zsvo/pkg/loader/package_loader.go

package loader

import (
	"bufio"
	"compress/gzip"
	"fmt"
	"io"
	"net/http"
	"path"
	"strings"
	"time"

	"zsvo/pkg/cache"
	"zsvo/pkg/resolver"

	"github.com/ulikunitz/xz"
)

// PackageLoader populates the cache from Debian Sources files
type PackageLoader struct {
	client       *http.Client
	cache        *cache.IndexCache
	cacheDir     string
	fastResolver *resolver.FastResolver
}

// NewPackageLoader creates a new package loader
func NewPackageLoader(cacheDir string) *PackageLoader {
	return &PackageLoader{
		client: &http.Client{
			Timeout: 30 * time.Second,
		},
		cache:    cache.NewIndexCache(cacheDir),
		cacheDir: cacheDir,
		// Note: fastResolver will be initialized lazily via GetGlobalResolver
	}
}

// LoadSources loads and parses Sources.xz from the given repository
func (l *PackageLoader) LoadSources(mirror, suite, component string) error {
	fmt.Printf("📦 Loading package index for %s/%s/%s...\n", mirror, suite, component)

	// Load or create index
	idx, err := l.cache.GetIndex(mirror, suite, component)
	if err != nil {
		return fmt.Errorf("failed to get index: %w", err)
	}

	// Check if cache is fresh (less than 24 hours old)
	if !idx.IsExpired(24 * time.Hour) {
		pkgCount, binCount := idx.GetStats()
		fmt.Printf("✅ Using fresh cache: %d packages, %d binaries\n", pkgCount, binCount)

		// Load into global resolver (singleton)
		if _, err := resolver.GetResolver(l.cacheDir, mirror, suite, component); err != nil {
			return fmt.Errorf("failed to load global resolver: %w", err)
		}

		return nil
	}

	fmt.Printf("🔄 Cache expired or missing, downloading fresh data...\n")

	// Download Sources.xz
	if err := l.downloadAndParseSources(mirror, suite, component, idx); err != nil {
		return fmt.Errorf("failed to download and parse sources: %w", err)
	}

	// Save to disk
	cacheFile := l.cache.CacheFilePath(mirror, suite, component)
	if err := idx.Save(cacheFile); err != nil {
		return fmt.Errorf("failed to save cache: %w", err)
	}

	pkgCount, binCount := idx.GetStats()
	fmt.Printf("✅ Loaded %d packages, %d binaries\n", pkgCount, binCount)

	// Load into global resolver (singleton)
	if _, err := resolver.GetResolver(l.cacheDir, mirror, suite, component); err != nil {
		return fmt.Errorf("failed to load global resolver: %w", err)
	}

	return nil
}

// downloadAndParseSources downloads and parses Sources.xz file
func (l *PackageLoader) downloadAndParseSources(mirror, suite, component string, idx *cache.PackageIndex) error {
	// Try different compression formats
	formats := []struct {
		ext     string
		decoder func(io.Reader) (io.Reader, error)
	}{
		{"xz", func(r io.Reader) (io.Reader, error) { return xz.NewReader(r) }},
		{"gz", func(r io.Reader) (io.Reader, error) { return gzip.NewReader(r) }},
		{"", func(r io.Reader) (io.Reader, error) { return r, nil }},
	}

	for _, format := range formats {
		url := l.buildSourcesURL(mirror, suite, component, format.ext)

		resp, err := l.client.Get(url)
		if err != nil {
			continue // Try next format
		}
		defer resp.Body.Close()

		if resp.StatusCode != http.StatusOK {
			continue // Try next format
		}

		// Decode and parse
		reader, err := format.decoder(resp.Body)
		if err != nil {
			continue // Try next format
		}

		return l.parseSourcesFile(reader, idx)
	}

	return fmt.Errorf("failed to download Sources file in any format")
}

// parseSourcesFile parses a Sources file and populates the index
func (l *PackageLoader) parseSourcesFile(reader io.Reader, idx *cache.PackageIndex) error {
	scanner := bufio.NewScanner(reader)
	scanner.Buffer(make([]byte, 0, 64*1024), 8*1024*1024)

	paragraph := make([]string, 0, 32)
	count := 0

	flush := func() error {
		if len(paragraph) == 0 {
			return nil
		}

		entry, err := l.parseSourcesParagraph(paragraph)
		paragraph = paragraph[:0]
		if err != nil {
			return nil // Skip invalid entries
		}

		// Add to index
		cacheEntry := &cache.PackageEntry{
			Package:      entry.Package,
			Version:      entry.Version,
			Directory:    entry.Directory,
			DSCName:      entry.DSCName,
			DSCSHA256:    entry.DSCSHA256,
			Binaries:     entry.Binaries,
			BuildDepends: entry.BuildDepends,
		}

		idx.AddPackage(cacheEntry)
		count++

		// Progress indicator
		if count%1000 == 0 {
			fmt.Printf("  Parsed %d packages...\n", count)
		}

		return nil
	}

	for scanner.Scan() {
		line := scanner.Text()
		if strings.TrimSpace(line) == "" {
			if err := flush(); err != nil {
				return err
			}
			continue
		}
		paragraph = append(paragraph, line)
	}

	// Flush last paragraph
	if err := flush(); err != nil {
		return err
	}

	if err := scanner.Err(); err != nil {
		return err
	}

	fmt.Printf("  Parsed %d packages total\n", count)
	return nil
}

// parseSourcesParagraph parses a single package paragraph
func (l *PackageLoader) parseSourcesParagraph(lines []string) (*SourceEntry, error) {
	fields := make(map[string]string)
	var currentKey string

	for _, raw := range lines {
		if strings.HasPrefix(raw, " ") || strings.HasPrefix(raw, "\t") {
			if currentKey == "" {
				continue
			}
			fields[currentKey] += "\n" + strings.TrimSpace(raw)
			continue
		}

		idx := strings.IndexByte(raw, ':')
		if idx <= 0 {
			continue
		}
		key := strings.TrimSpace(raw[:idx])
		value := strings.TrimSpace(raw[idx+1:])
		fields[key] = value
		currentKey = key
	}

	pkg := fields["Package"]
	ver := fields["Version"]
	dir := fields["Directory"]
	if pkg == "" || ver == "" || dir == "" {
		return nil, fmt.Errorf("missing required fields")
	}

	dscName, dscHash := l.parseChecksumsForDSC(fields["Checksums-Sha256"])
	if dscName == "" {
		return nil, fmt.Errorf("no dsc in checksums")
	}

	dir = strings.Trim(strings.TrimSpace(dir), "/")

	return &SourceEntry{
		Package:      pkg,
		Version:      ver,
		Directory:    dir,
		DSCName:      dscName,
		DSCSHA256:    dscHash,
		Binaries:     l.parseCommaSeparatedField(fields["Binary"]),
		BuildDepends: l.parseCommaSeparatedField(fields["Build-Depends"]),
	}, nil
}

// SourceEntry represents a parsed source package entry
type SourceEntry struct {
	Package      string
	Version      string
	Directory    string
	DSCName      string
	DSCSHA256    string
	Binaries     []string
	BuildDepends []string
}

// parseCommaSeparatedField parses a comma-separated field
func (l *PackageLoader) parseCommaSeparatedField(raw string) []string {
	parts := strings.Split(strings.TrimSpace(raw), ",")
	out := make([]string, 0, len(parts))
	seen := make(map[string]struct{}, len(parts))

	for _, part := range parts {
		part = strings.TrimSpace(strings.ToLower(part))
		if part == "" {
			continue
		}
		if _, exists := seen[part]; exists {
			continue
		}
		seen[part] = struct{}{}
		out = append(out, part)
	}

	return out
}

// parseChecksumsForDSC extracts DSC file info from checksums
func (l *PackageLoader) parseChecksumsForDSC(raw string) (string, string) {
	for _, line := range strings.Split(raw, "\n") {
		fields := strings.Fields(strings.TrimSpace(line))
		if len(fields) < 3 {
			continue
		}
		name := fields[2]
		if strings.HasSuffix(strings.ToLower(name), ".dsc") {
			return name, strings.ToLower(fields[0])
		}
	}
	return "", ""
}

// buildSourcesURL builds the URL for Sources file
func (l *PackageLoader) buildSourcesURL(mirror, suite, component, ext string) string {
	mirror = strings.TrimRight(strings.TrimSpace(mirror), "/")
	suite = strings.Trim(strings.TrimSpace(suite), "/")
	component = strings.Trim(strings.TrimSpace(component), "/")

	name := "Sources"
	if ext != "" {
		name += "." + ext
	}

	return mirror + "/" + path.Join("dists", suite, component, "source", name)
}

// GetFastResolver returns the loaded fast resolver
func (l *PackageLoader) GetFastResolver() *resolver.FastResolver {
	return l.fastResolver
}

// LoadDefaultRepository loads the default Debian repository
func (l *PackageLoader) LoadDefaultRepository() error {
	mirror := "https://deb.debian.org/debian"
	suite := "stable"
	component := "main"

	return l.LoadSources(mirror, suite, component)
}