[arch-projects] [netctl] [PATCH] wifi-menu: Support UTF-8 encoded SSIDs (FS#45384)
Strictly speaking, we should check with the SSIDEncoding value sent out by the station, as specified in the 2012 version of 802.11 (page 566), but wpa_supplicant does not (yet) expose this information and stations may not set the field to UTF-8 and still encode their SSID accordingly. We therefore assume SSIDs to be UTF-8 encoded (the only real alternative is treating SSIDs as raw byte string) in wifi-menu. Note that wifi-menu is only provided as a convenience and many/most use cases involve writing profile files manually. When the current character map is not set to UTF-8 , wifi-menu will only use ASCII characters in its screen output. This can be forced using for example `LC_CTYPE=C wifi-menu`. --- src/wifi-menu | 40 ++++++++++++++++++++++++++++++---------- 1 file changed, 30 insertions(+), 10 deletions(-) diff --git a/src/wifi-menu b/src/wifi-menu index 42c4c53..525830f 100755 --- a/src/wifi-menu +++ b/src/wifi-menu @@ -20,6 +20,11 @@ Arguments: END } +# Undo printf escaping in $1 +printf_decode() { + printf -- "${1//%/%%}" +} + # Prepare $1 for use in a special quoting context quote_safe() { if [[ "$1" = \"* ]]; then @@ -38,6 +43,9 @@ init_profiles() { unset Interface ESSID source "$PROFILE_DIR/$profile" > /dev/null if [[ "$Interface" = "$1" && -n "$ESSID" ]]; then + if [[ "$ESSID" = \"\"*\" ]]; then + ESSID=${ESSID:2:-1} + fi printf "%s" "$ESSID" if [[ "$Description" =~ "Automatically generated" ]]; then return 2 @@ -62,14 +70,19 @@ init_profiles() { # Build ENTRIES as an argument list for dialog based on scan results in $1 init_entries() { - local i=0 sep=$'\t' flags signal ssid + local i=0 sep=$'\t' decoded flags signal ssid while IFS=$'\t' read -r signal flags ssid; do + decoded=$(printf_decode "$ssid") ENTRIES[i++]="--" # the SSID might look like an option to dialog - ENTRIES[i++]=$ssid - if is_yes "${CONNECTED:-no}" && [[ "$ssid" = "$CONNECTION" ]]; then + if [[ "$CHARMAP" = "UTF-8" ]]; then + ENTRIES[i++]=$decoded + else + ENTRIES[i++]=$ssid + fi + if is_yes "${CONNECTED:-no}" && [[ "$decoded" = "$CONNECTION" ]]; then ENTRIES[i]="*" # Currently connected - elif in_array "$ssid" "${ESSIDS[@]}"; then - if in_array "$(ssid_to_profile "$ssid")" "${GENERATED[@]}"; then + elif in_array "$decoded" "${ESSIDS[@]}"; then + if in_array "$(ssid_to_profile "$decoded")" "${GENERATED[@]}"; then ENTRIES[i]="." # Automatically generated else ENTRIES[i]=":" # Handmade @@ -115,11 +128,14 @@ Do you want to overwrite it?" # Create a profile for ssid $1 create_profile() { - local box flags key msg security - PROFILE="$INTERFACE-${1//\//_}" + local box flags key msg security signal ssid + PROFILE=$(iconv -c -f UTF-8 -t //TRANSLIT <<< "$1") + PROFILE="$INTERFACE-${PROFILE//[?\/]/_}" [[ -e "$PROFILE_DIR/$PROFILE" ]] && PROFILE+=".wifi-menu" confirm_profile || return $? - flags=$(grep -m 1 $'\t'"$1\$" "$NETWORKS" | cut -f 2) + while IFS=$'\t' read -r signal flags ssid; do + [[ "$(printf_decode "$ssid")" != "$1" ]] || break + done < "$NETWORKS" if [[ "$flags" =~ WPA|WEP ]]; then security=${BASH_REMATCH[0],,} else @@ -221,6 +237,8 @@ ensure_root "$(basename "$0")" if ! type dialog &> /dev/null; then exit_error "Please install 'dialog' to use wifi-menu" fi +CHARMAP=$(locale charmap) +cd / # We do not want to spawn anything that can block unmounting INTERFACE=$1 if [[ -z "$INTERFACE" ]]; then @@ -237,7 +255,6 @@ elif ! is_interface "$INTERFACE"; then fi load_interface_config "$INTERFACE" -cd / # We do not want to spawn anything that can block unmounting if [[ "$RFKill" && "$(rf_status "$INTERFACE" "$RFKill")" ]]; then if ! rf_enable "$INTERFACE" "$RFKill"; then exit_error "Could not unblock transmission on interface '$INTERFACE'" @@ -247,7 +264,7 @@ fi echo -n "Scanning for networks... " if CONNECTION=$(wpa_call "$INTERFACE" status 2> /dev/null | grep -m 1 "^ssid="); then - CONNECTION=${CONNECTION#ssid=} + CONNECTION=$(printf_decode "${CONNECTION#ssid=}") CONNECTED=yes fi NETWORKS=$(wpa_supplicant_scan "$INTERFACE" 3,4,5) @@ -270,6 +287,9 @@ Flags description: CHOICE=$(dialog --menu "$MSG" 24 50 12 "${ENTRIES[@]}" --stdout) RETURN=$? if (( RETURN == 0 )); then + if [[ "$CHARMAP" != "UTF-8" ]]; then + CHOICE=$(printf_decode "$CHOICE") + fi connect_to_ssid "$CHOICE" RETURN=$? fi -- 2.18.0
Hi Jouke, Thanks for fixing this! Sharing a couple of ideas that come to mind. On 23 June 2018 at 12:25, Jouke Witteveen via arch-projects <arch-projects@archlinux.org> wrote:
Strictly speaking, we should check with the SSIDEncoding value sent out by the station, as specified in the 2012 version of 802.11 (page 566), but wpa_supplicant does not (yet) expose this information and stations may not set the field to UTF-8 and still encode their SSID accordingly.
I wonder if iwd (meant to be wpa_supplicant replacement) could help here. Be that by exposing SSIDEncoding or setting it, as it detects UTF-8 in the SSID.
@@ -38,6 +43,9 @@ init_profiles() { unset Interface ESSID source "$PROFILE_DIR/$profile" > /dev/null if [[ "$Interface" = "$1" && -n "$ESSID" ]]; then + if [[ "$ESSID" = \"\"*\" ]]; then + ESSID=${ESSID:2:-1} + fi FYI, I will send out a small patch which simplifying init_profiles. It will cause a trivial conflict so I can rebase it on top of this patch, if you prefer.
@@ -221,6 +237,8 @@ ensure_root "$(basename "$0")" if ! type dialog &> /dev/null; then exit_error "Please install 'dialog' to use wifi-menu" fi +CHARMAP=$(locale charmap) +cd / # We do not want to spawn anything that can block unmounting
Was the "cd /" line moved by mistake? Other than the cd nitpick, the patch looks great. FWIW Reviewed-by: Emil Velikov <emil.l.velikov@gmail.com> HTH Emil
On Tue, Jun 26, 2018 at 12:17 PM Emil Velikov <emil.l.velikov@gmail.com> wrote:
On 23 June 2018 at 12:25, Jouke Witteveen via arch-projects <arch-projects@archlinux.org> wrote:
Strictly speaking, we should check with the SSIDEncoding value sent out by the station, as specified in the 2012 version of 802.11 (page 566), but wpa_supplicant does not (yet) expose this information and stations may not set the field to UTF-8 and still encode their SSID accordingly.
I wonder if iwd (meant to be wpa_supplicant replacement) could help here. Be that by exposing SSIDEncoding or setting it, as it detects UTF-8 in the SSID.
As far as I can tell, only developers of iwd think of iwd as a wpa_supplicant replacement. Currently, it is lacking a CLI and documentation. Moreover, wpa_supplicant is tried and tested. While I try to not tie netctl in too strongly with any specific client, wpa_supplicant will likely be the client of choice for netctl for the foreseeable future. I submitted a patch to wpa_supplicant for exposing the UTF-8 SSID bit: http://lists.infradead.org/pipermail/hostap/2018-June/038658.html However, experimentation showed that it is not really used (yet?).
@@ -38,6 +43,9 @@ init_profiles() { unset Interface ESSID source "$PROFILE_DIR/$profile" > /dev/null if [[ "$Interface" = "$1" && -n "$ESSID" ]]; then + if [[ "$ESSID" = \"\"*\" ]]; then + ESSID=${ESSID:2:-1} + fi FYI, I will send out a small patch which simplifying init_profiles. It will cause a trivial conflict so I can rebase it on top of this patch, if you prefer.
Note that the above code is more about matching multiple encodings of ESSIDs than about the UTF-8 conversion per se. Thanks for being so quick with your patch!
@@ -221,6 +237,8 @@ ensure_root "$(basename "$0")" if ! type dialog &> /dev/null; then exit_error "Please install 'dialog' to use wifi-menu" fi +CHARMAP=$(locale charmap) +cd / # We do not want to spawn anything that can block unmounting
Was the "cd /" line moved by mistake?
No. It was moved because it fits better (semantically) in this section of the code.
Other than the cd nitpick, the patch looks great. FWIW Reviewed-by: Emil Velikov <emil.l.velikov@gmail.com>
I'll reply to your patch in the other mail. Regards, - Jouke
On 26 June 2018 at 11:34, Jouke Witteveen <j.witteveen@gmail.com> wrote:
On Tue, Jun 26, 2018 at 12:17 PM Emil Velikov <emil.l.velikov@gmail.com> wrote:
On 23 June 2018 at 12:25, Jouke Witteveen via arch-projects <arch-projects@archlinux.org> wrote:
Strictly speaking, we should check with the SSIDEncoding value sent out by the station, as specified in the 2012 version of 802.11 (page 566), but wpa_supplicant does not (yet) expose this information and stations may not set the field to UTF-8 and still encode their SSID accordingly.
I wonder if iwd (meant to be wpa_supplicant replacement) could help here. Be that by exposing SSIDEncoding or setting it, as it detects UTF-8 in the SSID.
As far as I can tell, only developers of iwd think of iwd as a wpa_supplicant replacement. Currently, it is lacking a CLI and documentation. Moreover, wpa_supplicant is tried and tested. While I try to not tie netctl in too strongly with any specific client, wpa_supplicant will likely be the client of choice for netctl for the foreseeable future.
I've noticed the iwctl and iwmon utils provided, although did not play around with them. You're on point though, sticking with wpa_supplicant for the foreseeable future is fine.
I submitted a patch to wpa_supplicant for exposing the UTF-8 SSID bit: http://lists.infradead.org/pipermail/hostap/2018-June/038658.html However, experimentation showed that it is not really used (yet?).
Vendors may start using it... one day Thanks again for the comprehensive answers! Emil
participants (2)
-
Emil Velikov
-
Jouke Witteveen