From e07cf743f930ecff3a610f1a38702a699a273a02 Mon Sep 17 00:00:00 2001
From: Andriy Oblivantsev <eslider@gmail.com>
Date: Sun, 26 Jul 2026 11:34:42 +0100
Subject: [PATCH] feat(catalog): mbox header scan, company legal-suffix match,
 apply order

Add --mbox-headers for Thunderbird folder files; strip GmbH/& Co. KG in
CompanyGroupingKey; FindCompany falls back to full scan; apply companies
before persons. Tightens role-mailbox noise filters.
---
 crm.go                | 18 +++++++++++++++++-
 crm_normalize.go      | 21 +++++++++++++++++++--
 crm_normalize_test.go |  4 ++++
 3 files changed, 40 insertions(+), 3 deletions(-)

diff --git a/crm.go b/crm.go
index 0807c6b..54501c5 100644
--- a/crm.go
+++ b/crm.go
@@ -46,9 +46,12 @@ func (c *Client) GetContact(ctx context.Context, contactID string) (map[string]a
 }
 
 // FindCompany searches for a company contact with an exact (case-insensitive)
-// displayName match. Returns nil when not found.
+// displayName match after slogan/legal-suffix normalization. Returns nil when not found.
 func (c *Client) FindCompany(ctx context.Context, name string) (map[string]any, error) {
 	needle := CompanyGroupingKey(name)
+	if needle == "" {
+		return nil, nil
+	}
 	const page = 50
 	for start := 0; ; start += page {
 		items, total, err := c.ListContacts(ctx, page, start, name)
@@ -67,6 +70,19 @@ func (c *Client) FindCompany(ctx context.Context, name string) (map[string]any,
 			break
 		}
 	}
+	// Search filter may miss longer legal names; fall back to a full company scan.
+	all, err := c.ListAllContacts(ctx)
+	if err != nil {
+		return nil, err
+	}
+	for _, co := range all {
+		if !isCompany(co) {
+			continue
+		}
+		if CompanyGroupingKey(fmt.Sprint(co["displayName"])) == needle {
+			return co, nil
+		}
+	}
 	return nil, nil
 }
 
diff --git a/crm_normalize.go b/crm_normalize.go
index 82d4af8..7b2dcce 100644
--- a/crm_normalize.go
+++ b/crm_normalize.go
@@ -10,6 +10,9 @@ var multiSpace = regexp.MustCompile(`\s+`)
 // sloganSeparators split a company name from a trailing tagline/slogan.
 var sloganSeparators = []string{" — ", " – ", " - ", "—", "–"}
 
+// companyLegalSuffix matches trailing legal-form tokens (GmbH, & Co. KG, …).
+var companyLegalSuffix = regexp.MustCompile(`(?i)[\s,]*(&\s*co\.?\s*kg\.?|co\.?\s*kg\.?|gmbh|ag|ltd\.?|llc\.?|inc\.?|plc\.?|s\.?l\.?|sarl|b\.?v\.?)\s*$`)
+
 // StripSloganSuffix returns the part before an em/en dash tagline, e.g.
 // "Affirm — Fraud Engineering" → "Affirm".
 func StripSloganSuffix(s string) string {
@@ -22,9 +25,23 @@ func StripSloganSuffix(s string) string {
 	return s
 }
 
-// CompanyGroupingKey normalizes a company name for dedupe (ignores slogans).
+// StripCompanyLegalSuffix removes trailing GmbH / & Co. KG / Ltd / … tokens.
+func StripCompanyLegalSuffix(s string) string {
+	s = strings.TrimSpace(s)
+	s = strings.TrimRight(s, ",; ")
+	for {
+		next := companyLegalSuffix.ReplaceAllString(s, "")
+		next = strings.TrimRight(strings.TrimSpace(next), ",; ")
+		if next == s || next == "" {
+			return s
+		}
+		s = next
+	}
+}
+
+// CompanyGroupingKey normalizes a company name for dedupe (ignores slogans and legal suffixes).
 func CompanyGroupingKey(s string) string {
-	return NormalizeCompanyName(StripSloganSuffix(s))
+	return NormalizeCompanyName(StripCompanyLegalSuffix(StripSloganSuffix(s)))
 }
 
 // NormalizeCompanyName lowercases and collapses whitespace for grouping.
diff --git a/crm_normalize_test.go b/crm_normalize_test.go
index f02c3a6..79b0f8f 100644
--- a/crm_normalize_test.go
+++ b/crm_normalize_test.go
@@ -65,6 +65,10 @@ func TestCompanyGroupingKey(t *testing.T) {
 	if a != b {
 		t.Fatalf("%q != %q", a, b)
 	}
+	if CompanyGroupingKey("Acme") != CompanyGroupingKey("Acme GmbH & Co. KG,") {
+		t.Fatalf("legal suffix should not change grouping key: %q vs %q",
+			CompanyGroupingKey("Acme"), CompanyGroupingKey("Acme GmbH & Co. KG,"))
+	}
 }
 
 func TestOpportunityTitlesMatchSlogan(t *testing.T) {