Add files via upload

internal/openai/normalize_streaming_delta_test.go

@@ -0,0 +1,56 @@
+package openai
+
+import "testing"
+
+func TestNormalizeStreamingDelta_RepeatedCharBoundary(t *testing.T) {
+	// 流式在重复数字边界分片：不得把 "43" 的首字符与 "194" 尾字符误合并。
+	cur, d := normalizeStreamingDelta("https://x:194", "43")
+	if want := "https://x:19443"; cur != want {
+		t.Fatalf("next: want %q got %q", want, cur)
+	}
+	if d != "43" {
+		t.Fatalf("delta: want %q got %q", "43", d)
+	}
+}
+
+func TestNormalizeStreamingDelta_CumulativePrefix(t *testing.T) {
+	cur, d := normalizeStreamingDelta("今天", "今天天气")
+	if cur != "今天天气" || d != "天气" {
+		t.Fatalf("got cur=%q d=%q", cur, d)
+	}
+}
+
+func TestNormalizeStreamingDelta_FullRetransmit(t *testing.T) {
+	cur, d := normalizeStreamingDelta("今天", "今天")
+	if d != "" || cur != "今天" {
+		t.Fatalf("got cur=%q d=%q", cur, d)
+	}
+}
+
+func TestNormalizeStreamingDelta_SingleRuneRepeated(t *testing.T) {
+	cur, d := normalizeStreamingDelta("呀", "呀")
+	if want := "呀呀"; cur != want {
+		t.Fatalf("next: want %q got %q", want, cur)
+	}
+	if d != "呀" {
+		t.Fatalf("delta: want %q got %q", "呀", d)
+	}
+	cur, d = normalizeStreamingDelta("4", "4")
+	if want := "44"; cur != want {
+		t.Fatalf("next: want %q got %q", want, cur)
+	}
+	if d != "4" {
+		t.Fatalf("delta: want %q got %q", "4", d)
+	}
+}
+
+func TestNormalizeStreamingDelta_CumulativeExtendsNumber(t *testing.T) {
+	// 已缓冲 "194" 后收到累计串 "19443"（注意 "1943" 并非 "19443" 的前缀，不能靠误写的中间态测 HasPrefix）。
+	cur, d := normalizeStreamingDelta("194", "19443")
+	if want := "19443"; cur != want {
+		t.Fatalf("next: want %q got %q", want, cur)
+	}
+	if d != "43" {
+		t.Fatalf("delta: want %q got %q", "43", d)
+	}
+}

internal/openai/openai.go

@@ -10,6 +10,7 @@ import (
 	"net/http"
 	"strings"
 	"time"
+	"unicode/utf8"
 
 	"cyberstrike-ai/internal/config"
 
@@ -34,7 +35,15 @@ func (e *APIError) Error() string {
 }
 
 // normalizeStreamingDelta 将可能是“累计片段/重发片段”的内容归一化为“纯增量”。
-// 部分兼容网关会返回累计 content；若直接 append 会出现重复文本（结巴）。
+// 部分兼容网关会返回累计 content；若直接 append 会出现重复文本。
+//
+// 注意：
+//   - 不做「任意后缀与前缀重叠」合并；流式可能在重复字符边界分片（"194"+"43"→"19443"）。
+//   - HasPrefix 仅在 incoming 严格长于 current 时视为累计全文，否则会把分片产生的第二个相同
+//     单字/单码点（叠字、44、22 等）误判为「整段重复」而吞字。
+//   - incoming==current 仅当 current 长度 >1 个码点时才视为整包重发；单码点重复必须走拼接。
+//   - 不再使用「current 以 incoming 结尾则丢弃」：否则 "1943"+"43" 会误吞增量（19443 显示成 1943）。
+//     若网关重复发送尾部片段，应重复送完整累计串，由 HasPrefix 分支去重。
 func normalizeStreamingDelta(current, incoming string) (next, delta string) {
 	if incoming == "" {
 		return current, ""
@@ -42,26 +51,12 @@ func normalizeStreamingDelta(current, incoming string) (next, delta string) {
 	if current == "" {
 		return incoming, incoming
 	}
-	if incoming == current {
-		return current, ""
-	}
-	if strings.HasPrefix(incoming, current) {
+	if strings.HasPrefix(incoming, current) && len(incoming) > len(current) {
 		return incoming, incoming[len(current):]
 	}
-	if strings.HasSuffix(current, incoming) {
+	if incoming == current && utf8.RuneCountInString(current) > 1 {
 		return current, ""
 	}
-
-	// 边界重叠：current 后缀与 incoming 前缀重合，仅追加非重叠部分。
-	max := len(current)
-	if len(incoming) < max {
-		max = len(incoming)
-	}
-	for overlap := max; overlap > 0; overlap-- {
-		if current[len(current)-overlap:] == incoming[:overlap] {
-			return current + incoming[overlap:], incoming[overlap:]
-		}
-	}
 	return current + incoming, incoming
 }