multiarch-docker-images/archivist/config.jsonc

{
  // Example configuration for the Archivist pipeline.
  // Keys in English; extracted values and rendered file names are in
  // French, as they appear in the source documents.

  // ==========================================================================
  // fields: declaration of every field handled by the rules.
  // A missing mandatory field at the end of the pipeline marks the document
  // invalid and routes it to pending/ for human review.
  // "template" is a regular field: rules select the naming template with it.
  // ==========================================================================
  "fields": [
    { "name": "emittingCompany", "type": "text", "mandatory": true  },
    { "name": "documentType",    "type": "enum", "mandatory": true  },
    { "name": "emissionDate",    "type": "date",  "mandatory": true  },
    { "name": "template",        "type": "text", "mandatory": false },
    { "name": "bankAccount",     "type": "enum", "mandatory": false },
    { "name": "address",         "type": "text", "mandatory": false },
    { "name": "reference",       "type": "text", "mandatory": false }
  ],

  // ==========================================================================
  // rules: an ordered tree evaluated against the first-page text
  // (firstPageText) and the fields already set or deleted by earlier rules.
  //
  // Each rule contains four keys:
  //   when  (required): expr-lang expression; the rest of the rule runs only
  //                     if it evaluates to true. String literals inside
  //                     expressions are written as raw strings with backticks.
  //   set   : list of assignments
  //             { "field": "x", "value": <fixed value> }  constant
  //             { "field": "x", "eval": "<expression>" }  evaluated in the
  //                                                        same context
  //   delete: list of field names to unset
  //   rules : child rules, evaluated in order only if the parent "when"
  //           matched; nesting is recursive
  //
  // Built-in helper: firstMatch(regex) returns the first match of the regex
  // in the first-page text (first capture group when present).
  // Strings are parsed into native dates with expr's standard date(value,
  // layout) builtin (Go layout, e.g. `02/01/2006` for 15/01/2026); the
  // format is carried by the rule so it can differ per document type.
  // ==========================================================================
  "rules": [
    // --- Generic extraction, runs for every document -------------------------
    // Emission date, e.g. "Paris, le 15/01/2026" -> "15/01/2026" (parsed
    // into a native date; named emissionDate to avoid clashing with the
    // expr date() builtin)
    {
      "when": "true",
      "set": [
        { "field": "emissionDate", "eval": "date(firstMatch(`[0-9]{2}/[0-9]{2}/[0-9]{4}`), `02/01/2006`)" }
      ]
    },

    // --- Bank statements (relevés de compte) ----------------------------------
    {
      "when": "firstPageText contains `Relevé de compte`",
      "set": [
        { "field": "documentType", "value": "Relevé de compte" },
        { "field": "template",     "value": "bank_statement" }
      ],
      "rules": [
        // Issuing bank
        {
          "when": "firstPageText contains `BRED`",
          "set": [ { "field": "emittingCompany", "value": "BRED" } ]
        },
        {
          "when": "firstPageText contains `Fortunéo`",
          "set": [ { "field": "emittingCompany", "value": "Fortunéo" } ]
        },
        {
          "when": "firstPageText contains `Crédit Mutuel`",
          "set": [ { "field": "emittingCompany", "value": "Crédit Mutuel" } ]
        },
        // Personal accounts, matched on the account number printed on page 1
        {
          "when": "firstPageText contains `12345678901`",
          "set": [ { "field": "bankAccount", "value": "Compte Thomas" } ]
        },
        {
          "when": "firstPageText contains `98765432109`",
          "set": [ { "field": "bankAccount", "value": "Compte joint" } ]
        }
      ]
    },

    // --- Taxe d'habitation, one child rule per address ------------------------
    {
      "when": "firstPageText contains `Taxe d'habitation`",
      "set": [
        { "field": "documentType",    "value": "Taxe d'habitation" },
        { "field": "emittingCompany", "value": "DGFiP" },
        { "field": "template",        "value": "tax_notice_housing" }
      ],
      "rules": [
        {
          "when": "firstPageText contains `12 rue des Lilas`",
          "set": [ { "field": "address", "value": "12 rue des Lilas, Paris" } ]
        },
        {
          "when": "firstPageText contains `34 avenue Jean Jaurès`",
          "set": [ { "field": "address", "value": "34 avenue Jean Jaurès, Bordeaux" } ]
        }
      ]
    },

    // --- Avis d'impôt sur le revenu --------------------------------------------
    {
      "when": "firstPageText contains `Impôt sur le revenu` && firstPageText contains `avis d'imposition`",
      "set": [
        { "field": "documentType",    "value": "Avis d'impôt sur le revenu" },
        { "field": "emittingCompany", "value": "DGFiP" },
        { "field": "template",        "value": "tax_notice_income" }
      ]
    },

    // --- Generic invoices (e-commerce and others) -----------------------------
    {
      "when": "firstPageText contains `Facture` && firstPageText contains `TVA`",
      "set": [
        { "field": "documentType", "value": "Facture" },
        { "field": "template",     "value": "invoice" },
        // Invoice number, e.g. "Facture n° F-2026-0123" -> "F-2026-0123"
        { "field": "reference",    "eval": "firstMatch(`[Ff]acture n° ?([A-Za-z0-9-]+)`)" }
      ],
      // Fields that do not apply to invoices, in case an earlier rule
      // set them
      "delete": [ "bankAccount", "address" ],
      "rules": [
        {
          "when": "firstPageText contains `Amazon`",
          "set": [ { "field": "emittingCompany", "value": "Amazon" } ]
        },
        {
          "when": "firstPageText contains `LDLC`",
          "set": [ { "field": "emittingCompany", "value": "LDLC" } ]
        },
        {
          "when": "firstPageText contains `Cdiscount`",
          "set": [ { "field": "emittingCompany", "value": "Cdiscount" } ]
        }
      ]
    }
  ],

  // ==========================================================================
  // templates: named naming templates, selected by the "template" field.
  // A template is a literal string containing {expr} placeholders: the
  // inside of each placeholder is an expr-lang/expr expression evaluated in
  // the document context (fields, helpers, builtins) and replaced by its
  // result, e.g. {emissionDate.Format(`2006-01-02`)} or
  // {upper(emittingCompany)}. Unset optional fields evaluate to an empty
  // string. The rendered template is the full object path under files/.
  // If no rule sets "template", the "default" template is used.
  // ==========================================================================
  "templates": {
    "bank_statement":     "{emittingCompany}/{emissionDate.Format(`2006-01-02`)}_{documentType}_{bankAccount}.pdf",
    "tax_notice_housing": "Impots/{emissionDate.Format(`2006-01-02`)}_{documentType}_{address}.pdf",
    "tax_notice_income":  "Impots/{emissionDate.Format(`2006-01-02`)}_{documentType}.pdf",
    "invoice":            "Factures/{emissionDate.Format(`2006-01-02`)}_{emittingCompany}_{reference}.pdf",
    "default":            "Divers/{documentType}.pdf"
  }
}