html-to-markdown 3.14.3 → 3.15.0
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- checksums.yaml +4 -4
- data/ext/html_to_markdown_rb/native/Cargo.lock +293 -19
- data/ext/html_to_markdown_rb/native/Cargo.toml +4 -4
- data/ext/html_to_markdown_rb/src/lib.rs +212 -39
- data/lib/html_to_markdown/native.rb +24 -47
- data/lib/html_to_markdown/version.rb +2 -2
- data/sig/types.rbs +65 -90
- metadata +2 -2
|
@@ -1,5 +1,5 @@
|
|
|
1
1
|
# This file is auto-generated by alef — DO NOT EDIT.
|
|
2
|
-
# alef:hash:
|
|
2
|
+
# alef:hash:b035df027077810826d9d805f32ef7ac142e025791c1098a52026f8500fe0fbc
|
|
3
3
|
# To regenerate: alef generate
|
|
4
4
|
# To verify freshness: alef verify
|
|
5
5
|
# frozen_string_literal: true
|
|
@@ -354,15 +354,13 @@ module HtmlToMarkdown
|
|
|
354
354
|
# written in the source. Callers that need an absolute URL must resolve it against the
|
|
355
355
|
# document base URL themselves.
|
|
356
356
|
sig { returns(String) }
|
|
357
|
-
|
|
358
|
-
def url = super
|
|
357
|
+
def url = Data.instance_method(:to_h).bind_call(self).fetch(:url)
|
|
359
358
|
# The `title` attribute of the `<a>` element, if present.
|
|
360
359
|
#
|
|
361
360
|
# `None` when the `<a>` tag has no `title="..."` attribute. When present, the value
|
|
362
361
|
# is copied verbatim — HTML entities within the title are not decoded.
|
|
363
362
|
sig { returns(T.nilable(String)) }
|
|
364
|
-
|
|
365
|
-
def title = super
|
|
363
|
+
def title = Data.instance_method(:to_h).bind_call(self).fetch(:title)
|
|
366
364
|
sig { returns(T::Boolean) }
|
|
367
365
|
def bold? = false
|
|
368
366
|
sig { returns(T::Boolean) }
|
|
@@ -445,12 +443,10 @@ module HtmlToMarkdown
|
|
|
445
443
|
|
|
446
444
|
# Heading level (1-6).
|
|
447
445
|
sig { returns(Integer) }
|
|
448
|
-
|
|
449
|
-
def level = super
|
|
446
|
+
def level = Data.instance_method(:to_h).bind_call(self).fetch(:level)
|
|
450
447
|
# The heading text content.
|
|
451
448
|
sig { returns(String) }
|
|
452
|
-
|
|
453
|
-
def text = super
|
|
449
|
+
def text = Data.instance_method(:to_h).bind_call(self).fetch(:text)
|
|
454
450
|
sig { returns(T::Boolean) }
|
|
455
451
|
def heading? = true
|
|
456
452
|
sig { returns(T::Boolean) }
|
|
@@ -491,8 +487,7 @@ module HtmlToMarkdown
|
|
|
491
487
|
|
|
492
488
|
# The paragraph text content.
|
|
493
489
|
sig { returns(String) }
|
|
494
|
-
|
|
495
|
-
def text = super
|
|
490
|
+
def text = Data.instance_method(:to_h).bind_call(self).fetch(:text)
|
|
496
491
|
sig { returns(T::Boolean) }
|
|
497
492
|
def heading? = false
|
|
498
493
|
sig { returns(T::Boolean) }
|
|
@@ -533,8 +528,7 @@ module HtmlToMarkdown
|
|
|
533
528
|
|
|
534
529
|
# Whether this is an ordered list.
|
|
535
530
|
sig { returns(T::Boolean) }
|
|
536
|
-
|
|
537
|
-
def ordered = super
|
|
531
|
+
def ordered = Data.instance_method(:to_h).bind_call(self).fetch(:ordered)
|
|
538
532
|
sig { returns(T::Boolean) }
|
|
539
533
|
def heading? = false
|
|
540
534
|
sig { returns(T::Boolean) }
|
|
@@ -575,8 +569,7 @@ module HtmlToMarkdown
|
|
|
575
569
|
|
|
576
570
|
# The list item text content.
|
|
577
571
|
sig { returns(String) }
|
|
578
|
-
|
|
579
|
-
def text = super
|
|
572
|
+
def text = Data.instance_method(:to_h).bind_call(self).fetch(:text)
|
|
580
573
|
sig { returns(T::Boolean) }
|
|
581
574
|
def heading? = false
|
|
582
575
|
sig { returns(T::Boolean) }
|
|
@@ -617,8 +610,7 @@ module HtmlToMarkdown
|
|
|
617
610
|
|
|
618
611
|
# The table grid structure.
|
|
619
612
|
sig { returns(TableGrid) }
|
|
620
|
-
|
|
621
|
-
def grid = super
|
|
613
|
+
def grid = Data.instance_method(:to_h).bind_call(self).fetch(:grid)
|
|
622
614
|
sig { returns(T::Boolean) }
|
|
623
615
|
def heading? = false
|
|
624
616
|
sig { returns(T::Boolean) }
|
|
@@ -659,16 +651,13 @@ module HtmlToMarkdown
|
|
|
659
651
|
|
|
660
652
|
# Alt text or caption.
|
|
661
653
|
sig { returns(T.nilable(String)) }
|
|
662
|
-
|
|
663
|
-
def description = super
|
|
654
|
+
def description = Data.instance_method(:to_h).bind_call(self).fetch(:description)
|
|
664
655
|
# Image source URL.
|
|
665
656
|
sig { returns(T.nilable(String)) }
|
|
666
|
-
|
|
667
|
-
def src = super
|
|
657
|
+
def src = Data.instance_method(:to_h).bind_call(self).fetch(:src)
|
|
668
658
|
# Index into `ConversionResult.images` when image extraction is enabled.
|
|
669
659
|
sig { returns(T.nilable(Integer)) }
|
|
670
|
-
|
|
671
|
-
def image_index = super
|
|
660
|
+
def image_index = Data.instance_method(:to_h).bind_call(self).fetch(:image_index)
|
|
672
661
|
sig { returns(T::Boolean) }
|
|
673
662
|
def heading? = false
|
|
674
663
|
sig { returns(T::Boolean) }
|
|
@@ -713,12 +702,10 @@ module HtmlToMarkdown
|
|
|
713
702
|
|
|
714
703
|
# The code text content.
|
|
715
704
|
sig { returns(String) }
|
|
716
|
-
|
|
717
|
-
def text = super
|
|
705
|
+
def text = Data.instance_method(:to_h).bind_call(self).fetch(:text)
|
|
718
706
|
# Programming language (from class="language-*" or similar).
|
|
719
707
|
sig { returns(T.nilable(String)) }
|
|
720
|
-
|
|
721
|
-
def language = super
|
|
708
|
+
def language = Data.instance_method(:to_h).bind_call(self).fetch(:language)
|
|
722
709
|
sig { returns(T::Boolean) }
|
|
723
710
|
def heading? = false
|
|
724
711
|
sig { returns(T::Boolean) }
|
|
@@ -835,12 +822,10 @@ module HtmlToMarkdown
|
|
|
835
822
|
|
|
836
823
|
# The term being defined.
|
|
837
824
|
sig { returns(String) }
|
|
838
|
-
|
|
839
|
-
def term = super
|
|
825
|
+
def term = Data.instance_method(:to_h).bind_call(self).fetch(:term)
|
|
840
826
|
# The definition text.
|
|
841
827
|
sig { returns(String) }
|
|
842
|
-
|
|
843
|
-
def definition = super
|
|
828
|
+
def definition = Data.instance_method(:to_h).bind_call(self).fetch(:definition)
|
|
844
829
|
sig { returns(T::Boolean) }
|
|
845
830
|
def heading? = false
|
|
846
831
|
sig { returns(T::Boolean) }
|
|
@@ -881,12 +866,10 @@ module HtmlToMarkdown
|
|
|
881
866
|
|
|
882
867
|
# The format of the raw content (e.g. "html", "css", "javascript").
|
|
883
868
|
sig { returns(String) }
|
|
884
|
-
|
|
885
|
-
def format = super
|
|
869
|
+
def format = Data.instance_method(:to_h).bind_call(self).fetch(:format)
|
|
886
870
|
# The raw content.
|
|
887
871
|
sig { returns(String) }
|
|
888
|
-
|
|
889
|
-
def content = super
|
|
872
|
+
def content = Data.instance_method(:to_h).bind_call(self).fetch(:content)
|
|
890
873
|
sig { returns(T::Boolean) }
|
|
891
874
|
def heading? = false
|
|
892
875
|
sig { returns(T::Boolean) }
|
|
@@ -927,8 +910,7 @@ module HtmlToMarkdown
|
|
|
927
910
|
|
|
928
911
|
# Key-value metadata pairs.
|
|
929
912
|
sig { returns(T::Array[MetadataEntry]) }
|
|
930
|
-
|
|
931
|
-
def entries = super
|
|
913
|
+
def entries = Data.instance_method(:to_h).bind_call(self).fetch(:entries)
|
|
932
914
|
sig { returns(T::Boolean) }
|
|
933
915
|
def heading? = false
|
|
934
916
|
sig { returns(T::Boolean) }
|
|
@@ -969,16 +951,13 @@ module HtmlToMarkdown
|
|
|
969
951
|
|
|
970
952
|
# Optional section label.
|
|
971
953
|
sig { returns(T.nilable(String)) }
|
|
972
|
-
|
|
973
|
-
def label = super
|
|
954
|
+
def label = Data.instance_method(:to_h).bind_call(self).fetch(:label)
|
|
974
955
|
# The heading level that created this group.
|
|
975
956
|
sig { returns(T.nilable(Integer)) }
|
|
976
|
-
|
|
977
|
-
def heading_level = super
|
|
957
|
+
def heading_level = Data.instance_method(:to_h).bind_call(self).fetch(:heading_level)
|
|
978
958
|
# The heading text that created this group.
|
|
979
959
|
sig { returns(T.nilable(String)) }
|
|
980
|
-
|
|
981
|
-
def heading_text = super
|
|
960
|
+
def heading_text = Data.instance_method(:to_h).bind_call(self).fetch(:heading_text)
|
|
982
961
|
sig { returns(T::Boolean) }
|
|
983
962
|
def heading? = false
|
|
984
963
|
sig { returns(T::Boolean) }
|
|
@@ -1084,8 +1063,7 @@ module HtmlToMarkdown
|
|
|
1084
1063
|
|
|
1085
1064
|
# @return [String]
|
|
1086
1065
|
sig { returns(String) }
|
|
1087
|
-
|
|
1088
|
-
def value = super
|
|
1066
|
+
def value = Data.instance_method(:to_h).bind_call(self).fetch(:value)
|
|
1089
1067
|
sig { returns(T::Boolean) }
|
|
1090
1068
|
def continue? = false
|
|
1091
1069
|
sig { returns(T::Boolean) }
|
|
@@ -1160,8 +1138,7 @@ module HtmlToMarkdown
|
|
|
1160
1138
|
|
|
1161
1139
|
# @return [String]
|
|
1162
1140
|
sig { returns(String) }
|
|
1163
|
-
|
|
1164
|
-
def value = super
|
|
1141
|
+
def value = Data.instance_method(:to_h).bind_call(self).fetch(:value)
|
|
1165
1142
|
sig { returns(T::Boolean) }
|
|
1166
1143
|
def continue? = false
|
|
1167
1144
|
sig { returns(T::Boolean) }
|
|
@@ -1,10 +1,10 @@
|
|
|
1
1
|
# This file is auto-generated by alef — DO NOT EDIT.
|
|
2
|
-
# alef:hash:
|
|
2
|
+
# alef:hash:d3db8f2a519dfe0fdc4abb8529bf0e8c4990d6581fa1a04122cce92a730c65d5
|
|
3
3
|
# To regenerate: alef generate
|
|
4
4
|
# To verify freshness: alef verify
|
|
5
5
|
# frozen_string_literal: true
|
|
6
6
|
|
|
7
7
|
module HtmlToMarkdown
|
|
8
8
|
## The version string for this package.
|
|
9
|
-
VERSION = "3.
|
|
9
|
+
VERSION = "3.15.0"
|
|
10
10
|
end
|
data/sig/types.rbs
CHANGED
|
@@ -1,5 +1,5 @@
|
|
|
1
1
|
# This file is auto-generated by alef — DO NOT EDIT.
|
|
2
|
-
# alef:hash:
|
|
2
|
+
# alef:hash:c5ae0579cda5776e922483b37b6e2a326aa2211bf97a44c9155bb8478aa502d5
|
|
3
3
|
# To regenerate: alef generate
|
|
4
4
|
# To verify freshness: alef verify
|
|
5
5
|
|
|
@@ -10,8 +10,8 @@ module HtmlToMarkdown
|
|
|
10
10
|
type json_value = Hash[String, untyped] | Array[untyped] | String | Integer | Float | bool | nil
|
|
11
11
|
|
|
12
12
|
class ConversionOptions
|
|
13
|
-
attr_reader heading_style:
|
|
14
|
-
attr_reader list_indent_type:
|
|
13
|
+
attr_reader heading_style: enum_HeadingStyle
|
|
14
|
+
attr_reader list_indent_type: enum_ListIndentType
|
|
15
15
|
attr_reader list_indent_width: Integer
|
|
16
16
|
attr_reader bullets: String
|
|
17
17
|
attr_reader strong_em_symbol: String
|
|
@@ -24,17 +24,17 @@ type json_value = Hash[String, untyped] | Array[untyped] | String | Integer | Fl
|
|
|
24
24
|
attr_reader default_title: bool
|
|
25
25
|
attr_reader br_in_tables: bool
|
|
26
26
|
attr_reader compact_tables: bool
|
|
27
|
-
attr_reader highlight_style:
|
|
27
|
+
attr_reader highlight_style: enum_HighlightStyle
|
|
28
28
|
attr_reader extract_metadata: bool
|
|
29
|
-
attr_reader whitespace_mode:
|
|
29
|
+
attr_reader whitespace_mode: enum_WhitespaceMode
|
|
30
30
|
attr_reader strip_newlines: bool
|
|
31
31
|
attr_reader wrap: bool
|
|
32
32
|
attr_reader wrap_width: Integer
|
|
33
33
|
attr_reader convert_as_inline: bool
|
|
34
34
|
attr_reader sub_symbol: String
|
|
35
35
|
attr_reader sup_symbol: String
|
|
36
|
-
attr_reader newline_style:
|
|
37
|
-
attr_reader code_block_style:
|
|
36
|
+
attr_reader newline_style: enum_NewlineStyle
|
|
37
|
+
attr_reader code_block_style: enum_CodeBlockStyle
|
|
38
38
|
attr_reader keep_inline_images_in: Array[String]
|
|
39
39
|
attr_reader preprocessing: PreprocessingOptions
|
|
40
40
|
attr_reader encoding: String
|
|
@@ -42,9 +42,9 @@ type json_value = Hash[String, untyped] | Array[untyped] | String | Integer | Fl
|
|
|
42
42
|
attr_reader strip_tags: Array[String]
|
|
43
43
|
attr_reader preserve_tags: Array[String]
|
|
44
44
|
attr_reader skip_images: bool
|
|
45
|
-
attr_reader url_escape_style:
|
|
46
|
-
attr_reader link_style:
|
|
47
|
-
attr_reader output_format:
|
|
45
|
+
attr_reader url_escape_style: enum_UrlEscapeStyle
|
|
46
|
+
attr_reader link_style: enum_LinkStyle
|
|
47
|
+
attr_reader output_format: enum_OutputFormat
|
|
48
48
|
attr_reader include_document_structure: bool
|
|
49
49
|
attr_reader extract_images: bool
|
|
50
50
|
attr_reader max_image_size: Integer
|
|
@@ -52,15 +52,16 @@ type json_value = Hash[String, untyped] | Array[untyped] | String | Integer | Fl
|
|
|
52
52
|
attr_reader infer_dimensions: bool
|
|
53
53
|
attr_reader max_depth: Integer?
|
|
54
54
|
attr_reader exclude_selectors: Array[String]
|
|
55
|
-
attr_reader tier_strategy:
|
|
55
|
+
attr_reader tier_strategy: enum_TierStrategy
|
|
56
|
+
attr_reader base_url: String?
|
|
56
57
|
attr_reader visitor: VisitorHandle?
|
|
57
58
|
|
|
58
|
-
def initialize: (?heading_style:
|
|
59
|
+
def initialize: (?heading_style: enum_HeadingStyle, ?list_indent_type: enum_ListIndentType, ?list_indent_width: Integer, ?bullets: String, ?strong_em_symbol: String, ?escape_asterisks: bool, ?escape_underscores: bool, ?escape_misc: bool, ?escape_ascii: bool, ?code_language: String, ?autolinks: bool, ?default_title: bool, ?br_in_tables: bool, ?compact_tables: bool, ?highlight_style: enum_HighlightStyle, ?extract_metadata: bool, ?whitespace_mode: enum_WhitespaceMode, ?strip_newlines: bool, ?wrap: bool, ?wrap_width: Integer, ?convert_as_inline: bool, ?sub_symbol: String, ?sup_symbol: String, ?newline_style: enum_NewlineStyle, ?code_block_style: enum_CodeBlockStyle, ?keep_inline_images_in: Array[String], ?preprocessing: PreprocessingOptions, ?encoding: String, ?debug: bool, ?strip_tags: Array[String], ?preserve_tags: Array[String], ?skip_images: bool, ?url_escape_style: enum_UrlEscapeStyle, ?link_style: enum_LinkStyle, ?output_format: enum_OutputFormat, ?include_document_structure: bool, ?extract_images: bool, ?max_image_size: Integer, ?capture_svg: bool, ?infer_dimensions: bool, ?max_depth: Integer, ?exclude_selectors: Array[String], ?tier_strategy: enum_TierStrategy, ?base_url: String, ?visitor: VisitorHandle) -> void
|
|
59
60
|
end
|
|
60
61
|
|
|
61
62
|
class ConversionOptionsUpdate
|
|
62
|
-
attr_reader heading_style:
|
|
63
|
-
attr_reader list_indent_type:
|
|
63
|
+
attr_reader heading_style: enum_HeadingStyle?
|
|
64
|
+
attr_reader list_indent_type: enum_ListIndentType?
|
|
64
65
|
attr_reader list_indent_width: Integer?
|
|
65
66
|
attr_reader bullets: String?
|
|
66
67
|
attr_reader strong_em_symbol: String?
|
|
@@ -73,17 +74,17 @@ def initialize: (?heading_style: HeadingStyle, ?list_indent_type: ListIndentType
|
|
|
73
74
|
attr_reader default_title: bool?
|
|
74
75
|
attr_reader br_in_tables: bool?
|
|
75
76
|
attr_reader compact_tables: bool?
|
|
76
|
-
attr_reader highlight_style:
|
|
77
|
+
attr_reader highlight_style: enum_HighlightStyle?
|
|
77
78
|
attr_reader extract_metadata: bool?
|
|
78
|
-
attr_reader whitespace_mode:
|
|
79
|
+
attr_reader whitespace_mode: enum_WhitespaceMode?
|
|
79
80
|
attr_reader strip_newlines: bool?
|
|
80
81
|
attr_reader wrap: bool?
|
|
81
82
|
attr_reader wrap_width: Integer?
|
|
82
83
|
attr_reader convert_as_inline: bool?
|
|
83
84
|
attr_reader sub_symbol: String?
|
|
84
85
|
attr_reader sup_symbol: String?
|
|
85
|
-
attr_reader newline_style:
|
|
86
|
-
attr_reader code_block_style:
|
|
86
|
+
attr_reader newline_style: enum_NewlineStyle?
|
|
87
|
+
attr_reader code_block_style: enum_CodeBlockStyle?
|
|
87
88
|
attr_reader keep_inline_images_in: Array[String]?
|
|
88
89
|
attr_reader preprocessing: PreprocessingOptionsUpdate?
|
|
89
90
|
attr_reader encoding: String?
|
|
@@ -91,9 +92,9 @@ def initialize: (?heading_style: HeadingStyle, ?list_indent_type: ListIndentType
|
|
|
91
92
|
attr_reader strip_tags: Array[String]?
|
|
92
93
|
attr_reader preserve_tags: Array[String]?
|
|
93
94
|
attr_reader skip_images: bool?
|
|
94
|
-
attr_reader url_escape_style:
|
|
95
|
-
attr_reader link_style:
|
|
96
|
-
attr_reader output_format:
|
|
95
|
+
attr_reader url_escape_style: enum_UrlEscapeStyle?
|
|
96
|
+
attr_reader link_style: enum_LinkStyle?
|
|
97
|
+
attr_reader output_format: enum_OutputFormat?
|
|
97
98
|
attr_reader include_document_structure: bool?
|
|
98
99
|
attr_reader extract_images: bool?
|
|
99
100
|
attr_reader max_image_size: Integer?
|
|
@@ -101,10 +102,11 @@ def initialize: (?heading_style: HeadingStyle, ?list_indent_type: ListIndentType
|
|
|
101
102
|
attr_reader infer_dimensions: bool?
|
|
102
103
|
attr_reader max_depth: Integer?
|
|
103
104
|
attr_reader exclude_selectors: Array[String]?
|
|
104
|
-
attr_reader tier_strategy:
|
|
105
|
+
attr_reader tier_strategy: enum_TierStrategy?
|
|
106
|
+
attr_reader base_url: String?
|
|
105
107
|
attr_reader visitor: VisitorHandle?
|
|
106
108
|
|
|
107
|
-
def initialize: (?heading_style:
|
|
109
|
+
def initialize: (?heading_style: enum_HeadingStyle, ?list_indent_type: enum_ListIndentType, ?list_indent_width: Integer, ?bullets: String, ?strong_em_symbol: String, ?escape_asterisks: bool, ?escape_underscores: bool, ?escape_misc: bool, ?escape_ascii: bool, ?code_language: String, ?autolinks: bool, ?default_title: bool, ?br_in_tables: bool, ?compact_tables: bool, ?highlight_style: enum_HighlightStyle, ?extract_metadata: bool, ?whitespace_mode: enum_WhitespaceMode, ?strip_newlines: bool, ?wrap: bool, ?wrap_width: Integer, ?convert_as_inline: bool, ?sub_symbol: String, ?sup_symbol: String, ?newline_style: enum_NewlineStyle, ?code_block_style: enum_CodeBlockStyle, ?keep_inline_images_in: Array[String], ?preprocessing: PreprocessingOptionsUpdate, ?encoding: String, ?debug: bool, ?strip_tags: Array[String], ?preserve_tags: Array[String], ?skip_images: bool, ?url_escape_style: enum_UrlEscapeStyle, ?link_style: enum_LinkStyle, ?output_format: enum_OutputFormat, ?include_document_structure: bool, ?extract_images: bool, ?max_image_size: Integer, ?capture_svg: bool, ?infer_dimensions: bool, ?max_depth: Integer?, ?exclude_selectors: Array[String], ?tier_strategy: enum_TierStrategy, ?base_url: String?, ?visitor: VisitorHandle) -> void
|
|
108
110
|
end
|
|
109
111
|
|
|
110
112
|
class ConversionResult
|
|
@@ -125,12 +127,12 @@ def initialize: (?content: String, ?document: DocumentStructure, ?metadata: Html
|
|
|
125
127
|
attr_reader canonical_url: String?
|
|
126
128
|
attr_reader base_href: String?
|
|
127
129
|
attr_reader language: String?
|
|
128
|
-
attr_reader text_direction:
|
|
130
|
+
attr_reader text_direction: enum_TextDirection?
|
|
129
131
|
attr_reader open_graph: Hash[String, String]
|
|
130
132
|
attr_reader twitter_card: Hash[String, String]
|
|
131
133
|
attr_reader meta_tags: Hash[String, String]
|
|
132
134
|
|
|
133
|
-
def initialize: (?title: String, ?description: String, ?keywords: Array[String], ?author: String, ?canonical_url: String, ?base_href: String, ?language: String, ?text_direction:
|
|
135
|
+
def initialize: (?title: String, ?description: String, ?keywords: Array[String], ?author: String, ?canonical_url: String, ?base_href: String, ?language: String, ?text_direction: enum_TextDirection, ?open_graph: Hash[String, String], ?twitter_card: Hash[String, String], ?meta_tags: Hash[String, String]) -> void
|
|
134
136
|
end
|
|
135
137
|
|
|
136
138
|
class DocumentNode
|
|
@@ -195,21 +197,21 @@ def initialize: (?document: DocumentMetadata, ?headers: Array[HeaderMetadata], ?
|
|
|
195
197
|
attr_reader alt: String?
|
|
196
198
|
attr_reader title: String?
|
|
197
199
|
attr_reader dimensions: ImageDimensions?
|
|
198
|
-
attr_reader image_type:
|
|
200
|
+
attr_reader image_type: enum_ImageType
|
|
199
201
|
attr_reader attributes: Hash[String, String]
|
|
200
202
|
|
|
201
|
-
def initialize: (src: String, ?alt: String, ?title: String, ?dimensions: ImageDimensions, image_type:
|
|
203
|
+
def initialize: (src: String, ?alt: String, ?title: String, ?dimensions: ImageDimensions, image_type: enum_ImageType, attributes: Hash[String, String]) -> void
|
|
202
204
|
end
|
|
203
205
|
|
|
204
206
|
class LinkMetadata
|
|
205
207
|
attr_reader href: String
|
|
206
208
|
attr_reader text: String
|
|
207
209
|
attr_reader title: String?
|
|
208
|
-
attr_reader link_type:
|
|
210
|
+
attr_reader link_type: enum_LinkType
|
|
209
211
|
attr_reader rel: Array[String]
|
|
210
212
|
attr_reader attributes: Hash[String, String]
|
|
211
213
|
|
|
212
|
-
def initialize: (href: String, text: String, ?title: String, link_type:
|
|
214
|
+
def initialize: (href: String, text: String, ?title: String, link_type: enum_LinkType, rel: Array[String], attributes: Hash[String, String]) -> void
|
|
213
215
|
end
|
|
214
216
|
|
|
215
217
|
class MetadataEntry
|
|
@@ -220,49 +222,49 @@ def initialize: (href: String, text: String, ?title: String, link_type: LinkType
|
|
|
220
222
|
end
|
|
221
223
|
|
|
222
224
|
class NodeContext
|
|
223
|
-
attr_reader node_type:
|
|
225
|
+
attr_reader node_type: enum_NodeType
|
|
224
226
|
attr_reader tag_name: String
|
|
225
227
|
attr_reader depth: Integer
|
|
226
228
|
attr_reader index_in_parent: Integer
|
|
227
229
|
attr_reader parent_tag: String?
|
|
228
230
|
attr_reader is_inline: bool
|
|
229
231
|
|
|
230
|
-
def initialize: (node_type:
|
|
232
|
+
def initialize: (node_type: enum_NodeType, tag_name: String, depth: Integer, index_in_parent: Integer, ?parent_tag: String, is_inline: bool) -> void
|
|
231
233
|
def attributes: () -> Hash[String, String]
|
|
232
234
|
def into_owned: () -> NodeContext
|
|
233
235
|
end
|
|
234
236
|
|
|
235
237
|
class PreprocessingOptions
|
|
236
238
|
attr_reader enabled: bool
|
|
237
|
-
attr_reader preset:
|
|
239
|
+
attr_reader preset: enum_PreprocessingPreset
|
|
238
240
|
attr_reader remove_navigation: bool
|
|
239
241
|
attr_reader remove_forms: bool
|
|
240
242
|
|
|
241
|
-
def initialize: (?enabled: bool, ?preset:
|
|
243
|
+
def initialize: (?enabled: bool, ?preset: enum_PreprocessingPreset, ?remove_navigation: bool, ?remove_forms: bool) -> void
|
|
242
244
|
end
|
|
243
245
|
|
|
244
246
|
class PreprocessingOptionsUpdate
|
|
245
247
|
attr_reader enabled: bool?
|
|
246
|
-
attr_reader preset:
|
|
248
|
+
attr_reader preset: enum_PreprocessingPreset?
|
|
247
249
|
attr_reader remove_navigation: bool?
|
|
248
250
|
attr_reader remove_forms: bool?
|
|
249
251
|
|
|
250
|
-
def initialize: (?enabled: bool, ?preset:
|
|
252
|
+
def initialize: (?enabled: bool, ?preset: enum_PreprocessingPreset, ?remove_navigation: bool, ?remove_forms: bool) -> void
|
|
251
253
|
end
|
|
252
254
|
|
|
253
255
|
class ProcessingWarning
|
|
254
256
|
attr_reader message: String
|
|
255
|
-
attr_reader kind:
|
|
257
|
+
attr_reader kind: enum_WarningKind
|
|
256
258
|
|
|
257
|
-
def initialize: (message: String, kind:
|
|
259
|
+
def initialize: (message: String, kind: enum_WarningKind) -> void
|
|
258
260
|
end
|
|
259
261
|
|
|
260
262
|
class StructuredData
|
|
261
|
-
attr_reader data_type:
|
|
263
|
+
attr_reader data_type: enum_StructuredDataType
|
|
262
264
|
attr_reader raw_json: String
|
|
263
265
|
attr_reader schema_type: String?
|
|
264
266
|
|
|
265
|
-
def initialize: (data_type:
|
|
267
|
+
def initialize: (data_type: enum_StructuredDataType, raw_json: String, ?schema_type: String) -> void
|
|
266
268
|
end
|
|
267
269
|
|
|
268
270
|
class TableData
|
|
@@ -294,80 +296,53 @@ def initialize: (?rows: Integer, ?cols: Integer, ?cells: Array[GridCell]) -> voi
|
|
|
294
296
|
class AnnotationKind
|
|
295
297
|
end
|
|
296
298
|
|
|
297
|
-
|
|
298
|
-
type value = :Indented | :Backticks | :Tildes
|
|
299
|
-
end
|
|
299
|
+
type enum_CodeBlockStyle = :Indented | :Backticks | :Tildes
|
|
300
300
|
|
|
301
|
-
|
|
302
|
-
type value = :Underlined | :Atx | :AtxClosed
|
|
303
|
-
end
|
|
301
|
+
type enum_HeadingStyle = :Underlined | :Atx | :AtxClosed
|
|
304
302
|
|
|
305
|
-
|
|
306
|
-
type value = :DoubleEqual | :Html | :Bold | :None
|
|
307
|
-
end
|
|
303
|
+
type enum_HighlightStyle = :DoubleEqual | :Html | :Bold | :None
|
|
308
304
|
|
|
309
|
-
|
|
310
|
-
type value = :data_uri | :inline_svg | :external | :relative
|
|
311
|
-
end
|
|
305
|
+
type enum_ImageType = :data_uri | :inline_svg | :external | :relative
|
|
312
306
|
|
|
313
|
-
|
|
314
|
-
type value = :Inline | :Reference
|
|
315
|
-
end
|
|
307
|
+
type enum_LinkStyle = :Inline | :Reference
|
|
316
308
|
|
|
317
|
-
|
|
318
|
-
type value = :anchor | :internal | :external | :email | :phone | :other
|
|
319
|
-
end
|
|
309
|
+
type enum_LinkType = :anchor | :internal | :external | :email | :phone | :other
|
|
320
310
|
|
|
321
|
-
|
|
322
|
-
type value = :Spaces | :Tabs
|
|
323
|
-
end
|
|
311
|
+
type enum_ListIndentType = :Spaces | :Tabs
|
|
324
312
|
|
|
325
|
-
|
|
326
|
-
type value = :Spaces | :Backslash
|
|
327
|
-
end
|
|
313
|
+
type enum_NewlineStyle = :Spaces | :Backslash
|
|
328
314
|
|
|
329
315
|
class NodeContent
|
|
330
316
|
end
|
|
331
317
|
|
|
332
|
-
|
|
333
|
-
type value = :Text | :Element | :Heading | :Paragraph | :Div | :Blockquote | :Pre | :Hr | :List | :ListItem | :DefinitionList | :DefinitionTerm | :DefinitionDescription | :Table | :TableRow | :TableCell | :TableHeader | :TableBody | :TableHead | :TableFoot | :Link | :Image | :Strong | :Em | :Code | :Strikethrough | :Underline | :Subscript | :Superscript | :Mark | :Small | :Br | :Span | :Article | :Section | :Nav | :Aside | :Header | :Footer | :Main | :Figure | :Figcaption | :Time | :Details | :Summary | :Form | :Input | :Select | :Option | :Button | :Textarea | :Label | :Fieldset | :Legend | :Audio | :Video | :Picture | :Source | :Iframe | :Svg | :Canvas | :Ruby | :Rt | :Rp | :Abbr | :Kbd | :Samp | :Var | :Cite | :Q | :Del | :Ins | :Data | :Meter | :Progress | :Output | :Template | :Slot | :Html | :Head | :Body | :Title | :Meta | :LinkTag | :Style | :Script | :Base | :Custom
|
|
334
|
-
end
|
|
318
|
+
type enum_NodeType = :Text | :Element | :Heading | :Paragraph | :Div | :Blockquote | :Pre | :Hr | :List | :ListItem | :DefinitionList | :DefinitionTerm | :DefinitionDescription | :Table | :TableRow | :TableCell | :TableHeader | :TableBody | :TableHead | :TableFoot | :Link | :Image | :Strong | :Em | :Code | :Strikethrough | :Underline | :Subscript | :Superscript | :Mark | :Small | :Br | :Span | :Article | :Section | :Nav | :Aside | :Header | :Footer | :Main | :Figure | :Figcaption | :Time | :Details | :Summary | :Form | :Input | :Select | :Option | :Button | :Textarea | :Label | :Fieldset | :Legend | :Audio | :Video | :Picture | :Source | :Iframe | :Svg | :Canvas | :Ruby | :Rt | :Rp | :Abbr | :Kbd | :Samp | :Var | :Cite | :Q | :Del | :Ins | :Data | :Meter | :Progress | :Output | :Template | :Slot | :Html | :Head | :Body | :Title | :Meta | :LinkTag | :Style | :Script | :Base | :Custom
|
|
335
319
|
|
|
336
|
-
|
|
337
|
-
type value = :Markdown | :Djot | :Plain
|
|
338
|
-
end
|
|
320
|
+
type enum_OutputFormat = :Markdown | :Djot | :Plain
|
|
339
321
|
|
|
340
|
-
|
|
341
|
-
type value = :Minimal | :Standard | :Aggressive
|
|
342
|
-
end
|
|
322
|
+
type enum_PreprocessingPreset = :Minimal | :Standard | :Aggressive
|
|
343
323
|
|
|
344
|
-
|
|
345
|
-
type value = :json_ld | :microdata | :rdfa
|
|
346
|
-
end
|
|
324
|
+
type enum_StructuredDataType = :json_ld | :microdata | :rdfa
|
|
347
325
|
|
|
348
|
-
|
|
349
|
-
type value = :ltr | :rtl | :auto
|
|
350
|
-
end
|
|
326
|
+
type enum_TextDirection = :ltr | :rtl | :auto
|
|
351
327
|
|
|
352
|
-
|
|
353
|
-
type value = :auto | :tier2 | :tier1
|
|
354
|
-
end
|
|
328
|
+
type enum_TierStrategy = :auto | :tier2 | :tier1
|
|
355
329
|
|
|
356
|
-
|
|
357
|
-
type value = :Angle | :Percent
|
|
358
|
-
end
|
|
330
|
+
type enum_UrlEscapeStyle = :Angle | :Percent
|
|
359
331
|
|
|
360
332
|
class VisitResult
|
|
361
333
|
end
|
|
362
334
|
|
|
363
|
-
|
|
364
|
-
type value = :image_extraction_failed | :encoding_fallback | :truncated_input | :malformed_html | :sanitization_applied | :depth_limit_exceeded
|
|
365
|
-
end
|
|
335
|
+
type enum_WarningKind = :image_extraction_failed | :encoding_fallback | :truncated_input | :malformed_html | :sanitization_applied | :depth_limit_exceeded
|
|
366
336
|
|
|
367
|
-
|
|
368
|
-
type value = :Normalized | :Strict
|
|
369
|
-
end
|
|
337
|
+
type enum_WhitespaceMode = :Normalized | :Strict
|
|
370
338
|
|
|
371
339
|
def self.convert: (String html, ?ConversionOptions options) -> ConversionResult
|
|
372
340
|
|
|
341
|
+
interface _HtmlVisitor
|
|
342
|
+
# Optional methods the bridge calls when the object defines them (the
|
|
343
|
+
# trait's Rust default behavior applies otherwise): visit_text, visit_element_start, visit_element_end, visit_link, visit_image, visit_heading, visit_code_block, visit_code_inline, visit_list_item, visit_list_start, visit_list_end, visit_table_start, visit_table_row, visit_table_end, visit_blockquote, visit_strong, visit_emphasis, visit_strikethrough, visit_underline, visit_subscript, visit_superscript, visit_mark, visit_line_break, visit_horizontal_rule, visit_custom_element, visit_definition_list_start, visit_definition_term, visit_definition_description, visit_definition_list_end, visit_form, visit_input, visit_button, visit_audio, visit_video, visit_iframe, visit_details, visit_summary, visit_figure_start, visit_figcaption, visit_figure_end.
|
|
344
|
+
# The lifecycle hooks `initialize`/`shutdown` are likewise optional (note:
|
|
345
|
+
# Ruby constructors are private and never count as the lifecycle hook).
|
|
346
|
+
end
|
|
347
|
+
|
|
373
348
|
end
|
metadata
CHANGED
|
@@ -1,14 +1,14 @@
|
|
|
1
1
|
--- !ruby/object:Gem::Specification
|
|
2
2
|
name: html-to-markdown
|
|
3
3
|
version: !ruby/object:Gem::Version
|
|
4
|
-
version: 3.
|
|
4
|
+
version: 3.15.0
|
|
5
5
|
platform: ruby
|
|
6
6
|
authors:
|
|
7
7
|
- Na'aman Hirschfeld <naaman@xberg.io>
|
|
8
8
|
autorequire:
|
|
9
9
|
bindir: bin
|
|
10
10
|
cert_chain: []
|
|
11
|
-
date: 2026-09-
|
|
11
|
+
date: 2026-09-26 00:00:00.000000000 Z
|
|
12
12
|
dependencies:
|
|
13
13
|
- !ruby/object:Gem::Dependency
|
|
14
14
|
name: rb_sys
|