symforge 11.3.0

SymForge — in-memory code intelligence MCP server for Claude Code, Codex, and Gemini CLI
Documentation
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
use super::{ConfigExtractor, EditCapability, ExtractionOutcome, ExtractionResult};
use crate::domain::{SymbolKind, SymbolRecord};
use std::collections::HashMap;

use super::parse_diagnostic;

/// Detects a CommonMark fenced-code-block delimiter line.
///
/// Returns `Some((fence_char, run_length, info_empty))` when `line`, after up to
/// three leading spaces, begins with a run of at least three identical backticks
/// (`` ` ``) or tildes (`~`). `info_empty` is true when only whitespace follows
/// the run — a closing fence carries no info string, so this distinguishes a
/// valid closer from an opener like ```` ```ruby ````. Returns `None` otherwise.
fn parse_code_fence(line: &str) -> Option<(u8, usize, bool)> {
    // Up to three leading spaces are allowed before a fence; four or more make
    // the line an indented code block, which is never a fence delimiter.
    let leading_spaces = line.bytes().take_while(|&b| b == b' ').count();
    if leading_spaces > 3 {
        return None;
    }
    let trimmed = &line[leading_spaces..];

    let fence_char = match trimmed.bytes().next()? {
        b'`' => b'`',
        b'~' => b'~',
        _ => return None,
    };
    let run_length = trimmed.bytes().take_while(|&b| b == fence_char).count();
    if run_length < 3 {
        return None;
    }
    let info = &trimmed[run_length..];
    let info_empty = info.trim().is_empty();
    Some((fence_char, run_length, info_empty))
}

pub struct MarkdownExtractor;

impl ConfigExtractor for MarkdownExtractor {
    fn extract(&self, content: &[u8]) -> ExtractionResult {
        let decoded = match crate::knowledge::decode_searchable_text(content) {
            Ok(decoded) => decoded,
            Err(_) => {
                return ExtractionResult {
                    symbols: vec![],
                    outcome: ExtractionOutcome::Failed(parse_diagnostic(
                        "utf-8",
                        "Invalid UTF-8",
                        None,
                        None,
                        None,
                        false,
                    )),
                };
            }
        };
        let text = decoded.text;

        if text.is_empty() {
            return ExtractionResult {
                symbols: vec![],
                outcome: ExtractionOutcome::Ok,
            };
        }

        // Collect (byte_offset, line_text) pairs, skipping YAML frontmatter.
        let mut lines: Vec<(usize, &str)> = Vec::new();
        {
            let raw: Vec<&str> = text.split('\n').collect();
            let mut i = 0usize;
            let mut byte_offset = decoded.leading_bytes as usize;
            fn clean_line(line: &str) -> &str {
                line.strip_suffix('\r').unwrap_or(line)
            }

            // Check for frontmatter
            if raw.first().copied().map(clean_line) == Some("---") {
                byte_offset += raw[0].len() + 1; // skip opening ---\n
                i = 1;
                let mut closed = false;
                while i < raw.len() {
                    let line_bytes = raw[i].len() + 1;
                    if clean_line(raw[i]) == "---" {
                        byte_offset += line_bytes;
                        i += 1;
                        closed = true;
                        break;
                    }
                    byte_offset += line_bytes;
                    i += 1;
                }
                if !closed {
                    return ExtractionResult {
                        symbols: vec![],
                        outcome: ExtractionOutcome::Ok,
                    };
                }
            }

            while i < raw.len() {
                lines.push((byte_offset, clean_line(raw[i])));
                byte_offset += raw[i].len();
                // Only add 1 for the newline delimiter if this isn't the last
                // segment or the original text actually ends with a newline.
                if i + 1 < raw.len() || text.ends_with('\n') {
                    byte_offset += 1;
                }
                i += 1;
            }
        }

        // Parse ATX and Setext headers from collected lines.
        struct HeaderInfo {
            level: u32,
            text: String,
            byte_start: usize,
            line_index: usize,
        }

        let mut headers: Vec<HeaderInfo> = Vec::new();
        // Track fenced code blocks so that '#' comment lines inside ``` or ~~~
        // fences (e.g. shell/ruby comments) are not mistaken for ATX headings,
        // per CommonMark. Holds (fence_char, opener_run_length) while open.
        let mut in_fence: Option<(u8, usize)> = None;
        for (li, &(byte_off, line)) in lines.iter().enumerate() {
            if let Some(fence) = parse_code_fence(line) {
                let (fence_char, fence_len, info_empty) = fence;
                match in_fence {
                    // Opening fence: remember its char and run length. An info
                    // string (e.g. ```ruby) is allowed on the opener.
                    None => in_fence = Some((fence_char, fence_len)),
                    // Closing fence: same char, run >= opener, and nothing but
                    // whitespace after the run (a closing fence has no info).
                    Some((open_char, open_len))
                        if fence_char == open_char && fence_len >= open_len && info_empty =>
                    {
                        in_fence = None;
                    }
                    // Otherwise the line stays inside the current fence as code.
                    Some(_) => {}
                }
                continue;
            }

            // Lines inside a fence are code, never headings.
            if in_fence.is_some() {
                continue;
            }

            let leading_spaces = line.bytes().take_while(|&byte| byte == b' ').count();
            let trimmed = &line[leading_spaces.min(line.len())..];

            if leading_spaces <= 3 && trimmed.starts_with('#') {
                let hashes = trimmed.bytes().take_while(|&byte| byte == b'#').count();
                if hashes <= 6 {
                    let rest = &trimmed[hashes..];
                    if rest.is_empty()
                        || rest
                            .bytes()
                            .next()
                            .is_some_and(|byte| byte == b' ' || byte == b'\t')
                    {
                        let mut title = rest.trim().to_string();
                        let closing_hashes = title.bytes().rev().take_while(|&b| b == b'#').count();
                        if closing_hashes > 0 {
                            let before = title.len() - closing_hashes;
                            if before == 0
                                || title.as_bytes()[before - 1].is_ascii_whitespace()
                            {
                                title.truncate(before);
                                title = title.trim_end().to_string();
                            }
                        }
                        headers.push(HeaderInfo {
                            level: hashes as u32,
                            text: title,
                            byte_start: byte_off,
                            line_index: li,
                        });
                        continue;
                    }
                }
            }

            let underline = trimmed.trim_end();
            let setext_level = if leading_spaces <= 3
                && !underline.is_empty()
                && underline.bytes().all(|byte| byte == b'=')
            {
                Some(1)
            } else if leading_spaces <= 3
                && !underline.is_empty()
                && underline.bytes().all(|byte| byte == b'-')
            {
                Some(2)
            } else {
                None
            };
            if let Some(level) = setext_level
                && li > 0
            {
                let (title_byte_start, title_line) = lines[li - 1];
                let title = title_line.trim();
                let prior_is_heading = headers.iter().any(|header| header.line_index == li - 1);
                if !title.is_empty()
                    && !prior_is_heading
                    && parse_code_fence(title_line).is_none()
                    && !title.starts_with('#')
                {
                    headers.push(HeaderInfo {
                        level,
                        text: title.to_string(),
                        byte_start: title_byte_start,
                        line_index: li - 1,
                    });
                }
            }
        }

        if headers.is_empty() {
            return ExtractionResult {
                symbols: vec![],
                outcome: ExtractionOutcome::Ok,
            };
        }

        let total_bytes = content.len() as u32;

        // Build symbols. Stack holds (level, segment) for path building.
        let mut stack: Vec<(u32, String)> = Vec::new();
        // Duplicate counter keyed by base path (before disambiguation suffix).
        let mut seen_paths: HashMap<String, u32> = HashMap::new();
        let mut symbols: Vec<SymbolRecord> = Vec::new();

        for (hi, header) in headers.iter().enumerate() {
            let level = header.level;

            // Pop entries at same or deeper level.
            while stack.last().is_some_and(|&(l, _)| l >= level) {
                stack.pop();
            }

            // Build dot-joined path from the raw heading text. Section names
            // are display-facing and looked up by exact full-name match only —
            // nothing splits them back into segments (unlike TOML key paths),
            // so `.`/`[`/`]` in a heading stay human-readable instead of being
            // escaped to `~1`/`~2`/`~3`. A heading like "A.B" colliding with a
            // nested A > B path falls into the existing #N duplicate handling.
            let segment = header.text.clone();
            let base_path = if stack.is_empty() {
                segment.clone()
            } else {
                let parent: String = stack
                    .iter()
                    .map(|(_, n)| n.as_str())
                    .collect::<Vec<_>>()
                    .join(".");
                format!("{}.{}", parent, segment)
            };

            // Disambiguate duplicates.
            let count = seen_paths.entry(base_path.clone()).or_insert(0);
            *count += 1;
            let name = if *count == 1 {
                base_path.clone()
            } else {
                format!("{}#{}", base_path, count)
            };

            stack.push((level, segment));

            // Byte range: this header's start -> byte before next header at same or higher level (or EOF).
            let byte_end: u32 = headers[hi + 1..]
                .iter()
                .find(|h| h.level <= level)
                .map_or(total_bytes, |h| h.byte_start as u32);
            let byte_range = (header.byte_start as u32, byte_end);

            // Line range: same logic using line indices.
            let line_start = header.line_index as u32;
            let line_end: u32 = headers[hi + 1..]
                .iter()
                .find(|h| h.level <= level)
                .map_or(lines.len() as u32, |h| h.line_index as u32);

            symbols.push(SymbolRecord {
                name,
                kind: SymbolKind::Section,
                depth: level - 1,
                sort_order: hi as u32,
                byte_range,
                item_byte_range: Some(byte_range),
                line_range: (line_start, line_end),
                doc_byte_range: None,
            });
        }

        ExtractionResult {
            symbols,
            outcome: ExtractionOutcome::Ok,
        }
    }

    fn edit_capability(&self) -> EditCapability {
        EditCapability::TextEditSafe
    }
}

#[cfg(test)]
mod tests {
    use super::*;

    fn extract(content: &[u8]) -> Vec<SymbolRecord> {
        MarkdownExtractor.extract(content).symbols
    }

    #[test]
    fn test_single_header() {
        let syms = extract(b"# Title\nSome text\n");
        assert_eq!(syms.len(), 1);
        assert_eq!(syms[0].name, "Title");
        assert_eq!(syms[0].kind, SymbolKind::Section);
        assert_eq!(syms[0].depth, 0);
    }

    #[test]
    fn test_nested_headers() {
        let syms = extract(b"# Top\n## Sub\n### Deep\n");
        assert_eq!(syms.len(), 3);
        assert_eq!(syms[0].name, "Top");
        assert_eq!(syms[1].name, "Top.Sub");
        assert_eq!(syms[2].name, "Top.Sub.Deep");
    }

    #[test]
    fn test_section_byte_range_spans_to_next_header() {
        // "# A\n" = 4, "line1\n" = 6, "line2\n" = 6  → "# B" starts at byte 16
        let content = b"# A\nline1\nline2\n# B\nline3\n";
        let syms = extract(content);
        assert_eq!(syms.len(), 2);
        assert_eq!(syms[0].byte_range.0, 0);
        assert_eq!(syms[0].byte_range.1, 16);
        assert_eq!(syms[1].byte_range.0, 16);
        assert_eq!(syms[1].byte_range.1, content.len() as u32);
    }

    #[test]
    fn test_duplicate_headers_disambiguated() {
        let syms = extract(b"## Install\ntext\n## Install\ntext\n");
        assert_eq!(syms.len(), 2);
        assert_eq!(syms[0].name, "Install");
        assert_eq!(syms[1].name, "Install#2");
    }

    // Dogfood B2: heading punctuation must stay human-readable — no `~1`/`~2`/`~3`
    // key-escaping leaking into display names.
    #[test]
    fn test_heading_with_dots_and_brackets_stays_readable() {
        let syms = extract(b"# SymForge 8.13.0 beta issues\n## [FIXED upstream] note\n");
        assert_eq!(syms.len(), 2);
        assert_eq!(syms[0].name, "SymForge 8.13.0 beta issues");
        assert_eq!(
            syms[1].name,
            "SymForge 8.13.0 beta issues.[FIXED upstream] note"
        );
        assert!(
            !syms.iter().any(|s| s.name.contains("~1")
                || s.name.contains("~2")
                || s.name.contains("~3")),
            "escape encoding must not leak into section names"
        );
    }

    #[test]
    fn test_frontmatter_skipped() {
        let syms = extract(b"---\ntitle: Hello\n---\n# Real Header\n");
        assert_eq!(syms.len(), 1);
        assert_eq!(syms[0].name, "Real Header");
    }

    #[test]
    fn test_empty_file() {
        let syms = extract(b"");
        assert_eq!(syms.len(), 0);
    }

    #[test]
    fn test_edit_capability() {
        assert_eq!(
            MarkdownExtractor.edit_capability(),
            EditCapability::TextEditSafe
        );
    }

    // ---- SF-STRESS-019: fenced code blocks must not produce phantom headings ----

    #[test]
    fn test_hash_inside_backtick_fence_not_heading() {
        let syms = extract(b"# Real\n\n```ruby\n# config/ci.rb\n# not a heading\n```\n");
        assert_eq!(syms.len(), 1, "only the real heading is a section");
        assert_eq!(syms[0].name, "Real");
    }

    #[test]
    fn test_hash_inside_tilde_fence_not_heading() {
        let syms = extract(b"# Real\n\n~~~python\n# also a comment\n~~~\n");
        assert_eq!(syms.len(), 1);
        assert_eq!(syms[0].name, "Real");
    }

    #[test]
    fn test_fence_with_info_string_opens_and_closes() {
        // The rails ```ruby#6-7 case: info strings on the opener must be ignored
        // and must not be treated as a closer.
        let syms = extract(b"# Real\n\n```ruby#6-7\n# TODO: phantom\n```\n## Sub\n");
        assert_eq!(syms.len(), 2);
        assert_eq!(syms[0].name, "Real");
        assert_eq!(syms[1].name, "Real.Sub", "real sub parents to the real heading");
    }

    #[test]
    fn test_unclosed_fence_at_eof_swallows_rest() {
        // An unclosed fence keeps the remainder of the file as code (CommonMark).
        let syms = extract(b"# Real\n\n```\n# inside unclosed fence\n## also inside\n");
        assert_eq!(syms.len(), 1);
        assert_eq!(syms[0].name, "Real");
    }

    #[test]
    fn test_closing_run_longer_than_opener_closes() {
        // Closer run length >= opener run length closes the fence.
        let syms = extract(b"# Real\n\n```\n# inside\n`````\n## After\n");
        assert_eq!(syms.len(), 2);
        assert_eq!(syms[0].name, "Real");
        assert_eq!(syms[1].name, "Real.After");
    }

    #[test]
    fn test_shorter_closing_run_does_not_close() {
        // A run shorter than the opener does not close the fence, so the '#'
        // line after it stays code.
        let syms = extract(b"# Real\n\n````\n# inside\n```\n# still inside\n````\n## After\n");
        assert_eq!(syms.len(), 2);
        assert_eq!(syms[0].name, "Real");
        assert_eq!(syms[1].name, "Real.After");
    }

    #[test]
    fn test_real_heading_after_closed_fence_keeps_parent_path() {
        // Hierarchy repair: a real '## Sub' after a phantom-producing fence must
        // parent under the preceding real heading, not a phantom level-1.
        let syms =
            extract(b"# Top\n\n```ruby\n# config/ci.rb\nclass Foo\nend\n```\n\n## Real Sub\n");
        assert_eq!(syms.len(), 2);
        assert_eq!(syms[0].name, "Top");
        assert_eq!(syms[1].name, "Top.Real Sub");
        assert_eq!(syms[1].depth, 1);
    }

    #[test]
    fn test_indented_fence_within_three_spaces_tracked() {
        // Up to three leading spaces still opens a fence.
        let syms = extract(b"# Real\n\n   ```\n# inside indented fence\n   ```\n## After\n");
        assert_eq!(syms.len(), 2);
        assert_eq!(syms[1].name, "Real.After");
    }

    #[test]
    fn markdown_hit_preserves_section_and_line_pointer() {
        let content = b"\xEF\xBB\xBFpreface\r\n# Section\r\nline \xCE\xB1\r\n## Child\r\nlast";
        let syms = extract(content);

        assert_eq!(syms.len(), 2);
        assert_eq!(syms[0].name, "Section");
        assert_eq!(syms[0].byte_range, (12, content.len() as u32));
        assert_eq!(syms[0].line_range, (1, 5));
        assert_eq!(syms[1].name, "Section.Child");
        assert_eq!(syms[1].byte_range, (32, content.len() as u32));
        assert_eq!(syms[1].line_range, (3, 5));
    }

    #[test]
    fn markdown_commonmark_corpus_handles_atx_setext_fence_frontmatter_table_and_link() {
        let content = br##"---
title: "# frontmatter is not a heading"
---
# ATX [Guide](guide.md) #
| Name | Value |
| --- | --- |

```md
Hidden setext
=============
# Hidden ATX
```

Setext [Top](top.md)
====================

Setext Child
-------------
body
"##;
        let syms = extract(content);

        assert_eq!(syms.len(), 3);
        assert_eq!(syms[0].name, "ATX [Guide](guide.md)");
        assert_eq!(syms[0].depth, 0);
        assert_eq!(syms[1].name, "Setext [Top](top.md)");
        assert_eq!(syms[1].depth, 0);
        assert_eq!(syms[2].name, "Setext [Top](top.md).Setext Child");
        assert_eq!(syms[2].depth, 1);
        assert!(syms.iter().all(|symbol| !symbol.name.contains("Hidden")));
    }
}