intelwatch 1.1.1 → 1.1.3

package/src/scrapers/brave-search.js

@@ -148,11 +148,24 @@ export async function searchPressMentions(brandName, options = {}) {
     }
   }
 
+  // ── Relevance filter: drop results that don't actually mention the brand ──
+  const brandLower = brandName.toLowerCase().trim();
+  const brandWords = brandLower.split(/\s+/);
+  const filtered = mentions.filter(m => {
+    const text = ((m.title || '') + ' ' + (m.snippet || '') + ' ' + (m.domain || '')).toLowerCase();
+    // Must contain the exact brand name OR all words of the brand
+    if (text.includes(brandLower)) return true;
+    if (brandWords.length > 1 && brandWords.every(w => text.includes(w))) return true;
+    // Fuzzy: allow 1 char difference for short names (e.g. "Endrix" vs "Endrick" should be EXCLUDED)
+    return false;
+  });
+
   return {
     brandName,
     checkedAt: new Date().toISOString(),
-    mentions,
-    mentionCount: mentions.length,
+    mentions: filtered,
+    mentionCount: filtered.length,
+    unfilteredCount: mentions.length,
     error: news.error || web.error || null,
   };
 }

package/src/scrapers/pappers.js

@@ -1,4 +1,7 @@
 import axios from 'axios';
+import { mkdirSync, readFileSync, writeFileSync, existsSync } from 'fs';
+import { join } from 'path';
+import { homedir } from 'os';
 
 const PAPPERS_API = 'https://api.pappers.fr/v1';
 const PAPPERS_API_V2 = 'https://api.pappers.fr/v2';
@@ -78,6 +81,10 @@ export async function pappersLookup(companyName) {
  * and collective procedures.
  */
 export async function pappersGetFullDossier(siren) {
+  // Check cache first
+  const cached = getCached(siren);
+  if (cached) return { data: cached, error: null, fromCache: true };
+
   const apiKey = getApiKey();
   if (!apiKey) return { data: null, error: 'No PAPPERS_API_KEY set' };
 
@@ -96,6 +103,17 @@ export async function pappersGetFullDossier(siren) {
       resultat: f.resultat ?? null,
       capitauxPropres: f.capitaux_propres ?? null,
       effectif: f.effectif ?? null,
+      ebitda: f.excedent_brut_exploitation ?? null,
+      margeEbitda: f.taux_marge_EBITDA ?? null,
+      dettesFinancieres: f.dettes_financieres ?? null,
+      tresorerie: f.tresorerie ?? null,
+      fondsPropres: f.fonds_propres ?? null,
+      bfr: f.BFR ?? null,
+      ratioEndettement: f.ratio_endettement ?? null,
+      autonomieFinanciere: f.autonomie_financiere ?? null,
+      rentabiliteFP: f.rentabilite_fonds_propres ?? null,
+      margeNette: f.marge_nette ?? null,
+      capaciteAutofinancement: f.capacite_autofinancement ?? null,
     }));
 
     // UBO — bénéficiaires effectifs
@@ -108,14 +126,35 @@ export async function pappersGetFullDossier(siren) {
       pourcentageVotes: b.pourcentage_votes ?? null,
     }));
 
-    // BODACC publications — last 10
-    const bodacc = (d.publications_bodacc || []).slice(0, 10).map(p => ({
-      date: p.date,
-      type: p.type,
-      tribunal: p.tribunal || null,
-      numero: p.numero_annonce || null,
-      description: p.acte?.actes_publies?.[0]?.type_acte || p.type || null,
-    }));
+    // BODACC publications — last 50 (captures M&A activity)
+    const bodacc = (d.publications_bodacc || []).slice(0, 50).map(p => {
+      // Build rich description from all available fields
+      const parts = [];
+      if (p.description && p.description !== p.type) parts.push(p.description);
+      if (p.administration) parts.push(p.administration);
+      if (p.capital) parts.push(`Capital: ${(p.capital / 1e3).toFixed(0)}K€`);
+      if (p.date_cloture) parts.push(`Clôture: ${p.date_cloture}`);
+      if (p.type_depot) parts.push(p.type_depot);
+      if (p.activite) parts.push(p.activite);
+      const actes = (p.acte?.actes_publies || []).map(a => a.type_acte).filter(Boolean);
+      if (actes.length) parts.push(actes.join(', '));
+      // Build BODACC URL: format id:{letter}{parution}{annonce}
+      const bodaccLetter = p.bodacc || (p.type?.toLowerCase().includes('comptes') ? 'C' : 'B');
+      const bodaccUrl = p.numero_parution && p.numero_annonce
+        ? `https://www.bodacc.fr/pages/annonces-commerciales-detail/?q.id=id:${bodaccLetter}${p.numero_parution}${p.numero_annonce}`
+        : null;
+      return {
+        date: p.date,
+        type: p.type,
+        tribunal: p.greffe || p.tribunal || null,
+        numero: p.numero_annonce || null,
+        description: parts.length ? parts.join('. ') : p.type || null,
+        details: p.acte?.descriptif || null,
+        url: bodaccUrl,
+        capital: p.capital || null,
+        rcs: p.rcs || null,
+      };
+    });
 
     // Dirigeants with their mandats in other companies
     const dirigeants = (d.dirigeants || []).map(dir => ({
@@ -161,10 +200,53 @@ export async function pappersGetFullDossier(siren) {
       dateRadiation: d.date_radiation || null,
     };
 
-    return {
-      data: { identity, financialHistory, ubo, bodacc, dirigeants, proceduresCollectives, raw: d },
-      error: null,
-    };
+    // Consolidated financials (group level)
+    const consolidatedFinances = (d.finances_consolidees || []).slice(0, 5).map(f => ({
+      annee: f.annee,
+      ca: f.chiffre_affaires ?? null,
+      resultat: f.resultat ?? null,
+      capitauxPropres: f.capitaux_propres ?? null,
+      effectif: f.effectif ?? null,
+      ebitda: f.excedent_brut_exploitation ?? null,
+      margeEbitda: f.taux_marge_EBITDA ?? null,
+      dettesFinancieres: f.dettes_financieres ?? null,
+      tresorerie: f.tresorerie ?? null,
+      fondsPropres: f.fonds_propres ?? null,
+      bfr: f.BFR ?? null,
+      ratioEndettement: f.ratio_endettement ?? null,
+      autonomieFinanciere: f.autonomie_financiere ?? null,
+      rentabiliteFP: f.rentabilite_fonds_propres ?? null,
+      margeNette: f.marge_nette ?? null,
+      capaciteAutofinancement: f.capacite_autofinancement ?? null,
+    }));
+
+    // Representants (dirigeants + corporate entities with mandats)
+    const representants = (d.representants || []).map(r => ({
+      nom: r.nom_complet || r.denomination || [r.prenom, r.nom].filter(Boolean).join(' ') || '?',
+      qualite: r.qualite || '',
+      siren: r.siren || null,
+      personneMorale: !!r.personne_morale,
+    }));
+
+    // Etablissements
+    const etablissements = (d.etablissements || []).map(e => ({
+      siret: e.siret,
+      type: e.type_etablissement,
+      adresse: [e.adresse_ligne_1, e.code_postal, e.ville].filter(Boolean).join(' '),
+      actif: !e.etablissement_cesse,
+    }));
+
+    // Extra fields
+    identity.objetSocial = d.objet_social || null;
+    identity.tvaIntra = d.numero_tva_intracommunautaire || null;
+    identity.rcs = d.numero_rcs || null;
+    identity.greffe = d.greffe || null;
+    identity.conventionCollective = d.conventions_collectives?.[0]?.nom || null;
+    identity.effectifTexte = d.effectif || null;
+
+      const result = { identity, financialHistory, consolidatedFinances, ubo, bodacc, dirigeants, representants, etablissements, proceduresCollectives };
+      setCache(siren, result);
+      return { data: result, error: null };
   } catch (err) {
     return { data: null, error: err.message };
   }
@@ -216,3 +298,156 @@ function formatPappersDetail(d) {
     formeJuridique: d.forme_juridique || null,
   };
 }
+
+/**
+ * Search for subsidiaries/related entities.
+ * Strategy 1: recherche-dirigeants — finds companies where the parent acts as corporate director.
+ * Strategy 2: name-search fallback.
+ * Returns array of entities with their latest financials, sorted by CA desc.
+ */
+export async function pappersSearchSubsidiaries(parentName, parentSiren) {
+  const apiKey = getApiKey();
+  if (!apiKey) return { subsidiaries: [], error: 'No PAPPERS_API_KEY set' };
+
+  // Check cache for subsidiary search results
+  const subsCacheKey = `subs_${parentSiren}`;
+  const cachedSubs = getCached(subsCacheKey);
+  if (cachedSubs?.subsidiaries) {
+    return { subsidiaries: cachedSubs.subsidiaries, total: cachedSubs.total || cachedSubs.subsidiaries.length, error: null, fromCache: true };
+  }
+
+  const searchName = parentName.replace(/\s*(GRP|SAS|SARL|SA|SCI|EURL|GROUP|GROUPE|HOLDING|SNC|SASU)\s*/gi, ' ').trim();
+  const nameNorm = searchName.toLowerCase();
+
+  // ── Strategy 1: recherche-dirigeants ──────────────────────────────────────
+  // Finds companies where an entity named like the parent is listed as dirigeant
+  try {
+    const resp = await axios.get(`${PAPPERS_API}/recherche-dirigeants`, {
+      params: { api_token: apiKey, q: searchName, par_page: 20 },
+      timeout: 15000,
+    });
+
+    const resultats = resp.data.resultats || [];
+    const subsidiaryMap = new Map();
+
+    for (const r of resultats) {
+      // Only match the PARENT entity as dirigeant (by SIREN if available, else exact name)
+      const dirigeantSiren = r.siren || '';
+      const dirigeantName = (r.nom_entreprise || r.denomination || r.nom_complet || '').toLowerCase().trim();
+      
+      // Strict filter: must be the parent SIREN, or exact parent name match
+      const isParent = (dirigeantSiren === parentSiren) || (dirigeantName === nameNorm) || (dirigeantName === searchName.toLowerCase());
+      if (!isParent) continue;
+
+      for (const e of (r.entreprises || [])) {
+        if (e.siren && e.siren !== parentSiren && !subsidiaryMap.has(e.siren)) {
+          subsidiaryMap.set(e.siren, e);
+        }
+      }
+    }
+
+    if (subsidiaryMap.size > 0) {
+      // Limit to 30 subsidiaries to save API credits
+      const entities = Array.from(subsidiaryMap.values()).slice(0, 30);
+      const subsidiaries = await fetchSubsidiaryDetails(apiKey, entities);
+      // Cache subsidiary search results
+      setCache(subsCacheKey, { subsidiaries, total: subsidiaryMap.size });
+      return { subsidiaries, total: subsidiaryMap.size, error: null };
+    }
+  } catch (_) {
+    // Fall through to name-search
+  }
+
+  // ── Strategy 2: name-search fallback ──────────────────────────────────────
+  try {
+    const resp = await axios.get(`${PAPPERS_API}/recherche`, {
+      params: { api_token: apiKey, q: searchName, par_page: 20 },
+      timeout: 15000,
+    });
+
+    const entities = (resp.data.entreprises || resp.data.resultats || [])
+      .filter(e => e.siren !== parentSiren);
+
+    const subsidiaries = await fetchSubsidiaryDetails(apiKey, entities);
+    // Cache fallback search results too
+    setCache(subsCacheKey, { subsidiaries, total: subsidiaries.length });
+    return { subsidiaries, error: null };
+  } catch (err) {
+    return { subsidiaries: [], error: err.message };
+  }
+}
+
+async function fetchSubsidiaryDetails(apiKey, entities) {
+  const subsidiaries = [];
+  for (const e of entities) {
+    try {
+      // Check cache first to save API credits
+      const cached = getCached(e.siren);
+      let d;
+      if (cached?.identity) {
+        // Reconstruct from cached full dossier
+        d = { nom_entreprise: cached.identity.name, code_naf: cached.identity.nafCode, libelle_code_naf: cached.identity.nafLabel, siege: { ville: cached.identity.ville }, effectif: cached.identity.effectifTexte, entreprise_cessee: false, date_creation: cached.identity.dateCreation, finances: cached.financialHistory?.map(f => ({ chiffre_affaires: f.ca, resultat: f.resultat, annee: f.annee })) || [] };
+      } else {
+        const det = await axios.get(`${PAPPERS_API}/entreprise`, {
+          params: { api_token: apiKey, siren: e.siren },
+          timeout: 10000,
+        });
+        d = det.data;
+        // Cache subsidiary data to avoid re-fetching
+        setCache(e.siren, {
+          identity: { name: d.nom_entreprise, nafCode: d.code_naf, nafLabel: d.libelle_code_naf, ville: d.siege?.ville, effectifTexte: d.effectif, dateCreation: d.date_creation },
+          financialHistory: (d.finances || []).map(f => ({ ca: f.chiffre_affaires, resultat: f.resultat, annee: f.annee })),
+          _subCache: true,
+        });
+      }
+      const fin = (d.finances || [])[0] || {};
+      subsidiaries.push({
+        siren: e.siren,
+        name: d.nom_entreprise || d.denomination || e.nom_entreprise || '?',
+        naf: d.code_naf || '',
+        nafLabel: d.libelle_code_naf || '',
+        ville: d.siege?.ville || '',
+        effectif: d.effectif || '',
+        ca: fin.chiffre_affaires ?? null,
+        resultat: fin.resultat ?? null,
+        annee: fin.annee || null,
+        status: d.entreprise_cessee ? 'Cessée' : 'Active',
+        dateCreation: d.date_creation || null,
+      });
+    } catch (_) {
+      subsidiaries.push({
+        siren: e.siren,
+        name: e.nom_entreprise || e.denomination || '?',
+        ville: e.siege?.ville || '',
+        ca: null, resultat: null, effectif: '', annee: null, status: '?',
+      });
+    }
+  }
+  subsidiaries.sort((a, b) => (b.ca || 0) - (a.ca || 0));
+  return subsidiaries;
+}
+
+// ── Local cache to save API credits ──────────────────────────────────────────
+const CACHE_DIR = join(homedir(), '.intelwatch', 'cache', 'pappers');
+const CACHE_TTL = 7 * 24 * 3600 * 1000; // 7 days
+
+function ensureCacheDir() {
+  if (!existsSync(CACHE_DIR)) mkdirSync(CACHE_DIR, { recursive: true });
+}
+
+export function getCached(siren) {
+  try {
+    const file = join(CACHE_DIR, `${siren}.json`);
+    if (!existsSync(file)) return null;
+    const data = JSON.parse(readFileSync(file, 'utf8'));
+    if (Date.now() - data._cachedAt > CACHE_TTL) return null;
+    return data;
+  } catch { return null; }
+}
+
+export function setCache(siren, data) {
+  try {
+    ensureCacheDir();
+    writeFileSync(join(CACHE_DIR, `${siren}.json`), JSON.stringify({ ...data, _cachedAt: Date.now() }));
+  } catch { /* silent */ }
+}