velum-raycast/src/lib/project-review.ts

import type { PlaceholderMapping } from "../types";
import { whitelistKey } from "./project-whitelist";
import type { ProjectWhitelistEntry } from "./project-whitelist";

const SEPARATOR = "\n\n\n\n";
export const MAX_BATCH_CODEPOINTS = 24000;

export type BatchItem = { key: string; text: string };
export type BatchSegment = BatchItem & { start: number; end: number };
export type TextBatch = { text: string; segments: BatchSegment[] };

/** Filename separators look like spaces to NER, except inside known literal terms. */
export function filenameScanText(stem: string, knownTerms: Iterable<string>): string {
  const protectedOffsets = new Set<number>();
  for (const term of knownTerms) {
    if (!term.includes("_")) continue;
    const escaped = term.replace(/[.*+?^${}()|[\]\\]/g, "\\$&");
    const pattern = new RegExp(`(?<![\\p{L}\\p{N}])${escaped}(?![\\p{L}\\p{N}])`, "giu");
    for (const match of stem.matchAll(pattern)) {
      for (let offset = match.index; offset < match.index + match[0].length; offset++) protectedOffsets.add(offset);
    }
  }
  return stem.replace(/_/g, (_character, offset: number) => protectedOffsets.has(offset) ? "_" : " ");
}

/** Keep file boundaries while reducing many small text/path API calls to a few batches. */
export function makeTextBatches(items: BatchItem[], limit = MAX_BATCH_CODEPOINTS): TextBatch[] {
  const batches: TextBatch[] = [];
  let segments: BatchSegment[] = [];
  let text = "";
  let length = 0;
  const separatorLength = [...SEPARATOR].length;
  const flush = () => {
    if (segments.length) batches.push({ text, segments });
    segments = [];
    text = "";
    length = 0;
  };
  for (const item of items) {
    if (!item.text) continue;
    const itemLength = [...item.text].length;
    if (segments.length && length + separatorLength + itemLength > limit) flush();
    if (segments.length) { text += SEPARATOR; length += separatorLength; }
    segments.push({ ...item, start: length, end: length + itemLength });
    text += item.text;
    length += itemLength;
  }
  flush();
  return batches;
}

export function spansBySegment<T extends { start: number; end: number }>(
  batch: TextBatch, spans: T[],
): Map<string, T[]> {
  const result = new Map<string, T[]>(batch.segments.map((segment) => [segment.key, []]));
  let segmentIndex = 0;
  for (const span of [...spans].sort((a, b) => a.start - b.start)) {
    while (segmentIndex < batch.segments.length && span.start >= batch.segments[segmentIndex].end) segmentIndex++;
    const segment = batch.segments[segmentIndex];
    if (!segment || span.start < segment.start || span.end > segment.end) continue;
    result.get(segment.key)!.push({ ...span, start: span.start - segment.start, end: span.end - segment.start });
  }
  return result;
}

export type GroupableHit = { id: string; type: string; original: string; placeholder: string };
export type ReviewGroup<T extends GroupableHit> = {
  key: string; type: string; original: string; placeholder: string; hits: T[];
};

export function groupReviewHits<T extends GroupableHit>(hits: T[]): ReviewGroup<T>[] {
  const groups = new Map<string, ReviewGroup<T>>();
  for (const hit of hits) {
    const key = JSON.stringify([hit.type, hit.original, hit.placeholder]);
    let group = groups.get(key);
    if (!group) {
      group = { key, type: hit.type, original: hit.original, placeholder: hit.placeholder,
        hits: [] };
      groups.set(key, group);
    }
    group.hits.push(hit);
  }
  return [...groups.values()].sort((a, b) => a.type.localeCompare(b.type) || a.original.localeCompare(b.original));
}

export function withoutWhitelistedHits<T extends GroupableHit>(hits: T[], whitelist: ProjectWhitelistEntry[]): T[] {
  const excluded = new Set(whitelist.map(whitelistKey));
  return hits.filter((hit) => !excluded.has(whitelistKey(hit)));
}

/** A later run must not turn an existing Velum placeholder plus adjacent text into a new entity. */
export function withoutPlaceholderHits<T extends { original: string; type: string }>(
  hits: T[], knownPlaceholders: Iterable<string> = [],
): T[] {
  const placeholder = /(?<![\p{L}\p{N}])[A-Z][A-Z0-9]*_\d{2,}(?![\p{L}\p{N}])/u;
  const known = [...knownPlaceholders].filter(Boolean);
  return hits.filter((hit) => hit.type === "PASSWORT" || (
    !placeholder.test(hit.original) && !known.some((key) => hit.original.includes(key))
  ));
}

export function withoutWhitelistedRules<T extends { type: string; value: string }>(
  rules: T[], whitelist: ProjectWhitelistEntry[],
): T[] {
  const excluded = new Set(whitelist.map(whitelistKey));
  return rules.filter((rule) => !excluded.has(whitelistKey({ type: rule.type, original: rule.value })));
}

export function canonicalPerson(placeholder: string, merges: Record<string, string>): string {
  const seen = new Set<string>();
  let current = placeholder;
  while (merges[current] && !seen.has(current)) {
    seen.add(current);
    current = merges[current];
  }
  return current;
}

export function mergePeople(merges: Record<string, string>, source: string, target: string): Record<string, string> {
  const from = canonicalPerson(source, merges);
  const into = canonicalPerson(target, merges);
  if (from === into) return merges;
  const next = { ...merges };
  for (const person of new Set([...Object.keys(next), ...Object.values(next), from])) {
    if (canonicalPerson(person, merges) === from && person !== into) next[person] = into;
  }
  delete next[into];
  return next;
}

export function mappingForSelection(
  preview: PlaceholderMapping, previous: PlaceholderMapping, accepted: GroupableHit[],
  merges: Record<string, string>,
): PlaceholderMapping {
  const keys = new Set([...Object.keys(previous), ...accepted.map((hit) => hit.placeholder)]);
  const mapping: PlaceholderMapping = {};
  for (const key of keys) {
    const entry = preview[key] ?? previous[key];
    if (entry) mapping[key] = { ...entry };
  }
  for (const [source, target] of Object.entries(merges)) {
    const root = canonicalPerson(target, merges);
    if (source === root || mapping[source]?.type !== "PERSON" || mapping[root]?.type !== "PERSON") {
      throw new Error("Ungültige Personen-Zusammenführung.");
    }
    mapping[source] = { ...mapping[source], merged_into: root };
  }
  return mapping;
}