html-to-markdown 3.4.0 → 3.6.0.pre.rc.24

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.
Files changed (188) hide show
  1. checksums.yaml +4 -4
  2. data/LICENSE +21 -0
  3. data/README.md +347 -0
  4. data/Steepfile +10 -2
  5. data/ext/html_to_markdown_rb/Cargo.toml +3 -2
  6. data/ext/html_to_markdown_rb/extconf.rb +5 -5
  7. data/ext/html_to_markdown_rb/native/Cargo.lock +962 -0
  8. data/ext/html_to_markdown_rb/native/Cargo.toml +6 -11
  9. data/ext/html_to_markdown_rb/native/extconf.rb +14 -0
  10. data/ext/html_to_markdown_rb/src/lib.rs +1715 -646
  11. data/lib/html_to_markdown/native.rb +913 -37
  12. data/lib/html_to_markdown/version.rb +3 -3
  13. data/lib/html_to_markdown.rb +9 -4
  14. data/lib/html_to_markdown_rb.so +0 -0
  15. data/sig/types.rbs +59 -292
  16. metadata +32 -179
  17. data/ext/html_to_markdown_rb/Makefile +0 -592
  18. data/lib/bin/html-to-markdown +0 -0
  19. data/vendor/Cargo.toml +0 -33
  20. data/vendor/html-to-markdown-rs/Cargo.toml +0 -54
  21. data/vendor/html-to-markdown-rs/README.md +0 -278
  22. data/vendor/html-to-markdown-rs/examples/basic.rs +0 -24
  23. data/vendor/html-to-markdown-rs/examples/table.rs +0 -25
  24. data/vendor/html-to-markdown-rs/examples/test_deser.rs +0 -12
  25. data/vendor/html-to-markdown-rs/examples/test_escape.rs +0 -58
  26. data/vendor/html-to-markdown-rs/examples/test_inline_formatting.rs +0 -113
  27. data/vendor/html-to-markdown-rs/examples/test_lists.rs +0 -39
  28. data/vendor/html-to-markdown-rs/examples/test_semantic_tags.rs +0 -89
  29. data/vendor/html-to-markdown-rs/examples/test_tables.rs +0 -100
  30. data/vendor/html-to-markdown-rs/examples/test_task_lists.rs +0 -61
  31. data/vendor/html-to-markdown-rs/examples/test_whitespace.rs +0 -34
  32. data/vendor/html-to-markdown-rs/src/convert_api.rs +0 -349
  33. data/vendor/html-to-markdown-rs/src/converter/block/blockquote.rs +0 -178
  34. data/vendor/html-to-markdown-rs/src/converter/block/container.rs +0 -114
  35. data/vendor/html-to-markdown-rs/src/converter/block/div.rs +0 -149
  36. data/vendor/html-to-markdown-rs/src/converter/block/heading.rs +0 -428
  37. data/vendor/html-to-markdown-rs/src/converter/block/horizontal_rule.rs +0 -103
  38. data/vendor/html-to-markdown-rs/src/converter/block/line_break.rs +0 -89
  39. data/vendor/html-to-markdown-rs/src/converter/block/mod.rs +0 -10
  40. data/vendor/html-to-markdown-rs/src/converter/block/paragraph.rs +0 -140
  41. data/vendor/html-to-markdown-rs/src/converter/block/preformatted.rs +0 -298
  42. data/vendor/html-to-markdown-rs/src/converter/block/table/builder.rs +0 -453
  43. data/vendor/html-to-markdown-rs/src/converter/block/table/caption.rs +0 -44
  44. data/vendor/html-to-markdown-rs/src/converter/block/table/cell.rs +0 -276
  45. data/vendor/html-to-markdown-rs/src/converter/block/table/cells.rs +0 -336
  46. data/vendor/html-to-markdown-rs/src/converter/block/table/layout.rs +0 -58
  47. data/vendor/html-to-markdown-rs/src/converter/block/table/mod.rs +0 -266
  48. data/vendor/html-to-markdown-rs/src/converter/block/table/scanner.rs +0 -146
  49. data/vendor/html-to-markdown-rs/src/converter/block/table/utils.rs +0 -34
  50. data/vendor/html-to-markdown-rs/src/converter/block/unknown.rs +0 -138
  51. data/vendor/html-to-markdown-rs/src/converter/context.rs +0 -208
  52. data/vendor/html-to-markdown-rs/src/converter/dom_context.rs +0 -337
  53. data/vendor/html-to-markdown-rs/src/converter/form/elements.rs +0 -770
  54. data/vendor/html-to-markdown-rs/src/converter/form/mod.rs +0 -82
  55. data/vendor/html-to-markdown-rs/src/converter/format/djot.rs +0 -64
  56. data/vendor/html-to-markdown-rs/src/converter/format/markdown.rs +0 -59
  57. data/vendor/html-to-markdown-rs/src/converter/format/mod.rs +0 -43
  58. data/vendor/html-to-markdown-rs/src/converter/handlers/blockquote.rs +0 -173
  59. data/vendor/html-to-markdown-rs/src/converter/handlers/code_block.rs +0 -434
  60. data/vendor/html-to-markdown-rs/src/converter/handlers/graphic.rs +0 -234
  61. data/vendor/html-to-markdown-rs/src/converter/handlers/image.rs +0 -282
  62. data/vendor/html-to-markdown-rs/src/converter/handlers/link.rs +0 -316
  63. data/vendor/html-to-markdown-rs/src/converter/handlers/mod.rs +0 -26
  64. data/vendor/html-to-markdown-rs/src/converter/inline/code.rs +0 -306
  65. data/vendor/html-to-markdown-rs/src/converter/inline/emphasis.rs +0 -345
  66. data/vendor/html-to-markdown-rs/src/converter/inline/link.rs +0 -428
  67. data/vendor/html-to-markdown-rs/src/converter/inline/mod.rs +0 -237
  68. data/vendor/html-to-markdown-rs/src/converter/inline/ruby.rs +0 -337
  69. data/vendor/html-to-markdown-rs/src/converter/inline/semantic/marks.rs +0 -566
  70. data/vendor/html-to-markdown-rs/src/converter/inline/semantic/mod.rs +0 -86
  71. data/vendor/html-to-markdown-rs/src/converter/inline/semantic/typography.rs +0 -558
  72. data/vendor/html-to-markdown-rs/src/converter/list/definition.rs +0 -232
  73. data/vendor/html-to-markdown-rs/src/converter/list/item.rs +0 -332
  74. data/vendor/html-to-markdown-rs/src/converter/list/mod.rs +0 -70
  75. data/vendor/html-to-markdown-rs/src/converter/list/ordered.rs +0 -201
  76. data/vendor/html-to-markdown-rs/src/converter/list/unordered.rs +0 -195
  77. data/vendor/html-to-markdown-rs/src/converter/list/utils.rs +0 -314
  78. data/vendor/html-to-markdown-rs/src/converter/main.rs +0 -710
  79. data/vendor/html-to-markdown-rs/src/converter/main_helpers.rs +0 -452
  80. data/vendor/html-to-markdown-rs/src/converter/media/embedded.rs +0 -393
  81. data/vendor/html-to-markdown-rs/src/converter/media/graphic.rs +0 -4
  82. data/vendor/html-to-markdown-rs/src/converter/media/image.rs +0 -183
  83. data/vendor/html-to-markdown-rs/src/converter/media/mod.rs +0 -87
  84. data/vendor/html-to-markdown-rs/src/converter/media/svg.rs +0 -280
  85. data/vendor/html-to-markdown-rs/src/converter/metadata.rs +0 -220
  86. data/vendor/html-to-markdown-rs/src/converter/mod.rs +0 -156
  87. data/vendor/html-to-markdown-rs/src/converter/plain_text.rs +0 -516
  88. data/vendor/html-to-markdown-rs/src/converter/preprocessing_helpers.rs +0 -201
  89. data/vendor/html-to-markdown-rs/src/converter/reference_collector.rs +0 -69
  90. data/vendor/html-to-markdown-rs/src/converter/semantic/attributes.rs +0 -269
  91. data/vendor/html-to-markdown-rs/src/converter/semantic/definition_list.rs +0 -266
  92. data/vendor/html-to-markdown-rs/src/converter/semantic/figure.rs +0 -391
  93. data/vendor/html-to-markdown-rs/src/converter/semantic/mod.rs +0 -112
  94. data/vendor/html-to-markdown-rs/src/converter/semantic/sectioning.rs +0 -85
  95. data/vendor/html-to-markdown-rs/src/converter/semantic/summary.rs +0 -324
  96. data/vendor/html-to-markdown-rs/src/converter/text/mod.rs +0 -8
  97. data/vendor/html-to-markdown-rs/src/converter/text/processing.rs +0 -56
  98. data/vendor/html-to-markdown-rs/src/converter/text_node.rs +0 -269
  99. data/vendor/html-to-markdown-rs/src/converter/utility/attributes.rs +0 -151
  100. data/vendor/html-to-markdown-rs/src/converter/utility/caching.rs +0 -74
  101. data/vendor/html-to-markdown-rs/src/converter/utility/content.rs +0 -271
  102. data/vendor/html-to-markdown-rs/src/converter/utility/mod.rs +0 -17
  103. data/vendor/html-to-markdown-rs/src/converter/utility/preprocessing.rs +0 -1002
  104. data/vendor/html-to-markdown-rs/src/converter/utility/serialization.rs +0 -126
  105. data/vendor/html-to-markdown-rs/src/converter/utility/siblings.rs +0 -97
  106. data/vendor/html-to-markdown-rs/src/converter/visitor_hooks.rs +0 -189
  107. data/vendor/html-to-markdown-rs/src/error.rs +0 -43
  108. data/vendor/html-to-markdown-rs/src/exports.rs +0 -24
  109. data/vendor/html-to-markdown-rs/src/inline_images.rs +0 -336
  110. data/vendor/html-to-markdown-rs/src/lib.rs +0 -139
  111. data/vendor/html-to-markdown-rs/src/metadata/collector.rs +0 -457
  112. data/vendor/html-to-markdown-rs/src/metadata/config.rs +0 -394
  113. data/vendor/html-to-markdown-rs/src/metadata/extraction.rs +0 -398
  114. data/vendor/html-to-markdown-rs/src/metadata/mod.rs +0 -288
  115. data/vendor/html-to-markdown-rs/src/metadata/types.rs +0 -477
  116. data/vendor/html-to-markdown-rs/src/options/conversion.rs +0 -559
  117. data/vendor/html-to-markdown-rs/src/options/inline_image.rs +0 -111
  118. data/vendor/html-to-markdown-rs/src/options/mod.rs +0 -20
  119. data/vendor/html-to-markdown-rs/src/options/preprocessing.rs +0 -201
  120. data/vendor/html-to-markdown-rs/src/options/validation.rs +0 -416
  121. data/vendor/html-to-markdown-rs/src/prelude.rs +0 -1
  122. data/vendor/html-to-markdown-rs/src/rcdom.rs +0 -487
  123. data/vendor/html-to-markdown-rs/src/text.rs +0 -358
  124. data/vendor/html-to-markdown-rs/src/types/document.rs +0 -191
  125. data/vendor/html-to-markdown-rs/src/types/mod.rs +0 -17
  126. data/vendor/html-to-markdown-rs/src/types/result.rs +0 -54
  127. data/vendor/html-to-markdown-rs/src/types/structure_builder.rs +0 -791
  128. data/vendor/html-to-markdown-rs/src/types/structure_collector.rs +0 -483
  129. data/vendor/html-to-markdown-rs/src/types/tables.rs +0 -52
  130. data/vendor/html-to-markdown-rs/src/types/warnings.rs +0 -33
  131. data/vendor/html-to-markdown-rs/src/validation.rs +0 -158
  132. data/vendor/html-to-markdown-rs/src/visitor/default_impl.rs +0 -63
  133. data/vendor/html-to-markdown-rs/src/visitor/mod.rs +0 -41
  134. data/vendor/html-to-markdown-rs/src/visitor/traits.rs +0 -370
  135. data/vendor/html-to-markdown-rs/src/visitor/types.rs +0 -319
  136. data/vendor/html-to-markdown-rs/src/visitor_helpers/helpers/callbacks/mod.rs +0 -1
  137. data/vendor/html-to-markdown-rs/src/visitor_helpers/helpers/content.rs +0 -126
  138. data/vendor/html-to-markdown-rs/src/visitor_helpers/helpers/mod.rs +0 -27
  139. data/vendor/html-to-markdown-rs/src/visitor_helpers/helpers/state.rs +0 -110
  140. data/vendor/html-to-markdown-rs/src/visitor_helpers/helpers/traversal.rs +0 -250
  141. data/vendor/html-to-markdown-rs/src/visitor_helpers.rs +0 -597
  142. data/vendor/html-to-markdown-rs/src/wrapper/sync.rs +0 -413
  143. data/vendor/html-to-markdown-rs/src/wrapper/utils.rs +0 -290
  144. data/vendor/html-to-markdown-rs/src/wrapper.rs +0 -9
  145. data/vendor/html-to-markdown-rs/tests/br_in_inline_test.rs +0 -87
  146. data/vendor/html-to-markdown-rs/tests/commonmark_compliance_test.rs +0 -297
  147. data/vendor/html-to-markdown-rs/tests/djot_output_test.rs +0 -153
  148. data/vendor/html-to-markdown-rs/tests/exclude_selectors_test.rs +0 -132
  149. data/vendor/html-to-markdown-rs/tests/integration_test.rs +0 -631
  150. data/vendor/html-to-markdown-rs/tests/issue_121_regressions.rs +0 -49
  151. data/vendor/html-to-markdown-rs/tests/issue_127_regressions.rs +0 -58
  152. data/vendor/html-to-markdown-rs/tests/issue_128_regressions.rs +0 -17
  153. data/vendor/html-to-markdown-rs/tests/issue_131_regressions.rs +0 -41
  154. data/vendor/html-to-markdown-rs/tests/issue_134_regressions.rs +0 -40
  155. data/vendor/html-to-markdown-rs/tests/issue_139_regressions.rs +0 -26
  156. data/vendor/html-to-markdown-rs/tests/issue_140_regressions.rs +0 -185
  157. data/vendor/html-to-markdown-rs/tests/issue_143_regressions.rs +0 -100
  158. data/vendor/html-to-markdown-rs/tests/issue_145_regressions.rs +0 -133
  159. data/vendor/html-to-markdown-rs/tests/issue_146_regressions.rs +0 -144
  160. data/vendor/html-to-markdown-rs/tests/issue_176_regressions.rs +0 -62
  161. data/vendor/html-to-markdown-rs/tests/issue_190_regressions.rs +0 -128
  162. data/vendor/html-to-markdown-rs/tests/issue_199_regressions.rs +0 -20
  163. data/vendor/html-to-markdown-rs/tests/issue_200_regressions.rs +0 -62
  164. data/vendor/html-to-markdown-rs/tests/issue_212_regressions.rs +0 -68
  165. data/vendor/html-to-markdown-rs/tests/issue_216_217_regressions.rs +0 -87
  166. data/vendor/html-to-markdown-rs/tests/issue_336_regressions.rs +0 -74
  167. data/vendor/html-to-markdown-rs/tests/issue_339_regressions.rs +0 -92
  168. data/vendor/html-to-markdown-rs/tests/issue_347_regressions.rs +0 -154
  169. data/vendor/html-to-markdown-rs/tests/issue_348_visitor_plain.rs +0 -93
  170. data/vendor/html-to-markdown-rs/tests/json_ld_script_extraction.rs +0 -44
  171. data/vendor/html-to-markdown-rs/tests/lists_test.rs +0 -199
  172. data/vendor/html-to-markdown-rs/tests/plain_output_test.rs +0 -273
  173. data/vendor/html-to-markdown-rs/tests/preprocessing_tests.rs +0 -61
  174. data/vendor/html-to-markdown-rs/tests/reference_links_test.rs +0 -169
  175. data/vendor/html-to-markdown-rs/tests/sectioning_elements_test.rs +0 -137
  176. data/vendor/html-to-markdown-rs/tests/skip_images_test.rs +0 -522
  177. data/vendor/html-to-markdown-rs/tests/tables_test.rs +0 -743
  178. data/vendor/html-to-markdown-rs/tests/test_custom_elements.rs +0 -41
  179. data/vendor/html-to-markdown-rs/tests/test_issue_187.rs +0 -204
  180. data/vendor/html-to-markdown-rs/tests/test_issue_218.rs +0 -68
  181. data/vendor/html-to-markdown-rs/tests/test_issue_277.rs +0 -77
  182. data/vendor/html-to-markdown-rs/tests/test_max_depth.rs +0 -82
  183. data/vendor/html-to-markdown-rs/tests/test_nested_simple.rs +0 -45
  184. data/vendor/html-to-markdown-rs/tests/test_script_style_stripping.rs +0 -396
  185. data/vendor/html-to-markdown-rs/tests/test_spa_bisect.rs +0 -34
  186. data/vendor/html-to-markdown-rs/tests/visitor_code_integration_test.rs +0 -121
  187. data/vendor/html-to-markdown-rs/tests/visitor_integration_test.rs +0 -1190
  188. data/vendor/html-to-markdown-rs/tests/xml_tables_test.rs +0 -372
@@ -1,770 +0,0 @@
1
- //! Handlers for HTML form elements.
2
- //!
3
- //! This module provides comprehensive handling for all HTML form-related elements,
4
- //! including form containers, input controls, and measurement elements.
5
- //!
6
- //! Processed elements:
7
- //! - **Form containers**: `<form>`, `<fieldset>`, `<legend>`, `<label>`
8
- //! - **Text inputs**: `<input>`, `<textarea>`, `<button>`
9
- //! - **Select controls**: `<select>`, `<option>`, `<optgroup>`
10
- //! - **Measurement**: `<progress>`, `<meter>`, `<output>`, `<datalist>`
11
- //!
12
- //! In Markdown, forms are typically not fully representable, so the handlers
13
- //! extract and format the content in a readable manner.
14
-
15
- // Note: Context and DomContext are defined in converter.rs
16
- // walk_node is also defined there and must be called via the parent module
17
- use super::walk_node;
18
- use std::borrow::Cow;
19
-
20
- /// Handles the `<form>` element.
21
- ///
22
- /// A form element is a container for form controls. In Markdown, it's rendered
23
- /// as a block container with its content visible.
24
- ///
25
- /// # Behavior
26
- ///
27
- /// - **Inline mode**: Children are processed inline without block spacing
28
- /// - **Block mode**: Content is collected, trimmed, and wrapped with blank lines
29
- /// - **Empty content**: Skipped entirely
30
- #[cfg_attr(not(feature = "visitor"), allow(unused_variables))]
31
- pub fn handle_form(
32
- _tag_name: &str,
33
- node_handle: &tl::NodeHandle,
34
- parser: &tl::Parser,
35
- output: &mut String,
36
- options: &crate::options::ConversionOptions,
37
- ctx: &super::Context,
38
- depth: usize,
39
- dom_ctx: &super::DomContext,
40
- ) {
41
- if let Some(tl::Node::Tag(tag)) = node_handle.get(parser) {
42
- #[cfg(feature = "visitor")]
43
- if let Some(ref visitor_handle) = ctx.visitor {
44
- use crate::converter::utility::content::collect_tag_attributes;
45
- use crate::visitor::{NodeContext, NodeType, VisitResult};
46
-
47
- let attributes = collect_tag_attributes(tag);
48
- let node_id = node_handle.get_inner();
49
- let parent_tag = dom_ctx.parent_tag_name(node_id, parser);
50
- let index_in_parent = dom_ctx.get_sibling_index(node_id).unwrap_or(0);
51
- let action = tag
52
- .attributes()
53
- .get("action")
54
- .flatten()
55
- .map(|v| v.as_utf8_str().into_owned());
56
- let method = tag
57
- .attributes()
58
- .get("method")
59
- .flatten()
60
- .map(|v| v.as_utf8_str().into_owned());
61
- let node_ctx = NodeContext {
62
- node_type: NodeType::Form,
63
- tag_name: "form".to_string(),
64
- attributes,
65
- depth,
66
- index_in_parent,
67
- parent_tag,
68
- is_inline: false,
69
- };
70
- let visit_result = {
71
- let mut visitor = visitor_handle.borrow_mut();
72
- visitor.visit_form(&node_ctx, action.as_deref(), method.as_deref())
73
- };
74
- match visit_result {
75
- VisitResult::Continue => {}
76
- VisitResult::Skip => return,
77
- VisitResult::Custom(custom) => {
78
- if !output.is_empty() && !output.ends_with("\n\n") {
79
- output.push_str("\n\n");
80
- }
81
- output.push_str(&custom);
82
- output.push_str("\n\n");
83
- return;
84
- }
85
- VisitResult::PreserveHtml => {
86
- use crate::converter::utility::serialization::serialize_node;
87
- output.push_str(&serialize_node(node_handle, parser));
88
- return;
89
- }
90
- VisitResult::Error(err) => {
91
- if ctx.visitor_error.borrow().is_none() {
92
- *ctx.visitor_error.borrow_mut() = Some(err);
93
- }
94
- return;
95
- }
96
- }
97
- }
98
-
99
- // In inline context, just process children inline
100
- if ctx.convert_as_inline {
101
- let children = tag.children();
102
- {
103
- for child_handle in children.top().iter() {
104
- super::walk_node(child_handle, parser, output, options, ctx, depth, dom_ctx);
105
- }
106
- }
107
- return;
108
- }
109
-
110
- // Collect content
111
- let mut content = String::new();
112
- let children = tag.children();
113
- {
114
- for child_handle in children.top().iter() {
115
- walk_node(child_handle, parser, &mut content, options, ctx, depth, dom_ctx);
116
- }
117
- }
118
-
119
- let trimmed = content.trim();
120
- if !trimmed.is_empty() {
121
- // Add spacing before if needed
122
- if !output.is_empty() && !output.ends_with("\n\n") {
123
- output.push_str("\n\n");
124
- }
125
-
126
- // Output content
127
- output.push_str(trimmed);
128
- output.push_str("\n\n");
129
- }
130
- }
131
- }
132
-
133
- /// Handles the `<fieldset>` element.
134
- ///
135
- /// A fieldset element groups form controls with an optional legend.
136
- /// In Markdown, it's rendered as a block container.
137
- ///
138
- /// # Behavior
139
- ///
140
- /// - **Inline mode**: Children are processed inline without block spacing
141
- /// - **Block mode**: Content is collected, trimmed, and wrapped with blank lines
142
- /// - **Empty content**: Skipped entirely
143
- pub fn handle_fieldset(
144
- _tag_name: &str,
145
- node_handle: &tl::NodeHandle,
146
- parser: &tl::Parser,
147
- output: &mut String,
148
- options: &crate::options::ConversionOptions,
149
- ctx: &super::Context,
150
- depth: usize,
151
- dom_ctx: &super::DomContext,
152
- ) {
153
- if let Some(tl::Node::Tag(tag)) = node_handle.get(parser) {
154
- // In inline context, just process children inline
155
- if ctx.convert_as_inline {
156
- let children = tag.children();
157
- {
158
- for child_handle in children.top().iter() {
159
- super::walk_node(child_handle, parser, output, options, ctx, depth, dom_ctx);
160
- }
161
- }
162
- return;
163
- }
164
-
165
- // Collect content
166
- let mut content = String::new();
167
- let children = tag.children();
168
- {
169
- for child_handle in children.top().iter() {
170
- walk_node(child_handle, parser, &mut content, options, ctx, depth, dom_ctx);
171
- }
172
- }
173
-
174
- let trimmed = content.trim();
175
- if !trimmed.is_empty() {
176
- // Add spacing before if needed
177
- if !output.is_empty() && !output.ends_with("\n\n") {
178
- output.push_str("\n\n");
179
- }
180
-
181
- // Output content
182
- output.push_str(trimmed);
183
- output.push_str("\n\n");
184
- }
185
- }
186
- }
187
-
188
- /// Handles the `<legend>` element.
189
- ///
190
- /// A legend element provides a caption for a fieldset. It's rendered as
191
- /// strong (bold) text to distinguish it from regular content.
192
- ///
193
- /// # Behavior
194
- ///
195
- /// - **Block mode**: Content is wrapped in strong markers (e.g., `**text**`)
196
- /// - **Inline mode**: Content is rendered without emphasis
197
- /// - Uses the configured strong/emphasis symbol from ConversionOptions
198
- pub fn handle_legend(
199
- _tag_name: &str,
200
- node_handle: &tl::NodeHandle,
201
- parser: &tl::Parser,
202
- output: &mut String,
203
- options: &crate::options::ConversionOptions,
204
- ctx: &super::Context,
205
- depth: usize,
206
- dom_ctx: &super::DomContext,
207
- ) {
208
- if let Some(tl::Node::Tag(tag)) = node_handle.get(parser) {
209
- let mut content = String::new();
210
-
211
- // Set strong context for nested content
212
- let mut legend_ctx = ctx.clone();
213
- if !ctx.convert_as_inline {
214
- legend_ctx.in_strong = true;
215
- }
216
-
217
- let children = tag.children();
218
- {
219
- for child_handle in children.top().iter() {
220
- super::walk_node(
221
- child_handle,
222
- parser,
223
- &mut content,
224
- options,
225
- &legend_ctx,
226
- depth + 1,
227
- dom_ctx,
228
- );
229
- }
230
- }
231
-
232
- let trimmed = content.trim();
233
- if !trimmed.is_empty() {
234
- if ctx.convert_as_inline {
235
- output.push_str(trimmed);
236
- } else {
237
- let mut symbol = String::with_capacity(2);
238
- symbol.push(options.strong_em_symbol);
239
- symbol.push(options.strong_em_symbol);
240
- output.push_str(&symbol);
241
- output.push_str(trimmed);
242
- output.push_str(&symbol);
243
- output.push_str("\n\n");
244
- }
245
- }
246
- }
247
- }
248
-
249
- /// Handles the `<label>` element.
250
- ///
251
- /// A label element associates text with a form control. It's rendered as
252
- /// a block or inline element depending on context.
253
- ///
254
- /// # Behavior
255
- ///
256
- /// - Content is collected from children
257
- /// - Non-empty content is output followed by blank lines (in block mode)
258
- /// - Blank lines are suppressed in inline mode
259
- pub fn handle_label(
260
- _tag_name: &str,
261
- node_handle: &tl::NodeHandle,
262
- parser: &tl::Parser,
263
- output: &mut String,
264
- options: &crate::options::ConversionOptions,
265
- ctx: &super::Context,
266
- depth: usize,
267
- dom_ctx: &super::DomContext,
268
- ) {
269
- if let Some(tl::Node::Tag(tag)) = node_handle.get(parser) {
270
- let mut content = String::new();
271
- let children = tag.children();
272
- {
273
- for child_handle in children.top().iter() {
274
- super::walk_node(child_handle, parser, &mut content, options, ctx, depth + 1, dom_ctx);
275
- }
276
- }
277
-
278
- let trimmed = content.trim();
279
- if !trimmed.is_empty() {
280
- output.push_str(trimmed);
281
- if !ctx.convert_as_inline {
282
- output.push_str("\n\n");
283
- }
284
- }
285
- }
286
- }
287
-
288
- /// Handles the `<input>` element.
289
- ///
290
- /// An input element represents a form control for user input. Since input
291
- /// elements typically have no text content, this handler produces no output.
292
- #[cfg_attr(not(feature = "visitor"), allow(unused_variables))]
293
- #[allow(clippy::ptr_arg, clippy::needless_pass_by_ref_mut)]
294
- pub fn handle_input(
295
- _tag_name: &str,
296
- node_handle: &tl::NodeHandle,
297
- parser: &tl::Parser,
298
- output: &mut String,
299
- _options: &crate::options::ConversionOptions,
300
- ctx: &super::Context,
301
- depth: usize,
302
- dom_ctx: &super::DomContext,
303
- ) {
304
- #[cfg(feature = "visitor")]
305
- if let Some(ref visitor_handle) = ctx.visitor {
306
- use crate::converter::utility::content::collect_tag_attributes;
307
- use crate::visitor::{NodeContext, NodeType, VisitResult};
308
-
309
- if let Some(tl::Node::Tag(tag)) = node_handle.get(parser) {
310
- let attributes = collect_tag_attributes(tag);
311
- let node_id = node_handle.get_inner();
312
- let parent_tag = dom_ctx.parent_tag_name(node_id, parser);
313
- let index_in_parent = dom_ctx.get_sibling_index(node_id).unwrap_or(0);
314
- let input_type = tag
315
- .attributes()
316
- .get("type")
317
- .flatten()
318
- .map_or_else(|| std::borrow::Cow::Borrowed("text"), |v| v.as_utf8_str())
319
- .into_owned();
320
- let name = tag
321
- .attributes()
322
- .get("name")
323
- .flatten()
324
- .map(|v| v.as_utf8_str().into_owned());
325
- let value = tag
326
- .attributes()
327
- .get("value")
328
- .flatten()
329
- .map(|v| v.as_utf8_str().into_owned());
330
- let node_ctx = NodeContext {
331
- node_type: NodeType::Input,
332
- tag_name: "input".to_string(),
333
- attributes,
334
- depth,
335
- index_in_parent,
336
- parent_tag,
337
- is_inline: true,
338
- };
339
- let visit_result = {
340
- let mut visitor = visitor_handle.borrow_mut();
341
- visitor.visit_input(&node_ctx, &input_type, name.as_deref(), value.as_deref())
342
- };
343
- match visit_result {
344
- VisitResult::Continue => {}
345
- VisitResult::Skip => (),
346
- VisitResult::Custom(custom) => {
347
- output.push_str(&custom);
348
- }
349
- VisitResult::PreserveHtml => {
350
- use crate::converter::utility::serialization::serialize_node;
351
- output.push_str(&serialize_node(node_handle, parser));
352
- }
353
- VisitResult::Error(err) => {
354
- if ctx.visitor_error.borrow().is_none() {
355
- *ctx.visitor_error.borrow_mut() = Some(err);
356
- }
357
- }
358
- }
359
- }
360
- }
361
- // Input elements have no text content; render nothing
362
- }
363
-
364
- /// Handles the `<textarea>` element.
365
- ///
366
- /// A textarea element represents a multi-line text input. Its content is
367
- /// rendered as plain text, with blank lines added after in block mode.
368
- ///
369
- /// # Behavior
370
- ///
371
- /// - Content is collected from children
372
- /// - Blank lines are added after content in block mode only
373
- pub fn handle_textarea(
374
- _tag_name: &str,
375
- node_handle: &tl::NodeHandle,
376
- parser: &tl::Parser,
377
- output: &mut String,
378
- options: &crate::options::ConversionOptions,
379
- ctx: &super::Context,
380
- depth: usize,
381
- dom_ctx: &super::DomContext,
382
- ) {
383
- if let Some(tl::Node::Tag(tag)) = node_handle.get(parser) {
384
- let start_len = output.len();
385
- let children = tag.children();
386
- {
387
- for child_handle in children.top().iter() {
388
- walk_node(child_handle, parser, output, options, ctx, depth + 1, dom_ctx);
389
- }
390
- }
391
-
392
- if !ctx.convert_as_inline && output.len() > start_len {
393
- output.push_str("\n\n");
394
- }
395
- }
396
- }
397
-
398
- /// Handles the `<select>` element.
399
- ///
400
- /// A select element represents a dropdown list of options. Its options are
401
- /// rendered as a list, with newlines between options.
402
- ///
403
- /// # Behavior
404
- ///
405
- /// - Content (options) is collected from children
406
- /// - A single newline is added after the select in block mode
407
- pub fn handle_select(
408
- _tag_name: &str,
409
- node_handle: &tl::NodeHandle,
410
- parser: &tl::Parser,
411
- output: &mut String,
412
- options: &crate::options::ConversionOptions,
413
- ctx: &super::Context,
414
- depth: usize,
415
- dom_ctx: &super::DomContext,
416
- ) {
417
- if let Some(tl::Node::Tag(tag)) = node_handle.get(parser) {
418
- let start_len = output.len();
419
- let children = tag.children();
420
- {
421
- for child_handle in children.top().iter() {
422
- walk_node(child_handle, parser, output, options, ctx, depth + 1, dom_ctx);
423
- }
424
- }
425
-
426
- if !ctx.convert_as_inline && output.len() > start_len {
427
- output.push('\n');
428
- }
429
- }
430
- }
431
-
432
- /// Handles the `<option>` element.
433
- ///
434
- /// An option element represents a choice within a select element.
435
- /// Selected options are marked with a bullet point (`*`) in block mode.
436
- ///
437
- /// # Behavior
438
- ///
439
- /// - Content is collected from children
440
- /// - If the option has the `selected` attribute, it's prefixed with `* ` in block mode
441
- /// - A newline is added after each option in block mode
442
- pub fn handle_option(
443
- _tag_name: &str,
444
- node_handle: &tl::NodeHandle,
445
- parser: &tl::Parser,
446
- output: &mut String,
447
- options: &crate::options::ConversionOptions,
448
- ctx: &super::Context,
449
- depth: usize,
450
- dom_ctx: &super::DomContext,
451
- ) {
452
- if let Some(tl::Node::Tag(tag)) = node_handle.get(parser) {
453
- let selected = tag.attributes().iter().any(|(name, _)| name.as_ref() == "selected");
454
-
455
- let mut text = String::new();
456
- let children = tag.children();
457
- {
458
- for child_handle in children.top().iter() {
459
- super::walk_node(child_handle, parser, &mut text, options, ctx, depth + 1, dom_ctx);
460
- }
461
- }
462
-
463
- let trimmed = text.trim();
464
- if !trimmed.is_empty() {
465
- if selected && !ctx.convert_as_inline {
466
- output.push_str("* ");
467
- }
468
- output.push_str(trimmed);
469
- if !ctx.convert_as_inline {
470
- output.push('\n');
471
- }
472
- }
473
- }
474
- }
475
-
476
- /// Handles the `<optgroup>` element.
477
- ///
478
- /// An optgroup element groups options within a select element with an optional label.
479
- /// The label is rendered as strong (bold) text, followed by the grouped options.
480
- ///
481
- /// # Behavior
482
- ///
483
- /// - The `label` attribute is output as strong text (if present)
484
- /// - Options within the group are rendered normally
485
- pub fn handle_optgroup(
486
- _tag_name: &str,
487
- node_handle: &tl::NodeHandle,
488
- parser: &tl::Parser,
489
- output: &mut String,
490
- options: &crate::options::ConversionOptions,
491
- ctx: &super::Context,
492
- depth: usize,
493
- dom_ctx: &super::DomContext,
494
- ) {
495
- if let Some(tl::Node::Tag(tag)) = node_handle.get(parser) {
496
- let label = tag
497
- .attributes()
498
- .get("label")
499
- .flatten()
500
- .map_or(Cow::Borrowed(""), |v| v.as_utf8_str());
501
-
502
- if !label.is_empty() {
503
- let mut symbol = String::with_capacity(2);
504
- symbol.push(options.strong_em_symbol);
505
- symbol.push(options.strong_em_symbol);
506
- output.push_str(&symbol);
507
- output.push_str(&label);
508
- output.push_str(&symbol);
509
- output.push('\n');
510
- }
511
-
512
- let children = tag.children();
513
- {
514
- for child_handle in children.top().iter() {
515
- walk_node(child_handle, parser, output, options, ctx, depth + 1, dom_ctx);
516
- }
517
- }
518
- }
519
- }
520
-
521
- /// Handles the `<button>` element.
522
- ///
523
- /// A button element represents a clickable button. Its text content is rendered
524
- /// as plain text, with blank lines added in block mode.
525
- ///
526
- /// # Behavior
527
- ///
528
- /// - Content is collected from children
529
- /// - Blank lines are added after content in block mode only
530
- #[cfg_attr(not(feature = "visitor"), allow(unused_variables))]
531
- pub fn handle_button(
532
- _tag_name: &str,
533
- node_handle: &tl::NodeHandle,
534
- parser: &tl::Parser,
535
- output: &mut String,
536
- options: &crate::options::ConversionOptions,
537
- ctx: &super::Context,
538
- depth: usize,
539
- dom_ctx: &super::DomContext,
540
- ) {
541
- if let Some(tl::Node::Tag(tag)) = node_handle.get(parser) {
542
- #[cfg(feature = "visitor")]
543
- if let Some(ref visitor_handle) = ctx.visitor {
544
- use crate::converter::get_text_content;
545
- use crate::converter::utility::content::collect_tag_attributes;
546
- use crate::visitor::{NodeContext, NodeType, VisitResult};
547
-
548
- let text = get_text_content(node_handle, parser, dom_ctx);
549
- let attributes = collect_tag_attributes(tag);
550
- let node_id = node_handle.get_inner();
551
- let parent_tag = dom_ctx.parent_tag_name(node_id, parser);
552
- let index_in_parent = dom_ctx.get_sibling_index(node_id).unwrap_or(0);
553
- let node_ctx = NodeContext {
554
- node_type: NodeType::Button,
555
- tag_name: "button".to_string(),
556
- attributes,
557
- depth,
558
- index_in_parent,
559
- parent_tag,
560
- is_inline: true,
561
- };
562
- let visit_result = {
563
- let mut visitor = visitor_handle.borrow_mut();
564
- visitor.visit_button(&node_ctx, &text)
565
- };
566
- match visit_result {
567
- VisitResult::Continue => {}
568
- VisitResult::Skip => return,
569
- VisitResult::Custom(custom) => {
570
- output.push_str(&custom);
571
- if !ctx.convert_as_inline && !custom.ends_with('\n') {
572
- output.push_str("\n\n");
573
- }
574
- return;
575
- }
576
- VisitResult::PreserveHtml => {
577
- use crate::converter::utility::serialization::serialize_node;
578
- output.push_str(&serialize_node(node_handle, parser));
579
- return;
580
- }
581
- VisitResult::Error(err) => {
582
- if ctx.visitor_error.borrow().is_none() {
583
- *ctx.visitor_error.borrow_mut() = Some(err);
584
- }
585
- return;
586
- }
587
- }
588
- }
589
-
590
- let start_len = output.len();
591
- let children = tag.children();
592
- {
593
- for child_handle in children.top().iter() {
594
- walk_node(child_handle, parser, output, options, ctx, depth + 1, dom_ctx);
595
- }
596
- }
597
-
598
- if !ctx.convert_as_inline && output.len() > start_len {
599
- output.push_str("\n\n");
600
- }
601
- }
602
- }
603
-
604
- /// Handles the `<progress>` element.
605
- ///
606
- /// A progress element represents a progress bar. It typically has no visible
607
- /// text content, but we render any child content present.
608
- ///
609
- /// # Behavior
610
- ///
611
- /// - Content is collected from children (usually empty)
612
- /// - Blank lines are added after content in block mode only
613
- pub fn handle_progress(
614
- _tag_name: &str,
615
- node_handle: &tl::NodeHandle,
616
- parser: &tl::Parser,
617
- output: &mut String,
618
- options: &crate::options::ConversionOptions,
619
- ctx: &super::Context,
620
- depth: usize,
621
- dom_ctx: &super::DomContext,
622
- ) {
623
- if let Some(tl::Node::Tag(tag)) = node_handle.get(parser) {
624
- let start_len = output.len();
625
- let children = tag.children();
626
- {
627
- for child_handle in children.top().iter() {
628
- walk_node(child_handle, parser, output, options, ctx, depth + 1, dom_ctx);
629
- }
630
- }
631
-
632
- if !ctx.convert_as_inline && output.len() > start_len {
633
- output.push_str("\n\n");
634
- }
635
- }
636
- }
637
-
638
- /// Handles the `<meter>` element.
639
- ///
640
- /// A meter element represents a scalar measurement (e.g., disk usage, temperature).
641
- /// It typically has no visible text content, but we render any child content.
642
- ///
643
- /// # Behavior
644
- ///
645
- /// - Content is collected from children (usually empty)
646
- /// - Blank lines are added after content in block mode only
647
- pub fn handle_meter(
648
- _tag_name: &str,
649
- node_handle: &tl::NodeHandle,
650
- parser: &tl::Parser,
651
- output: &mut String,
652
- options: &crate::options::ConversionOptions,
653
- ctx: &super::Context,
654
- depth: usize,
655
- dom_ctx: &super::DomContext,
656
- ) {
657
- if let Some(tl::Node::Tag(tag)) = node_handle.get(parser) {
658
- let start_len = output.len();
659
- let children = tag.children();
660
- {
661
- for child_handle in children.top().iter() {
662
- walk_node(child_handle, parser, output, options, ctx, depth + 1, dom_ctx);
663
- }
664
- }
665
-
666
- if !ctx.convert_as_inline && output.len() > start_len {
667
- output.push_str("\n\n");
668
- }
669
- }
670
- }
671
-
672
- /// Handles the `<output>` element.
673
- ///
674
- /// An output element represents the result of a calculation. It renders its
675
- /// text content as plain output, with blank lines in block mode.
676
- ///
677
- /// # Behavior
678
- ///
679
- /// - Content is collected from children
680
- /// - Blank lines are added after content in block mode only
681
- pub fn handle_output(
682
- _tag_name: &str,
683
- node_handle: &tl::NodeHandle,
684
- parser: &tl::Parser,
685
- output: &mut String,
686
- options: &crate::options::ConversionOptions,
687
- ctx: &super::Context,
688
- depth: usize,
689
- dom_ctx: &super::DomContext,
690
- ) {
691
- if let Some(tl::Node::Tag(tag)) = node_handle.get(parser) {
692
- let start_len = output.len();
693
- let children = tag.children();
694
- {
695
- for child_handle in children.top().iter() {
696
- walk_node(child_handle, parser, output, options, ctx, depth + 1, dom_ctx);
697
- }
698
- }
699
-
700
- if !ctx.convert_as_inline && output.len() > start_len {
701
- output.push_str("\n\n");
702
- }
703
- }
704
- }
705
-
706
- /// Handles the `<datalist>` element.
707
- ///
708
- /// A datalist element provides a list of predefined options for an input element.
709
- /// Options are rendered as a list with newlines between them.
710
- ///
711
- /// # Behavior
712
- ///
713
- /// - Content (options) is collected from children
714
- /// - A single newline is added after the datalist in block mode
715
- pub fn handle_datalist(
716
- _tag_name: &str,
717
- node_handle: &tl::NodeHandle,
718
- parser: &tl::Parser,
719
- output: &mut String,
720
- options: &crate::options::ConversionOptions,
721
- ctx: &super::Context,
722
- depth: usize,
723
- dom_ctx: &super::DomContext,
724
- ) {
725
- if let Some(tl::Node::Tag(tag)) = node_handle.get(parser) {
726
- let start_len = output.len();
727
- let children = tag.children();
728
- {
729
- for child_handle in children.top().iter() {
730
- walk_node(child_handle, parser, output, options, ctx, depth + 1, dom_ctx);
731
- }
732
- }
733
-
734
- if !ctx.convert_as_inline && output.len() > start_len {
735
- output.push('\n');
736
- }
737
- }
738
- }
739
-
740
- /// Dispatcher for form elements.
741
- ///
742
- /// Routes all form-related elements to their respective handlers.
743
- pub fn handle(
744
- tag_name: &str,
745
- node_handle: &tl::NodeHandle,
746
- parser: &tl::Parser,
747
- output: &mut String,
748
- options: &crate::options::ConversionOptions,
749
- ctx: &super::Context,
750
- depth: usize,
751
- dom_ctx: &super::DomContext,
752
- ) {
753
- match tag_name {
754
- "form" => handle_form(tag_name, node_handle, parser, output, options, ctx, depth, dom_ctx),
755
- "fieldset" => handle_fieldset(tag_name, node_handle, parser, output, options, ctx, depth, dom_ctx),
756
- "legend" => handle_legend(tag_name, node_handle, parser, output, options, ctx, depth, dom_ctx),
757
- "label" => handle_label(tag_name, node_handle, parser, output, options, ctx, depth, dom_ctx),
758
- "input" => handle_input(tag_name, node_handle, parser, output, options, ctx, depth, dom_ctx),
759
- "textarea" => handle_textarea(tag_name, node_handle, parser, output, options, ctx, depth, dom_ctx),
760
- "select" => handle_select(tag_name, node_handle, parser, output, options, ctx, depth, dom_ctx),
761
- "option" => handle_option(tag_name, node_handle, parser, output, options, ctx, depth, dom_ctx),
762
- "optgroup" => handle_optgroup(tag_name, node_handle, parser, output, options, ctx, depth, dom_ctx),
763
- "button" => handle_button(tag_name, node_handle, parser, output, options, ctx, depth, dom_ctx),
764
- "progress" => handle_progress(tag_name, node_handle, parser, output, options, ctx, depth, dom_ctx),
765
- "meter" => handle_meter(tag_name, node_handle, parser, output, options, ctx, depth, dom_ctx),
766
- "output" => handle_output(tag_name, node_handle, parser, output, options, ctx, depth, dom_ctx),
767
- "datalist" => handle_datalist(tag_name, node_handle, parser, output, options, ctx, depth, dom_ctx),
768
- _ => {}
769
- }
770
- }