llm-intelligence/scripts/import_bedrock_pricing_test.go

//go:build llm_script

package main

import (
	"bytes"
	"os"
	"path/filepath"
	"strings"
	"testing"
)

func TestParseBedrockPricingCatalogBuildsRecords(t *testing.T) {
	raw, err := os.ReadFile(filepath.Join("testdata", "bedrock_pricing_sample.html"))
	if err != nil {
		t.Fatalf("读取 fixture 失败: %v", err)
	}

	records, err := parseBedrockPricingCatalog(string(raw))
	if err != nil {
		t.Fatalf("parseBedrockPricingCatalog 返回错误: %v", err)
	}
	if len(records) != 4 {
		t.Fatalf("期望 4 条 Bedrock 价格记录，实际 %d", len(records))
	}
	if records[0].ProviderName != "Amazon" {
		t.Fatalf("Nova provider 归一化错误: %q", records[0].ProviderName)
	}
	if records[1].OutputPrice != 0.4 {
		t.Fatalf("Nova Lite 输出价错误: %v", records[1].OutputPrice)
	}
	if records[2].Region != "Europe (Frankfurt) and Asia Pacific (Jakarta)" {
		t.Fatalf("Qwen region 错误: %q", records[2].Region)
	}
}

func TestRunBedrockPricingImportDryRunPrintsSummary(t *testing.T) {
	var out bytes.Buffer
	err := runBedrockPricingImport(bedrockPricingImportConfig{
		URL:     defaultBedrockPricingURL,
		Fixture: filepath.Join("testdata", "bedrock_pricing_sample.html"),
		DryRun:  true,
	}, nil, &out)
	if err != nil {
		t.Fatalf("runBedrockPricingImport 返回错误: %v", err)
	}
	output := out.String()
	for _, want := range []string{
		"source=bedrock-pricing-import",
		"models=4",
		"operator=Amazon Bedrock",
		"dry_run=true",
	} {
		if !strings.Contains(output, want) {
			t.Fatalf("输出缺少 %q，实际: %q", want, output)
		}
	}
}

func TestParseBedrockPricingCatalogPrefersFirstTierForSameModelAndRegion(t *testing.T) {
	raw := `
<h3 id="Model_Pricing" class="lb-txt-none lb-txt-32 lb-h3 lb-title">Model Pricing</h3>
<h2 id="Anthropic" class="lb-txt-none lb-h2 lb-title">Anthropic</h2>
<p><b>Regions:&nbsp;US East (N. Virginia)</b></p>
<table>
  <tbody>
    <tr>
      <td><b>Anthropic models</b></td>
      <td><b>Price per 1M input tokens</b></td>
      <td><b>Price per 1M output tokens</b></td>
    </tr>
    <tr>
      <td>Claude Sonnet 4.5</td>
      <td>$3.00</td>
      <td>$15.00</td>
    </tr>
  </tbody>
</table>
<h3 id="Priority">Priority</h3>
<table>
  <tbody>
    <tr>
      <td><b>Anthropic models</b></td>
      <td><b>Price per 1M input tokens</b></td>
      <td><b>Price per 1M output tokens</b></td>
    </tr>
    <tr>
      <td>Claude Sonnet 4.5</td>
      <td>$6.00</td>
      <td>$30.00</td>
    </tr>
  </tbody>
</table>
<h3 id="Pricing_examples" class="lb-txt-none lb-txt-48 lb-h2 lb-title">Pricing examples</h3>
`

	records, err := parseBedrockPricingCatalog(raw)
	if err != nil {
		t.Fatalf("parseBedrockPricingCatalog 返回错误: %v", err)
	}
	if len(records) != 1 {
		t.Fatalf("期望仅保留默认价一条记录，实际 %d", len(records))
	}
	if records[0].InputPrice != 3 || records[0].OutputPrice != 15 {
		t.Fatalf("默认价被后续价阶污染: %+v", records[0])
	}
}

func TestParseBedrockPricingTextFallbackBuildsRecords(t *testing.T) {
	raw := cleanHTMLText(`
<h2 id="Qwen" class="lb-txt-none lb-h2 lb-title">Qwen</h2>
<p><b>Regions:&nbsp;Europe (Frankfurt) and&nbsp;Asia Pacific (Jakarta)</b></p>
Qwen models Price per 1M input tokens Price per 1M output tokens
Qwen3 Coder Next $ 0.60 $ 1.44
<p><b>Region: Asia Pacific (Sydney)</b></p>
Qwen models Price per 1M input tokens Price per 1M output tokens
Qwen3 Next 80B A3B $ 0.1545 $ 1.2360
`)

	records := parseBedrockPricingTextFallback(raw)
	if len(records) != 2 {
		t.Fatalf("期望 fallback 解析 2 条记录，实际 %d", len(records))
	}
	if records[0].Region != "Europe (Frankfurt) and Asia Pacific (Jakarta)" {
		t.Fatalf("fallback region 错误: %q", records[0].Region)
	}
	if records[1].InputPrice != 0.1545 || records[1].OutputPrice != 1.2360 {
		t.Fatalf("fallback 价格错误: %+v", records[1])
	}
}