From 5147d0970ea9239b17a2d00e90122d47d410bc95 Mon Sep 17 00:00:00 2001
From: Jonathon Anderson <janderson@ciq.com>
Date: Tue, 10 Mar 2026 15:17:40 -0600
Subject: [PATCH] Simplify warewulfd parser

Signed-off-by: Jonathon Anderson <janderson@ciq.com>
---
 internal/pkg/warewulfd/parser.go | 127 ++++++++++++++-----------------
 1 file changed, 57 insertions(+), 70 deletions(-)

diff --git a/internal/pkg/warewulfd/parser.go b/internal/pkg/warewulfd/parser.go
index 47efe6f4..1fdbeda1 100644
--- a/internal/pkg/warewulfd/parser.go
+++ b/internal/pkg/warewulfd/parser.go
@@ -90,76 +90,85 @@ type parsedRequest struct {
 	compress   string
 }
 
+func parseHwaddr(hwaddr string) string {
+	hwaddr = strings.ReplaceAll(hwaddr, "-", ":")
+	hwaddr = strings.ToLower(hwaddr)
+	return hwaddr
+}
+
 func parseRequest(req *http.Request) (parsedRequest, error) {
 	var ret parsedRequest
 
-	url := strings.Split(req.URL.Path, "?")[0]
-	path_parts := strings.Split(url, "/")
+	path_parts := strings.Split(req.URL.Path, "/")
 
-	if len(path_parts) < 3 {
-		return ret, errors.New("unknown path components in GET")
+	// initial stage passed in the url path /[stage]/hwaddr
+	if len(path_parts) < 2 {
+		return ret, errors.New("path missing initial stage: " + req.URL.Path)
+	}
+	ret.stage = path_parts[1]
+
+	// prefer stage from query string for provision stage
+	if ret.stage == "provision" && len(req.URL.Query()["stage"]) > 0 {
+		ret.stage = req.URL.Query()["stage"][0]
 	}
 
-	// handle when stage was passed in the url path /[stage]/hwaddr
-	stage := path_parts[1]
-	hwaddr := ""
-	switch stage {
-	case "efiboot":
-		// /efiboot/{file}: no hwaddr in path; identified via ARP
-		if len(path_parts) > 3 {
+	// map the requested stage to a known stage
+	switch ret.stage {
+	case "provision":
+		ret.stage = "ipxe"
+	case "container":
+		ret.stage = "image"
+	case "overlay-system":
+		ret.stage = "system"
+	case "overlay-runtime":
+		ret.stage = "runtime"
+	}
+
+	if ret.stage == "" {
+		return ret, errors.New("no stage specified: " + req.URL.RawQuery)
+	}
+
+	if len(path_parts) > 2 {
+		if ret.stage == "efiboot" {
+			// /efiboot/{file}: no wwid in path; identified via ARP
 			ret.efifile = strings.Join(path_parts[2:], "/")
 		} else {
-			ret.efifile = path_parts[2]
+			ret.hwaddr = parseHwaddr(path_parts[2])
 		}
-	case "grub":
-		// /grub/{hwaddr}: hwaddr explicit in path
-		hwaddr = path_parts[2]
-		hwaddr = strings.ReplaceAll(hwaddr, "-", ":")
-		hwaddr = strings.ToLower(hwaddr)
-	default:
-		hwaddr = path_parts[2]
-		hwaddr = strings.ReplaceAll(hwaddr, "-", ":")
-		hwaddr = strings.ToLower(hwaddr)
 	}
-	ret.hwaddr = hwaddr
+
 	remoteAddrPort, err := netip.ParseAddrPort(req.RemoteAddr)
 	if err != nil {
 		return ret, errors.New("could not parse remote address")
 	}
 	ret.ipaddr = remoteAddrPort.Addr().String()
+	if ret.ipaddr == "" {
+		return ret, errors.New("could not obtain ipaddr from HTTP request")
+	}
 	ret.remoteport = int(remoteAddrPort.Port())
+	if ret.remoteport == 0 {
+		return ret, errors.New("could not obtain remote port from HTTP request: " + req.RemoteAddr)
+	}
+
+	if ret.hwaddr == "" && len(req.URL.Query()["wwid"]) > 0 {
+		ret.hwaddr = parseHwaddr(req.URL.Query()["wwid"][0])
+	}
+	if ret.hwaddr == "" {
+		if hwaddr := parseHwaddr(ArpFind(ret.ipaddr)); hwaddr != "" {
+			ret.hwaddr = hwaddr
+			wwlog.Verbose("using %s from arp cache for %s", ret.hwaddr, ret.ipaddr)
+		}
+	}
+	if ret.hwaddr == "" {
+		return ret, errors.New("unable to determine wwid: " + req.URL.RawQuery)
+	}
+
 	if len(req.URL.Query()["assetkey"]) > 0 {
 		ret.assetkey = req.URL.Query()["assetkey"][0]
 	}
-
 	if len(req.URL.Query()["uuid"]) > 0 {
 		ret.uuid = req.URL.Query()["uuid"][0]
 	}
-
-	if len(req.URL.Query()["stage"]) > 0 {
-		ret.stage = req.URL.Query()["stage"][0]
-	} else {
-
-		switch stage {
-		case "ipxe", "provision":
-			ret.stage = "ipxe"
-		case "kernel":
-			ret.stage = "kernel"
-		case "image", "container":
-			ret.stage = "image"
-		case "overlay-system", "system":
-			ret.stage = "system"
-		case "overlay-runtime", "runtime":
-			ret.stage = "runtime"
-		case "efiboot":
-			ret.stage = "efiboot"
-		case "grub":
-			ret.stage = "grub"
-		case "initramfs":
-			ret.stage = "initramfs"
-		}
-	}
-
 	if len(req.URL.Query()["overlay"]) > 0 {
 		ret.overlay = req.URL.Query()["overlay"][0]
 	}
@@ -169,28 +178,6 @@ func parseRequest(req *http.Request) (parsedRequest, error) {
 	if ret.efifile == "" && len(req.URL.Query()["file"]) > 0 {
 		ret.efifile = req.URL.Query()["file"][0]
 	}
-	if ret.stage == "" {
-		return ret, errors.New("no stage encoded in GET")
-	}
-	if ret.hwaddr == "" {
-		if len(req.URL.Query()["wwid"]) > 0 {
-			ret.hwaddr = req.URL.Query()["wwid"][0]
-			ret.hwaddr = strings.ReplaceAll(ret.hwaddr, "-", ":")
-			ret.hwaddr = strings.ToLower(ret.hwaddr)
-		} else {
-			ret.hwaddr = ArpFind(ret.ipaddr)
-			wwlog.Verbose("node mac not encoded, arp cache got %s for %s", ret.hwaddr, ret.ipaddr)
-			if ret.hwaddr == "" {
-				return ret, errors.New("no hwaddr encoded in GET")
-			}
-		}
-	}
-	if ret.ipaddr == "" {
-		return ret, errors.New("could not obtain ipaddr from HTTP request")
-	}
-	if ret.remoteport == 0 {
-		return ret, errors.New("could not obtain remote port from HTTP request: " + req.RemoteAddr)
-	}
 
 	return ret, nil
 }