Segments model

1/**
2 * Shared model for page-segmented document results — Retab "partition" (keyed
3 * chunks) and "split" (named subdocuments). Both reduce to the same `Segment[]`:
4 * a label, the pages it owns, a deterministic color, and optional confidence.
5 *
6 * The legend, sidebar, and timeline primitives all consume `Segment[]`, so a
7 * partition and a split render through the exact same components.
8 */
9 
10// Tableau 20 — the palette Retab's dashboard uses, so colors match.
11export const SEGMENT_PALETTE = [
12  "#4E79A7",
13  "#A0CBE8",
14  "#F28E2B",
15  "#FFBE7D",
16  "#59A14F",
17  "#8CD17D",
18  "#B6992D",
19  "#F1CE63",
20  "#499894",
21  "#86BCB6",
22  "#E15759",
23  "#FF9D9A",
24  "#79706E",
25  "#BAB0AC",
26  "#D37295",
27  "#FABFD2",
28  "#B07AA1",
29  "#D4A6C8",
30  "#9D7660",
31  "#D7B5A6",
32] as const;
33 
34export interface Segment {
35  /** Stable id (label + occurrence, since a split name can repeat). */
36  id: string;
37  /** Display label — the partition key or the subdocument name. */
38  label: string;
39  /** Sorted, de-duplicated 1-based pages owned by this segment. */
40  pages: number[];
41  /** Deterministic color (by label, so the same label is always one color). */
42  color: string;
43  /** Render order. */
44  index: number;
45  /** Optional consensus confidence in [0, 1]. */
46  confidence?: number | null;
47}
48 
49export interface SegmentChunk {
50  key?: string;
51  name?: string;
52  pages: number[];
53}
54 
55export function segmentDisplayLabel(label: string): string {
56  const trimmed = typeof label === "string" ? label.trim() : "";
57  return trimmed || "unnamed";
58}
59 
60export function normalizePageCount(value: number): number {
61  return Number.isFinite(value) && value > 0 ? Math.floor(value) : 0;
62}
63 
64/** Assign one palette color per distinct display label, ordered by sorted label. */
65export function buildColorMap(labels: string[]): Map<string, string> {
66  const distinct = Array.from(new Set(labels.map(segmentDisplayLabel))).sort(
67    (a, b) => a.localeCompare(b),
68  );
69  const map = new Map<string, string>();
70  distinct.forEach((label, i) => {
71    map.set(label, SEGMENT_PALETTE[i % SEGMENT_PALETTE.length]);
72  });
73  return map;
74}
75 
76function normalizePages(pages: number[]): number[] {
77  const rawPages = Array.isArray(pages) ? pages : [];
78  return Array.from(
79    new Set(rawPages.filter((p) => Number.isInteger(p) && p > 0)),
80  ).sort((a, b) => a - b);
81}
82 
83export function segmentPageCount(pages: number[]): number {
84  return normalizePages(pages).length;
85}
86 
87export function firstSegmentPage(pages: number[]): number | null {
88// … truncated