Explicitly choose GNU AWK were necessary

This avoids unexpected failures with no or poor error messages.

README.md

@@ -30,6 +30,9 @@ or build `doc/bibtools.pdf` yourself with `make doc`.
 
 Setup
 -----
+GNU AWK is required.  For Ubuntu you may install the `gawk` package:
+
+    sudo apt-get install gawk
 
 Install scripts to `$HOME/bin`:
 

bin/bib-separate

@@ -49,7 +49,7 @@ do
 
     # Write records to single files.
     sed '1s/^/\n\n/;N;s/\([^\n]\)\n@/\1\n\n@/;P;D' "$bibfile" \
-        | awk -v src="$bibfile" -v useFile=$USEFILE -v override=$OVERRIDE '
+        | gawk -v src="$bibfile" -v useFile=$USEFILE -v override=$OVERRIDE '
 BEGIN {
     # Split at new lines preceding an "@" sign.
     RS="\n\n@"
@@ -60,12 +60,12 @@ BEGIN {
 {
     if ((useFile) && ($0 ~ /\<File\> += {:[^}]+:[A-Z]+}/)) {
         # If exists, obtain pdf filename from JabRef style field File.
-        match($0, "File += {:[^}]+:[A-Z]+}", a)
+        match($0, "File += {:[^}]+:[A-Z]+}", a) # Requires GNU AWK.
         sub("^File += {:", "", a[0])
         sub(":[A-Z]+}$", "", a[0])
     } else {
         # If not, derive a filename from the citation key.
-        match($0, "[A-Za-z][^{]*{[^,]*", a)
+        match($0, "[A-Za-z][^{]*{[^,]*", a) # Requires GNU AWK.
         sub("^[A-Za-z][^{]*{", "", a[0])
         if (a[0] != "") {
             cmd = sprintf("echo \"%s\" | sed -r \

bin/bib-unite

@@ -32,12 +32,12 @@ fi
 
 # Concatenate and sort all files.
 sed '1s/^/\n\n/;$s/$/\n@/;N;s/\([^\n]\)\n@/\1\n\n@/;P;D' "$@" \
-    | awk '
+    | gawk '
 BEGIN {
     RS="\n\n@"
     ORS="\n"
     FS="\n"
-    OFS="\0"
+    OFS="\0" # Requires GNU AWK.
 }
 {
     # Print one line records with fields delimited by NUL.
@@ -56,11 +56,11 @@ BEGIN {
     | sort -sn -k2,2 $RYEAR \
     | sort -sf -k1,1 \
     | sed 's/^[^\x00]*\x00\x00/\x00/' \
-    | awk '
+    | gawk '
 BEGIN {
     RS="\n"
     ORS="\n"
-    FS="\0"
+    FS="\0" # Requires GNU AWK.
     OFS="\n"
 }
 {