llm-intelligence/scripts/bedrock_pricing_lib.go

//go:build llm_script

package main

import (
	"fmt"
	"regexp"
	"strings"
)

const defaultBedrockPricingURL = "https://aws.amazon.com/bedrock/pricing/"

var (
	bedrockRegionPattern = regexp.MustCompile(`(?s)<p><b>Regions?:&nbsp;([^<]+)</b></p>`)
	bedrockTablePattern  = regexp.MustCompile(`(?s)<table[^>]*>(.*?)</table>`)
	bedrockRowPattern    = regexp.MustCompile(`(?s)<tr>(.*?)</tr>`)
	bedrockCellPattern   = regexp.MustCompile(`(?s)<t[dh][^>]*>(.*?)</t[dh]>`)
)

func parseBedrockPricingCatalog(raw string) ([]officialPricingRecord, error) {
	section := extractBetween(raw, `<h3 id="Model_Pricing"`, `<h2 id="Pricing_examples"`)
	if strings.TrimSpace(section) == "" {
		section = raw
	}

	blocks := splitBedrockProviderBlocks(section)
	records := make([]officialPricingRecord, 0)
	for _, block := range blocks {
		records = append(records, parseBedrockProviderBlock(block.providerLabel, block.content)...)
	}
	if len(records) == 0 {
		records = append(records, parseBedrockPricingTextFallback(cleanHTMLText(section))...)
	}
	if len(records) == 0 {
		return nil, fmt.Errorf("no bedrock pricing rows found")
	}
	return records, nil
}

func parseBedrockProviderBlock(providerLabel string, raw string) []officialPricingRecord {
	providerName := normalizeBedrockProvider(providerLabel)
	providerNameCn, providerCountry, providerWebsite := providerMetadata(providerName)

	regionMatches := bedrockRegionPattern.FindAllStringSubmatchIndex(raw, -1)
	tables := bedrockTablePattern.FindAllStringSubmatchIndex(raw, -1)
	records := make([]officialPricingRecord, 0)
	seenModelRegion := make(map[string]struct{})
	for _, tableIndex := range tables {
		tableHTML := raw[tableIndex[2]:tableIndex[3]]
		if !strings.Contains(tableHTML, "Price per 1M input tokens") || !strings.Contains(tableHTML, "$") {
			continue
		}
		region := "global"
		for _, regionIndex := range regionMatches {
			if regionIndex[0] < tableIndex[0] {
				region = cleanHTMLText(raw[regionIndex[2]:regionIndex[3]])
			}
		}
		rows := parseBedrockTableRows(tableHTML)
		for _, row := range rows {
			dedupeKey := strings.Join([]string{region, row.ModelName}, "|")
			if _, exists := seenModelRegion[dedupeKey]; exists {
				continue
			}
			record := officialPricingRecord{
				ModelID:         normalizeExternalID("bedrock", providerName, row.ModelName),
				ModelName:       row.ModelName,
				ProviderName:    providerName,
				ProviderNameCn:  providerNameCn,
				ProviderCountry: providerCountry,
				ProviderWebsite: providerWebsite,
				OperatorName:    "Amazon Bedrock",
				OperatorNameCn:  "Amazon Bedrock",
				OperatorCountry: "US",
				OperatorWebsite: "https://aws.amazon.com/bedrock/",
				OperatorType:    "cloud",
				Region:          region,
				Currency:        "USD",
				InputPrice:      row.InputPrice,
				OutputPrice:     row.OutputPrice,
				SourceURL:       defaultBedrockPricingURL,
				ModelSourceURL:  defaultBedrockPricingURL,
				DateConfidence:  "unknown",
				DateSourceKind:  "official_pricing",
				Modality:        detectModality(row.ModelName),
			}
			record.IsFree = false
			seenModelRegion[dedupeKey] = struct{}{}
			records = append(records, record)
		}
	}
	return records
}

type bedrockProviderBlock struct {
	providerLabel string
	content       string
}

func splitBedrockProviderBlocks(raw string) []bedrockProviderBlock {
	marker := `<h2 id="`
	indices := make([]int, 0)
	for offset := 0; ; {
		next := strings.Index(raw[offset:], marker)
		if next == -1 {
			break
		}
		indices = append(indices, offset+next)
		offset += next + len(marker)
	}
	blocks := make([]bedrockProviderBlock, 0, len(indices))
	for i, start := range indices {
		end := len(raw)
		if i+1 < len(indices) {
			end = indices[i+1]
		}
		chunk := raw[start:end]
		h2End := strings.Index(chunk, "</h2>")
		if h2End == -1 {
			continue
		}
		openEnd := strings.Index(chunk, ">")
		if openEnd == -1 || openEnd >= h2End {
			continue
		}
		label := cleanHTMLText(chunk[openEnd+1 : h2End])
		if strings.TrimSpace(label) == "" {
			continue
		}
		blocks = append(blocks, bedrockProviderBlock{
			providerLabel: label,
			content:       chunk,
		})
	}
	return blocks
}

func extractBetween(raw string, startMarker string, endMarker string) string {
	start := strings.Index(raw, startMarker)
	if start == -1 {
		return ""
	}
	segment := raw[start:]
	if endMarker == "" {
		return segment
	}
	end := strings.Index(segment, endMarker)
	if end == -1 {
		return segment
	}
	return segment[:end]
}

type bedrockPriceRow struct {
	ModelName   string
	InputPrice  float64
	OutputPrice float64
}

func parseBedrockTableRows(tableHTML string) []bedrockPriceRow {
	rows := bedrockRowPattern.FindAllStringSubmatch(tableHTML, -1)
	parsed := make([]bedrockPriceRow, 0)
	for _, row := range rows {
		cells := bedrockCellPattern.FindAllStringSubmatch(row[1], -1)
		if len(cells) < 3 {
			continue
		}
		values := make([]string, 0, len(cells))
		for _, cell := range cells {
			values = append(values, cleanHTMLText(cell[1]))
		}
		if strings.Contains(strings.ToLower(values[0]), "models") {
			continue
		}

		modelName := values[0]
		inputCell := values[1]
		outputCell := values[2]
		if len(values) >= 6 && strings.Contains(strings.ToLower(values[5]), "$") {
			outputCell = values[5]
		}
		inputPrice, ok := firstDollarPrice(inputCell)
		if !ok {
			continue
		}
		outputPrice, ok := firstDollarPrice(outputCell)
		if !ok {
			continue
		}
		parsed = append(parsed, bedrockPriceRow{
			ModelName:   modelName,
			InputPrice:  inputPrice,
			OutputPrice: outputPrice,
		})
	}
	return parsed
}

func normalizeBedrockProvider(raw string) string {
	switch strings.TrimSpace(raw) {
	case "Amazon Nova":
		return "Amazon"
	case "Anthropic":
		return "Anthropic"
	case "Cohere":
		return "Cohere"
	case "DeepSeek":
		return "DeepSeek"
	case "Meta":
		return "Meta"
	case "Mistral AI":
		return "Mistral AI"
	case "Moonshot AI":
		return "Moonshot AI"
	case "Kimi":
		return "Moonshot AI"
	case "NVIDIA":
		return "NVIDIA"
	case "OpenAI OSS Models":
		return "OpenAI"
	case "Qwen":
		return "Qwen"
	case "Writer":
		return "Writer"
	case "Z AI":
		return "Zhipu AI"
	default:
		return strings.TrimSpace(raw)
	}
}

var bedrockTextProviderHeaderPattern = regexp.MustCompile(`([A-Za-z][A-Za-z0-9 .&-]+)\s+models\s+Pr(?:i)?ce per 1M input tokens`)
var bedrockTextRowPattern = regexp.MustCompile(`([A-Za-z0-9 .:+-]+?)\s+\$\s*([0-9.]+)\s+\$\s*([0-9.]+)`)

func parseBedrockPricingTextFallback(raw string) []officialPricingRecord {
	matches := bedrockTextProviderHeaderPattern.FindAllStringSubmatchIndex(raw, -1)
	records := make([]officialPricingRecord, 0)
	seen := make(map[string]struct{})
	for i, match := range matches {
		if len(match) < 4 {
			continue
		}
		start := match[0]
		end := len(raw)
		if i+1 < len(matches) {
			end = matches[i+1][0]
		}
		block := raw[start:end]
		region := normalizeBedrockRegionText(findBedrockTextRegion(raw, start))
		providerName := normalizeBedrockProvider(raw[match[2]:match[3]])
		providerNameCn, providerCountry, providerWebsite := providerMetadata(providerName)
		rows := bedrockTextRowPattern.FindAllStringSubmatch(block, -1)
		for _, row := range rows {
			if len(row) != 4 {
				continue
			}
			modelName := strings.TrimSpace(row[1])
			key := strings.Join([]string{providerName, region, modelName}, "|")
			if _, exists := seen[key]; exists {
				continue
			}
			seen[key] = struct{}{}
			records = append(records, officialPricingRecord{
				ModelID:         normalizeExternalID("bedrock", providerName, modelName),
				ModelName:       modelName,
				ProviderName:    providerName,
				ProviderNameCn:  providerNameCn,
				ProviderCountry: providerCountry,
				ProviderWebsite: providerWebsite,
				OperatorName:    "Amazon Bedrock",
				OperatorNameCn:  "Amazon Bedrock",
				OperatorCountry: "US",
				OperatorWebsite: "https://aws.amazon.com/bedrock/",
				OperatorType:    "cloud",
				Region:          region,
				Currency:        "USD",
				InputPrice:      mustParseSubscriptionPrice(row[2]),
				OutputPrice:     mustParseSubscriptionPrice(row[3]),
				SourceURL:       defaultBedrockPricingURL,
				ModelSourceURL:  defaultBedrockPricingURL,
				DateConfidence:  "unknown",
				DateSourceKind:  "official_pricing",
				Modality:        detectModality(modelName),
			})
		}
	}
	return records
}

func findBedrockTextRegion(raw string, headerStart int) string {
	prefixStart := headerStart - 300
	if prefixStart < 0 {
		prefixStart = 0
	}
	prefix := raw[prefixStart:headerStart]
	lastPlural := strings.LastIndex(prefix, "Regions:")
	lastSingular := strings.LastIndex(prefix, "Region:")
	lastIndex := lastPlural
	marker := "Regions:"
	if lastSingular > lastIndex {
		lastIndex = lastSingular
		marker = "Region:"
	}
	if lastIndex == -1 {
		return ""
	}
	region := strings.TrimSpace(prefix[lastIndex+len(marker):])
	for _, stopMarker := range []string{" Priority ", " Flex ", " Batch ", " models "} {
		if stop := strings.Index(region, stopMarker); stop != -1 {
			region = strings.TrimSpace(region[:stop])
		}
	}
	return region
}

func normalizeBedrockRegionText(raw string) string {
	trimmed := strings.TrimSpace(raw)
	if trimmed == "" {
		return "global"
	}
	trimmed = strings.TrimSuffix(trimmed, ",")
	return strings.Join(strings.Fields(trimmed), " ")
}