Add gopher support; clean up

1 files changed, 271 insertions, 104 deletions

@@ -2,14 +2,11 @@
 # bollux: a bash gemini client
 # Author: Case Duckworth
 # License: MIT
-# Version: 0.3
+# Version: 0.4.0
 
 # Program information
 PRGN="${0##*/}"
-VRSN=0.3
-# State
-REDIRECTS=0
-set -f
+VRSN=0.4.0
 
 bollux_usage() {
         cat <<END
@@ -33,7 +30,7 @@ run() {
 }
 
 die() {
-        ec="$1"
+        local ec="$1"
         shift
         log error "$*"
         exit "$ec"
@@ -48,6 +45,8 @@ trim() {
 
 log() {
         [[ "$BOLLUX_LOGLEVEL" == QUIET ]] && return
+        local fmt
+
         case "$1" in
         [dD]*) # debug
                 [[ "$BOLLUX_LOGLEVEL" == DEBUG ]] || return
@@ -59,14 +58,15 @@ log() {
         *) fmt=1 ;;
         esac
         shift
+
         printf >&2 '\e[%sm%s:\e[0m\t%s\n' "$fmt" "$PRGN" "$*"
 }
 
 # main entry point
 bollux() {
-        run bollux_config
-        run bollux_args "$@"
-        run history_init
+        run bollux_config    # TODO: figure out better config method
+        run bollux_args "$@" # and argument parsing
+        run bollux_init
 
         if [[ ! "${BOLLUX_URL:+x}" ]]; then
                 run prompt GO BOLLUX_URL
@@ -105,6 +105,7 @@ bollux_config() {
         fi
 
         ## behavior
+        : "${BOLLUX_TIMEOUT:=30}"                     # connection timeout
         : "${BOLLUX_MAXREDIR:=5}"                     # max redirects
         : "${BOLLUX_PORT:=1965}"                      # port number
         : "${BOLLUX_PROTO:=gemini}"                   # default protocol
@@ -142,12 +143,12 @@ set_title() {
 }
 
 prompt() { # prompt [-u] PROMPT [READ_ARGS...]
-        read_cmd=(read -e -r)
+        local read_cmd=(read -e -r)
         if [[ "$1" == "-u" ]]; then
                 read_cmd+=(-i "$BOLLUX_URL")
                 shift
         fi
-        prompt="$1"
+        local prompt="$1"
         shift
         read_cmd+=(-p "$prompt> ")
         "${read_cmd[@]}" </dev/tty "$@"
@@ -155,30 +156,42 @@ prompt() { # prompt [-u] PROMPT [READ_ARGS...]
 
 blastoff() { # load a url
         local well_formed=true
+        local proto url
         if [[ "$1" == "-u" ]]; then
                 well_formed=false
                 shift
         fi
-        URL="$1"
+        url="$1"
 
         if $well_formed && [[ "$1" != "$BOLLUX_URL" ]]; then
-                URL="$(run transform_resource "$BOLLUX_URL" "$1")"
+                url="$(run transform_resource "$BOLLUX_URL" "$1")"
         fi
-        [[ "$URL" != *://* ]] && URL="$BOLLUX_PROTO://$URL"
-        URL="$(trim "$URL")"
+        [[ "$url" != *://* ]] && url="$BOLLUX_PROTO://$url"
+        url="$(trim "$url")"
+        proto="${url%://*}"
 
-        server="${URL#*://}"
-        server="${server%%/*}"
+        log d "PROTO='$proto' URL='$url'"
 
-        log d "URL='$URL' server='$server'"
-
-        run request_url "$server" "$BOLLUX_PORT" "$URL" |
-                run normalize_crlf |
-                run handle_response "$URL"
+        {
+                if declare -Fp "${proto}_request" >/dev/null; then
+                        run "${proto}_request" "$url"
+                else
+                        log d "No request handler for '$proto'; trying gemini"
+                        run gemini_request "$url"
+                fi
+        } | run normalize |
+                {
+                        if declare -Fp "${proto}_response" >/dev/null; then
+                                run "${proto}_response" "$url"
+                        else
+                                log d "No response handler for '$proto'; handling raw response"
+                                raw_response
+                        fi
+                }
 }
 
 transform_resource() { # transform_resource BASE_URL REFERENCE_URL
-        declare -A R B T # reference, base url, target
+        local -A R B T # reference, base url, target
         eval "$(run parse_url B "$1")"
         eval "$(run parse_url R "$2")"
         # A non-strict parser may ignore a scheme in the reference
@@ -223,7 +236,7 @@ transform_resource() { # transform_resource BASE_URL REFERENCE_URL
         fi
         isdefined R[fragment] && T[fragment]="${R[fragment]}"
         # cf. 5.3 -- recomposition
-        local r=""
+        local r
         isdefined "T[scheme]" && r="$r${T[scheme]}:"
         # remove the port from the authority
         isdefined "T[authority]" && r="$r//${T[authority]%:*}"
@@ -235,9 +248,9 @@ transform_resource() { # transform_resource BASE_URL REFERENCE_URL
 
 merge_paths() { # 5.2.3
         # shellcheck disable=2034
-        B_authority="$1"
-        B_path="$2"
-        R_path="$3"
+        local B_authority="$1"
+        local B_path="$2"
+        local R_path="$3"
         # if R_path is empty, get rid of // in B_path
         if [[ -z "$R_path" ]]; then
                 printf '%s\n' "${B_path//\/\//\//}"
@@ -258,8 +271,7 @@ merge_paths() { # 5.2.3
 
 remove_dot_segments() { # 5.2.4
         local input="$1"
-        local output=
-        # ^/\.(/|$) - BASH_REMATCH[0]
+        local output
         while [[ "$input" ]]; do
                 if [[ "$input" =~ ^\.\.?/ ]]; then
                         input="${input#${BASH_REMATCH[0]}}"
@@ -306,40 +318,54 @@ parse_url() { # eval "$(split_url NAME STRING)" => NAME[...]
 isdefined() { [[ "${!1+x}" ]]; } # isdefined NAME
 # is a NAME defined AND empty?
 isempty() { [[ ! "${!1-x}" ]]; } # isempty NAME
+# split a string -- see pure bash bible
+split() { # split STRING DELIMITER
+        local -a arr
+        IFS=$'\n' read -d "" -ra arr <<<"${1//$2/$'\n'}"
+        printf '%s\n' "${arr[@]}"
+}
+
+# GEMINI
+# https://gemini.circumlunar.space/docs/spec-spec.txt
+gemini_request() {
+        local url port server
+        local ssl_cmd
+        url="$1"
+        port=1965
+        server="${url#*://}"
+        server="${server%%/*}"
 
-request_url() {
-        local server="$1"
-        local port="$2"
-        local url="$3"
-
-        # support for TLS v1.3 and v1.2
         ssl_cmd=(openssl s_client -crlf -quiet -connect "$server:$port")
-        ssl_cmd+=(-servername "$server") # SNI
-        ssl_cmd_tls1_2=("${ssl_cmd[@]}" -tls1_2)
-        ssl_cmd_tls1_3=("${ssl_cmd[@]}" -tls1_3)
+        # disable old TLS/SSL versions (thanks makeworld!)
+        ssl_cmd+=(-no_ssl3 -no_tls1 -no_tls1_1)
 
         # always try to connect with TLS v1.3 first
-        run "${ssl_cmd_tls1_3[@]}" <<<"$url" 2>/dev/null ||
-                run "${ssl_cmd_tls1_2[@]}" <<<"$url" 2>/dev/null
+        run "${ssl_cmd[@]}" <<<"$url" 2>/dev/null
 }
 
-handle_response() {
-        local URL="$1" code meta
+gemini_response() {
+        local url code meta
+        local title
+        url="$1"
+
+        # we need a loop here so it waits for the first line
+        while read -t "3" -r code meta ||
+                { (($? > 128)) && die 99 "Timeout."; }; do
+                break
+        done
 
-        read -r code meta
         log d "[$code] $meta"
 
         case "$code" in
-        1*)
+        1*) # input
                 REDIRECTS=0
-                run history_append "$URL" "$meta"
                 run prompt "$meta"
                 run blastoff "?$REPLY"
                 ;;
-        2*)
+        2*) # OK
                 REDIRECTS=0
                 # read ahead to find a title
-                pretitle=
+                local pretitle
                 while read -r; do
                         pretitle="$pretitle$REPLY"$'\n'
                         if [[ "$REPLY" =~ ^#[[:space:]]*(.*) ]]; then
@@ -347,59 +373,194 @@ handle_response() {
                                 break
                         fi
                 done
-                run history_append "$URL" "${title:-}"
+                run history_append "$url" "${title:-}"
+                # read the body out and pipe it to display
                 {
                         printf '%s' "$pretitle"
-                        while read -r; do
-                                printf '%s\n' "$REPLY"
-                        done
-                } | run display "$meta"
+                        passthru
+                } | run display "$meta" "${title:-}"
                 ;;
-        3*)
+        3*) # redirect
                 ((REDIRECTS += 1))
                 if ((REDIRECTS > BOLLUX_MAXREDIR)); then
                         die $((100 + code)) "Too many redirects!"
                 fi
-                run blastoff "$meta"
+                run blastoff "$meta" # TODO: confirm redirect
                 ;;
-        4*)
+        4*) # temporary error
                 REDIRECTS=0
-                die "$((100 + code))" "Temporary error: $code"
+                die "$((100 + code))" "Temporary error [$code]: $meta"
                 ;;
-        5*)
+        5*) # permanent error
                 REDIRECTS=0
-                die "$((100 + code))" "Permanent error: $code"
+                die "$((100 + code))" "Permanent error [$code]: $meta"
                 ;;
-        6*)
+        6*) # certificate error
                 REDIRECTS=0
-                die "$((100 + code))" "Certificate error: $code"
+                log d "Not implemented: Client certificates"
+                # TODO: recheck the speck
+                die "$((100 + code))" "[$code] $meta"
                 ;;
         *)
                 [[ -z "${code-}" ]] && die 100 "Empty response code."
-                die "$((100 + code)) Unknown response code: $code."
+                die "$((100 + code))" "Unknown response code: $code."
+                ;;
+        esac
+}
+
+# GOPHER
+# https://tools.ietf.org/html/rfc1436 protocol
+# https://tools.ietf.org/html/rfc4266 url
+gopher_request() {
+        local url server port type path
+        url="$1"
+        port=70
+
+        # RFC 4266
+        [[ "$url" =~ gopher://([^/?#:]*)(:([0-9]+))?(/((.))?(/?.*))?$ ]]
+        server="${BASH_REMATCH[1]}"
+        port="${BASH_REMATCH[3]:-70}"
+        type="${BASH_REMATCH[6]:-1}"
+        path="${BASH_REMATCH[7]}"
+
+        log d "URL='$url' SERVER='$server' TYPE='$type' PATH='$path'"
+
+        exec 9<>"/dev/tcp/$server/$port"
+        printf '%s\r\n' "$path" >&9
+        passthru <&9
+}
+
+gopher_response() {
+        local url pre type cur_server
+        pre=false
+        url="$1"
+        # RFC 4266
+        [[ "$url" =~ gopher://([^/?#:]*)(:([0-9]+))?(/((.))?(/?.*))?$ ]]
+        cur_server="${BASH_REMATCH[1]}"
+        type="${BASH_REMATCH[6]:-1}"
+
+        run history_append "$url" "" # TODO: get the title ??
+
+        log d "TYPE='$type'"
+
+        case "$type" in
+        0) # text
+                run display text/plain
+                ;;
+        1) # menu
+                run gopher_convert | run display text/gemini
+                ;;
+        3) # failure
+                die 203 "GOPHER: failed"
+                ;;
+        7) # search
+                die 207 "Not implemented"
+                ;;
+        *) # something else
+                die "$((200 + ${type:-0}))" "Not implemented"
                 ;;
         esac
 }
 
-display() {
+passthru() {
+        while IFS= read -r; do
+                printf '%s\n' "$REPLY"
+        done
+}
+
+gopher_convert() {
+        local type label path server port regex
+        # cf. https://github.com/jamestomasino/dotfiles-minimal/blob/master/bin/gophermap2gemini.awk
+        while IFS= read -r; do
+                printf -v regex '(.)([^\t]*)(\t([^\t]*)\t([^\t]*)\t([^\t]*))?'
+                if [[ "$REPLY" =~ $regex ]]; then
+                        type="${BASH_REMATCH[1]}"
+                        label="${BASH_REMATCH[2]}"
+                        path="${BASH_REMATCH[4]:-/}"
+                        server="${BASH_REMATCH[5]:-$cur_server}"
+                        port="${BASH_REMATCH[6]}"
+                else
+                        log e "CAN'T PARSE LINE"
+                        printf '%s\n' "$REPLY"
+                        continue
+                fi
+                case "$type" in
+                .) # end of file
+                        printf '.\n'
+                        break
+                        ;;
+                i) # label
+                        case "$label" in
+                        '#'* | '*'[[:space:]]*)
+                                if $pre; then
+                                        printf '%s\n' '```'
+                                        pre=false
+                                fi
+                                ;;
+                        *)
+                                if ! $pre; then
+                                        printf '%s\n' '```'
+                                        pre=true
+                                fi
+                                ;;
+                        esac
+                        printf '%s\n' "$label"
+                        ;;
+                h) # html link
+                        if $pre; then
+                                printf '%s\n' '```'
+                                pre=false
+                        fi
+                        printf '=> %s %s\n' "${path:4}" "$label"
+                        ;;
+                T) # telnet link
+                        if $pre; then
+                                printf '%s\n' '```'
+                                pre=false
+                        fi
+                        printf '=> telnet://%s:%s/%s%s %s\n' \
+                                "$server" "$port" "$type" "$path" "$label"
+                        ;;
+                *) # other type
+                        if $pre; then
+                                printf '%s\n' '```'
+                                pre=false
+                        fi
+                        printf '=> gopher://%s:%s/%s%s %s\n' \
+                                "$server" "$port" "$type" "$path" "$label"
+                        ;;
+                esac
+        done
+        if $pre; then
+                printf '%s\n' '```'
+        fi
+        # close the connection
+        exec 9<&-
+        exec 9>&-
+}
+
+display() { # display METADATA [TITLE]
+        local -a less_cmd
+        local i mime charset
         # split header line
         local -a hdr
-        local i mime charset h
-        IFS=$'\n' read -d "" -ra hdr <<<"${1//;/$'\n'}"
+        IFS=';' read -ra hdr <<<"$1"
+        # title is optional but nice looking
+        local title
+        if (($# == 2)); then
+                title="$2"
+        fi
 
         mime="$(trim "${hdr[0],,}")"
         for ((i = 1; i <= "${#hdr[@]}"; i++)); do
-                h="$(trim "${hdr[$i]}")"
+                h="${hdr[$i]}"
                 case "$h" in
-                charset=*) charset="${h#charset=}" ;;
-                        # add mime-extensions here
+                *charset=*) charset="${h#*=}" ;;
                 esac
         done
 
         [[ -z "$mime" ]] && mime="text/gemini"
-        if [[ -z "$charset" ]]; then
-                charset="utf-8"
-        fi
+        [[ -z "$charset" ]] && charset="utf-8"
 
         log debug "mime='$mime'; charset='$charset'"
 
@@ -409,28 +570,25 @@ display() {
                 less_cmd=(less -R)
                 mklesskey "$BOLLUX_LESSKEY" && less_cmd+=(-k "$BOLLUX_LESSKEY")
                 less_cmd+=(
-                        -Pm'bollux$'
+                        -Pm"$title${title:+ - }bollux$"
                         -PM'o\:open, g\:goto, [\:back, ]\:forward, r\:refresh$'
-                        -M
+                        -m
                 )
 
-                submime="${mime#*/}"
-                if declare -F | grep -q "$submime"; then
-                        log d "typeset_$submime"
-                        {
-                                iconv -f "${charset^^}" -t "UTF-8" |
-                                        tee "$BOLLUX_PAGESRC" |
-                                        run "typeset_$submime" |
-                                        run "${less_cmd[@]}" && bollux_quit
-                        } || run handle_keypress "$?"
+                local typeset
+                local submime="${mime#*/}"
+                if declare -Fp "typeset_$submime" >/dev/null; then
+                        typeset="typeset_$submime"
                 else
-                        log "cat"
-                        {
-                                iconv -f "${charset^^}" -t "UTF-8" |
-                                        tee "$BOLLUX_PAGESRC" |
-                                        run "${less_cmd[@]}" && bollux_quit
-                        } || run handle_keypress "$?"
+                        typeset="passthru"
                 fi
+
+                {
+                        run iconv -f "${charset^^}" -t "UTF-8" |
+                                run tee "$BOLLUX_PAGESRC" |
+                                run "$typeset" |
+                                run "${less_cmd[@]}" && bollux_quit
+                } || run handle_keypress "$?"
                 ;;
         *) run download "$BOLLUX_URL" ;;
         esac
@@ -450,7 +608,7 @@ mklesskey() {
         END
 }
 
-normalize_crlf() {
+normalize() {
         shopt -s extglob
         while IFS= read -r; do
                 printf '%s\n' "${REPLY//$'\r'?($'\n')/}"
@@ -480,7 +638,7 @@ typeset_gemini() {
 
         while IFS= read -r; do
                 case "$REPLY" in
-                '```' | '```'*)
+                '```'*)
                         if $pre; then
                                 pre=false
                         else
@@ -493,12 +651,8 @@ typeset_gemini() {
                         gemini_link "$REPLY" $pre "$ln"
                         ;;
                 '#'*) gemini_header "$REPLY" $pre ;;
-                '*'*)
-                        if [[ "$REPLY" =~ ^\*[[:space:]]+ ]]; then
-                                gemini_list "$REPLY" $pre
-                        else
-                                gemini_text "$REPLY" $pre
-                        fi
+                '*'[[:space:]]*)
+                        gemini_list "$REPLY" $pre
                         ;;
                 *) gemini_text "$REPLY" $pre ;;
                 esac
@@ -607,8 +761,8 @@ handle_keypress() {
                 run select_url "$BOLLUX_PAGESRC"
                 ;;
         49) # g - goto a url -- input a new url
-                prompt GO URL
-                run blastoff -u "$URL"
+                prompt GO
+                run blastoff -u "$REPLY"
                 ;;
         50) # [ - back in the history
                 run history_back || {
@@ -626,10 +780,11 @@ handle_keypress() {
                 run blastoff "$BOLLUX_URL"
                 ;;
         53) # G - goto a url (pre-filled with current)
-                prompt -u GO URL
-                run blastoff -u "$URL"
+                prompt -u GO
+                run blastoff -u "$REPLY"
                 ;;
-        *) # 53-57 -- still available for binding
+        *) # 54-57 -- still available for binding
+                die "$?" "less(1) error"
                 ;;
         esac
 }
@@ -640,6 +795,7 @@ select_url() {
         select u in "${MAPFILE[@]}"; do
                 case "$REPLY" in
                 q) bollux_quit ;;
+                [^0-9]*) run blastoff -u "$REPLY" && break ;;
                 esac
                 run blastoff "${u%%[[:space:]]*}" && break
         done </dev/tty
@@ -675,13 +831,25 @@ download() {
         fi
 }
 
-history_init() {
+bollux_init() {
+        # Trap cleanup
+        trap bollux_cleanup INT QUIT EXIT
+        # State
+        REDIRECTS=0
+        set -f
+        # History
         declare -a HISTORY # history is kept in an array
         HN=0               # position of history in the array
         run mkdir -p "${BOLLUX_HISTFILE%/*}"
 }
 
-history_append() { # history_append URL TITLE
+bollux_cleanup() {
+        # XXX
+        :
+        #kill $(jobs -p)
+}
+
+history_append() { # history_append url TITLE
         BOLLUX_URL="$1"
         # date/time, url, title (best guess)
         run printf '%(%FT%T)T\t%s\t%s\n' -1 "$1" "$2" >>"$BOLLUX_HISTFILE"
@@ -699,6 +867,7 @@ history_back() {
         fi
         run blastoff "${HISTORY[$HN]}"
 }
+
 history_forward() {
         log d "HN=$HN"
         if ((HN >= ${#HISTORY[@]})); then
@@ -711,6 +880,4 @@ history_forward() {
 
 if [[ "${BASH_SOURCE[0]}" == "$0" ]]; then
         run bollux "$@"
-else
-        BOLLUX_LOGLEVEL=DEBUG
 fi