From 17f83c6f122db849a8877ef0ca641d4743f352cd Mon Sep 17 00:00:00 2001
From: Brendan Jurd <direvus@gmail.com>
Date: Sun, 15 Aug 2021 21:30:57 +1000
Subject: Add support for Degree Sign on input (#2791)

This commit adds support in dsmtor() for a Degree Sign (U+00B0), encoded
as UTF-8 (`\xc2\xb0`) or as a single byte (`\xb0`) (in ISO 8859 parts 1-4, 7-10,
13, 15, 16, and several Windows code pages), as an alternative symbol to
`D`/`d` to designate the degree unit.

Fixes #2712.
---
 docs/source/apps/cs2cs.rst   |  2 +-
 src/dmstor.cpp               | 63 +++++++++++++++++++++++++++++++-------------
 test/unit/gie_self_tests.cpp |  6 +++++
 3 files changed, 51 insertions(+), 20 deletions(-)

diff --git a/docs/source/apps/cs2cs.rst b/docs/source/apps/cs2cs.rst
index 909912df..abddae9e 100644
--- a/docs/source/apps/cs2cs.rst
+++ b/docs/source/apps/cs2cs.rst
@@ -261,7 +261,7 @@ The following script
 ::
 
     cs2cs +proj=latlong +datum=NAD83 +to +proj=utm +zone=10 +datum=NAD27 -r <<EOF
-    45d15'33.1" 111.5W
+    45°15'33.1" 111.5W
     45d15.551666667N -111d30
     +45.25919444444 111d30'000w
     EOF
diff --git a/src/dmstor.cpp b/src/dmstor.cpp
index e65d4884..eeff1841 100644
--- a/src/dmstor.cpp
+++ b/src/dmstor.cpp
@@ -20,6 +20,12 @@ vm[] = {
 	.0002908882086657216,
 	.0000048481368110953599
 };
+/* byte sequence for Degree Sign U+00B0 in UTF-8. */
+	static constexpr char
+DEG_SIGN1 = '\xc2';
+	static constexpr char
+DEG_SIGN2 = '\xb0';
+
 	double
 dmstor(const char *is, char **rs) {
 	return dmstor_ctx( pj_get_default_ctx(), is, rs );
@@ -29,57 +35,76 @@ dmstor(const char *is, char **rs) {
 dmstor_ctx(PJ_CONTEXT *ctx, const char *is, char **rs) {
 	int n, nl;
 	char *s, work[MAX_WORK];
-        const char* p;
+	const char* p;
 	double v, tv;
 
 	if (rs)
 		*rs = (char *)is;
-	/* copy sting into work space */
+	/* copy string into work space */
 	while (isspace(*is)) ++is;
 	n = MAX_WORK;
 	s = work;
 	p = (char *)is;
-	while (isgraph(*p) && --n)
+
+	/*
+	 * Copy characters into work until we hit a non-printable character or run
+	 * out of space in the buffer.  Make a special exception for the bytes of
+	 * the Degree Sign in UTF-8.
+	 *
+	 * It is possible that a really odd input (like lots of leading zeros)
+	 * could be truncated in copying into work.  But ...
+	 */
+	while ((isgraph(*p) || *p == DEG_SIGN1 || *p == DEG_SIGN2) && --n)
 		*s++ = *p++;
 	*s = '\0';
-	/* it is possible that a really odd input (like lots of leading
-		zeros) could be truncated in copying into work.  But ... */
 	int sign = *(s = work);
 	if (sign == '+' || sign == '-') s++;
 	else sign = '+';
 	v = 0.;
-	for (nl = 0 ; nl < 3 ; nl = n + 1 ) {
+	for (nl = 0 ; nl < 3 ; nl = n + 1) {
 		if (!(isdigit(*s) || *s == '.')) break;
 		if ((tv = proj_strtod(s, &s)) == HUGE_VAL)
 			return tv;
-		switch (*s) {
-		case 'D': case 'd':
-			n = 0; break;
-		case '\'':
-			n = 1; break;
-		case '"':
-			n = 2; break;
-		case 'r': case 'R':
+		int adv = 1;
+
+		if (*s == 'D' || *s == 'd' || *s == DEG_SIGN2) {
+			/*
+			 * Accept \xb0 as a single-byte degree symbol. This byte is the
+			 * degree symbol in various single-byte encodings: multiple ISO
+			 * 8859 parts, several Windows code pages and others.
+			 */
+			n = 0;
+		} else if (*s == '\'') {
+			n = 1;
+		} else if (*s == '"') {
+			n = 2;
+		} else if (s[0] == DEG_SIGN1 && s[1] == DEG_SIGN2) {
+			/* degree symbol in UTF-8 */
+			n = 0;
+			adv = 2;
+		} else if (*s == 'r' || *s == 'R') {
 			if (nl) {
 				proj_context_errno_set( ctx, PROJ_ERR_INVALID_OP_ILLEGAL_ARG_VALUE );
 				return HUGE_VAL;
 			}
 			++s;
 			v = tv;
-			goto skip;
-		default:
+			n = 4;
+			continue;
+		} else {
 			v += tv * vm[nl];
-		skip:	n = 4;
+			n = 4;
 			continue;
 		}
+
 		if (n < nl) {
 			proj_context_errno_set( ctx, PROJ_ERR_INVALID_OP_ILLEGAL_ARG_VALUE );
 			return HUGE_VAL;
 		}
 		v += tv * vm[n];
-		++s;
+		s += adv;
 	}
-		/* postfix sign */
+	/* postfix sign */
 	if (*s && (p = strchr(sym, *s))) {
 		sign = (p - sym) >= 4 ? '-' : '+';
 		++s;
diff --git a/test/unit/gie_self_tests.cpp b/test/unit/gie_self_tests.cpp
index c7c1ddf5..f0d4ea86 100644
--- a/test/unit/gie_self_tests.cpp
+++ b/test/unit/gie_self_tests.cpp
@@ -442,6 +442,12 @@ TEST(gie, info_functions) {
     /* we can't expect perfect numerical accuracy so testing with a tolerance */
     ASSERT_NEAR(-2.0, proj_dmstor(&buf[0], NULL), 1e-7);
 
+    /* test UTF-8 degree sign on DMS input */
+    ASSERT_NEAR(0.34512432, proj_dmstor("19°46'27\"E", NULL), 1e-7);
+
+    /* test ISO 8859-1, cp1252, et al. degree sign on DMS input */
+    ASSERT_NEAR(0.34512432, proj_dmstor("19" "\260" "46'27\"E", NULL), 1e-7);
+
     /* test proj_derivatives_retrieve() and proj_factors_retrieve() */
     P = proj_create(PJ_DEFAULT_CTX, "+proj=merc +ellps=WGS84");
     a = proj_coord(0, 0, 0, 0);
-- 
cgit v1.2.3