FEAT: enhex/uri and dehex/uri for a special short variant of space encoding

Author: Oldes

src/boot/natives.reb

@@ -511,7 +511,7 @@ dehex: native [
 	{Converts URL-style hex encoded (%xx) strings. If input is UTF-8 encode, you should first convert it to binary!}
 	value [any-string! binary!] {The string to dehex}
 	/escape char [char!] {Can be used to change the default escape char #"%"}
-	/url {Decode + as a space}
+	/uri {Decode space from a special char (#"+" by default or #"_" when escape char is #"=")}
 ]
 
 enhex: native [
@@ -521,6 +521,7 @@ enhex: native [
 	 char [char!] 
 	/except {Can be used to specify, which chars can be left unescaped}
 	 unescaped [bitset!] {By default it is URI bitset when value is file or url, else URI-Component}
+	/uri {Encode space using a special char (#"+" by default or #"_" when escape char is #"=")}
 ]
 
 get: native [

src/core/n-strings.c

@@ -613,12 +613,13 @@ static struct digest {
 	REBVAL *arg        = D_ARG(1);
 //	REBOOL  ref_escape = D_REF(2);
 //	REBVAL *val_escape = D_ARG(3);
-	REBOOL as_url      = D_REF(4);
+	REBOOL as_uri      = D_REF(4);
 	REBINT len = (REBINT)VAL_LEN(arg); // due to len -= 2 below
 	REBUNI n;
 	REBSER *ser;
 
 	const REBCHR escape_char = D_REF(2) ? VAL_CHAR(D_ARG(3)) : '%';
+	const REBCHR space_char = escape_char == '=' ? '_' : '+';
 
 	if (VAL_BYTE_SIZE(arg)) {
 		REBYTE *bp = VAL_BIN_DATA(arg);
@@ -629,7 +630,12 @@ static struct digest {
 				*dp++ = (REBYTE)n;
 				bp += 3;
 				len -= 2;
-			} else {
+			}
+			else if (*bp == space_char && as_uri) {
+				*dp++ = ' ';
+				bp++;
+			}
+			else {
 				*dp++ = *bp++;
 			}
 		}
@@ -680,10 +686,12 @@ static struct digest {
 //	REBVAL *val_escape = D_ARG(3);
 	REBOOL  ref_bitset = D_REF(4);
 	REBVAL *val_bitset = D_ARG(5);
+	REBOOL  no_space   = D_REF(6);
 	REBYTE encoded[4];
 	REBCNT n, encoded_size;
 	REBSER *ser;
 	const REBCHR escape_char = D_REF(2) ? VAL_CHAR(D_ARG(3)) : '%';
+	const REBCHR space_char = escape_char == '=' ? '_' : '+';
 
 	if (!ref_bitset) {
 		// use bitset value from system/catalog/bitsets
@@ -710,10 +718,18 @@ static struct digest {
 		while (bp < ep) {
 			REBYTE c = bp[0];
 			bp++;
+			if (no_space) {
+				if (c == ' ') {
+					*dp++ = space_char;
+					continue;
+				}
+				if (c == space_char) goto escaped;
+			}
 			if (Check_Bit_Cased(VAL_SERIES(val_bitset), c)) {
 				*dp++ = c;
 				continue;
 			}
+		escaped:
 			*dp++ = escape_char;
 			*dp++ = Hex_Digits[(c & 0xf0) >> 4];
 			*dp++ = Hex_Digits[ c & 0xf];

src/tests/units/series-test.r3

@@ -1720,6 +1720,18 @@ Rebol [
 	--assert "12%20%61%62" = enhex/except "12 ab" charset "12"
 	--assert "12 %61%62"   = enhex/except "12 ab" charset "12 "
 
+--test-- "ENHEX/uri"
+	--assert "a%20b%2B" = enhex "a b+"
+	--assert "a+b%2B" = enhex/uri "a b+"
+	; quoted-printable:
+	--assert "a=20b_" = enhex/escape "a b_" #"="
+	--assert "a_b=5F" = enhex/uri/escape "a b_" #"=" 
+--test-- "DEHEX/uri"
+	--assert "a+b+"   = dehex "a+b%2B"
+	--assert "a b+"   = dehex/uri "a+b%2B"
+	; quoted-printable:
+	--assert "a_b_"   = dehex/escape"a_b=5F" #"="
+	--assert "a b_"   = dehex/uri/escape"a_b=5F" #"="
 
 ===end-group===