Skip to main content

ChapterSage Connection

Chapter 5 — Checking the mappings

A software stores its data in tables and columns. The Mappings screen says which ones IntegraSys reads. The starting values are those of a standard installation: checking them before the first extraction saves a failed one.

In this chapter

  • Read a card: the source table and its columns
  • Correct a mapping
  • The case of Sage Paie
  • Let the AI propose the employee form
  1. Software
  2. Card
  3. Source table
  4. Columns
  5. Save
Book contents

One card per data set

First choose your software in the list at the top right: Sage 100 (SQL Server), Sage Paie (SQL Server) or SQL générique (mappings). The page then shows one card per data set.

The four Sage 100 cards on the Mappings page (French screen)
  1. 1. The software whose mappings you are reading.
  2. 2. One card per data set, with its Source table in the Sage database, here F_COMPTEG for the chart of accounts.
  3. 3. The pencil opens the card to correct it.
  4. 4. On the left, the IntegraSys field; on the right, the Sage column that feeds it.
  5. 5. An entry's date is read from periode and jour.
  6. 6. The reminder: the starting values are those of the software's standard schema.

For Sage 100, there are four cards: Plan comptable (chart of accounts), Plan tiers (third parties), Journaux (journals) and Écritures comptables (accounting entries). On each, the Source table line gives the Sage table (F_COMPTEG, F_COMPTET, F_JOURNAUX, F_ECRITUREC), then each line links an IntegraSys field, on the left, to a Sage column, on the right.

A few fields worth knowing:

  • Journaux: type (purchases, sales, cash/bank, general) and compte_contrepartie (the bank or cash account of a cash/bank journal) give the journals their type in IntegraSys (chapter 8).
  • Écritures comptables: IntegraSys dates an entry with periode (the journal's month) and jour (day). The date column, the keying date in Sage, is only a fallback.
  • montant and sens: Sage keeps an amount and a direction (debit or credit); IntegraSys turns them back into a debit and a credit.

Correct a mapping

Compare the cards with the real names of the client's database, which their IT person or Sage reseller can read out. If a name differs:

  1. Click the card's pencil.
  2. In the window, correct the Source table, or a field's column.
  3. Click Save. « Mapping saved » appears and the card shows custom.

A wrong table, or a required column (number, name, amount, direction…), makes its data set end as Failed at the extraction, with its reason; the other data sets carry on (chapter 6). An optional column that cannot be found, for its part, is ignored without a message, and its field stays empty; a periode or jour column that cannot be found makes the entry be dated by date. To test a correction, run an extraction of that data set alone and look at the file.

With SQL générique (mappings), for software other than Sage, the four accounting cards start empty: fill in their table and columns.

The case of Sage Paie

Choose Sage Paie (SQL Server): one card appears for each payroll data set, including Salaries, the employee records.

Let the AI propose the employee form

The employee form has many fields: payroll number, title, surname, first name, family status, date of birth, social security number, bank, base salary… Rather than linking each column by hand, the AI can propose the mapping.

The Salaries card window and its AI proposal block (French screen)
  1. 1. Source table: the employees table in Sage Paie.
  2. 2. Propose the mapping with AI: it reads the column names, never the data.
  3. 3. With no Sage Paie database declared, this orange message replaces the button.
  4. 4. A field of the employee form, and the Sage column that fills it.
  1. On the Salaries card, click the pencil.
  2. In the Propose the mapping with AI block, choose the Sage Paie database to read. Its agent must be Online.
  3. Click List columns and propose. IntegraSys reads the table's real columns, through the agent, then the AI proposes a column for each field.
  4. Read the result: the number of fields mapped, out of the table's number of columns, then « Review each mapping before saving. ».
  5. Check each field. A badge gives the AI's confidence: high confidence, to check or uncertain, with its reason in italics under the field. Each box offers the list of the table's real columns.
  6. Click Save. Nothing is applied before.
  • What leaves your network: only the table's name and its column names. No row, no salary, no employee name.
  • AI credits: the proposal uses the firm's AI credits.
  • If the AI does not answer (AI credits used up, AI refused by the firm's policy, service unavailable), an orange message says so, with the reason, in French: « L'IA n'a pas pu proposer la correspondance (crédits IA épuisés) : la proposition ci-dessous rapproche seulement les noms des colonnes. Vérifiez-la. » (the AI could not propose the mapping: the proposal below only matches the column names; check it). A proposal based on the names alone then replaces the AI's.
  • If the table does not exist in the database: « Table “Salarie” has no column in this database: check the source table name. ».