marked_text.rs

  1use std::{cmp::Ordering, collections::HashMap, ops::Range};
  2
  3/// Construct a string and a list of offsets within that string using a single
  4/// string containing embedded position markers.
  5pub fn marked_text_offsets_by(
  6    marked_text: &str,
  7    markers: Vec<char>,
  8) -> (String, HashMap<char, Vec<usize>>) {
  9    let mut extracted_markers: HashMap<char, Vec<usize>> = Default::default();
 10    let mut unmarked_text = String::new();
 11
 12    for char in marked_text.chars() {
 13        if markers.contains(&char) {
 14            let char_offsets = extracted_markers.entry(char).or_insert(Vec::new());
 15            char_offsets.push(unmarked_text.len());
 16        } else {
 17            unmarked_text.push(char);
 18        }
 19    }
 20
 21    (unmarked_text, extracted_markers)
 22}
 23
 24/// Construct a string and a list of ranges within that string using a single
 25/// string containing embedded range markers, using arbitrary characters as
 26/// range markers. By using multiple different range markers, you can construct
 27/// ranges that overlap each other.
 28///
 29/// The returned ranges will be grouped by their range marking characters.
 30pub fn marked_text_ranges_by(
 31    marked_text: &str,
 32    markers: Vec<TextRangeMarker>,
 33) -> (String, HashMap<TextRangeMarker, Vec<Range<usize>>>) {
 34    let all_markers = markers.iter().flat_map(|m| m.markers()).collect();
 35
 36    let (unmarked_text, mut marker_offsets) = marked_text_offsets_by(marked_text, all_markers);
 37    let range_lookup = markers
 38        .into_iter()
 39        .map(|marker| {
 40            (
 41                marker.clone(),
 42                match marker {
 43                    TextRangeMarker::Empty(empty_marker_char) => marker_offsets
 44                        .remove(&empty_marker_char)
 45                        .unwrap_or_default()
 46                        .into_iter()
 47                        .map(|empty_index| empty_index..empty_index)
 48                        .collect::<Vec<Range<usize>>>(),
 49                    TextRangeMarker::Range(start_marker, end_marker) => {
 50                        let starts = marker_offsets.remove(&start_marker).unwrap_or_default();
 51                        let ends = marker_offsets.remove(&end_marker).unwrap_or_default();
 52                        assert_eq!(starts.len(), ends.len(), "marked ranges are unbalanced");
 53                        starts
 54                            .into_iter()
 55                            .zip(ends)
 56                            .map(|(start, end)| {
 57                                assert!(end >= start, "marked ranges must be disjoint");
 58                                start..end
 59                            })
 60                            .collect::<Vec<Range<usize>>>()
 61                    }
 62                    TextRangeMarker::ReverseRange(start_marker, end_marker) => {
 63                        let starts = marker_offsets.remove(&start_marker).unwrap_or_default();
 64                        let ends = marker_offsets.remove(&end_marker).unwrap_or_default();
 65                        assert_eq!(starts.len(), ends.len(), "marked ranges are unbalanced");
 66                        starts
 67                            .into_iter()
 68                            .zip(ends)
 69                            .map(|(start, end)| {
 70                                assert!(end >= start, "marked ranges must be disjoint");
 71                                end..start
 72                            })
 73                            .collect::<Vec<Range<usize>>>()
 74                    }
 75                },
 76            )
 77        })
 78        .collect();
 79
 80    (unmarked_text, range_lookup)
 81}
 82
 83/// Construct a string and a list of ranges within that string using a single
 84/// string containing embedded range markers. The characters used to mark the
 85/// ranges are as follows:
 86///
 87/// 1. To mark a range of text, surround it with the `«` and `»` angle brackets,
 88///    which can be typed on a US keyboard with the `alt-|` and `alt-shift-|` keys.
 89///
 90///    ```text
 91///    foo «selected text» bar
 92///    ```
 93///
 94/// 2. To mark a single position in the text, use the `ˇ` caron,
 95///    which can be typed on a US keyboard with the `alt-shift-t` key.
 96///
 97///    ```text
 98///    the cursors are hereˇ and hereˇ.
 99///    ```
100///
101/// 3. To mark a range whose direction is meaningful (like a selection),
102///    put a caron character beside one of its bounds, on the inside:
103///
104///    ```text
105///    one «ˇreversed» selection and one «forwardˇ» selection
106///    ```
107pub fn marked_text_ranges(
108    marked_text: &str,
109    ranges_are_directed: bool,
110) -> (String, Vec<Range<usize>>) {
111    let mut unmarked_text = String::with_capacity(marked_text.len());
112    let mut ranges = Vec::new();
113    let mut prev_marked_ix = 0;
114    let mut current_range_start = None;
115    let mut current_range_cursor = None;
116
117    for (marked_ix, marker) in marked_text.match_indices(&['«', '»', 'ˇ']) {
118        unmarked_text.push_str(&marked_text[prev_marked_ix..marked_ix]);
119        let unmarked_len = unmarked_text.len();
120        let len = marker.len();
121        prev_marked_ix = marked_ix + len;
122
123        match marker {
124            "ˇ" => {
125                if current_range_start.is_some() {
126                    if current_range_cursor.is_some() {
127                        panic!("duplicate point marker 'ˇ' at index {marked_ix}");
128                    } else {
129                        current_range_cursor = Some(unmarked_len);
130                    }
131                } else {
132                    ranges.push(unmarked_len..unmarked_len);
133                }
134            }
135            "«" => {
136                if current_range_start.is_some() {
137                    panic!("unexpected range start marker '«' at index {marked_ix}");
138                }
139                current_range_start = Some(unmarked_len);
140            }
141            "»" => {
142                let current_range_start = if let Some(start) = current_range_start.take() {
143                    start
144                } else {
145                    panic!("unexpected range end marker '»' at index {marked_ix}");
146                };
147
148                let mut reversed = false;
149                if let Some(current_range_cursor) = current_range_cursor.take() {
150                    if current_range_cursor == current_range_start {
151                        reversed = true;
152                    } else if current_range_cursor != unmarked_len {
153                        panic!("unexpected 'ˇ' marker in the middle of a range");
154                    }
155                } else if ranges_are_directed {
156                    panic!("missing 'ˇ' marker to indicate range direction");
157                }
158
159                ranges.push(if reversed {
160                    unmarked_len..current_range_start
161                } else {
162                    current_range_start..unmarked_len
163                });
164            }
165            _ => unreachable!(),
166        }
167    }
168
169    unmarked_text.push_str(&marked_text[prev_marked_ix..]);
170    (unmarked_text, ranges)
171}
172
173pub fn marked_text_offsets(marked_text: &str) -> (String, Vec<usize>) {
174    let (text, ranges) = marked_text_ranges(marked_text, false);
175    (
176        text,
177        ranges
178            .into_iter()
179            .map(|range| {
180                assert_eq!(range.start, range.end);
181                range.start
182            })
183            .collect(),
184    )
185}
186
187pub fn generate_marked_text(
188    unmarked_text: &str,
189    ranges: &[Range<usize>],
190    indicate_cursors: bool,
191) -> String {
192    let mut marked_text = unmarked_text.to_string();
193    for range in ranges.iter().rev() {
194        if indicate_cursors {
195            match range.start.cmp(&range.end) {
196                Ordering::Less => {
197                    marked_text.insert_str(range.end, "ˇ»");
198                    marked_text.insert(range.start, '«');
199                }
200                Ordering::Equal => {
201                    marked_text.insert(range.start, 'ˇ');
202                }
203                Ordering::Greater => {
204                    marked_text.insert(range.start, '»');
205                    marked_text.insert_str(range.end, "«ˇ");
206                }
207            }
208        } else {
209            marked_text.insert(range.end, '»');
210            marked_text.insert(range.start, '«');
211        }
212    }
213    marked_text
214}
215
216#[derive(Clone, Eq, PartialEq, Hash)]
217pub enum TextRangeMarker {
218    Empty(char),
219    Range(char, char),
220    ReverseRange(char, char),
221}
222
223impl TextRangeMarker {
224    fn markers(&self) -> Vec<char> {
225        match self {
226            Self::Empty(m) => vec![*m],
227            Self::Range(l, r) => vec![*l, *r],
228            Self::ReverseRange(l, r) => vec![*l, *r],
229        }
230    }
231}
232
233impl From<char> for TextRangeMarker {
234    fn from(marker: char) -> Self {
235        Self::Empty(marker)
236    }
237}
238
239impl From<(char, char)> for TextRangeMarker {
240    fn from((left_marker, right_marker): (char, char)) -> Self {
241        Self::Range(left_marker, right_marker)
242    }
243}
244
245#[cfg(test)]
246mod tests {
247    use super::{generate_marked_text, marked_text_ranges};
248
249    #[test]
250    fn test_marked_text() {
251        let (text, ranges) = marked_text_ranges("one «ˇtwo» «threeˇ» «ˇfour» fiveˇ six", true);
252
253        assert_eq!(text, "one two three four five six");
254        assert_eq!(ranges.len(), 4);
255        assert_eq!(ranges[0], 7..4);
256        assert_eq!(ranges[1], 8..13);
257        assert_eq!(ranges[2], 18..14);
258        assert_eq!(ranges[3], 23..23);
259
260        assert_eq!(
261            generate_marked_text(&text, &ranges, true),
262            "one «ˇtwo» «threeˇ» «ˇfour» fiveˇ six"
263        );
264    }
265}