snapper_fmt/parser/
org.rs

1use regex::Regex;
2use std::sync::LazyLock;
3
4use crate::parser::{FormatParser, Region};
5
6static HEADLINE_RE: LazyLock<Regex> =
7    LazyLock::new(|| Regex::new(r"^(\*+\s+(?:TODO\s+|DONE\s+|NEXT\s+|WAIT\s+)?)(.*)$").unwrap());
8
9static LIST_ITEM_RE: LazyLock<Regex> =
10    LazyLock::new(|| Regex::new(r"^(\s*(?:[-+]|\d+[.)]) )(.*)$").unwrap());
11
12pub struct OrgParser;
13
14impl OrgParser {
15    /// Check if a line starts a block (#+BEGIN_...)
16    fn is_block_begin(line: &str) -> bool {
17        let trimmed = line.trim_start();
18        trimmed.to_ascii_uppercase().starts_with("#+BEGIN_")
19    }
20
21    /// Check if a line ends a block (#+END_...)
22    fn is_block_end(line: &str) -> bool {
23        let trimmed = line.trim_start();
24        trimmed.to_ascii_uppercase().starts_with("#+END_")
25    }
26
27    /// Check if a line starts a property drawer
28    fn is_drawer_begin(line: &str) -> bool {
29        let trimmed = line.trim();
30        trimmed.starts_with(':') && trimmed.ends_with(':') && trimmed.len() > 2
31    }
32
33    /// Check if a line ends a drawer
34    fn is_drawer_end(line: &str) -> bool {
35        line.trim().eq_ignore_ascii_case(":END:")
36    }
37
38    /// Check if a line is a keyword/directive (#+KEYWORD:)
39    fn is_keyword(line: &str) -> bool {
40        let trimmed = line.trim_start();
41        trimmed.starts_with("#+") && !Self::is_block_begin(line) && !Self::is_block_end(line)
42    }
43
44    /// Check if a line is a comment (starts with #, but not #+)
45    fn is_comment(line: &str) -> bool {
46        let trimmed = line.trim_start();
47        trimmed.starts_with('#') && !trimmed.starts_with("#+")
48    }
49
50    /// Check if a line is a table row
51    fn is_table_row(line: &str) -> bool {
52        line.trim_start().starts_with('|')
53    }
54}
55
56impl FormatParser for OrgParser {
57    fn parse(&self, input: &str) -> Vec<Region> {
58        let mut regions: Vec<Region> = Vec::new();
59        let mut current_prose = String::new();
60        let mut in_block = false;
61        let mut in_drawer = false;
62        let mut pragma_off = false;
63        // Track list item context: indent level of the marker text.
64        // Continuation lines indented at or beyond this level belong to the item.
65        let mut list_item_indent: Option<usize> = None;
66
67        let flush_prose = |prose: &mut String, regions: &mut Vec<Region>| {
68            if !prose.is_empty() {
69                regions.push(Region::Prose(prose.clone()));
70                prose.clear();
71            }
72        };
73
74        for line in input.lines() {
75            // Check for snapper:off/on pragmas
76            if let Some(on) = super::check_pragma(line) {
77                flush_prose(&mut current_prose, &mut regions);
78                pragma_off = !on;
79                regions.push(Region::Structure(format!("{line}\n")));
80                continue;
81            }
82
83            // Inside pragma-off region: pass through unchanged
84            if pragma_off {
85                flush_prose(&mut current_prose, &mut regions);
86                regions.push(Region::Structure(format!("{line}\n")));
87                continue;
88            }
89
90            // Inside a block -- everything is structure
91            if in_block {
92                flush_prose(&mut current_prose, &mut regions);
93                if Self::is_block_end(line) {
94                    in_block = false;
95                }
96                regions.push(Region::Structure(format!("{line}\n")));
97                continue;
98            }
99
100            // Inside a drawer -- everything is structure
101            if in_drawer {
102                flush_prose(&mut current_prose, &mut regions);
103                if Self::is_drawer_end(line) {
104                    in_drawer = false;
105                }
106                regions.push(Region::Structure(format!("{line}\n")));
107                continue;
108            }
109
110            // Block begin
111            if Self::is_block_begin(line) {
112                flush_prose(&mut current_prose, &mut regions);
113                in_block = true;
114                regions.push(Region::Structure(format!("{line}\n")));
115                continue;
116            }
117
118            // Drawer begin
119            if Self::is_drawer_begin(line) {
120                flush_prose(&mut current_prose, &mut regions);
121                in_drawer = true;
122                regions.push(Region::Structure(format!("{line}\n")));
123                continue;
124            }
125
126            // Blank line
127            if line.trim().is_empty() {
128                flush_prose(&mut current_prose, &mut regions);
129                list_item_indent = None;
130                regions.push(Region::BlankLines(format!("{line}\n")));
131                continue;
132            }
133
134            // Keyword/directive
135            if Self::is_keyword(line) {
136                flush_prose(&mut current_prose, &mut regions);
137                regions.push(Region::Structure(format!("{line}\n")));
138                continue;
139            }
140
141            // Comment
142            if Self::is_comment(line) {
143                flush_prose(&mut current_prose, &mut regions);
144                regions.push(Region::Structure(format!("{line}\n")));
145                continue;
146            }
147
148            // Table row
149            if Self::is_table_row(line) {
150                flush_prose(&mut current_prose, &mut regions);
151                regions.push(Region::Structure(format!("{line}\n")));
152                continue;
153            }
154
155            // Bare file/http links on their own line -- treat as structure
156            if line.trim_start().starts_with("file:")
157                || line.trim_start().starts_with("http://")
158                || line.trim_start().starts_with("https://")
159            {
160                flush_prose(&mut current_prose, &mut regions);
161                regions.push(Region::Structure(format!("{line}\n")));
162                continue;
163            }
164
165            // Headline: stars + optional keyword are structure, rest is prose
166            if let Some(caps) = HEADLINE_RE.captures(line) {
167                flush_prose(&mut current_prose, &mut regions);
168                let prefix = caps.get(1).unwrap().as_str();
169                let text = caps.get(2).unwrap().as_str();
170                regions.push(Region::Structure(prefix.to_string()));
171                if !text.is_empty() {
172                    regions.push(Region::Prose(text.to_string()));
173                }
174                regions.push(Region::Structure("\n".to_string()));
175                continue;
176            }
177
178            // List item: marker is structure, rest is prose
179            if let Some(caps) = LIST_ITEM_RE.captures(line) {
180                flush_prose(&mut current_prose, &mut regions);
181                let marker = caps.get(1).unwrap().as_str();
182                let text = caps.get(2).unwrap().as_str();
183                // Track indent for continuation detection: text starts at marker length
184                list_item_indent = Some(marker.len());
185                regions.push(Region::Structure(marker.to_string()));
186                if !text.is_empty() {
187                    regions.push(Region::Prose(text.to_string()));
188                }
189                regions.push(Region::Structure("\n".to_string()));
190                continue;
191            }
192
193            // List item continuation: indented line following a list item
194            if let Some(indent) = list_item_indent {
195                let leading = line.len() - line.trim_start().len();
196                if leading >= indent && !line.trim().is_empty() {
197                    // Append to the previous Prose region of the list item.
198                    // The last three regions are Structure(marker), Prose(text), Structure(\n)
199                    // We want to extend the Prose region.
200                    if let Some(Region::Structure(s)) = regions.last() {
201                        if s == "\n" {
202                            regions.pop(); // remove the \n
203                            if let Some(Region::Prose(prose)) = regions.last_mut() {
204                                prose.push(' ');
205                                prose.push_str(line.trim());
206                            }
207                            regions.push(Region::Structure("\n".to_string()));
208                            continue;
209                        }
210                    }
211                }
212                // Not a continuation: leave list context
213                list_item_indent = None;
214            }
215
216            // Regular prose line -- accumulate
217            if !current_prose.is_empty() {
218                current_prose.push(' ');
219            }
220            current_prose.push_str(line.trim());
221        }
222
223        // Flush remaining
224        flush_prose(&mut current_prose, &mut regions);
225
226        regions
227    }
228}
229
230#[cfg(test)]
231mod tests {
232    use super::*;
233
234    #[test]
235    fn simple_prose() {
236        let input = "Hello world. This is a test.\nAnother line here.";
237        let regions = OrgParser.parse(input);
238        assert_eq!(
239            regions,
240            vec![Region::Prose(
241                "Hello world. This is a test. Another line here.".to_string()
242            )]
243        );
244    }
245
246    #[test]
247    fn preserves_blocks() {
248        let input = "Some prose.\n#+BEGIN_SRC python\nprint('hello')\n#+END_SRC\nMore prose.";
249        let regions = OrgParser.parse(input);
250        assert_eq!(regions.len(), 5);
251        assert!(matches!(&regions[0], Region::Prose(_)));
252        assert!(matches!(&regions[1], Region::Structure(_))); // BEGIN
253        assert!(matches!(&regions[2], Region::Structure(_))); // code
254        assert!(matches!(&regions[3], Region::Structure(_))); // END
255        assert!(matches!(&regions[4], Region::Prose(_)));
256    }
257
258    #[test]
259    fn preserves_keywords() {
260        let input = "#+TITLE: My Document\n#+AUTHOR: Someone\n\nSome text here.";
261        let regions = OrgParser.parse(input);
262        assert!(matches!(&regions[0], Region::Structure(_)));
263        assert!(matches!(&regions[1], Region::Structure(_)));
264    }
265
266    #[test]
267    fn headline_split() {
268        let input = "* TODO This is a headline";
269        let regions = OrgParser.parse(input);
270        assert_eq!(regions.len(), 3);
271        assert_eq!(regions[0], Region::Structure("* TODO ".to_string()));
272        assert_eq!(regions[1], Region::Prose("This is a headline".to_string()));
273        assert_eq!(regions[2], Region::Structure("\n".to_string()));
274    }
275
276    #[test]
277    fn table_preserved() {
278        let input = "| Name | Age |\n|------+-----|\n| Alice | 30 |";
279        let regions = OrgParser.parse(input);
280        assert!(regions.iter().all(|r| matches!(r, Region::Structure(_))));
281    }
282
283    #[test]
284    fn list_item_split() {
285        let input = "- First item text\n- Second item text";
286        let regions = OrgParser.parse(input);
287        // Each list item: Structure(marker) + Prose(text) + Structure(\n)
288        assert_eq!(regions.len(), 6);
289        assert_eq!(regions[0], Region::Structure("- ".to_string()));
290        assert_eq!(regions[1], Region::Prose("First item text".to_string()));
291    }
292
293    #[test]
294    fn list_item_continuation() {
295        let input = "- First sentence of item.\n  Continuation of the same item.\n- Second item";
296        let regions = OrgParser.parse(input);
297        // First item: Structure("- ") + Prose("First sentence of item. Continuation of the same item.") + Structure("\n")
298        assert_eq!(regions[0], Region::Structure("- ".to_string()));
299        assert_eq!(
300            regions[1],
301            Region::Prose("First sentence of item. Continuation of the same item.".to_string())
302        );
303        assert_eq!(regions[2], Region::Structure("\n".to_string()));
304        // Second item: Structure("- ") + Prose("Second item") + Structure("\n")
305        assert_eq!(regions[3], Region::Structure("- ".to_string()));
306        assert_eq!(regions[4], Region::Prose("Second item".to_string()));
307    }
308
309    #[test]
310    fn drawer_preserved() {
311        let input = ":PROPERTIES:\n:ID: abc123\n:END:\nSome text.";
312        let regions = OrgParser.parse(input);
313        assert!(matches!(&regions[0], Region::Structure(_))); // :PROPERTIES:
314        assert!(matches!(&regions[1], Region::Structure(_))); // :ID:
315        assert!(matches!(&regions[2], Region::Structure(_))); // :END:
316    }
317}
snapper_fmt/parser/org.rs

snapper_fmt/parser/
org.rs
