+
154
+
@see <https://html.spec.whatwg.org/multipage/parsing.html#parse-error-missing-doctype-public-identifier> *)
155
+
156
+
| Missing_doctype_system_identifier
157
+
(** [>] where system identifier expected; sets quirks mode.
158
+
159
+
@see <https://html.spec.whatwg.org/multipage/parsing.html#parse-error-missing-doctype-system-identifier> *)
160
+
161
+
| Missing_end_tag_name
162
+
(** [>] where end tag name expected ([</>]); sequence is ignored.
163
+
164
+
@see <https://html.spec.whatwg.org/multipage/parsing.html#parse-error-missing-end-tag-name> *)
165
+
166
+
| Missing_quote_before_doctype_public_identifier
167
+
(** Public identifier lacks preceding quote; sets quirks mode.
168
+
169
+
@see <https://html.spec.whatwg.org/multipage/parsing.html#parse-error-missing-quote-before-doctype-public-identifier> *)
170
+
171
+
| Missing_quote_before_doctype_system_identifier
172
+
(** System identifier lacks preceding quote; sets quirks mode.
173
+
174
+
@see <https://html.spec.whatwg.org/multipage/parsing.html#parse-error-missing-quote-before-doctype-system-identifier> *)
175
+
176
+
| Missing_semicolon_after_character_reference
177
+
(** Character reference lacks terminating [;]; behaves as if
178
+
semicolon were present.
179
+
180
+
@see <https://html.spec.whatwg.org/multipage/parsing.html#parse-error-missing-semicolon-after-character-reference> *)
181
+
182
+
| Missing_whitespace_after_doctype_public_keyword
183
+
(** No whitespace between "PUBLIC" and identifier; treated as
184
+
if whitespace were present.
185
+
186
+
@see <https://html.spec.whatwg.org/multipage/parsing.html#parse-error-missing-whitespace-after-doctype-public-keyword> *)
187
+
188
+
| Missing_whitespace_after_doctype_system_keyword
189
+
(** No whitespace between "SYSTEM" and identifier; treated as
190
+
if whitespace were present.
191
+
192
+
@see <https://html.spec.whatwg.org/multipage/parsing.html#parse-error-missing-whitespace-after-doctype-system-keyword> *)
193
+
194
+
| Missing_whitespace_before_doctype_name
195
+
(** No whitespace between "DOCTYPE" and name; treated as if
196
+
whitespace were present.
197
+
198
+
@see <https://html.spec.whatwg.org/multipage/parsing.html#parse-error-missing-whitespace-before-doctype-name> *)
199
+
200
+
| Missing_whitespace_between_attributes
201
+
(** Adjacent attributes lack separating whitespace; treated as
202
+
if whitespace were present.
203
+
204
+
@see <https://html.spec.whatwg.org/multipage/parsing.html#parse-error-missing-whitespace-between-attributes> *)
205
+
206
+
| Missing_whitespace_between_doctype_public_and_system_identifiers
207
+
(** Public and system identifiers not separated by whitespace;
208
+
treated as if whitespace were present.
209
+
210
+
@see <https://html.spec.whatwg.org/multipage/parsing.html#parse-error-missing-whitespace-between-doctype-public-and-system-identifiers> *)
211
+
212
+
| Nested_comment
213
+
(** Nested [<!--] detected within comment; comment still closes
214
+
at first [-->].
215
+
216
+
@see <https://html.spec.whatwg.org/multipage/parsing.html#parse-error-nested-comment> *)
217
+
218
+
| Noncharacter_character_reference
219
+
(** Numeric reference to a Unicode noncharacter; resolved as-is
220
+
(not replaced).
221
+
222
+
@see <https://html.spec.whatwg.org/multipage/parsing.html#parse-error-noncharacter-character-reference> *)
223
+
224
+
| Noncharacter_in_input_stream
225
+
(** Unicode noncharacter code point in input; parsed as-is.
226
+
227
+
@see <https://html.spec.whatwg.org/multipage/parsing.html#parse-error-noncharacter-in-input-stream> *)
228
+
229
+
| Non_void_html_element_start_tag_with_trailing_solidus
230
+
(** Non-void element start tag has [/] before [>] (like
231
+
[<div/>]); the [/] is ignored.
232
+
233
+
@see <https://html.spec.whatwg.org/multipage/parsing.html#parse-error-non-void-html-element-start-tag-with-trailing-solidus> *)
234
+
235
+
| Null_character_reference
236
+
(** Numeric reference to U+0000 (NULL); resolves to U+FFFD
237
+
REPLACEMENT CHARACTER.
238
+
239
+
@see <https://html.spec.whatwg.org/multipage/parsing.html#parse-error-null-character-reference> *)
240
+
241
+
| Surrogate_character_reference
242
+
(** Numeric reference to a surrogate code point (U+D800-U+DFFF);
243
+
resolves to U+FFFD REPLACEMENT CHARACTER.
244
+
245
+
@see <https://html.spec.whatwg.org/multipage/parsing.html#parse-error-surrogate-character-reference> *)
246
+
247
+
| Surrogate_in_input_stream
248
+
(** Surrogate code point in input stream; parsed as-is.
249
+
250
+
@see <https://html.spec.whatwg.org/multipage/parsing.html#parse-error-surrogate-in-input-stream> *)
251
+
252
+
| Unexpected_character_after_doctype_system_identifier
253
+
(** Non-whitespace/non-[>] character after system identifier;
254
+
the character is ignored.
255
+
256
+
@see <https://html.spec.whatwg.org/multipage/parsing.html#parse-error-unexpected-character-after-doctype-system-identifier> *)
257
+
258
+
| Unexpected_character_in_attribute_name
259
+
(** Double quote, single quote, or less-than sign in attribute name;
260
+
included in the attribute name.
261
+
262
+
@see <https://html.spec.whatwg.org/multipage/parsing.html#parse-error-unexpected-character-in-attribute-name> *)
263
+
264
+
| Unexpected_character_in_unquoted_attribute_value
265
+
(** Double quote, equals sign, backtick, or less-than sign in
266
+
unquoted attribute value; included in the value.
267
+
268
+
@see <https://html.spec.whatwg.org/multipage/parsing.html#parse-error-unexpected-character-in-unquoted-attribute-value> *)
269
+
270
+
| Unexpected_equals_sign_before_attribute_name
271
+
(** [=] where attribute name expected; treated as first
272
+
character of attribute name.
273
+
274
+
@see <https://html.spec.whatwg.org/multipage/parsing.html#parse-error-unexpected-equals-sign-before-attribute-name> *)
275
+
276
+
| Unexpected_null_character
277
+
(** U+0000 (NULL) in various positions; ignored or replaced
278
+
with U+FFFD depending on context.
279
+
280
+
@see <https://html.spec.whatwg.org/multipage/parsing.html#parse-error-unexpected-null-character> *)
281
+
282
+
| Unexpected_question_mark_instead_of_tag_name
283
+
(** [?] where tag name expected (like [<?xml]); treated as
284
+
start of bogus comment.
285
+
286
+
@see <https://html.spec.whatwg.org/multipage/parsing.html#parse-error-unexpected-question-mark-instead-of-tag-name> *)
287
+
288
+
| Unexpected_solidus_in_tag
289
+
(** [/] in tag not immediately before [>]; treated as
290
+
whitespace.
291
+
292
+
@see <https://html.spec.whatwg.org/multipage/parsing.html#parse-error-unexpected-solidus-in-tag> *)
293
+
294
+
| Unknown_named_character_reference
295
+
(** Ambiguous ampersand: [&] followed by characters that don't
296
+
match any named reference; not resolved as reference.
297
+
298
+
@see <https://html.spec.whatwg.org/multipage/parsing.html#parse-error-unknown-named-character-reference> *)
299
+
300
+
| Tree_construction_error of string
301
+
(** Tree construction error not defined in the WHATWG specification.
302
+
303
+
These are informative errors produced during tree construction
304
+
to indicate various issues like unexpected tags, missing closing
305
+
tags, etc. The string contains a descriptive error code. *)
306
+
307
+
val to_string : t -> string
308
+
(** Convert an error code to its WHATWG specification string representation.
309
+
310
+
The returned string is lowercase with hyphens, matching the WHATWG
311
+
specification naming convention. For example:
312
+
- [Abrupt_closing_of_empty_comment] becomes ["abrupt-closing-of-empty-comment"]
313
+
- [Eof_in_tag] becomes ["eof-in-tag"] *)
314
+
315
+
val of_string : string -> t
316
+
(** Parse an error code from its WHATWG specification string representation.
317
+
318
+
If the string matches a known WHATWG error code, returns that variant.
319
+
Otherwise, returns [Tree_construction_error s]. *)
320
+
321
+
val of_string_opt : string -> t option
322
+
(** Parse an error code from its WHATWG specification string representation.
323
+
324
+
Always returns [Some code]. For unrecognized strings, returns
325
+
[Some (Tree_construction_error s)]. *)
326
+
327
+
val is_whatwg_standard : t -> bool
328
+
(** Check if an error code is defined in the WHATWG specification.
329
+
330
+
Returns [false] for [Tree_construction_error _], [true] for all others. *)
+1
lib/html5rw/parser/parser.ml
+1
lib/html5rw/parser/parser.ml
+23
-9
lib/html5rw/parser/parser.mli
+23
-9
lib/html5rw/parser/parser.mli
···
117
117
(** DOM types and manipulation. *)
118
118
module Dom = Dom
119
119
120
+
(** Parse error code types.
121
+
122
+
This module provides the {!Parse_error_code.t} variant type that represents
123
+
all WHATWG-defined parse errors plus tree construction errors.
124
+
125
+
@see <https://html.spec.whatwg.org/multipage/parsing.html#parse-errors>
126
+
WHATWG: Parse errors *)
127
+
module Parse_error_code = Parse_error_code
128
+
120
129
(** HTML5 tokenizer.
121
130
122
131
The tokenizer implements the first stage of HTML5 parsing, converting
···
242
251
WHATWG: Complete list of parse errors *)
243
252
type parse_error
244
253
245
-
(** Get the error code string.
254
+
(** Get the error code.
255
+
256
+
Returns the {!Parse_error_code.t} variant representing this error.
257
+
This allows pattern matching on specific error types:
258
+
259
+
{[
260
+
match Parser.error_code err with
261
+
| Parse_error_code.Unexpected_null_character -> (* handle *)
262
+
| Parse_error_code.Eof_in_tag -> (* handle *)
263
+
| Parse_error_code.Tree_construction_error msg -> (* handle tree error *)
264
+
| _ -> (* other *)
265
+
]}
246
266
247
-
Error codes are lowercase with hyphens, exactly matching the WHATWG
248
-
specification naming. Examples:
249
-
- ["unexpected-null-character"]
250
-
- ["eof-before-tag-name"]
251
-
- ["missing-end-tag-name"]
252
-
- ["duplicate-attribute"]
253
-
- ["missing-doctype"]
267
+
Use {!Parse_error_code.to_string} to convert to a string representation.
254
268
255
269
@see <https://html.spec.whatwg.org/multipage/parsing.html#parse-errors>
256
270
WHATWG: Parse error codes *)
257
-
val error_code : parse_error -> string
271
+
val error_code : parse_error -> Parse_error_code.t
258
272
259
273
(** Get the line number where the error occurred.
260
274
+2
-2
lib/html5rw/parser/parser_tree_builder.ml
+2
-2
lib/html5rw/parser/parser_tree_builder.ml
···
18
18
}
19
19
20
20
type parse_error = {
21
-
code : string;
21
+
code : Parse_error_code.t;
22
22
line : int;
23
23
column : int;
24
24
}
···
113
113
(* Error handling *)
114
114
let parse_error t code =
115
115
if t.collect_errors then
116
-
t.errors <- { code; line = 0; column = 0 } :: t.errors
116
+
t.errors <- { code = Parse_error_code.of_string code; line = 0; column = 0 } :: t.errors
117
117
118
118
(* Stack helpers *)
119
119
let current_node t =
+7
-1
lib/html5rw/tokenizer/tokenizer.mli
+7
-1
lib/html5rw/tokenizer/tokenizer.mli
···
140
140
(** Parse error types. *)
141
141
module Errors : sig
142
142
type t = Tokenizer_errors.t = {
143
-
code : string;
143
+
code : Parse_error_code.t;
144
144
line : int;
145
145
column : int;
146
146
}
147
147
148
148
val make : code:string -> line:int -> column:int -> t
149
+
(** Create an error from a string code. The string is converted to
150
+
{!Parse_error_code.t} using {!Parse_error_code.of_string}. *)
151
+
152
+
val make_with_code : code:Parse_error_code.t -> line:int -> column:int -> t
153
+
(** Create an error with a typed error code. *)
154
+
149
155
val to_string : t -> string
150
156
end
151
157
+7
-3
lib/html5rw/tokenizer/tokenizer_errors.ml
+7
-3
lib/html5rw/tokenizer/tokenizer_errors.ml
···
1
1
(* HTML5 parse error types *)
2
2
3
3
type t = {
4
-
code : string;
4
+
code : Parse_error_code.t;
5
5
line : int;
6
6
column : int;
7
7
}
8
8
9
-
let make ~code ~line ~column = { code; line; column }
9
+
let make ~code ~line ~column =
10
+
{ code = Parse_error_code.of_string code; line; column }
11
+
12
+
let make_with_code ~code ~line ~column = { code; line; column }
10
13
11
14
let to_string err =
12
-
Printf.sprintf "(%d,%d): %s" err.line err.column err.code
15
+
Printf.sprintf "(%d,%d): %s" err.line err.column
16
+
(Parse_error_code.to_string err.code)