refactor(make): Use awk to extract target names

bash_completion

@@ -3207,6 +3207,22 @@ _comp_xfunc()
     "$xfunc_name" "${@:3}"
 }
 
+# Call a POSIX-compatible awk.  Solaris awk is not POSIX-compliant, but Solaris
+# provides a POSIX-compatible version through /usr/xpg4/bin/awk.  We switch the
+# implementation to /usr/xpg4/bin/awk in Solaris if any.
+# @since 2.12
+if [[ $OSTYPE == *solaris* && -x /usr/xpg4/bin/awk ]]; then
+    _comp_awk()
+    {
+        /usr/xpg4/bin/awk "$@"
+    }
+else
+    _comp_awk()
+    {
+        command awk "$@"
+    }
+fi
+
 # source compat completion directory definitions
 _comp__init_compat_dirs=()
 if [[ ${BASH_COMPLETION_COMPAT_DIR-} ]]; then

completions/_nmcli

@@ -6,25 +6,25 @@
 _comp_cmd_nmcli__con_id()
 {
     _comp_compgen_split -l -- "$(nmcli con list 2>/dev/null |
-        tail -n +2 | awk -F ' {2,}' '{print $1 }')"
+        tail -n +2 | _comp_awk -F ' {2,}' '{print $1}')"
 }
 
 _comp_cmd_nmcli__con_uuid()
 {
     _comp_compgen_split -- "$(nmcli con list 2>/dev/null |
-        tail -n +2 | awk -F ' {2,}' '{print $2}')"
+        tail -n +2 | _comp_awk -F ' {2,}' '{print $2}')"
 }
 
 _comp_cmd_nmcli__ap_ssid()
 {
     _comp_compgen_split -l -- "$(nmcli dev wifi list 2>/dev/null |
-        tail -n +2 | awk -F ' {2,}' '{print $1}')"
+        tail -n +2 | _comp_awk -F ' {2,}' '{print $1}')"
 }
 
 _comp_cmd_nmcli__ap_bssid()
 {
     _comp_compgen_split -- "$(nmcli dev wifi list 2>/dev/null |
-        tail -n +2 | awk -F ' {2,}' '{print $2}')"
+        tail -n +2 | _comp_awk -F ' {2,}' '{print $2}')"
 }
 
 _comp_cmd_nmcli()

completions/convert

@@ -49,7 +49,7 @@ _comp_cmd_convert__common_options()
             return
             ;;
         -format)
-            _comp_compgen_split -- "$(convert -list format | awk \
+            _comp_compgen_split -- "$(convert -list format | _comp_awk \
                 '/ [r-][w-][+-] / { sub("[*]$","",$1); print tolower($1) }')"
             return
             ;;

completions/dpkg

@@ -7,7 +7,7 @@ _comp_xfunc_dpkg_compgen_installed_packages()
         grep-status -P -e "^${cur-}" -a \
             -FStatus 'ok installed' \
             -n -s Package 2>/dev/null ||
-            command awk -F '\n' -v RS="" "
+            _comp_awk -F '\n' -v RS="" "
             index(\$1, \"Package: ${cur-}\") == 1 &&
             \$2 ~ /ok installed|half-installed|unpacked|half-configured|^Essential: yes/ {
                 print(substr(\$1, 10));
@@ -22,7 +22,7 @@ _comp_xfunc_dpkg_compgen_purgeable_packages()
         grep-status -P -e "^${cur-}" -a \
             -FStatus 'ok installed' -o -FStatus 'ok config-files' \
             -n -s Package 2>/dev/null ||
-            command awk -F '\n' -v RS="" "
+            _comp_awk -F '\n' -v RS="" "
             index(\$1, \"Package: ${cur-}\") == 1 &&
             \$2 ~ /ok installed|half-installed|unpacked|half-configured|config-files|^Essential: yes/ {
                 print(substr(\$1, 10));

completions/iwconfig

@@ -15,7 +15,7 @@ _comp_cmd_iwconfig()
             _comp_compgen -- -W 'on off any'
             if [[ ${BASH_COMPLETION_CMD_IWCONFIG_SCAN-${COMP_IWLIST_SCAN-}} ]]; then
                 _comp_compgen -a split -- "$(iwlist "${words[1]}" scan |
-                    awk -F'\"' '/ESSID/ {print $2}')"
+                    _comp_awk -F '\"' '/ESSID/ {print $2}')"
             fi
             return
             ;;
@@ -38,7 +38,7 @@ _comp_cmd_iwconfig()
             _comp_compgen -- -W 'on off any'
             if [[ ${BASH_COMPLETION_CMD_IWCONFIG_SCAN-${COMP_IWLIST_SCAN-}} ]]; then
                 _comp_compgen -a split -- "$(iwlist "${words[1]}" scan |
-                    awk -F ': ' '/Address/ {print $2}')"
+                    _comp_awk -F ': ' '/Address/ {print $2}')"
             fi
             return
             ;;

completions/make

@@ -1,91 +1,21 @@
 # bash completion for GNU make                             -*- shell-script -*-
 
-# TODO: rename per API conventions, rework to use vars rather than outputting(?)
-_make_target_extract_script()
+# Extract the valid target names starting with PREFIX from the output of
+# `make -npq'
+# @param mode    If this is `-d', the directory names already specified in
+#                PREFIX are omitted in the output
+# @param prefix  Prefix of the target names
+_comp_cmd_make__extract_targets()
 {
-    local mode="$1"
-    shift
-
-    local prefix="$1"
-    local prefix_pat=$(command sed 's/[][\,.*^$(){}?+|/]/\\&/g' <<<"$prefix")
-    local basename=${prefix##*/}
-    local dirname_len=$((${#prefix} - ${#basename}))
-    # Avoid expressions like '\(.\{0\}\)', FreeBSD sed doesn't like them:
-    # > sed: 1: ...: RE error: empty (sub)expression
-    local dirname_re=
-    ((dirname_len > 0)) && dirname_re="\(.\{${dirname_len}\}\)"
-
-    if [[ ! $dirname_re ]]; then
-        local output="\1"
-    elif [[ $mode == -d ]]; then
-        # display mode, only output current path component to the next slash
-        local output="\2"
-    else
-        # completion mode, output full path to the next slash
-        local output="\1\2"
-    fi
-
-    cat <<EOF
-    1,/^# * Make data base/           d;        # skip any makefile output
-    /^# * Finished Make data base/,/^# * Make data base/{
-                                      d;        # skip any makefile output
-    }
-    /^# * Variables/,/^# * Files/     d;        # skip until files section
-    /^# * Not a target/,/^$/          d;        # skip not target blocks
-    /^${prefix_pat}/,/^$/!            d;        # skip anything user dont want
-
-    # The stuff above here describes lines that are not
-    #  explicit targets or not targets other than special ones
-    # The stuff below here decides whether an explicit target
-    #  should be output.
-
-    /^# * File is an intermediate prerequisite/ {
-      s/^.*$//;x;                               # unhold target
-      d;                                        # delete line
-    }
-
-    /^$/ {                                      # end of target block
-      x;                                        # unhold target
-      /^$/d;                                    # dont print blanks
-      s|^${dirname_re-}\(.\{${#basename}\}[^:]*\):.*$|${output}|p;
-      d;                                        # hide any bugs
-    }
-
-    # This pattern includes a literal tab character as \t is not a portable
-    # representation and fails with BSD sed
-    /^[^#	:%]\{1,\}:/ {         # found target block
-      /^\.PHONY:/                 d;            # special target
-      /^\.SUFFIXES:/              d;            # special target
-      /^\.DEFAULT:/               d;            # special target
-      /^\.PRECIOUS:/              d;            # special target
-      /^\.INTERMEDIATE:/          d;            # special target
-      /^\.SECONDARY:/             d;            # special target
-      /^\.SECONDEXPANSION:/       d;            # special target
-      /^\.DELETE_ON_ERROR:/       d;            # special target
-      /^\.IGNORE:/                d;            # special target
-      /^\.LOW_RESOLUTION_TIME:/   d;            # special target
-      /^\.SILENT:/                d;            # special target
-      /^\.EXPORT_ALL_VARIABLES:/  d;            # special target
-      /^\.NOTPARALLEL:/           d;            # special target
-      /^\.ONESHELL:/              d;            # special target
-      /^\.POSIX:/                 d;            # special target
-      /^\.NOEXPORT:/              d;            # special target
-      /^\.MAKE:/                  d;            # special target
-EOF
-
-    # don't complete with hidden targets unless we are doing a partial completion
-    if [[ ! ${prefix_pat} || ${prefix_pat} == */ ]]; then
-        cat <<EOF
-      /^${prefix_pat}[^a-zA-Z0-9]/d;            # convention for hidden tgt
-EOF
-    fi
+    local mode=$1
+    local -x prefix=$2
 
-    cat <<EOF
-      h;                                        # hold target
-      d;                                        # delete line
-    }
+    # display mode, only output current path component to the next slash
+    local -x prefix_replace=$prefix
+    [[ $mode == -d && $prefix == */* ]] &&
+        prefix_replace=${prefix##*/}
 
-EOF
+    _comp_awk -f "${BASH_SOURCE[0]%/*}/../helpers/make-extract-targets.awk"
 }
 
 # Truncate the non-unique filepaths in COMPREPLY to only generate unique
@@ -224,11 +154,11 @@ _comp_cmd_make()
         #     mode=-d # display-only mode
         # fi
 
-        local IFS=$' \t\n' script=$(_make_target_extract_script $mode "$cur")
+        local IFS=$' \t\n'
         COMPREPLY=($(LC_ALL=C \
             $1 -npq __BASH_MAKE_COMPLETION__=1 \
             ${makef+"${makef[@]}"} "${makef_dir[@]}" .DEFAULT 2>/dev/null |
-            command sed -ne "$script"))
+            _comp_cmd_make__extract_targets "$mode" "$cur"))
 
         _comp_cmd_make__truncate_non_unique_paths
 

completions/perl

@@ -132,7 +132,7 @@ _comp_cmd_perldoc()
             if [[ $cur == p* ]]; then
                 _comp_compgen -a split -- "$(PERLDOC_PAGER=cat "$1" -u perl |
                     command sed -ne '/perl.*Perl overview/,/perlwin32/p' |
-                    awk '$NF=2 && $1 ~ /^perl/ { print $1 }')"
+                    awk 'NF >= 2 && $1 ~ /^perl/ { print $1 }')"
             fi
         fi
         _comp_compgen -a filedir 'p@([lm]|od)'

completions/portinstall

@@ -16,7 +16,7 @@ _comp_cmd_portinstall()
         indexfile=$portsdir/INDEX
     [[ -f $indexfile && -r $indexfile ]] || return
 
-    _comp_compgen_split -l -- "$(awk -F '|' '
+    _comp_compgen_split -l -- "$(_comp_awk -F '|' '
         BEGIN { portsdir = ENVIRON["portsdir"]; len = length(portsdir) }
         { print $1 }
         substr($2, 1, len) == portsdir { print substr($2, len + 1) }

helpers/Makefile.am

@@ -1,4 +1,4 @@
 helpersdir = $(datadir)/$(PACKAGE)/helpers
-helpers_DATA = perl python
+helpers_DATA = perl python make-extract-targets.awk
 
 EXTRA_DIST = $(helpers_DATA)

helpers/make-extract-targets.awk

@@ -0,0 +1,93 @@
+# helper AWK script for GNU make                                    -*- awk -*-
+
+# This AWK script is used by the function `_comp_cmd_make__extract_targets` in
+# `completions/make`.  This script receives the output of `make -npq' as the
+# input file or stdin and outputs the list of targets matching the prefix.
+#
+# @env prefix         Specifies the prefix to match.
+# @env prefix_replace Specifies the string that replaces the prefix in the
+#   output.  This is used when we want to omit the directory name in showing
+#   the list of the completions.
+#
+
+BEGIN {
+  prefix = ENVIRON["prefix"];
+  prefix_replace = ENVIRON["prefix_replace"];
+  is_target_block = 0;
+  target = "";
+}
+
+function starts_with(str, prefix) {
+  return substr(str, 1, length(prefix)) == prefix;
+}
+
+# skip any makefile outputs
+NR == 1, /^# +Make data base/ { next; }
+/^# +Finished Make data base/,/^# +Make data base/ { next; }
+
+# skip until files section
+/^# +Variables/, /^# +Files/ { next; }
+
+# skip not-target blocks
+/^# +Not a target/, /^$/     { next; }
+
+# The stuff above here describes lines that are not
+#  explicit targets or not targets other than special ones
+# The stuff below here decides whether an explicit target
+#  should be output.
+
+# only process the targets the user wants.
+starts_with($0, prefix) { is_target_block = 1; }
+is_target_block == 0 { next; }
+
+/^# +File is an intermediate prerequisite/ { # cancel the block
+  is_target_block = 0;
+  target = "";
+  next;
+}
+
+# end of target block
+/^$/ {
+  is_target_block = 0;
+  if (target != "") {
+    print target;
+    target = "";
+  }
+  next;
+}
+
+# found target block
+/^[^#\t:%]+:/ {
+  # special targets
+  if (/^\.PHONY:/               ) next;
+  if (/^\.SUFFIXES:/            ) next;
+  if (/^\.DEFAULT:/             ) next;
+  if (/^\.PRECIOUS:/            ) next;
+  if (/^\.INTERMEDIATE:/        ) next;
+  if (/^\.SECONDARY:/           ) next;
+  if (/^\.SECONDEXPANSION:/     ) next;
+  if (/^\.DELETE_ON_ERROR:/     ) next;
+  if (/^\.IGNORE:/              ) next;
+  if (/^\.LOW_RESOLUTION_TIME:/ ) next;
+  if (/^\.SILENT:/              ) next;
+  if (/^\.EXPORT_ALL_VARIABLES:/) next;
+  if (/^\.NOTPARALLEL:/         ) next;
+  if (/^\.ONESHELL:/            ) next;
+  if (/^\.POSIX:/               ) next;
+  if (/^\.NOEXPORT:/            ) next;
+  if (/^\.MAKE:/                ) next;
+
+  # dont complete with hidden targets unless we are doing a partial completion
+  if (prefix == "" || prefix ~ /\/$/)
+    if (substr($0, length(prefix) + 1, 1) ~ /[^a-zA-Z0-9]/)
+      next;
+
+  target = $0;
+  sub(/:.*/, "", target);
+  if (prefix_replace != prefix)
+    target = prefix_replace substr(target, 1 + length(prefix));
+
+  next;
+}
+
+# ex: filetype=awk

test/runLint

@@ -21,7 +21,7 @@ filter_out=
 gitgrep $cmdstart"awk\b.*-F([[:space:]]|[[:space:]]*[\"'][^\"']{2,})" \
     'awk with -F char or -F ERE, use -Fchar instead (Solaris)'
 
-gitgrep $cmdstart"awk\b.*\[:[a-z]*:\]" \
+gitgrep $cmdstart"(_comp_)?awk\b.*\[:[a-z]*:\]" \
     'awk with POSIX character class not supported in mawk (Debian/Ubuntu)'
 
 gitgrep $cmdstart'sed\b.*\\[?+]' \