Add pack, cfl and safehtml

pack loads local product packs (format 1): t5de-product.json and the product's files, as product.chkn or
files/, with index.xml's rating, type, gender and parent product. cfl writes and reads IMVU's product
container, uncompressed. safehtml cleans product descriptions by an allowlist (bluemonday).

Co-Authored-By: Claude Opus 5.5 <[email protected]>

pack/index.go

@@ -0,0 +1,137 @@
+package pack
+
+import (
+	"bytes"
+	"encoding/xml"
+	"errors"
+	"fmt"
+	"io"
+	"strconv"
+	"strings"
+)
+
+// Index is what T5DE reads from a product's index.xml, IMVU's product template. Other elements are ignored.
+type Index struct {
+	Rating      string // ProductRating: "GA" (general) or "AP" (adult)
+	Name        string // ProductName
+	Gender      string // ProductGender
+	TypeID      int64  // ProductTypeID, 0 if absent
+	BodyPattern int64  // BodyPatternProductId, 0 if absent
+	Parent      int64  // the product it derives from (__DATAIMPORT product://<id>/index.xml), 0 if none
+}
+
+// ParseIndex parses index.xml.
+func ParseIndex(data []byte) (*Index, error) {
+	d := xml.NewDecoder(bytes.NewReader(data))
+	d.Strict = false
+	d.Entity = xml.HTMLEntity
+	d.CharsetReader = charsetReader
+
+	values := map[string]string{}
+	depth := 0
+	root := false
+	var field string // the top-level element being read, if wanted
+	var text strings.Builder
+	for {
+		tok, err := d.Token()
+		if err == io.EOF {
+			break
+		}
+		if err != nil {
+			return nil, err
+		}
+		switch t := tok.(type) {
+		case xml.StartElement:
+			depth++
+			if depth == 1 && t.Name.Local != "Template" {
+				return nil, fmt.Errorf("root element is <%s>, not <Template>", t.Name.Local)
+			}
+			root = true
+			if depth == 2 && wanted[t.Name.Local] {
+				if _, ok := values[t.Name.Local]; ok {
+					return nil, fmt.Errorf("<%s> appears twice", t.Name.Local)
+				}
+				field = t.Name.Local
+				text.Reset()
+			}
+		case xml.EndElement:
+			if depth == 2 && field != "" {
+				values[field] = strings.TrimSpace(text.String())
+				field = ""
+			}
+			depth--
+		case xml.CharData:
+			if field != "" {
+				text.Write(t)
+			}
+		}
+	}
+	if !root {
+		return nil, errors.New("no <Template>")
+	}
+
+	var x Index
+	x.Rating = values["ProductRating"]
+	x.Name = values["ProductName"]
+	x.Gender = values["ProductGender"]
+	if x.Rating != "GA" && x.Rating != "AP" {
+		return nil, fmt.Errorf("ProductRating: %q must be GA or AP", x.Rating)
+	}
+	var err error
+	if x.TypeID, err = id(values, "ProductTypeID"); err != nil {
+		return nil, err
+	}
+	if x.BodyPattern, err = id(values, "BodyPatternProductId"); err != nil {
+		return nil, err
+	}
+	if imp := values["__DATAIMPORT"]; imp != "" {
+		s, ok := strings.CutPrefix(imp, "product://")
+		if ok {
+			s, ok = strings.CutSuffix(s, "/index.xml")
+		}
+		n, err := strconv.ParseInt(s, 10, 64)
+		if !ok || err != nil || n < 1 {
+			return nil, fmt.Errorf("__DATAIMPORT: %q isn't product://<id>/index.xml", imp)
+		}
+		x.Parent = n
+	}
+	return &x, nil
+}
+
+var wanted = map[string]bool{
+	"ProductRating":        true,
+	"ProductName":          true,
+	"ProductGender":        true,
+	"ProductTypeID":        true,
+	"BodyPatternProductId": true,
+	"__DATAIMPORT":         true,
+}
+
+func id(values map[string]string, name string) (int64, error) {
+	s := values[name]
+	if s == "" {
+		return 0, nil
+	}
+	n, err := strconv.ParseInt(s, 10, 64)
+	if err != nil || n < 0 {
+		return 0, fmt.Errorf("%s: %q isn't a number", name, s)
+	}
+	return n, nil
+}
+
+// charsetReader reads Latin-1 (and windows-1252, approximately) as well as UTF-8.
+func charsetReader(charset string, r io.Reader) (io.Reader, error) {
+	switch strings.ToLower(charset) {
+	case "iso-8859-1", "latin1", "latin-1", "windows-1252", "cp1252", "us-ascii", "ascii":
+		data, err := io.ReadAll(r)
+		if err != nil {
+			return nil, err
+		}
+		var b strings.Builder
+		for _, c := range data {
+			b.WriteRune(rune(c))
+		}
+		return strings.NewReader(b.String()), nil
+	}
+	return nil, fmt.Errorf("unsupported encoding %q", charset)
+}