From 60eb89c12ee0661785395bff90940204b16dcb3a Mon Sep 17 00:00:00 2001
From: ax_rd <ax_rd@aisim.cn>
Date: Thu, 03 Sep 2026 13:21:39 +0800
Subject: [PATCH] docs: e2e 验证脚本 + README 更新（增量同步/explore/构建要求）
---
internal/search/explore_test.go | 61 +++++++++++++++++++++++++++++-
1 files changed, 58 insertions(+), 3 deletions(-)
diff --git a/internal/search/explore_test.go b/internal/search/explore_test.go
index 125fcac..bfe2677 100644
--- a/internal/search/explore_test.go
+++ b/internal/search/explore_test.go
@@ -1,7 +1,12 @@
package search
import (
+ "path/filepath"
+ "strings"
"testing"
+
+ "github.com/aisim/kb-cli/internal/graph"
+ "github.com/aisim/kb-cli/internal/index"
)
// TestExploreParagraphExtraction 段落截取：只输出命中关键词的段落，整段不截半句
@@ -22,10 +27,10 @@
}
}
-// TestExploreKeywordExpansion 长 CJK 词拆 bigram（原词保留，ASCII/短词不拆）
+// TestExploreKeywordExpansion 长 CJK 词拆 bigram（原词保留，ASCII/短词不拆；末尾去重保序）
func TestExploreKeywordExpansion(t *testing.T) {
- got := expandKeywords([]string{"电子秤补气失败", "补气", "abc"})
- want := []string{"电子秤补气失败", "电子", "子秤", "秤补", "补气", "气失", "失败", "补气", "abc"}
+ got := ExpandCJKKeywords([]string{"电子秤补气失败", "补气", "abc"})
+ want := []string{"电子秤补气失败", "电子", "子秤", "秤补", "补气", "气失", "失败", "abc"}
if len(got) != len(want) {
t.Fatalf("展开数量: got=%d want=%d (%v)", len(got), len(want), got)
}
@@ -43,3 +48,53 @@
t.Errorf("超预算段落应跳过: %q", got)
}
}
+
+// TestExploreFallbackToExpanded 回归（审查 Important）：单复合 CJK 词在大文档中
+// 原文不连续出现（bigram 召回入选，但原词在任何段落都不出现）时，
+// 段落截取必须回退到展开词，不能静默丢弃文档。
+func TestExploreFallbackToExpanded(t *testing.T) {
+ tmpDir := t.TempDir()
+ dbPath := filepath.Join(tmpDir, "test.db")
+ store, err := index.Open(dbPath)
+ if err != nil {
+ t.Fatalf("Open failed: %v", err)
+ }
+ defer store.Close()
+
+ // 大文档（> 2000 字节预算）：某段落只含「电子秤补气」片段，不含完整原词「电子秤补气失败」
+ filler := strings.Repeat("填充段落内容，用于把文档撑过预算。\n", 60)
+ content := "# 称重故障排查\n\n" + filler + "\n## 称重故障\n\n电子秤补气 时先检查阀门，失败则断电重启。\n"
+
+ n := &graph.Node{
+ ID: 1,
+ Path: "FAQ/称重/001-test.md",
+ Title: "称重故障排查",
+ Section: "FAQ",
+ Content: content,
+ }
+ if _, err := store.InsertNode(n); err != nil {
+ t.Fatalf("InsertNode failed: %v", err)
+ }
+ if err := store.CreateFTS(); err != nil {
+ t.Fatalf("CreateFTS failed: %v", err)
+ }
+ if err := store.PopulateFTS(); err != nil {
+ t.Fatalf("PopulateFTS failed: %v", err)
+ }
+
+ // 原词提取必为空（内容里没有完整的「电子秤补气失败」）
+ if b := extractRelevantParagraphs(content, []string{"电子秤补气失败"}, 2000); b != "" {
+ t.Fatalf("前提不成立：原词提取应为空，got=%q", b)
+ }
+
+ res, err := Explore(store, []string{"电子秤补气失败"}, ExploreOptions{Budget: 2000, TopN: 5})
+ if err != nil {
+ t.Fatalf("Explore failed: %v", err)
+ }
+ if len(res.Docs) != 1 {
+ t.Fatalf("应召回 1 篇文档（bigram 命中），got=%d", len(res.Docs))
+ }
+ if !strings.Contains(res.Docs[0].Body, "电子秤补气") {
+ t.Errorf("回退展开词后应提取含片段段落: body=%q", res.Docs[0].Body)
+ }
+}
--
Gitblit v1.10.0