| 1 | use markdown_it::parser::block::{BlockRule, BlockState}; |
| 2 | use markdown_it::parser::inline::{InlineRoot, InlineRule, InlineState}; |
| 3 | use markdown_it::{Node, NodeValue, Renderer}; |
| 4 | |
| 5 | use crate::marko_ast::OpenOrClose; |
| 6 | use crate::plugin::{get_line_raw, ErrorBlock}; |
| 7 | use crate::{marko, marko_ast}; |
| 8 | |
| 9 | /// An opening Marko tag: <div>, <if=cond>, <for|item| of=items>, etc. |
| 10 | #[derive(Debug)] |
| 11 | pub struct MarkoOpen { |
| 12 | pub open: marko_ast::OpenOwned, |
| 13 | } |
| 14 | |
| 15 | impl NodeValue for MarkoOpen { |
| 16 | fn render(&self, _node: &Node, fmt: &mut dyn Renderer) { |
| 17 | fmt.text_raw(&self.open.src); |
| 18 | fmt.text_raw("\n"); |
| 19 | } |
| 20 | } |
| 21 | |
| 22 | /// A closing Marko tag: </div>, </> |
| 23 | #[derive(Debug)] |
| 24 | pub struct MarkoClose { |
| 25 | pub content: String, |
| 26 | pub tag_name: Option<String>, // None for </> |
| 27 | /// Length of the tag name (for error span highlighting), None for </> |
| 28 | pub tag_name_len: Option<usize>, |
| 29 | } |
| 30 | |
| 31 | impl NodeValue for MarkoClose { |
| 32 | fn render(&self, _node: &Node, fmt: &mut dyn Renderer) { |
| 33 | fmt.text_raw(&self.content); |
| 34 | fmt.text_raw("\n"); |
| 35 | } |
| 36 | } |
| 37 | |
| 38 | /// A closing Marko tag with same-line text: </div>text here |
| 39 | /// The text is parsed as inline markdown (children) but not wrapped in <p> |
| 40 | #[derive(Debug)] |
| 41 | pub struct MarkoCloseWithText { |
| 42 | pub content: String, |
| 43 | pub tag_name: Option<String>, // None for </> |
| 44 | /// Length of the tag name (for error span highlighting), None for </> |
| 45 | pub tag_name_len: Option<usize>, |
| 46 | } |
| 47 | |
| 48 | impl NodeValue for MarkoCloseWithText { |
| 49 | fn render(&self, node: &Node, fmt: &mut dyn Renderer) { |
| 50 | fmt.text_raw(&self.content); |
| 51 | fmt.text_raw("\n"); |
| 52 | fmt.contents(&node.children); |
| 53 | fmt.text_raw("\n"); |
| 54 | } |
| 55 | } |
| 56 | |
| 57 | /// A self-closing tag: <input />, <Chart data=x /> |
| 58 | #[derive(Debug)] |
| 59 | pub struct MarkoSelfClosing { |
| 60 | pub content: String, |
| 61 | } |
| 62 | |
| 63 | impl NodeValue for MarkoSelfClosing { |
| 64 | fn render(&self, _node: &Node, fmt: &mut dyn Renderer) { |
| 65 | fmt.text_raw(&self.content); |
| 66 | fmt.text_raw("\n"); |
| 67 | } |
| 68 | } |
| 69 | |
| 70 | /// An opening Marko tag with same-line text: <div>text here |
| 71 | /// The text is parsed as inline markdown (children) but not wrapped in <p> |
| 72 | #[derive(Debug)] |
| 73 | pub struct MarkoOpenWithText { |
| 74 | pub open: marko_ast::OpenOwned, |
| 75 | } |
| 76 | |
| 77 | impl NodeValue for MarkoOpenWithText { |
| 78 | fn render(&self, node: &Node, fmt: &mut dyn Renderer) { |
| 79 | fmt.text_raw(&self.open.src); |
| 80 | fmt.text_raw("\n"); |
| 81 | fmt.contents(&node.children); |
| 82 | fmt.text_raw("\n"); |
| 83 | } |
| 84 | } |
| 85 | |
| 86 | // ============================================================================ |
| 87 | // Inline tag markers and types |
| 88 | // ============================================================================ |
| 89 | |
| 90 | /// Marker for an inline open tag, to be matched by post-processor |
| 91 | #[derive(Debug)] |
| 92 | pub struct MarkoInlineOpenMarker { |
| 93 | pub content: String, |
| 94 | pub tag_name: String, |
| 95 | pub tag_name_len: usize, |
| 96 | } |
| 97 | |
| 98 | impl NodeValue for MarkoInlineOpenMarker { |
| 99 | fn render(&self, _node: &Node, _fmt: &mut dyn Renderer) { |
| 100 | // Should be replaced by post-processor before rendering |
| 101 | panic!("MarkoInlineOpenMarker should be processed before rendering"); |
| 102 | } |
| 103 | } |
| 104 | |
| 105 | /// Marker for an inline close tag, to be matched by post-processor |
| 106 | #[derive(Debug)] |
| 107 | pub struct MarkoInlineCloseMarker { |
| 108 | pub content: String, |
| 109 | pub tag_name: Option<String>, // None for </> |
| 110 | pub tag_name_len: Option<usize>, |
| 111 | } |
| 112 | |
| 113 | impl NodeValue for MarkoInlineCloseMarker { |
| 114 | fn render(&self, _node: &Node, _fmt: &mut dyn Renderer) { |
| 115 | // Should be replaced by post-processor before rendering |
| 116 | panic!("MarkoInlineCloseMarker should be processed before rendering"); |
| 117 | } |
| 118 | } |
| 119 | |
| 120 | /// A matched inline Marko tag with content: <button>click me</button> |
| 121 | #[derive(Debug)] |
| 122 | pub struct MarkoInlineTag { |
| 123 | pub open_tag: String, |
| 124 | pub close_tag: String, |
| 125 | pub tag_name: String, |
| 126 | } |
| 127 | |
| 128 | impl NodeValue for MarkoInlineTag { |
| 129 | fn render(&self, node: &Node, fmt: &mut dyn Renderer) { |
| 130 | fmt.text_raw(&self.open_tag); |
| 131 | fmt.contents(&node.children); |
| 132 | fmt.text_raw(&self.close_tag); |
| 133 | } |
| 134 | } |
| 135 | |
| 136 | /// A complete block-level tag on one line: <tag>content</tag> |
| 137 | /// The content is parsed as inline markdown |
| 138 | #[derive(Debug)] |
| 139 | pub struct MarkoBlockComplete { |
| 140 | pub open: marko_ast::OpenOwned, |
| 141 | pub close_tag: String, |
| 142 | } |
| 143 | |
| 144 | impl NodeValue for MarkoBlockComplete { |
| 145 | fn render(&self, node: &Node, fmt: &mut dyn Renderer) { |
| 146 | fmt.text_raw(&self.open.src); |
| 147 | fmt.contents(&node.children); |
| 148 | fmt.text_raw(&self.close_tag); |
| 149 | fmt.text_raw("\n"); |
| 150 | } |
| 151 | } |
| 152 | |
| 153 | /// Find a matching close tag in content for the given tag name. |
| 154 | /// Returns (content_before_close, close_tag_text) if found. |
| 155 | /// |
| 156 | /// This function scans forward through the content, tracking nested tags. |
| 157 | /// A close tag only matches the outer tag when nesting depth is zero. |
| 158 | fn find_same_line_close<'a>(content: &'a str, tag_name: &str) -> Option<(&'a str, &'a str)> { |
| 159 | // Stack of open tag names for tracking nesting |
| 160 | let mut stack: Vec<&str> = Vec::new(); |
| 161 | let mut pos = 0; |
| 162 | |
| 163 | while pos < content.len() { |
| 164 | // Find next '<' |
| 165 | let Some(lt_pos) = content[pos..].find('<') else { |
| 166 | break; |
| 167 | }; |
| 168 | let lt_pos = pos + lt_pos; |
| 169 | |
| 170 | // Try to parse as a tag |
| 171 | match marko::parse_tag(&content[lt_pos..]) { |
| 172 | Ok(OpenOrClose::Open(open)) => { |
| 173 | if !open.self_closing { |
| 174 | stack.push(open.tag_name()); |
| 175 | } |
| 176 | pos = lt_pos + (open.src.len()); |
| 177 | } |
| 178 | Ok(OpenOrClose::Close(close)) => { |
| 179 | if stack.is_empty() { |
| 180 | if close.tag_name.map(|n| n == tag_name).unwrap_or(true) { |
| 181 | let content_before = &content[..lt_pos]; |
| 182 | let close_tag = &content[lt_pos..lt_pos + close.length as usize]; |
| 183 | return Some((content_before, close_tag)); |
| 184 | } |
| 185 | pos = lt_pos + close.length as usize; |
| 186 | } else { |
| 187 | if close.tag_name.is_none() || stack.last() == close.tag_name.as_ref() { |
| 188 | stack.pop(); |
| 189 | } |
| 190 | pos = lt_pos + close.length as usize; |
| 191 | } |
| 192 | } |
| 193 | Err(_) => { |
| 194 | // TODO: why no forward |
| 195 | pos = lt_pos + 1; |
| 196 | } |
| 197 | } |
| 198 | } |
| 199 | |
| 200 | None |
| 201 | } |
| 202 | |
| 203 | /// Parse Marko tags at block level |
| 204 | pub struct Rule; |
| 205 | |
| 206 | impl BlockRule for Rule { |
| 207 | fn run(state: &mut BlockState) -> Option<(Node, usize)> { |
| 208 | let line = get_line_raw(state, state.line); |
| 209 | let trimmed = line.trim_start(); |
| 210 | if !trimmed.starts_with("<") { |
| 211 | return None; |
| 212 | } |
| 213 | |
| 214 | let unbounded_src = &state.src[state.line_offsets[state.line].first_nonspace..]; |
| 215 | |
| 216 | match marko::parse_tag(unbounded_src) { |
| 217 | Ok(OpenOrClose::Open(open)) => { |
| 218 | // Count how many lines this tag spans |
| 219 | let lines_consumed = 1 + open.src.chars().filter(|&c| c == '\n').count(); |
| 220 | |
| 221 | if open.self_closing { |
| 222 | // <input /> - self-closing tag |
| 223 | return Some(( |
| 224 | Node::new(MarkoSelfClosing { |
| 225 | content: open.src.to_string(), |
| 226 | }), |
| 227 | lines_consumed, |
| 228 | )); |
| 229 | } |
| 230 | |
| 231 | // Check if there's content on the same line after the tag |
| 232 | // We need to look at unbounded_src, not trimmed, since tag may span multiple lines |
| 233 | let rest_after_tag = &unbounded_src[open.src.len()..]; |
| 234 | let rest_of_last_line = rest_after_tag.lines().next().unwrap_or(""); |
| 235 | if !rest_of_last_line.trim().is_empty() { |
| 236 | let text = rest_of_last_line.trim(); |
| 237 | |
| 238 | // Check if the same-line content contains a matching close tag |
| 239 | if let Some((content, close_tag)) = find_same_line_close(text, open.tag_name()) |
| 240 | { |
| 241 | // Complete tag on one line: <tag>content</tag>. |
| 242 | // If the previous line is non-empty prose (not another tag), this |
| 243 | // self-contained tag is inline content within a paragraph - let |
| 244 | // the paragraph rule handle it instead. |
| 245 | if state.line > 0 { |
| 246 | let prev_line = get_line_raw(state, state.line - 1).trim(); |
| 247 | if !prev_line.is_empty() && !prev_line.starts_with('<') { |
| 248 | return None; |
| 249 | } |
| 250 | } |
| 251 | |
| 252 | let content = content.trim(); |
| 253 | |
| 254 | // Calculate byte offset for source mapping |
| 255 | let text_start_in_line = open.src.len() |
| 256 | + (rest_of_last_line.len() - rest_of_last_line.trim_start().len()); |
| 257 | let line_start = state.line_offsets[state.line].first_nonspace; |
| 258 | let text_start = line_start + text_start_in_line; |
| 259 | |
| 260 | let mapping = vec![(0, text_start)]; |
| 261 | |
| 262 | let mut node = Node::new(MarkoBlockComplete { |
| 263 | open: open.to_owned(), |
| 264 | close_tag: close_tag.to_string(), |
| 265 | }); |
| 266 | |
| 267 | if !content.is_empty() { |
| 268 | node.children |
| 269 | .push(Node::new(InlineRoot::new(content.to_string(), mapping))); |
| 270 | } |
| 271 | |
| 272 | return Some((node, lines_consumed)); |
| 273 | } |
| 274 | |
| 275 | // No close tag - just content after open tag |
| 276 | // Parse the text as inline markdown (not wrapped in paragraph) |
| 277 | let text_start_in_line = open.src.len() |
| 278 | + (rest_of_last_line.len() - rest_of_last_line.trim_start().len()); |
| 279 | let line_start = state.line_offsets[state.line].first_nonspace; |
| 280 | let text_start = line_start + text_start_in_line; |
| 281 | |
| 282 | let mapping = vec![(0, text_start)]; |
| 283 | |
| 284 | let mut node = Node::new(MarkoOpenWithText { |
| 285 | open: open.to_owned(), |
| 286 | }); |
| 287 | node.children |
| 288 | .push(Node::new(InlineRoot::new(text.to_string(), mapping))); |
| 289 | |
| 290 | return Some((node, lines_consumed)); |
| 291 | } |
| 292 | |
| 293 | // Check if next line after the tag has content (non-blank) that isn't another tag |
| 294 | let next_line_idx = state.line + lines_consumed; |
| 295 | if next_line_idx < state.line_max { |
| 296 | let next_line = get_line_raw(state, next_line_idx); |
| 297 | let next_trimmed = next_line.trim(); |
| 298 | if !next_trimmed.is_empty() && !next_trimmed.starts_with("<") { |
| 299 | // Content immediately follows without blank line |
| 300 | // Parse as inline markdown (not wrapped in paragraph) |
| 301 | let text = next_trimmed; |
| 302 | |
| 303 | // Calculate byte offset of next line's text for source mapping |
| 304 | let next_line_start = state.line_offsets[next_line_idx].first_nonspace; |
| 305 | |
| 306 | let mapping = vec![(0, next_line_start)]; |
| 307 | |
| 308 | let mut node = Node::new(MarkoOpenWithText { |
| 309 | open: open.to_owned(), |
| 310 | }); |
| 311 | node.children |
| 312 | .push(Node::new(InlineRoot::new(text.to_string(), mapping))); |
| 313 | |
| 314 | // Consume the tag lines + the content line |
| 315 | return Some((node, lines_consumed + 1)); |
| 316 | } |
| 317 | } |
| 318 | |
| 319 | // Clean case: open tag on its own line(s), followed by blank or another tag |
| 320 | Some(( |
| 321 | Node::new(MarkoOpen { |
| 322 | open: open.to_owned(), |
| 323 | }), |
| 324 | lines_consumed, |
| 325 | )) |
| 326 | } |
| 327 | Ok(OpenOrClose::Close(close)) => { |
| 328 | // Close tags should be single line, but check anyway |
| 329 | let lines_consumed = 1 + close.length.saturating_sub(1).min( |
| 330 | unbounded_src[..close.length as usize] |
| 331 | .chars() |
| 332 | .filter(|&c| c == '\n') |
| 333 | .count() as u32, |
| 334 | ); |
| 335 | |
| 336 | // Check if there's content on the same line after the close tag |
| 337 | let close_text = &unbounded_src[..close.length as usize]; |
| 338 | let rest_after_tag = &unbounded_src[close.length as usize..]; |
| 339 | let rest_of_line = rest_after_tag.lines().next().unwrap_or(""); |
| 340 | if !rest_of_line.trim().is_empty() { |
| 341 | let text = rest_of_line.trim(); |
| 342 | |
| 343 | // Calculate byte offset of the text for source mapping |
| 344 | let text_start_in_line = close.length as usize |
| 345 | + (rest_of_line.len() - rest_of_line.trim_start().len()); |
| 346 | let line_start = state.line_offsets[state.line].first_nonspace; |
| 347 | let text_start = line_start + text_start_in_line; |
| 348 | |
| 349 | let mapping = vec![(0, text_start)]; |
| 350 | |
| 351 | let mut node = Node::new(MarkoCloseWithText { |
| 352 | content: close_text.to_string(), |
| 353 | tag_name: close.tag_name.map(|s| s.to_string()), |
| 354 | tag_name_len: close.tag_name.map(|s| s.len()), |
| 355 | }); |
| 356 | node.children |
| 357 | .push(Node::new(InlineRoot::new(text.to_string(), mapping))); |
| 358 | |
| 359 | return Some((node, lines_consumed as usize)); |
| 360 | } |
| 361 | |
| 362 | Some(( |
| 363 | Node::new(MarkoClose { |
| 364 | content: close_text.to_string(), |
| 365 | tag_name: close.tag_name.map(|s| s.to_string()), |
| 366 | tag_name_len: close.tag_name.map(|s| s.len()), |
| 367 | }), |
| 368 | lines_consumed as usize, |
| 369 | )) |
| 370 | } |
| 371 | Err(err) => Some((Node::new(ErrorBlock { errors: vec![err] }), 1)), |
| 372 | } |
| 373 | } |
| 374 | } |
| 375 | |
| 376 | impl InlineRule for Rule { |
| 377 | const MARKER: char = '<'; |
| 378 | |
| 379 | fn run(state: &mut InlineState) -> Option<(Node, usize)> { |
| 380 | let input = &state.src[state.pos..state.pos_max]; |
| 381 | if !input.starts_with("<") { |
| 382 | return None; |
| 383 | } |
| 384 | |
| 385 | match marko::parse_tag(input) { |
| 386 | Ok(OpenOrClose::Open(open)) => { |
| 387 | if open.self_closing { |
| 388 | return Some(( |
| 389 | Node::new(MarkoSelfClosing { |
| 390 | content: open.src.to_string(), |
| 391 | }), |
| 392 | open.src.len(), |
| 393 | )); |
| 394 | } |
| 395 | |
| 396 | Some(( |
| 397 | Node::new(MarkoInlineOpenMarker { |
| 398 | content: open.src.to_string(), |
| 399 | tag_name: open.tag_name().to_owned(), |
| 400 | tag_name_len: open.tag_name().len(), |
| 401 | }), |
| 402 | open.src.len(), |
| 403 | )) |
| 404 | } |
| 405 | Ok(OpenOrClose::Close(close)) => { |
| 406 | let close_text = &input[..close.length as usize]; |
| 407 | Some(( |
| 408 | Node::new(MarkoInlineCloseMarker { |
| 409 | content: close_text.to_string(), |
| 410 | tag_name: close.tag_name.map(|s| s.to_string()), |
| 411 | tag_name_len: close.tag_name.map(|s| s.len()), |
| 412 | }), |
| 413 | close.length as usize, |
| 414 | )) |
| 415 | } |
| 416 | // TODO: why this no reported |
| 417 | Err(_) => None, |
| 418 | } |
| 419 | } |
| 420 | } |
| 421 | |
| 422 | #[cfg(test)] |
| 423 | mod tests { |
| 424 | use super::find_same_line_close; |
| 425 | use crate::marko; |
| 426 | |
| 427 | #[test] |
| 428 | fn test_find_same_line_close_with_nested_tags() { |
| 429 | // Should find </> after the nested <code></code> |
| 430 | let content = "<code>aaa</code> mmmm</>"; |
| 431 | let result = find_same_line_close(content, "h3"); |
| 432 | assert!(result.is_some(), "Should find closing tag"); |
| 433 | let (before, close) = result.unwrap(); |
| 434 | assert_eq!(before, "<code>aaa</code> mmmm"); |
| 435 | assert_eq!(close, "</>"); |
| 436 | } |
| 437 | |
| 438 | #[test] |
| 439 | fn test_block_tag_with_inline_content() { |
| 440 | // the string |
| 441 | // 'bbbbbbb'><code>aaa</code> mmmm</> |
| 442 | // |
| 443 | // 'string' > <T>id < /regexp/ > |
| 444 | let src = "<h3 id='bbbbbbb'><code>aaa</code> mmmm</>"; |
| 445 | let open = marko::parse_open(src).unwrap(); |
| 446 | assert_eq!(open.tag_name(), "h3"); |
| 447 | assert_eq!(open.src, "<h3 id='bbbbbbb'>"); |
| 448 | |
| 449 | // The rest after the tag |
| 450 | let rest = &src[open.src.len()..]; |
| 451 | assert_eq!(rest, "<code>aaa</code> mmmm</>"); |
| 452 | |
| 453 | // find_same_line_close should find the </> at the end |
| 454 | let result = find_same_line_close(rest, "h3"); |
| 455 | assert!(result.is_some(), "Should find closing tag for h3"); |
| 456 | let (content, close) = result.unwrap(); |
| 457 | assert_eq!(content, "<code>aaa</code> mmmm"); |
| 458 | assert_eq!(close, "</>"); |
| 459 | } |
| 460 | |
| 461 | #[test] |
| 462 | fn test_nested_greater_than() { |
| 463 | let src = "<h3 id='bbbbbbb'><code>aaa</code> mmmm</>"; |
| 464 | let open = marko::parse_open(src).unwrap(); |
| 465 | assert_eq!(open.tag_name(), "h3"); |
| 466 | assert_eq!(open.src, "<h3 id='bbbbbbb'>"); |
| 467 | |
| 468 | // The rest after the tag |
| 469 | let rest = &src[open.src.len()..]; |
| 470 | assert_eq!(rest, "<code>aaa</code> mmmm</>"); |
| 471 | |
| 472 | // find_same_line_close should find the </> at the end |
| 473 | let result = find_same_line_close(rest, "h3"); |
| 474 | assert!(result.is_some(), "Should find closing tag for h3"); |
| 475 | let (content, close) = result.unwrap(); |
| 476 | assert_eq!(content, "<code>aaa</code> mmmm"); |
| 477 | assert_eq!(close, "</>"); |
| 478 | } |
| 479 | } |