Update sent folder regex (#14069)

2025-08-25 21:17:31 +05:30
parent 734b0bd2d2
commit 0c8631c1c8
3 changed files with 180 additions and 23 deletions
@@ -3,6 +3,8 @@ import { Injectable, Logger } from '@nestjs/common';
 import { isNumber } from 'class-validator';
 import { ListResponse, type ImapFlow } from 'imapflow';

+import { getImapSentFolderCandidatesByRegex } from 'src/modules/messaging/message-import-manager/drivers/imap/utils/get-sent-folder-candidates-by-regex.util';
+
 /**
 * Service to find sent folder using IMAP special-use flags
 *
@@ -89,7 +91,7 @@ export class ImapFindSentFolderService {
    client: ImapFlow,
    list: ListResponse[],
  ): Promise<string | null> {
-    const regexCandidateFolders = this.getSentFolderCandidatesByRegex(list);
+    const regexCandidateFolders = getImapSentFolderCandidatesByRegex(list);

    for (const folder of regexCandidateFolders) {
      const messageCount = await this.getFolderMessageCount(client, folder);
@@ -112,28 +114,6 @@ export class ImapFindSentFolderService {
    return null;
  }

-  private getSentFolderCandidatesByRegex(list: ListResponse[]): string[] {
-    // Comprehensive regex pattern for legacy IMAP servers
-    // Source: https://imapsync.lamiral.info/FAQ.d/FAQ.Folders_Mapping.txt
-    // Based on imapsync's regextrans2 examples (originally "Sent|Sent Messages|Gesendet")
-    // Extended with additional common localizations for broader provider/language support
-    const sentFolderPattern =
-      /^(.*\/)?(sent|sent[\s_-]?(items|mail|messages|elements)?|envoy[éê]s?|[ée]l[ée]ments[\s_-]?envoy[éê]s|gesendet|gesendete[\s_-]?elemente|enviados?|elementos[\s_-]?enviados|itens[\s_-]?enviados|posta[\s_-]?inviata|inviati|보낸편지함|\[gmail\]\/sent[\s_-]?mail)$/i;
-
-    const regexCandidateFolders = [];
-
-    for (const folder of list) {
-      if (sentFolderPattern.test(folder.path)) {
-        this.logger.debug(
-          `Found potential sent folder via pattern match: ${folder.path}`,
-        );
-        regexCandidateFolders.push(folder.path);
-      }
-    }
-
-    return regexCandidateFolders;
-  }
-
  private async getFolderMessageCount(
    client: ImapFlow,
    folderPath: string,
@@ -0,0 +1,124 @@
+import { type ListResponse } from 'imapflow';
+
+import { getImapSentFolderCandidatesByRegex } from 'src/modules/messaging/message-import-manager/drivers/imap/utils/get-sent-folder-candidates-by-regex.util';
+
+function makeList(paths: string[]): ListResponse[] {
+  return paths.map((p) => ({ path: p }) as ListResponse);
+}
+
+describe('getSentFolderCandidatesByRegex', () => {
+  it('matches English variants', () => {
+    const englishVariants = [
+      'Sent',
+      'Sent Items',
+      'Sent Mail',
+      'Sent Messages',
+      'sent-elements',
+    ];
+    const input = makeList(englishVariants);
+    const result = getImapSentFolderCandidatesByRegex(input);
+
+    expect(result).toEqual(expect.arrayContaining(englishVariants));
+  });
+
+  it('matches French variants', () => {
+    const frenchVariants = ['Envoyés', 'Éléments envoyés', 'Objets envoyés'];
+    const input = makeList(frenchVariants);
+    const result = getImapSentFolderCandidatesByRegex(input);
+
+    expect(result).toEqual(expect.arrayContaining(frenchVariants));
+  });
+
+  it('matches German variants', () => {
+    const germanVariants = ['Gesendet', 'Gesendete Elemente'];
+    const input = makeList(germanVariants);
+    const result = getImapSentFolderCandidatesByRegex(input);
+
+    expect(result).toEqual(expect.arrayContaining(germanVariants));
+  });
+
+  it('matches Spanish variants', () => {
+    const spanishVariants = ['Enviados', 'Elementos enviados'];
+    const input = makeList(spanishVariants);
+    const result = getImapSentFolderCandidatesByRegex(input);
+
+    expect(result).toEqual(expect.arrayContaining(spanishVariants));
+  });
+
+  it('matches Portuguese variants', () => {
+    const portugueseVariants = ['Itens enviados'];
+    const input = makeList(portugueseVariants);
+    const result = getImapSentFolderCandidatesByRegex(input);
+
+    expect(result).toEqual(expect.arrayContaining(portugueseVariants));
+  });
+
+  it('matches Italian variants', () => {
+    const italianVariants = ['Posta inviata', 'Inviati'];
+    const input = makeList(italianVariants);
+    const result = getImapSentFolderCandidatesByRegex(input);
+
+    expect(result).toEqual(expect.arrayContaining(italianVariants));
+  });
+
+  it('matches Korean variant', () => {
+    const koreanVariants = ['보낸편지함'];
+    const input = makeList(koreanVariants);
+    const result = getImapSentFolderCandidatesByRegex(input);
+
+    expect(result).toEqual(koreanVariants);
+  });
+
+  it('matches Japanese variants', () => {
+    const japaneseVariants = ['送信済みメール', '送信済み'];
+    const input = makeList(japaneseVariants);
+    const result = getImapSentFolderCandidatesByRegex(input);
+
+    expect(result).toEqual(expect.arrayContaining(japaneseVariants));
+  });
+
+  it('matches Polish variants', () => {
+    const polishVariants = ['Wysłane', 'Elementy wysłane'];
+    const input = makeList(polishVariants);
+    const result = getImapSentFolderCandidatesByRegex(input);
+
+    expect(result).toEqual(expect.arrayContaining(polishVariants));
+  });
+
+  it('matches Russian variants', () => {
+    const russianVariants = [
+      'Отправленные',
+      'Отправленные письма',
+      'Отправленные сообщения',
+      'Исходящие',
+    ];
+    const input = makeList(russianVariants);
+    const result = getImapSentFolderCandidatesByRegex(input);
+
+    expect(result).toEqual(expect.arrayContaining(russianVariants));
+  });
+
+  it('matches Gmail special folder', () => {
+    const gmailVariants = ['[Gmail]/Sent Mail', '[Gmail]\\Sent Mail'];
+    const input = makeList(gmailVariants);
+    const result = getImapSentFolderCandidatesByRegex(input);
+
+    expect(result).toEqual(expect.arrayContaining(gmailVariants));
+  });
+
+  it('does not match unrelated folders', () => {
+    const unrelatedFolders = [
+      'Inbox',
+      'Drafts',
+      'Trash',
+      'Archive',
+      'Junk',
+      'Important',
+      'RandomFolder',
+    ];
+    const input = makeList(unrelatedFolders);
+    const result = getImapSentFolderCandidatesByRegex(input);
+
+    expect(result).toEqual([]);
+  });
+});
@@ -0,0 +1,53 @@
+import { type ListResponse } from 'imapflow';
+
+export function getImapSentFolderCandidatesByRegex(
+  list: ListResponse[],
+): string[] {
+  const sentFolderPattern = new RegExp(
+    [
+      // EN
+      'sent([\\s_-]?(items|mail|messages|elements))?',
+      // FR
+      'envoy[éê]s?',
+      '[ée]l[ée]ments[\\s_-]?envoy[éê]s',
+      // DE
+      'gesendet',
+      'gesendete[\\s_-]?elemente',
+      // ES
+      'enviados?',
+      'elementos[\\s_-]?enviados',
+      // PT
+      'itens[\\s_-]?enviados',
+      // IT
+      'posta[\\s_-]?inviata',
+      'inviati',
+      // KO
+      '보낸편지함',
+      // JA
+      '送信済みメール',
+      '送信済み',
+      // PL
+      'elementy[\\s_-]?wysłane',
+      'wysłane',
+      // RU
+      'отправленные',
+      'отправленные[\\s_-]?(сообщения|письма)?',
+      'исходящие',
+      // GMAIL
+      '\\[gmail\\][\\/]+sent[\\s_-]?mail',
+    ]
+      .map((s) => `(${s})`)
+      .join('|'),
+    'i',
+  );
+
+  const regexCandidateFolders = [];
+
+  for (const folder of list) {
+    if (sentFolderPattern.test(folder.path)) {
+      regexCandidateFolders.push(folder.path);
+    }
+  }
+
+  return regexCandidateFolders;
+}