From 297d2c1839f0627c7a5b79b6d03db150b88dc3f6 Mon Sep 17 00:00:00 2001
From: Rain Seven <128443127+RAiNY7Study@users.noreply.github.com>
Date: Sun, 16 Aug 2026 17:09:08 +0800
Subject: [PATCH] fix: deduplicate cumulative IMS SMS segments (#34)

---
 internal/store/sms_reassembly.go      | 32 ++++++++++++++-
 internal/store/sms_reassembly_test.go | 56 +++++++++++++++++++++++++++
 2 files changed, 86 insertions(+), 2 deletions(-)

diff --git a/internal/store/sms_reassembly.go b/internal/store/sms_reassembly.go
index 573620c..6154d41 100644
--- a/internal/store/sms_reassembly.go
+++ b/internal/store/sms_reassembly.go
@@ -3,6 +3,7 @@ package store
 import (
 	"encoding/json"
 	"fmt"
+	"maps"
 	"sort"
 	"strconv"
 	"strings"
@@ -94,9 +95,15 @@ func mergeConcatSegment(
 			}
 		}
 	}
-	prior, alreadyHad := parts[sequence]
-	changed = !alreadyHad || prior != segmentBody
+	// Some IMS stacks hand us a cumulative segment: sequence 2 contains the
+	// already-decoded text of sequence 1 followed by its own payload. Keep a
+	// snapshot so normalizing that representation remains idempotent on a later
+	// redelivery of the same segment.
+	previousParts := maps.Clone(parts)
+	normalizeCumulativeConcatParts(previousParts)
 	parts[sequence] = segmentBody
+	normalizeCumulativeConcatParts(parts)
+	changed = !maps.Equal(previousParts, parts)
 
 	sequences := make([]int, 0, len(parts))
 	for n := range parts {
@@ -130,3 +137,24 @@ func mergeConcatSegment(
 	}
 	return joined.String(), json.RawMessage(encoded), changed, nil
 }
+
+// normalizeCumulativeConcatParts converts cumulative IMS segment bodies back
+// into ordinary per-segment bodies. It only removes an exact, non-empty prefix
+// assembled from every preceding sequence starting at 1, and only when the
+// current value also contains additional text. That deliberately leaves equal
+// repeated segments and incomplete/out-of-order prefixes untouched.
+func normalizeCumulativeConcatParts(parts map[int]string) {
+	var prefix strings.Builder
+	for sequence := 1; ; sequence++ {
+		text, ok := parts[sequence]
+		if !ok {
+			return
+		}
+		assembled := prefix.String()
+		if assembled != "" && len(text) > len(assembled) && strings.HasPrefix(text, assembled) {
+			text = strings.TrimPrefix(text, assembled)
+			parts[sequence] = text
+		}
+		prefix.WriteString(text)
+	}
+}
diff --git a/internal/store/sms_reassembly_test.go b/internal/store/sms_reassembly_test.go
index 9434050..9e75106 100644
--- a/internal/store/sms_reassembly_test.go
+++ b/internal/store/sms_reassembly_test.go
@@ -101,6 +101,62 @@ func TestMergeConcatSegmentRedeliveryIsIdempotent(t *testing.T) {
 	}
 }
 
+func TestMergeConcatSegmentNormalizesCumulativeIMSPart(t *testing.T) {
+	first := strings.Repeat("安全提醒", 17)
+	want := first + "请通过官方渠道核实。"
+	_, extra, _, err := mergeConcatSegment(nil, first, concatExtra(t, 8, 2, 1))
+	if err != nil {
+		t.Fatal(err)
+	}
+	body, normalized, changed, err := mergeConcatSegment(extra, want, concatExtra(t, 8, 2, 2))
+	if err != nil || !changed {
+		t.Fatalf("cumulative segment: body=%q changed=%v err=%v", body, changed, err)
+	}
+	if body != want {
+		t.Fatalf("body = %q, want cumulative text once %q", body, want)
+	}
+
+	// Redelivering the cumulative wire representation must compare equal to the
+	// normalized stored representation and must not churn the durable row id.
+	body, _, changed, err = mergeConcatSegment(normalized, want, concatExtra(t, 8, 2, 2))
+	if err != nil {
+		t.Fatal(err)
+	}
+	if changed || body != want {
+		t.Fatalf("redelivery: body=%q changed=%v, want %q/false", body, changed, want)
+	}
+}
+
+func TestMergeConcatSegmentNormalizesCumulativeIMSPartOutOfOrder(t *testing.T) {
+	first := strings.Repeat("甲", 67)
+	want := first + "尾段"
+	_, extra, _, err := mergeConcatSegment(nil, want, concatExtra(t, 12, 2, 2))
+	if err != nil {
+		t.Fatal(err)
+	}
+	body, _, changed, err := mergeConcatSegment(extra, first, concatExtra(t, 12, 2, 1))
+	if err != nil || !changed {
+		t.Fatalf("out-of-order segment: body=%q changed=%v err=%v", body, changed, err)
+	}
+	if body != want {
+		t.Fatalf("body = %q, want cumulative text once %q", body, want)
+	}
+}
+
+func TestMergeConcatSegmentKeepsEqualRepeatedPart(t *testing.T) {
+	_, extra, _, err := mergeConcatSegment(nil, "重复", concatExtra(t, 13, 2, 1))
+	if err != nil {
+		t.Fatal(err)
+	}
+	body, _, _, err := mergeConcatSegment(extra, "重复", concatExtra(t, 13, 2, 2))
+	if err != nil {
+		t.Fatal(err)
+	}
+	if body != "重复重复" {
+		t.Fatalf("body = %q, want intentional equal segments preserved", body)
+	}
+}
+
 func TestMergeConcatSegmentWithoutHeaderPassesThrough(t *testing.T) {
 	extra, err := json.Marshal(map[string]any{"encoding": "gsm7"})
 	if err != nil {