@erxes/api-shared 0.0.0-stage → 3.2.19-2

package/src/core-modules/import-export/utils/exportCursor.ts

@@ -0,0 +1,47 @@
+type BuildExportCursorQueryArgs = {
+  baseQuery?: Record<string, any>;
+  cursor?: string;
+  ids?: string[];
+  limit?: number;
+};
+
+type BuildExportCursorQueryResult = {
+  query: Record<string, any>;
+  isIdsMode: boolean;
+};
+
+export const normalizeExportLimit = (
+  limit?: number,
+  fallback = 100,
+): number => {
+  return limit && limit > 0 ? limit : fallback;
+};
+
+export const buildExportCursorQuery = ({
+  baseQuery = {},
+  cursor,
+  ids,
+  limit = 100,
+}: BuildExportCursorQueryArgs): BuildExportCursorQueryResult => {
+  const query = { ...baseQuery };
+  const hasIds = Boolean(ids?.length);
+
+  if (hasIds) {
+    const processedCount = cursor ? Number.parseInt(cursor, 10) || 0 : 0;
+    const remainingIds = ids!.slice(processedCount);
+
+    if (remainingIds.length === 0) {
+      query._id = { $in: [] };
+      return { query, isIdsMode: true };
+    }
+
+    query._id = { $in: remainingIds.slice(0, limit) };
+    return { query, isIdsMode: true };
+  }
+
+  if (cursor) {
+    query._id = query._id ? { ...query._id, $gt: cursor } : { $gt: cursor };
+  }
+
+  return { query, isIdsMode: false };
+};

package/src/core-modules/import-export/utils/headerMatcher.test.ts

@@ -0,0 +1,90 @@
+import { ImportHeaderDefinition } from '../types';
+import { matchImportHeaders, normalizeHeader } from './headerMatcher';
+
+const SYSTEM_FIELDS: ImportHeaderDefinition[] = [
+  { label: 'First Name', key: 'firstName' },
+  { label: 'Last Name', key: 'lastName' },
+  { label: 'Email', key: 'primaryEmail' },
+  { label: 'Phone', key: 'primaryPhone' },
+  { label: 'Code', key: 'code' },
+];
+
+const repeatingField = (rowIndex: number): ImportHeaderDefinition => ({
+  label: `Address ${rowIndex} / Street [street#${rowIndex}]`,
+  key: `propertiesData.g:group1/field1#${rowIndex}`,
+  aliases: [`Address ${rowIndex} / Street`],
+  type: 'customProperty',
+  code: 'street',
+  rowIndex,
+});
+
+describe('normalizeHeader', () => {
+  it('folds away case, punctuation and bracketed codes', () => {
+    expect(normalizeHeader('  First_Name  ')).toBe('first name');
+    expect(normalizeHeader('Street [street#2]')).toBe('street');
+  });
+
+  it('keeps non-latin scripts intact', () => {
+    expect(normalizeHeader('  Нэр,  Овог ')).toBe('нэр овог');
+  });
+});
+
+describe('matchImportHeaders', () => {
+  it('matches labels and keys exactly', () => {
+    const [label, key] = matchImportHeaders(
+      ['First Name', 'primaryEmail'],
+      SYSTEM_FIELDS,
+    );
+
+    expect(label).toMatchObject({ key: 'firstName', status: 'matched' });
+    expect(key).toMatchObject({ key: 'primaryEmail', status: 'matched' });
+  });
+
+  it('matches through case and punctuation differences', () => {
+    const [match] = matchImportHeaders(['first_name'], SYSTEM_FIELDS);
+
+    expect(match).toMatchObject({ key: 'firstName', status: 'matched' });
+  });
+
+  it('only suggests a near miss instead of wiring it up', () => {
+    const [match] = matchImportHeaders(['Frist Name'], SYSTEM_FIELDS);
+
+    expect(match.status).toBe('suggested');
+    expect(match.key).toBe('firstName');
+  });
+
+  it('leaves an unrelated header unmatched', () => {
+    const [match] = matchImportHeaders(['Legacy Id'], SYSTEM_FIELDS);
+
+    expect(match).toMatchObject({ status: 'unmatched', key: undefined });
+  });
+
+  it('keeps repeating property rows apart', () => {
+    const [first, second] = matchImportHeaders(
+      ['Address 1 / Street [street#1]', 'Address 2 / Street [street#2]'],
+      [repeatingField(1), repeatingField(2)],
+    );
+
+    expect(first.key).toBe('propertiesData.g:group1/field1#1');
+    expect(second.key).toBe('propertiesData.g:group1/field1#2');
+  });
+
+  it('sends a code without a row to the first row', () => {
+    const [match] = matchImportHeaders(
+      ['Street [street]'],
+      [repeatingField(1), repeatingField(2)],
+    );
+
+    expect(match.key).toBe('propertiesData.g:group1/field1#1');
+  });
+
+  it('claims a target field once and reports the duplicate', () => {
+    const [first, second] = matchImportHeaders(
+      ['Email', 'e-mail'],
+      SYSTEM_FIELDS,
+    );
+
+    expect(first).toMatchObject({ key: 'primaryEmail', status: 'matched' });
+    expect(second).toMatchObject({ key: undefined, status: 'unmatched' });
+  });
+});

package/src/core-modules/import-export/utils/headerMatcher.ts

@@ -0,0 +1,229 @@
+import { ImportHeaderDefinition } from '../types';
+
+export type ImportColumnMatchStatus = 'matched' | 'suggested' | 'unmatched';
+
+export interface ImportColumnMatch {
+  index: number;
+  header: string;
+  key?: string;
+  confidence: number;
+  status: ImportColumnMatchStatus;
+}
+
+/** At or above this score a column is wired up without asking. */
+export const AUTO_MATCH_CONFIDENCE = 0.8;
+
+/** Below this score the guess is too weak to show at all. */
+export const SUGGEST_CONFIDENCE = 0.55;
+
+const CODE_PATTERN = /\[([^\]]+)\]/g;
+
+/**
+ * Fold a header down to the letters and digits it carries.
+ *
+ * Case, spacing, punctuation and bracketed codes are all noise when a
+ * spreadsheet is edited by hand. The rule is deliberately language-agnostic:
+ * it strips anything that is not a letter or a number in any script, so it
+ * behaves the same for `First Name`, `first_name` and `Нэр  Овог`.
+ */
+export const normalizeHeader = (value: string): string =>
+  String(value ?? '')
+    .normalize('NFKC')
+    .toLowerCase()
+    .replace(/\[[^\]]*\]/g, ' ')
+    .replace(/[^\p{L}\p{N}]+/gu, ' ')
+    .trim();
+
+const extractCodes = (value: string): string[] =>
+  [...String(value ?? '').matchAll(CODE_PATTERN)]
+    .map((match) => match[1].trim().toLowerCase())
+    .filter(Boolean);
+
+const levenshtein = (left: string, right: string): number => {
+  if (left === right) return 0;
+  if (!left.length) return right.length;
+  if (!right.length) return left.length;
+
+  let previous = Array.from({ length: right.length + 1 }, (_, i) => i);
+
+  for (let i = 0; i < left.length; i++) {
+    const current = [i + 1];
+
+    for (let j = 0; j < right.length; j++) {
+      current.push(
+        Math.min(
+          previous[j + 1] + 1,
+          current[j] + 1,
+          previous[j] + (left[i] === right[j] ? 0 : 1),
+        ),
+      );
+    }
+
+    previous = current;
+  }
+
+  return previous[right.length];
+};
+
+const similarity = (left: string, right: string): number => {
+  if (!left || !right) return 0;
+
+  const longest = Math.max(left.length, right.length);
+
+  return 1 - levenshtein(left, right) / longest;
+};
+
+/** The code a repeating property column carries, e.g. `street#2`. */
+const qualifiedCode = (
+  definition: ImportHeaderDefinition,
+): string | undefined =>
+  definition.code
+    ? definition.rowIndex
+      ? `${definition.code}#${definition.rowIndex}`
+      : definition.code
+    : undefined;
+
+const scoreHeader = (
+  header: string,
+  definition: ImportHeaderDefinition,
+): number => {
+  if (header === definition.label.trim() || header === definition.key.trim()) {
+    return 1;
+  }
+
+  if (
+    definition.aliases?.some((alias) => String(alias || '').trim() === header)
+  ) {
+    return 0.95;
+  }
+
+  const headerCodes = extractCodes(header);
+
+  if (headerCodes.length && definition.code) {
+    const qualified = qualifiedCode(definition);
+
+    if (qualified && headerCodes.includes(qualified.toLowerCase())) {
+      return 0.95;
+    }
+
+    // A bare code carries no row, so it belongs to the group's first row.
+    if (
+      (!definition.rowIndex || definition.rowIndex === 1) &&
+      headerCodes.includes(definition.code.toLowerCase())
+    ) {
+      return 0.9;
+    }
+
+    return 0;
+  }
+
+  const normalizedHeader = normalizeHeader(header);
+
+  if (!normalizedHeader) {
+    return 0;
+  }
+
+  const candidates = [
+    definition.label,
+    definition.key,
+    ...(definition.aliases || []),
+  ]
+    .map(normalizeHeader)
+    .filter(Boolean);
+
+  if (candidates.includes(normalizedHeader)) {
+    return 0.85;
+  }
+
+  // Fuzzy hits stay below AUTO_MATCH_CONFIDENCE on purpose: a near-miss is a
+  // suggestion for a person to confirm, never something to wire up silently.
+  const closest = candidates.reduce(
+    (best, candidate) =>
+      Math.max(best, similarity(normalizedHeader, candidate)),
+    0,
+  );
+
+  return closest * 0.79;
+};
+
+/**
+ * Resolve a file's header row against a module's import fields.
+ *
+ * Every column comes back, including the ones nothing matched — the caller
+ * shows them so a mismatch is visible instead of silently dropping data. A
+ * target field is claimed at most once; the weaker duplicate is reported
+ * unmatched rather than overwriting its sibling.
+ */
+export const matchImportHeaders = (
+  fileHeaders: string[],
+  definitions: ImportHeaderDefinition[],
+): ImportColumnMatch[] => {
+  const matches: ImportColumnMatch[] = fileHeaders.map((rawHeader, index) => {
+    const header = String(rawHeader ?? '').trim();
+
+    if (!header) {
+      return {
+        index,
+        header,
+        key: undefined,
+        confidence: 0,
+        status: 'unmatched',
+      };
+    }
+
+    let bestKey: string | undefined;
+    let bestScore = 0;
+
+    for (const definition of definitions) {
+      const score = scoreHeader(header, definition);
+
+      if (score > bestScore) {
+        bestScore = score;
+        bestKey = definition.key;
+      }
+    }
+
+    if (!bestKey || bestScore < SUGGEST_CONFIDENCE) {
+      return {
+        index,
+        header,
+        key: undefined,
+        confidence: 0,
+        status: 'unmatched',
+      };
+    }
+
+    return {
+      index,
+      header,
+      key: bestKey,
+      confidence: Number(bestScore.toFixed(2)),
+      status: bestScore >= AUTO_MATCH_CONFIDENCE ? 'matched' : 'suggested',
+    };
+  });
+
+  const claimed = new Map<string, ImportColumnMatch>();
+
+  for (const match of matches) {
+    if (!match.key) continue;
+
+    const holder = claimed.get(match.key);
+
+    if (!holder) {
+      claimed.set(match.key, match);
+      continue;
+    }
+
+    const loser = holder.confidence >= match.confidence ? match : holder;
+
+    if (loser === holder) {
+      claimed.set(match.key, match);
+    }
+
+    loser.key = undefined;
+    loser.confidence = 0;
+    loser.status = 'unmatched';
+  }
+
+  return matches;
+};