llm-intelligence/scripts/discover_intraday_news_candidates_test.go

//go:build llm_script

package main

import (
	"context"
	"database/sql"
	"path/filepath"
	"strings"
	"testing"
)

func TestLoadIntradaySearchRecordsFromFixture(t *testing.T) {
	cfg := intradayProviderConfig{
		Mode:    "fixture",
		Fixture: filepath.Join("testdata", "intraday_discovery_search_sample.json"),
	}
	records, err := loadIntradaySearchRecords(cfg, "2026-05-25", []string{"OpenAI pricing release"})
	if err != nil {
		t.Fatalf("loadIntradaySearchRecords 返回错误: %v", err)
	}
	if len(records) != 2 {
		t.Fatalf("搜索样例条数错误: got=%d", len(records))
	}
	if records[0].URL == "" || records[0].Provider == "" {
		t.Fatalf("搜索样例未保留 URL/provider: %+v", records[0])
	}
}

func TestLoadIntradayLLMRecordsFromFixture(t *testing.T) {
	cfg := intradayProviderConfig{
		Mode:    "fixture",
		Fixture: filepath.Join("testdata", "intraday_discovery_llm_sample.json"),
	}
	records, err := loadIntradayLLMRecords(cfg, "2026-05-25", nil)
	if err != nil {
		t.Fatalf("loadIntradayLLMRecords 返回错误: %v", err)
	}
	if len(records) != 2 {
		t.Fatalf("LLM 样例条数错误: got=%d", len(records))
	}
	if records[0].EventType != "official_release" {
		t.Fatalf("LLM 事件类型错误: %+v", records[0])
	}
}

func TestNormalizeIntradayCandidatesDedupesEquivalentEvents(t *testing.T) {
	searchRecords := []intradaySearchRecord{{
		Title:       "OpenAI announces GPT-5.6 preview pricing update",
		Summary:     "Search summary",
		URL:         "https://openai.example.com/news/gpt-5-6-pricing",
		Provider:    "OpenAI",
		PublishedAt: "2026-05-25",
	}}
	llmRecords := []intradayLLMRecord{
		{
			EventType:       "official_release",
			ProviderName:    "OpenAI",
			ModelName:       "GPT-5.6",
			ProviderCountry: "US",
			Title:           "GPT-5.6 preview pricing update",
			Summary:         "First summary",
			CandidateURLs:   []string{"https://openai.example.com/news/gpt-5-6-pricing"},
		},
		{
			EventType:       "official_release",
			ProviderName:    "OpenAI",
			ModelName:       "GPT 5.6",
			ProviderCountry: "US",
			Title:           "OpenAI GPT 5.6 preview pricing update",
			Summary:         "Second summary",
			CandidateURLs:   []string{"https://openai.example.com/news/gpt-5-6-pricing"},
		},
	}
	candidates := normalizeIntradayCandidates("2026-05-25", searchRecords, llmRecords)
	if len(candidates) != 1 {
		t.Fatalf("期望去重后只剩 1 条候选, got=%d", len(candidates))
	}
	if candidates[0].DiscoverySource != "web_search+llm" {
		t.Fatalf("期望 discovery source 合并, got=%q", candidates[0].DiscoverySource)
	}
}

func TestNormalizeIntradayCandidatesDropsOutdatedSearchMatches(t *testing.T) {
	searchRecords := []intradaySearchRecord{{
		Title:       "Old DeepSeek pricing article",
		Summary:     "Yesterday record",
		URL:         "https://deepseek.example.com/pricing",
		Provider:    "DeepSeek",
		PublishedAt: "2026-05-24",
	}}
	llmRecords := []intradayLLMRecord{{
		EventType:       "price_cut",
		ProviderName:    "DeepSeek",
		ModelName:       "DeepSeek-V4-Flash",
		ProviderCountry: "CN",
		Title:           "DeepSeek V4 Flash price cut",
		Summary:         "Should be dropped because search evidence is stale",
		CandidateURLs:   []string{"https://deepseek.example.com/pricing"},
	}}
	candidates := normalizeIntradayCandidates("2026-05-25", searchRecords, llmRecords)
	if len(candidates) != 0 {
		t.Fatalf("旧闻搜索结果不应进入候选池, got=%d", len(candidates))
	}
}

func TestNormalizeIntradayCandidatesDropsURLlessRecords(t *testing.T) {
	llmRecords := []intradayLLMRecord{{
		EventType:    "promo_campaign",
		ProviderName: "DeepSeek",
		ModelName:    "DeepSeek-V4-Flash",
		Title:        "No URL candidate",
		Summary:      "Should be dropped",
	}}
	candidates := normalizeIntradayCandidates("2026-05-25", nil, llmRecords)
	if len(candidates) != 0 {
		t.Fatalf("无 URL 候选应被丢弃, got=%d", len(candidates))
	}
}

func TestSearchRecordMatchesLocalizedBingDate(t *testing.T) {
	record := intradaySearchRecord{PublishedAt: "周一, 25 5月 2026 14:08:00 GMT"}
	if !searchRecordMatchesDate(record, "2026-05-25") {
		t.Fatal("应识别本地化 Bing pubDate 为当天")
	}
}

func TestValidateIntradayProviderConfigRequiresCommandOrURLOrFixture(t *testing.T) {
	if err := validateIntradayProviderConfig("search", intradayProviderConfig{Mode: "command_json"}); err == nil {
		t.Fatal("缺少 command 时应报错")
	}
	if err := validateIntradayProviderConfig("llm", intradayProviderConfig{Mode: "http_json"}); err == nil {
		t.Fatal("缺少 url 时应报错")
	}
	if err := validateIntradayProviderConfig("search", intradayProviderConfig{Mode: "fixture", Fixture: "fixture.json"}); err != nil {
		t.Fatalf("fixture provider 不应报错: %v", err)
	}
}

func TestBuildIntradayNormalizedKeyUsesProviderModelAndDate(t *testing.T) {
	key := buildIntradayNormalizedKey(intradayNewsCandidate{
		CandidateDate: "2026-05-25",
		EventType:     "official_release",
		ProviderName:  "OpenAI",
		ModelName:     "GPT-5.6",
	})
	if !strings.Contains(key, "2026-05-25") || !strings.Contains(key, "openai") || !strings.Contains(key, "gpt-5-6") {
		t.Fatalf("normalized key 不符合预期: %q", key)
	}
}

func TestUpsertIntradayCandidatesRequiresDB(t *testing.T) {
	var db *sql.DB
	err := upsertIntradayCandidates(context.Background(), db, nil)
	if err == nil {
		t.Fatal("nil db 时应报错")
	}
}