* A segmenter that wraps Intl.Segmenter and merges graphemes that fall * within paste markers into single atomic segments. This makes cursor * movement, deletion, word-wrap, etc. treat paste markers as single units. * * Only markers whose numeric ID exists in `validIds` are merged.
( text: string, baseSegmenter: Intl.Segmenter, validIds: Set<number>, )
| 38 | * Only markers whose numeric ID exists in `validIds` are merged. |
| 39 | */ |
| 40 | function segmentWithMarkers( |
| 41 | text: string, |
| 42 | baseSegmenter: Intl.Segmenter, |
| 43 | validIds: Set<number>, |
| 44 | ): Iterable<Intl.SegmentData> { |
| 45 | // Fast path: no paste markers in the text or no valid IDs. |
| 46 | if (validIds.size === 0 || !text.includes("[paste #")) { |
| 47 | return baseSegmenter.segment(text); |
| 48 | } |
| 49 | |
| 50 | // Find all marker spans with valid IDs. |
| 51 | const markers: Array<{ start: number; end: number }> = []; |
| 52 | for (const m of text.matchAll(PASTE_MARKER_REGEX)) { |
| 53 | const id = Number.parseInt(m[1]!, 10); |
| 54 | if (!validIds.has(id)) continue; |
| 55 | markers.push({ start: m.index, end: m.index + m[0].length }); |
| 56 | } |
| 57 | if (markers.length === 0) { |
| 58 | return baseSegmenter.segment(text); |
| 59 | } |
| 60 | |
| 61 | // Build merged segment list. |
| 62 | const baseSegments = baseSegmenter.segment(text); |
| 63 | const result: Intl.SegmentData[] = []; |
| 64 | let markerIdx = 0; |
| 65 | |
| 66 | for (const seg of baseSegments) { |
| 67 | // Skip past markers that are entirely before this segment. |
| 68 | while (markerIdx < markers.length && markers[markerIdx]!.end <= seg.index) { |
| 69 | markerIdx++; |
| 70 | } |
| 71 | |
| 72 | const marker = markerIdx < markers.length ? markers[markerIdx]! : null; |
| 73 | |
| 74 | if (marker && seg.index >= marker.start && seg.index < marker.end) { |
| 75 | // This segment falls inside a marker. |
| 76 | // If this is the first segment of the marker, emit a merged segment. |
| 77 | if (seg.index === marker.start) { |
| 78 | const markerText = text.slice(marker.start, marker.end); |
| 79 | result.push({ |
| 80 | segment: markerText, |
| 81 | index: marker.start, |
| 82 | input: text, |
| 83 | }); |
| 84 | } |
| 85 | // Otherwise skip (already merged into the first segment). |
| 86 | } else { |
| 87 | result.push(seg); |
| 88 | } |
| 89 | } |
| 90 | |
| 91 | return result; |
| 92 | } |
| 93 | |
| 94 | /** |
| 95 | * Represents a chunk of text for word-wrap layout. |