fix(gnoweb): simplify url parsing system

gno.land/pkg/gnoweb/app_test.go

@@ -73,6 +73,7 @@ func TestRoutes(t *testing.T) {
 
 	for _, r := range routes {
 		t.Run(fmt.Sprintf("test route %s", r.route), func(t *testing.T) {
+			t.Logf("input: %q", r.route)
 			request := httptest.NewRequest(http.MethodGet, r.route, nil)
 			response := httptest.NewRecorder()
 			router.ServeHTTP(response, request)
@@ -125,7 +126,7 @@ func TestAnalytics(t *testing.T) {
 				request := httptest.NewRequest(http.MethodGet, route, nil)
 				response := httptest.NewRecorder()
 				router.ServeHTTP(response, request)
-				fmt.Println("HELLO:", response.Body.String())
+
 				assert.Contains(t, response.Body.String(), "sa.gno.services")
 			})
 		}
@@ -143,6 +144,7 @@ func TestAnalytics(t *testing.T) {
 				request := httptest.NewRequest(http.MethodGet, route, nil)
 				response := httptest.NewRecorder()
 				router.ServeHTTP(response, request)
+
 				assert.NotContains(t, response.Body.String(), "sa.gno.services")
 			})
 		}

gno.land/pkg/gnoweb/handler.go

@@ -140,9 +140,7 @@ func (h *WebHandler) renderPackage(w io.Writer, gnourl *GnoURL) (status int, err
 	switch {
 	case gnourl.WebQuery.Has("source"):
 		return h.renderRealmSource(w, gnourl)
-	case kind == KindPure,
-		strings.HasSuffix(gnourl.Path, "/"),
-		isFile(gnourl.Path):
+	case kind == KindPure, gnourl.IsFile(), gnourl.IsDir():
 		i := strings.LastIndexByte(gnourl.Path, '/')
 		if i < 0 {
 			return http.StatusInternalServerError, fmt.Errorf("unable to get ending slash for %q", gnourl.Path)
@@ -152,10 +150,13 @@ func (h *WebHandler) renderPackage(w io.Writer, gnourl *GnoURL) (status int, err
 		gnourl.WebQuery.Set("source", "") // set source
 
 		file := gnourl.Path[i+1:]
+		// If there nothing after the last slash that mean its a
+		// directory ...
 		if file == "" {
 			return h.renderRealmDirectory(w, gnourl)
 		}
 
+		// ... else, remaining part is a file
 		gnourl.WebQuery.Set("file", file)
 		gnourl.Path = gnourl.Path[:i]
 
@@ -370,10 +371,3 @@ func generateBreadcrumbPaths(path string) []components.BreadcrumbPart {
 
 	return parts
 }
-
-// IsFile checks if the last element of the path is a file (has an extension)
-func isFile(path string) bool {
-	base := filepath.Base(path)
-	ext := filepath.Ext(base)
-	return ext != ""
-}

gno.land/pkg/gnoweb/url.go

@@ -4,18 +4,22 @@
 	"errors"
 	"fmt"
 	"net/url"
+	"path/filepath"
 	"regexp"
 	"strings"
 )
 
 type PathKind byte
 
 const (
-	KindInvalid PathKind = 0
+	KindUnknown PathKind = 0
 	KindRealm   PathKind = 'r'
 	KindPure    PathKind = 'p'
 )
 
+// rePkgOrRealmPath matches and validates a realm or package path.
+var rePkgOrRealmPath = regexp.MustCompile(`^/[a-z][a-zA-Z0-9_/.]*$`)
+
 // GnoURL decomposes the parts of an URL to query a realm.
 type GnoURL struct {
 	// Example full path:
@@ -28,121 +32,147 @@
 	Query    url.Values // c=d
 }
 
-func (url GnoURL) EncodeArgs() string {
-	var urlstr strings.Builder
-	if url.Args != "" {
-		urlstr.WriteString(url.Args)
-	}
+// EncodeFlag is used to compose and encode URL components.
+type EncodeFlag int
 
-	if len(url.Query) > 0 {
-		urlstr.WriteString("?" + url.Query.Encode())
-	}
+const (
+	EncodePath EncodeFlag = 1 << iota
+	EncodeArgs
+	EncodeWebQuery
+	EncodeQuery
+	EncodeNoEscape // Disable escaping on arg
+)
 
-	return urlstr.String()
+// Has checks if the EncodeFlag contains all the specified flags.
+func (f EncodeFlag) Has(flags EncodeFlag) bool {
+	return f&flags != 0
 }
 
-func (url GnoURL) EncodePath() string {
+// Encode encodes the URL components based on the provided flags.
+// Encode assumes the URL is valid.
+func (gnoURL GnoURL) Encode(encodeFlags EncodeFlag) string {
 	var urlstr strings.Builder
-	urlstr.WriteString(url.Path)
-	if url.Args != "" {
-		urlstr.WriteString(":" + url.Args)
-	}
 
-	if len(url.Query) > 0 {
-		urlstr.WriteString("?" + url.Query.Encode())
+	if encodeFlags.Has(EncodePath) {
+		urlstr.WriteString(gnoURL.Path)
 	}
 
-	return urlstr.String()
-}
+	if encodeFlags.Has(EncodeArgs) && gnoURL.Args != "" {
+		if encodeFlags.Has(EncodePath) {
+			urlstr.WriteRune(':')
+		}
 
-func (url GnoURL) EncodeWebPath() string {
-	var urlstr strings.Builder
-	urlstr.WriteString(url.Path)
-	if url.Args != "" {
-		pathEscape := escapeDollarSign(url.Args)
-		urlstr.WriteString(":" + pathEscape)
+		// XXX: Arguments should ideally always be escaped,
+		// but this may require changes in some realms.
+		args := gnoURL.Args
+		if !encodeFlags.Has(EncodeNoEscape) {
+			args = escapeDollarSign(url.PathEscape(args))
+		}
+
+		urlstr.WriteString(args)
 	}
 
-	if len(url.WebQuery) > 0 {
-		urlstr.WriteString("$" + url.WebQuery.Encode())
+	if encodeFlags.Has(EncodeWebQuery) && len(gnoURL.WebQuery) > 0 {
+		urlstr.WriteRune('$')
+		urlstr.WriteString(gnoURL.WebQuery.Encode())
 	}
 
-	if len(url.Query) > 0 {
-		urlstr.WriteString("?" + url.Query.Encode())
+	if encodeFlags.Has(EncodeQuery) && len(gnoURL.Query) > 0 {
+		urlstr.WriteRune('?')
+		urlstr.WriteString(gnoURL.Query.Encode())
 	}
 
 	return urlstr.String()
 }
 
-func (url GnoURL) Kind() PathKind {
-	if len(url.Path) < 2 {
-		return KindInvalid
-	}
-	pk := PathKind(url.Path[1])
-	switch pk {
-	case KindPure, KindRealm:
-		return pk
+// EncodeArgs encodes the arguments and query parameters into a string.
+// This function is intended to be passed as a realm `Render` argument.
+func (gnoURL GnoURL) EncodeArgs() string {
+	return gnoURL.Encode(EncodeArgs | EncodeQuery | EncodeNoEscape)
+}
+
+// EncodeURL encodes the path, arguments, and query parameters into a string.
+// This function provides the full representation of the URL without the web query.
+func (gnoURL GnoURL) EncodeURL() string {
+	return gnoURL.Encode(EncodePath | EncodeArgs | EncodeQuery)
+}
+
+// EncodeWebURL encodes the path, package arguments, web query, and query into a string.
+// This function provides the full representation of the URL.
+func (gnoURL GnoURL) EncodeWebURL() string {
+	return gnoURL.Encode(EncodePath | EncodeArgs | EncodeWebQuery | EncodeQuery)
+}
+
+// Kind determines the kind of path (invalid, realm, or pure) based on the path structure.
+func (gnoURL GnoURL) Kind() PathKind {
+	if len(gnoURL.Path) > 2 && gnoURL.Path[0] == '/' && gnoURL.Path[2] == '/' {
+		switch k := PathKind(gnoURL.Path[1]); k {
+		case KindPure, KindRealm:
+			return k
+		}
 	}
-	return KindInvalid
+	return KindUnknown
 }
 
-var (
-	ErrURLMalformedPath   = errors.New("malformed URL path")
-	ErrURLInvalidPathKind = errors.New("invalid path kind")
-)
+func (gnoURL GnoURL) IsValid() bool {
+	return rePkgOrRealmPath.MatchString(gnoURL.Path)
+}
 
-// reRealName match a realm path
-// - matches[1]: path
-// - matches[2]: path args
-var reRealmPath = regexp.MustCompile(`^` +
-	`(/(?:[a-zA-Z0-9_-]+)/` + // path kind
-	`[a-zA-Z][a-zA-Z0-9_-]*` + // First path segment
-	`(?:/[a-zA-Z][.a-zA-Z0-9_-]*)*/?)` + // Additional path segments
-	`([:$](?:.*))?$`, // Remaining portions args, separate by `$` or `:`
-)
+// IsDir checks if the URL path represents a directory.
+func (gnoURL GnoURL) IsDir() bool {
+	return len(gnoURL.Path) > 0 && gnoURL.Path[len(gnoURL.Path)-1] == '/'
+}
 
+// IsFile checks if the URL path represents a file.
+func (gnoURL GnoURL) IsFile() bool {
+	return filepath.Ext(gnoURL.Path) != ""
+}
+
+var ErrURLInvalidPath = errors.New("invalid or malformed path")
+
+// ParseGnoURL parses a URL into a GnoURL structure, extracting and validating its components.
 func ParseGnoURL(u *url.URL) (*GnoURL, error) {
-	matches := reRealmPath.FindStringSubmatch(u.EscapedPath())
-	if len(matches) != 3 {
-		return nil, fmt.Errorf("%w: %s", ErrURLMalformedPath, u.Path)
+	var webargs string
+	path, args, found := strings.Cut(u.EscapedPath(), ":")
+	if found {
+		args, webargs, _ = strings.Cut(args, "$")
+	} else {
+		path, webargs, _ = strings.Cut(path, "$")
 	}
 
-	path := matches[1]
-	args := matches[2]
+	// NOTE: `PathUnescape` should already unescape dollar signs.
+	upath, err := url.PathUnescape(path)
+	if err != nil {
+		return nil, fmt.Errorf("unable to unescape path %q: %w", path, err)
+	}
 
-	if len(args) > 0 {
-		switch args[0] {
-		case ':':
-			args = args[1:]
-		case '$':
-		default:
-			return nil, fmt.Errorf("%w: %s", ErrURLMalformedPath, u.Path)
-		}
+	if !rePkgOrRealmPath.MatchString(upath) {
+		return nil, fmt.Errorf("%w: %q", ErrURLInvalidPath, upath)
 	}
 
-	var err error
 	webquery := url.Values{}
-	args, webargs, found := strings.Cut(args, "$")
-	if found {
-		if webquery, err = url.ParseQuery(webargs); err != nil {
-			return nil, fmt.Errorf("unable to parse webquery %q: %w ", webquery, err)
+	if len(webargs) > 0 {
+		var parseErr error
+		if webquery, parseErr = url.ParseQuery(webargs); parseErr != nil {
+			return nil, fmt.Errorf("unable to parse webquery %q: %w", webargs, parseErr)
 		}
 	}
 
 	uargs, err := url.PathUnescape(args)
 	if err != nil {
-		return nil, fmt.Errorf("unable to unescape path %q: %w", args, err)
+		return nil, fmt.Errorf("unable to unescape args %q: %w", args, err)
 	}
 
 	return &GnoURL{
-		Path:     path,
+		Path:     upath,
 		Args:     uargs,
 		WebQuery: webquery,
 		Query:    u.Query(),
 		Domain:   u.Hostname(),
 	}, nil
 }
 
+// escapeDollarSign replaces dollar signs with their URL-encoded equivalent.
 func escapeDollarSign(s string) string {
 	return strings.ReplaceAll(s, "$", "%24")
 }