feat: third-party import

packages/core/src/structures/synset/parseSynset.ts

@@ -3,7 +3,7 @@ import { Lemma } from '../lemma';
 import { isVerified, stripMetacharacters } from './metacharacters';
 
 export function parseSynset(rawString: string) {
-  const sanitized = sanitize(rawString);
+  const sanitized = sanitize(rawString || '');
   const verified = isVerified(sanitized);
   const annotations = sanitized.includes('(')
     ? new AnnotationHelper()

packages/engine/src/cli/synsets-cmd/argv.ts

@@ -5,6 +5,7 @@ export type PullArgv = {
   subcommand: 'pull';
   partial: boolean;
   only: boolean;
+  source: string;
   _: string[];
 };
 

packages/engine/src/cli/synsets-cmd/common.ts

@@ -5,33 +5,40 @@ import type { WordsAddLangSheet, WordsSheet } from '../../google';
 
 export * from './argv';
 
-export async function getGoogleGitSyncPrerequisites() {
+interface Overrides {
+  spreadsheetName?: string;
+  mainLanguagesSheet?: string;
+  additionalLanguagesSheet?: string;
+}
+
+export async function getGoogleGitSyncPrerequisites({
+  spreadsheetName = 'new_interslavic_words_list',
+  mainLanguagesSheet = 'words',
+  additionalLanguagesSheet = 'words_add_lang'
+}: Overrides = {}) {
   const { fileDatabase, googleAPIs } = await compose();
 
   const spreadsheetConfig = await fileDatabase.spreadsheets.findById(
-    'new_interslavic_words_list',
+    spreadsheetName,
   );
 
   if (!spreadsheetConfig) {
     throw new Error(
-      'Cannot find the spreadsheet config for "new_interslavic_words_list"',
+      `Cannot find the spreadsheet config with the name: ${spreadsheetName}`,
     );
   }
 
   const spreadsheet = googleAPIs.spreadsheet(spreadsheetConfig.google_id);
-  const words = await spreadsheet.getSheetByTitle('words');
+  const words = await spreadsheet.getSheetByTitle(mainLanguagesSheet);
   if (!words) {
-    throw new Error('Cannot find the sheet: words');
+    throw new Error(`Cannot find the sheet: ${mainLanguagesSheet}`);
   }
 
-  const wordsAddLang = await spreadsheet.getSheetByTitle('words_add_lang');
-  if (!wordsAddLang) {
-    throw new Error('Cannot find the sheet: words_add_lang');
-  }
+  const wordsAddLang = await spreadsheet.getSheetByTitle(additionalLanguagesSheet);
 
   return {
     words: words as unknown as WordsSheet,
-    wordsAddLang: wordsAddLang as unknown as WordsAddLangSheet,
+    wordsAddLang: wordsAddLang as unknown as (WordsAddLangSheet | undefined),
     multisynsets: fileDatabase.multisynsets,
   };
 }

packages/engine/src/cli/synsets-cmd/pull.ts

@@ -3,9 +3,21 @@ import { GSheets2Git } from '../../sync';
 import type { PullArgv } from './argv';
 import { getGoogleGitSyncPrerequisites, parseSelectedSynsets } from './common';
 
+function toUndefined(value: string): string | undefined {
+  return value === '' ? undefined : value;
+}
+
 export async function pull(argv: PullArgv) {
+  const [spreadsheetName, mainLanguagesSheet, additionalLanguagesSheet] = argv.source
+    ? argv.source.split(':').map(toUndefined)
+    : [];
+
   const { multisynsets, words, wordsAddLang } =
-    await getGoogleGitSyncPrerequisites();
+    await getGoogleGitSyncPrerequisites({
+      spreadsheetName,
+      mainLanguagesSheet,
+      additionalLanguagesSheet,
+    });
 
   const selectedIds = await parseSelectedSynsets(multisynsets, argv);
   if (argv.only && !selectedIds) {

packages/engine/src/cli/synsets-cmd/push.ts

@@ -7,6 +7,10 @@ export async function push(argv: PushArgv) {
   const { words, wordsAddLang, multisynsets } =
     await getGoogleGitSyncPrerequisites();
 
+  if (!wordsAddLang) {
+    throw new Error('Cannot find the sheet: words_add_lang');
+  }
+
   const selectedIds = await parseSelectedSynsets(multisynsets, argv);
   if (argv.only && !selectedIds) {
     console.log(

packages/engine/src/cli/synsets.ts

@@ -39,6 +39,10 @@ export const builder: CommandBuilder<subcommand.SynsetsArgvAny, any> = {
     description: 'Partial sync (disable deletion)',
     default: false,
   },
+  source: {
+    type: 'string',
+    description: 'Custom source spreadsheet',
+  },
   only: {
     type: 'boolean',
     description: 'Only selected synsets',

packages/engine/src/sync/words/GSheetsOp.ts

@@ -13,7 +13,7 @@ import { log } from '../../utils';
 
 export type GSheetsOpOptions = {
   readonly words: WordsSheet;
-  readonly wordsAddLang: WordsAddLangSheet;
+  readonly wordsAddLang?: WordsAddLangSheet;
   readonly selectedIds?: number[];
   readonly multisynsets: MultilingualSynsetRepository;
   /** Disables deletion of entities */
@@ -25,8 +25,9 @@ type TableDTO = WordsRecord | WordsAddLangRecord;
 export abstract class GSheetsOp extends IdSyncOperation<number> {
   private _words?: Promise<Map<number, WordsDTO>>;
   private _wordsAdd?: Promise<Map<number, WordsAddLangDTO>>;
+  private _maxId?: Promise<number>;
   protected readonly wordsSheet: WordsSheet;
-  protected readonly wordsAddLangSheet: WordsAddLangSheet;
+  protected readonly wordsAddLangSheet?: WordsAddLangSheet;
   protected readonly multisynsets: MultilingualSynsetRepository;
   protected readonly selectedIds?: Set<number>;
 
@@ -49,6 +50,15 @@ export abstract class GSheetsOp extends IdSyncOperation<number> {
     await this.wordsAdd();
   }
 
+  protected async maxFsId(): Promise<number> {
+    if (!this._maxId) {
+      // eslint-disable-next-line unicorn/no-array-reduce
+      this._maxId = this.multisynsets.keys().then((keys) => keys.reduce((a, b) => Math.max(Math.abs(a), Math.abs(b)), 0));
+    }
+
+    return this._maxId;
+  }
+
   protected async wordIds(): Promise<number[]> {
     return this.words().then((r) => [...r.keys()]);
   }
@@ -63,7 +73,9 @@ export abstract class GSheetsOp extends IdSyncOperation<number> {
 
   protected async wordsAdd(): Promise<Map<number, WordsAddLangDTO>> {
     if (!this._wordsAdd) {
-      this._wordsAdd = this._getRecords(this.wordsAddLangSheet);
+      this._wordsAdd = this.wordsAddLangSheet
+        ? this._getRecords(this.wordsAddLangSheet)
+        : Promise.resolve(new Map());
     }
 
     return this._wordsAdd;
@@ -76,10 +88,19 @@ export abstract class GSheetsOp extends IdSyncOperation<number> {
       { cat: ['gsheets'], tid: ['sync', sheet.id] },
       `fetch ${sheet.title}`,
       async () => {
+        let maxFsId = await this.maxFsId();
         const dtos = (await sheet.getValues()) as unknown as DTO[];
         const grecords = new Map(
           dtos
-            .map((dto) => [Math.abs(+dto.id), dto]),
+            .map((dto) => {
+              let id = Math.abs(+dto.id);
+              if (Number.isNaN(id)) {
+                id = ++maxFsId;
+                dto.id = id.toString();
+              }
+
+              return [id, dto];
+            }),
         );
 
         return grecords;

packages/engine/src/sync/words/Git2Gsheets.ts

@@ -61,7 +61,7 @@ export class Git2Gsheets extends GSheetsOp {
       notes,
     });
 
-    this.wordsAddLangSheet.batch.appendRows({
+    this.wordsAddLangSheet!.batch.appendRows({
       values: [[...dtoAddLang]],
     });
   }
@@ -79,7 +79,7 @@ export class Git2Gsheets extends GSheetsOp {
 
   protected async rollbackTransaction(): Promise<void> {
     this.wordsSheet.batch.clear();
-    this.wordsAddLangSheet.batch.clear();
+    this.wordsAddLangSheet!.batch.clear();
   }
 
   protected async commit(): Promise<void> {
@@ -91,7 +91,7 @@ export class Git2Gsheets extends GSheetsOp {
       'batch update',
       async () => {
         await this.wordsSheet.batch.flush();
-        await this.wordsAddLangSheet.batch.flush();
+        await this.wordsAddLangSheet!.batch.flush();
       },
     );
   }
@@ -139,7 +139,7 @@ export class Git2Gsheets extends GSheetsOp {
   }
 
   private _synset2dtoAddLang(ms: MultilingualSynset): WordsAddLangDTO {
-    const Mapper = this.wordsAddLangSheet
+    const Mapper = this.wordsAddLangSheet!
       .Mapper as ArrayMapper<WordsAddLangRecord>;
     const isv = ms.synsets.isv!;
     // TODO: this is a violation of synset vs lemma separation
@@ -199,7 +199,7 @@ export class Git2Gsheets extends GSheetsOp {
     if (!dto.isv.startsWith('!')) {
       dto.isv = '!' + dto.isv;
 
-      this.wordsAddLangSheet.batch.updateRows({
+      this.wordsAddLangSheet!.batch.updateRows({
         startRowIndex: dto.getIndex() + 1,
         startColumnIndex: dto.getColumnIndex('isv'),
         values: [[dto.isv]],
@@ -245,13 +245,13 @@ export class Git2Gsheets extends GSheetsOp {
     if (dtoOld) {
       if (this._hasChangesWords(dtoOld, dtoNew)) {
         const startRowIndex = dtoOld.getIndex() + 1;
-        this.wordsAddLangSheet.batch.updateRows({
+        this.wordsAddLangSheet!.batch.updateRows({
           startRowIndex,
           values: [[...dtoNew]],
         });
       }
     } else {
-      this.wordsAddLangSheet.batch.appendRows({
+      this.wordsAddLangSheet!.batch.appendRows({
         values: [[...dtoNew]],
       });
     }