labmcp/internal/packages/parser.go

package packages

import (
	"encoding/json"
	"fmt"
	"io"
	"strings"
)

// ParsePackages reads and parses a nix-env JSON output file.
func ParsePackages(r io.Reader) (map[string]*ParsedPackage, error) {
	var raw PackagesFile
	if err := json.NewDecoder(r).Decode(&raw); err != nil {
		return nil, fmt.Errorf("failed to decode packages JSON: %w", err)
	}

	packages := make(map[string]*ParsedPackage, len(raw))
	for attrPath, pkg := range raw {
		parsed := &ParsedPackage{
			AttrPath:        attrPath,
			Pname:           pkg.Pname,
			Version:         pkg.Version,
			Description:     pkg.Meta.Description,
			LongDescription: pkg.Meta.LongDescription,
			Homepage:        normalizeHomepage(pkg.Meta.Homepage),
			License:         normalizeLicense(pkg.Meta.License),
			Platforms:       normalizePlatforms(pkg.Meta.Platforms),
			Maintainers:     normalizeMaintainers(pkg.Meta.Maintainers),
			Broken:          pkg.Meta.Broken,
			Unfree:          pkg.Meta.Unfree,
			Insecure:        pkg.Meta.Insecure,
		}
		packages[attrPath] = parsed
	}

	return packages, nil
}

// normalizeHomepage converts homepage to a string.
func normalizeHomepage(v interface{}) string {
	if v == nil {
		return ""
	}
	switch hp := v.(type) {
	case string:
		return hp
	case []interface{}:
		if len(hp) > 0 {
			if s, ok := hp[0].(string); ok {
				return s
			}
		}
	}
	return ""
}

// normalizeLicense converts license to a JSON array string.
func normalizeLicense(v interface{}) string {
	if v == nil {
		return "[]"
	}

	licenses := make([]string, 0)

	switch l := v.(type) {
	case string:
		licenses = append(licenses, l)
	case map[string]interface{}:
		// Single license object
		if spdxID, ok := l["spdxId"].(string); ok {
			licenses = append(licenses, spdxID)
		} else if fullName, ok := l["fullName"].(string); ok {
			licenses = append(licenses, fullName)
		} else if shortName, ok := l["shortName"].(string); ok {
			licenses = append(licenses, shortName)
		}
	case []interface{}:
		for _, item := range l {
			switch li := item.(type) {
			case string:
				licenses = append(licenses, li)
			case map[string]interface{}:
				if spdxID, ok := li["spdxId"].(string); ok {
					licenses = append(licenses, spdxID)
				} else if fullName, ok := li["fullName"].(string); ok {
					licenses = append(licenses, fullName)
				} else if shortName, ok := li["shortName"].(string); ok {
					licenses = append(licenses, shortName)
				}
			}
		}
	}

	data, _ := json.Marshal(licenses)
	return string(data)
}

// normalizePlatforms converts platforms to a JSON array string.
func normalizePlatforms(v []interface{}) string {
	if v == nil {
		return "[]"
	}

	platforms := make([]string, 0, len(v))
	for _, p := range v {
		switch pv := p.(type) {
		case string:
			platforms = append(platforms, pv)
			// Skip complex platform specs (objects)
		}
	}

	data, _ := json.Marshal(platforms)
	return string(data)
}

// normalizeMaintainers converts maintainers to a JSON array string.
func normalizeMaintainers(maintainers []Maintainer) string {
	if len(maintainers) == 0 {
		return "[]"
	}

	names := make([]string, 0, len(maintainers))
	for _, m := range maintainers {
		name := m.Name
		if name == "" && m.Github != "" {
			name = "@" + m.Github
		}
		if name != "" {
			names = append(names, name)
		}
	}

	data, _ := json.Marshal(names)
	return string(data)
}

// ParsePackagesStream parses packages from a reader using streaming to reduce memory usage.
// It yields parsed packages through a callback function.
func ParsePackagesStream(r io.Reader, callback func(*ParsedPackage) error) (int, error) {
	dec := json.NewDecoder(r)

	// Read the opening brace
	t, err := dec.Token()
	if err != nil {
		return 0, fmt.Errorf("failed to read opening token: %w", err)
	}
	if delim, ok := t.(json.Delim); !ok || delim != '{' {
		return 0, fmt.Errorf("expected opening brace, got %v", t)
	}

	count := 0
	for dec.More() {
		// Read the key (attr path)
		t, err := dec.Token()
		if err != nil {
			return count, fmt.Errorf("failed to read attr path: %w", err)
		}
		attrPath, ok := t.(string)
		if !ok {
			return count, fmt.Errorf("expected string key, got %T", t)
		}

		// Read the value (package)
		var pkg RawPackage
		if err := dec.Decode(&pkg); err != nil {
			// Skip malformed packages
			continue
		}

		parsed := &ParsedPackage{
			AttrPath:        attrPath,
			Pname:           pkg.Pname,
			Version:         pkg.Version,
			Description:     pkg.Meta.Description,
			LongDescription: pkg.Meta.LongDescription,
			Homepage:        normalizeHomepage(pkg.Meta.Homepage),
			License:         normalizeLicense(pkg.Meta.License),
			Platforms:       normalizePlatforms(pkg.Meta.Platforms),
			Maintainers:     normalizeMaintainers(pkg.Meta.Maintainers),
			Broken:          pkg.Meta.Broken,
			Unfree:          pkg.Meta.Unfree,
			Insecure:        pkg.Meta.Insecure,
		}

		if err := callback(parsed); err != nil {
			return count, fmt.Errorf("callback error for %s: %w", attrPath, err)
		}
		count++
	}

	return count, nil
}

// SplitAttrPath splits an attribute path into its components.
// For example, "python312Packages.requests" returns ["python312Packages", "requests"].
func SplitAttrPath(attrPath string) []string {
	return strings.Split(attrPath, ".")
}