Make txt2enc smarter, so we don't have to add DELETE any more in all original tables, better keep them as-is.

19 files changed, 13 insertions, 20 deletions

diff --git a/tools/encoding/Makefile b/tools/encoding/Makefile
index a2122d5..ff19492 100644
--- a/tools/encoding/Makefile
+++ b/tools/encoding/Makefile
@@ -67,7 +67,7 @@ encodings: clean txt2enc $(EUC_ENCODINGS)
 	@for p in *.txt; do \
 	    enc=`echo $$p | sed 's/\..*$$/\.enc/'`; \
 	    echo $$enc; \
-	    ./txt2enc $$p > $$enc; \
+	    ./txt2enc -e 0 -u 1 $$p > $$enc; \
 	done
 	@echo
 	@echo Compiling special versions of encoding files.
diff --git a/tools/encoding/ascii.txt b/tools/encoding/ascii.txt
index 2afbaab..66ba6f3 100644
--- a/tools/encoding/ascii.txt
+++ b/tools/encoding/ascii.txt
@@ -93,4 +93,3 @@
 0x7C	0x007C	#	VERTICAL LINE
 0x7D	0x007D	#	RIGHT CURLY BRACKET
 0x7E	0x007E	#	TILDE
-0x7F	0x007F	#	DELETE
\ No newline at end of file
diff --git a/tools/encoding/big5.txt b/tools/encoding/big5.txt
index 06b0fac..58cdfe2 100644
--- a/tools/encoding/big5.txt
+++ b/tools/encoding/big5.txt
@@ -185,7 +185,6 @@
 0x7C	0x007C	# VERTICAL LINE
 0x7D	0x007D	# RIGHT CURLY BRACKET
 0x7E	0x007E	# TILDE
-0x7F	0x007F	# DELETE
 0xA140	0x3000	# IDEOGRAPHIC SPACE
 0xA141	0xFF0C	# FULLWIDTH COMMA
 0xA142	0x3001	# IDEOGRAPHIC COMMA
diff --git a/tools/encoding/dingbats.txt b/tools/encoding/dingbats.txt
index 93a6081..334f8d6 100644
--- a/tools/encoding/dingbats.txt
+++ b/tools/encoding/dingbats.txt
@@ -155,7 +155,6 @@
 0x7C	0x275C	# HEAVY SINGLE COMMA QUOTATION MARK ORNAMENT
 0x7D	0x275D	# HEAVY DOUBLE TURNED COMMA QUOTATION MARK ORNAMENT
 0x7E	0x275E	# HEAVY DOUBLE COMMA QUOTATION MARK ORNAMENT
-0x7F	0x275E	# DELETE
 0xA1	0x2761	# CURVED STEM PARAGRAPH SIGN ORNAMENT
 0xA2	0x2762	# HEAVY EXCLAMATION MARK ORNAMENT
 0xA3	0x2763	# HEAVY HEART EXCLAMATION MARK ORNAMENT
diff --git a/tools/encoding/gb1988.txt b/tools/encoding/gb1988.txt
index b9197e5..800cd68 100644
--- a/tools/encoding/gb1988.txt
+++ b/tools/encoding/gb1988.txt
@@ -93,7 +93,6 @@
 0x7C	0x007C	# VERTICAL LINE
 0x7D	0x007D	# RIGHT CURLY BRACKET
 0x7E	0x203E	# OVERLINE
-0x7F	0x007F	# DELETE
 0xA1	0xFF61	# HALFWIDTH IDEOGRAPHIC FULL STOP
 0xA2	0xFF62	# HALFWIDTH LEFT CORNER BRACKET
 0xA3	0xFF63	# HALFWIDTH RIGHT CORNER BRACKET
diff --git a/tools/encoding/macCentEuro.txt b/tools/encoding/macCentEuro.txt
index aa92908..bf424c1 100644
--- a/tools/encoding/macCentEuro.txt
+++ b/tools/encoding/macCentEuro.txt
@@ -188,7 +188,6 @@
 0x7C	0x007C	# VERTICAL LINE
 0x7D	0x007D	# RIGHT CURLY BRACKET
 0x7E	0x007E	# TILDE
-0x7F	0x007F	# DELETE
 #
 0x80	0x00C4	# LATIN CAPITAL LETTER A WITH DIAERESIS
 0x81	0x0100	# LATIN CAPITAL LETTER A WITH MACRON
diff --git a/tools/encoding/macCroatian.txt b/tools/encoding/macCroatian.txt
index 2eef246..538eda3 100644
--- a/tools/encoding/macCroatian.txt
+++ b/tools/encoding/macCroatian.txt
@@ -216,7 +216,6 @@
 0x7C	0x007C	# VERTICAL LINE
 0x7D	0x007D	# RIGHT CURLY BRACKET
 0x7E	0x007E	# TILDE
-0x7F	0x007F	# DELETE
 #
 0x80	0x00C4	# LATIN CAPITAL LETTER A WITH DIAERESIS
 0x81	0x00C5	# LATIN CAPITAL LETTER A WITH RING ABOVE
diff --git a/tools/encoding/macCyrillic.txt b/tools/encoding/macCyrillic.txt
index 2e9f8e2..695dade 100644
--- a/tools/encoding/macCyrillic.txt
+++ b/tools/encoding/macCyrillic.txt
@@ -213,7 +213,6 @@
 0x7C	0x007C	# VERTICAL LINE
 0x7D	0x007D	# RIGHT CURLY BRACKET
 0x7E	0x007E	# TILDE
-0x7F	0x007F	# DELETE
 #
 0x80	0x0410	# CYRILLIC CAPITAL LETTER A
 0x81	0x0411	# CYRILLIC CAPITAL LETTER BE
diff --git a/tools/encoding/macDingbats.txt b/tools/encoding/macDingbats.txt
index 4b815f4..273d526 100644
--- a/tools/encoding/macDingbats.txt
+++ b/tools/encoding/macDingbats.txt
@@ -151,7 +151,6 @@
 0x7C	0x275C	# HEAVY SINGLE COMMA QUOTATION MARK ORNAMENT
 0x7D	0x275D	# HEAVY DOUBLE TURNED COMMA QUOTATION MARK ORNAMENT
 0x7E	0x275E	# HEAVY DOUBLE COMMA QUOTATION MARK ORNAMENT
-0x7F	0x007F	# DELETE
 0x80	0xF8D7	# medium left parenthesis ornament
 0x81	0xF8D8	# medium right parenthesis ornament
 0x82	0xF8D9	# medium flattened left parenthesis ornament
diff --git a/tools/encoding/macGreek.txt b/tools/encoding/macGreek.txt
index b960d68..9783259 100644
--- a/tools/encoding/macGreek.txt
+++ b/tools/encoding/macGreek.txt
@@ -207,7 +207,6 @@
 0x7C	0x007C	# VERTICAL LINE
 0x7D	0x007D	# RIGHT CURLY BRACKET
 0x7E	0x007E	# TILDE
-0x7F	0x007F	# DELETE
 #
 0x80	0x00C4	# LATIN CAPITAL LETTER A WITH DIAERESIS
 0x81	0x00B9	# SUPERSCRIPT ONE
diff --git a/tools/encoding/macIceland.txt b/tools/encoding/macIceland.txt
index c60b8d2..0a0b27b 100644
--- a/tools/encoding/macIceland.txt
+++ b/tools/encoding/macIceland.txt
@@ -234,7 +234,6 @@
 0x7C	0x007C	# VERTICAL LINE
 0x7D	0x007D	# RIGHT CURLY BRACKET
 0x7E	0x007E	# TILDE
-0x7F	0x007F	# DELETE
 #
 0x80	0x00C4	# LATIN CAPITAL LETTER A WITH DIAERESIS
 0x81	0x00C5	# LATIN CAPITAL LETTER A WITH RING ABOVE
diff --git a/tools/encoding/macJapan.txt b/tools/encoding/macJapan.txt
index 3c48c4a..7121b3b 100644
--- a/tools/encoding/macJapan.txt
+++ b/tools/encoding/macJapan.txt
@@ -318,7 +318,6 @@
 0x7C	0x007C	# VERTICAL LINE
 0x7D	0x007D	# RIGHT CURLY BRACKET
 0x7E	0x007E	# TILDE # Apple change
-0x7F	0x007F	# DELETE
 #
 0x8140	0x3000	# IDEOGRAPHIC SPACE
 0x8141	0x3001	# IDEOGRAPHIC COMMA
diff --git a/tools/encoding/macRoman.txt b/tools/encoding/macRoman.txt
index 43ad44b..7ddcf8d 100644
--- a/tools/encoding/macRoman.txt
+++ b/tools/encoding/macRoman.txt
@@ -233,7 +233,6 @@
 0x7C	0x007C	# VERTICAL LINE
 0x7D	0x007D	# RIGHT CURLY BRACKET
 0x7E	0x007E	# TILDE
-0x7F	0x007F	# DELETE
 #
 0x80	0x00C4	# LATIN CAPITAL LETTER A WITH DIAERESIS
 0x81	0x00C5	# LATIN CAPITAL LETTER A WITH RING ABOVE
diff --git a/tools/encoding/macRomania.txt b/tools/encoding/macRomania.txt
index 36a0b68..2a84adc 100644
--- a/tools/encoding/macRomania.txt
+++ b/tools/encoding/macRomania.txt
@@ -154,7 +154,6 @@
 0x7C	0x007C	# VERTICAL LINE
 0x7D	0x007D	# RIGHT CURLY BRACKET
 0x7E	0x007E	# TILDE
-0x7F	0x007F	# DELETE
 #
 0x80	0x00C4	# LATIN CAPITAL LETTER A WITH DIAERESIS
 0x81	0x00C5	# LATIN CAPITAL LETTER A WITH RING ABOVE
diff --git a/tools/encoding/macThai.txt b/tools/encoding/macThai.txt
index 2043621..b991833 100644
--- a/tools/encoding/macThai.txt
+++ b/tools/encoding/macThai.txt
@@ -168,7 +168,6 @@
 0x7C	0x007C	# VERTICAL LINE
 0x7D	0x007D	# RIGHT CURLY BRACKET
 0x7E	0x007E	# TILDE
-0x7F	0x007F	# DELETE
 #
 0x80	0x00AB	# LEFT-POINTING DOUBLE ANGLE QUOTATION MARK
 0x81	0x00BB	# RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK
diff --git a/tools/encoding/macTurkish.txt b/tools/encoding/macTurkish.txt
index ca3cda3..4a1ddab 100644
--- a/tools/encoding/macTurkish.txt
+++ b/tools/encoding/macTurkish.txt
@@ -203,7 +203,6 @@
 0x7C	0x007C	# VERTICAL LINE
 0x7D	0x007D	# RIGHT CURLY BRACKET
 0x7E	0x007E	# TILDE
-0x7F	0x007F	# DELETE
 #
 0x80	0x00C4	# LATIN CAPITAL LETTER A WITH DIAERESIS
 0x81	0x00C5	# LATIN CAPITAL LETTER A WITH RING ABOVE
diff --git a/tools/encoding/macUkraine.txt b/tools/encoding/macUkraine.txt
index dc07cdc..dba4e10 100644
--- a/tools/encoding/macUkraine.txt
+++ b/tools/encoding/macUkraine.txt
@@ -148,7 +148,6 @@
 0x7C	0x007C	# VERTICAL LINE
 0x7D	0x007D	# RIGHT CURLY BRACKET
 0x7E	0x007E	# TILDE
-0x7F	0x007F	# DELETE
 #
 0x80	0x0410	# CYRILLIC CAPITAL LETTER A
 0x81	0x0411	# CYRILLIC CAPITAL LETTER BE
diff --git a/tools/encoding/symbol.txt b/tools/encoding/symbol.txt
index 13a3ed8..12dcae6 100644
--- a/tools/encoding/symbol.txt
+++ b/tools/encoding/symbol.txt
@@ -169,7 +169,6 @@
 0x7C	0x007C	# VERTICAL LINE
 0x7D	0x007D	# RIGHT CURLY BRACKET
 0x7E	0x223C	# TILDE OPERATOR
-0x7F	0x007F	# DELETE
 0xA1	0x03D2	# GREEK UPSILON WITH HOOK SYMBOL
 0xA2	0x2032	# PRIME
 0xA3	0x2264	# LESS-THAN OR EQUAL TO
diff --git a/tools/encoding/txt2enc.c b/tools/encoding/txt2enc.c
index 80b44b9..32c7344 100644
--- a/tools/encoding/txt2enc.c
+++ b/tools/encoding/txt2enc.c
@@ -26,7 +26,7 @@ main(int argc, char **argv)
 {
     FILE *fp;
     Rune *toUnicode[256];
-    int i, multiByte, enc, uni, hi, lo, used, maxEnc;
+    int i, multiByte, enc, uni, hi, lo, fixmissing, used, maxEnc;
     int ch, encColumn, uniColumn, fallbackKnown, width;
     char *fallbackString, *str, *rest, *dot;
     unsigned int magic, type, symbol, fallbackChar;
@@ -43,6 +43,7 @@ main(int argc, char **argv)
     fallbackKnown = 0;
     type = -1;
     symbol = 0;
+    fixmissing = 1;
 
     opterr = 0;
     while (1) {
@@ -88,6 +89,10 @@ main(int argc, char **argv)
 	    symbol = 1;
 	    break;
 
+	case 'm':
+	    fixmissing = 0;
+	    break;
+
 	default:
 	    goto usage;
 	}
@@ -101,7 +106,7 @@ main(int argc, char **argv)
 	fputs("    -f\tfallback character (default: QUESTION MARK)\n", stderr);
 	fputs("    -t\toverride implicit type with single, double, or multi\n", stderr);
 	fputs("    -s\tsymbol+ascii encoding\n", stderr);
-	fputs("    -m\tdon't implicitly include range 0080 to 00FF\n", stderr);
+	fputs("    -m\tdon't implicitly include 007F\n", stderr);
 	return 1;
     }
 
@@ -202,6 +207,11 @@ main(int argc, char **argv)
 	for (i = 0; i < 0x20; i++) {
 	    toUnicode[0][i] = i;
 	}
+	if (fixmissing) {
+	    if (toUnicode[0x7F] == NULL && toUnicode[0][0x7F] == 0) {
+		toUnicode[0][0x7F] = 0x7F;
+	    }
+	}
     }
 
     printf("# Encoding file: %s, %s-byte\n", argv[argc - 1], typeString[type]);