Unicode in progress

FossilOrigin-Name: 9ec5167660a62e37c261456bc1b5d371c403803fec309dd855aa4e128ab4cfef

5 files changed, 73 insertions, 14 deletions

diff --git a/Makefile b/Makefile
index 5f63c8e..5e3b63c 100644
--- a/Makefile
+++ b/Makefile
@@ -28,7 +28,6 @@ CFLAGS			 = -std=c89
 CFLAGS			+= -Wall
 CFLAGS			+= -Wextra
 CFLAGS			+= -Werror
-CFLAGS			+= -Wno-unicode
 CFLAGS			+= -pedantic
 
 RM				 = rm -rf
diff --git a/src/c_subst.c b/src/c_subst.c
index 116336c..cc9823c 100644
--- a/src/c_subst.c
+++ b/src/c_subst.c
@@ -209,13 +209,71 @@ c_unicode_subst(char filename[])
 	p = (unsigned char*)filename;
 	while (*p != 0x00) {
 		if (*p == 0xc3) {
-			if (u_isucharset((unsigned char)*(p + 1), "\xa0\xb6") == TRUE) {
-				printf(">>>>{%hhx}\n", *p);
-				printf(">>>>{%hhx}\n", *(p + 1));
+			if (
+				u_isucharset((unsigned char)*(p + 1), C_CHARSET_A_MAJ) == TRUE ||
+				u_isucharset((unsigned char)*(p + 1), C_CHARSET_A_MIN) == TRUE
+			) {
 				*p = 'a';
-				memmove(p + 1, p + 2, (strlen((const char*)p + 2) + 1) * sizeof(char));
+				memmove(
+					p + 1,
+					p + 2,
+					(strlen((const char*)p + 2) + 1) * sizeof(char)
+				);
 				c_unicode_subst(filename);
 			}
+			if (
+				u_isucharset((unsigned char)*(p + 1), C_CHARSET_O_MAJ) == TRUE ||
+				u_isucharset((unsigned char)*(p + 1), C_CHARSET_O_MIN) == TRUE
+			) {
+				*p = 'o';
+				memmove(
+					p + 1,
+					p + 2,
+					(strlen((const char*)p + 2) + 1) * sizeof(char)
+				);
+				c_unicode_subst(filename);
+			}
+			if (u_isucharset((unsigned char)*(p + 1), C_CHARSET_AE) == TRUE) {
+				*p = 'a';
+				*(p + 1) = 'e';
+				c_unicode_subst(filename);
+			}
+			if (u_isucharset((unsigned char)*(p + 1), C_CHARSET_N) == TRUE) {
+				*p = 'n';
+				memmove(
+					p + 1,
+					p + 2,
+					(strlen((const char*)p + 2) + 1) * sizeof(char)
+				);
+				c_unicode_subst(filename);
+			}		
+			if (*(p + 1) == 0x72) {
+				*p = '2';
+				memmove(
+					p + 1,
+					p + 2,
+					(strlen((const char*)p + 2) + 1) * sizeof(char)
+				);
+				c_unicode_subst(filename);
+			}		
+			if (*(p + 1) == 0x73) {
+				*p = '3';
+				memmove(
+					p + 1,
+					p + 2,
+					(strlen((const char*)p + 2) + 1) * sizeof(char)
+				);
+				c_unicode_subst(filename);
+			}		
+			if (*(p + 1) == 0x75) {
+				*p = 'u';
+				memmove(
+					p + 1,
+					p + 2,
+					(strlen((const char*)p + 2) + 1) * sizeof(char)
+				);
+				c_unicode_subst(filename);
+			}		
 		}
 		p++;
 	}
diff --git a/src/c_subst.h b/src/c_subst.h
index fb4f16a..1a64621 100644
--- a/src/c_subst.h
+++ b/src/c_subst.h
@@ -50,11 +50,13 @@
 
 #include "c_unixize.h"
 
-#define C_CHARSET_VALID "_-."
-#define C_CHARSET_A_MAJ "\u00c0\u00c1\u00c2\u00c3\u00c4\u00c5"
-#define C_CHARSET_A_MIN "\u00e0\u00e1\u00e2\u00e3\u00e4\u00e5"
-#define C_CHARSET_O_MAJ "\u00d2\u00d3\u00d4\u00d5\u00d6\u00d8"
-#define C_CHARSET_O_MIN "\u00f2\u00f3\u00f4\u00f5\u00f6\u00f8"
+#define C_CHARSET_VALID	"_-."
+#define C_CHARSET_A_MAJ	"\x80\x81\x82\x83\x84\x85"
+#define C_CHARSET_A_MIN	"\xa0\xa1\xa2\xa3\xa4\xa5"
+#define C_CHARSET_O_MAJ	"\x92\x93\x94\x95\x96\x98"
+#define C_CHARSET_O_MIN	"\xb2\xb3\xb4\xb5\xb6\xb8"
+#define C_CHARSET_AE	"\x86\xa6"
+#define C_CHARSET_N		"\x91\xb1"
 
 struct lfiles_s* c_subst_filenames(struct lfiles_s*, const bool_t,
 	const bool_t, const unsigned char);
diff --git a/src/u_utils.c b/src/u_utils.c
index 88f768c..a2ad0a4 100644
--- a/src/u_utils.c
+++ b/src/u_utils.c
@@ -260,11 +260,11 @@ u_ischarset
 bool_t
 u_isucharset
 (const unsigned char	c,
- const unsigned char	cs[])
+ const char				cs[])
 {
-	char *p;
+	unsigned char *p;
 
-	p = (char*)cs;
+	p = (unsigned char*)cs;
 	while (*p != 0x00 && *p != c) {
 		p++;
 	}
diff --git a/src/u_utils.h b/src/u_utils.h
index 9c04055..c966cf2 100644
--- a/src/u_utils.h
+++ b/src/u_utils.h
@@ -55,4 +55,4 @@ char**	u_get_nargv(struct opts_s*);
 void	u_increase_subpath(char[], const char[]);
 void	u_decrease_subpath(char[]);
 bool_t	u_ischarset(const int, const char[]);
-bool_t	u_isucharset(const unsigned char, const unsigned char[]);
+bool_t	u_isucharset(const unsigned char, const char[]);