Remove more old utf8 code

cf

@@ -54,8 +54,7 @@ MALLOC_CHECK=""
 CPPFLAGS=""
 LDFLAGS=""
 
-#CFLAGS="-g -O0 $CFLAGS"
-CFLAGS="-g $CFLAGS"
+CFLAGS="-g -O0 $CFLAGS -Wno-undefined"
 
 MAKE=make
 IGNORE="test 0 = 1"

cf.cmake

@@ -1,5 +1,5 @@
 # Visual Studio
-#VS=1
+VS=1
 
 # Is netcdf-4 and/or DAP enabled?
 NC4=1

include/ncutf8.h

@@ -6,6 +6,8 @@
 #ifndef NCUTF8_H
 #define NCUTF8_H 1
 
+#include "ncexternl.h"
+
 /* Provide a wrapper around whatever utf8 library we use. */
 
 /*
@@ -15,7 +17,7 @@
  * NC_ENOMEM -- out of memory
  * NC_EBADNAME-- not valid utf8
  */
-extern int nc_utf8_validate(const unsigned char * name);
+EXTERNL int nc_utf8_validate(const unsigned char * name);
 
 /*
  * Apply NFC normalization to a string.
@@ -28,7 +30,19 @@ extern int nc_utf8_validate(const unsigned char * name);
  * NC_ENOMEM -- out of memory
  * NC_EBADNAME -- other failure
  */
-extern int nc_utf8_normalize(const unsigned char* str, unsigned char** normalp);
+EXTERNL int nc_utf8_normalize(const unsigned char* str, unsigned char** normalp);
 
-#endif /*NCUTF8_H*/
+/*
+ * Convert a normalized utf8 string to utf16. This is approximate
+ * because it just does the truncation version of conversion for
+ * each 32-bit codepoint to get the corresponding utf16.
+ * Return codes:
+ * NC_NOERR -- success
+ * NC_ENOMEM -- out of memory
+ * NC_EINVAL -- invalid argument or internal error
+ * NC_EBADNAME-- not valid utf16
+ */
 
+EXTERNL int nc_utf8_to_utf16(const unsigned char* s8, unsigned short** utf16p, size_t* lenp);
+
+#endif /*NCUTF8_H*/

libdispatch/dutf8.c

@@ -4,6 +4,12 @@
  */
 
 #include "config.h"
+#ifdef HAVE_STDLIB_H
+#include <stdlib.h>
+#endif
+#ifdef HAVE_STRING_H
+#include <string.h>
+#endif
 #include "netcdf.h"
 #include "ncutf8.h"
 #include "utf8proc.h"
@@ -23,13 +29,13 @@ int nc_utf8_validate(const unsigned char* name)
 {
     int ncstat = NC_NOERR;
     const nc_utf8proc_uint8_t *str;
-    nc_utf8proc_ssize_t strlen = -1;
+    nc_utf8proc_ssize_t nchars = -1;
     nc_utf8proc_int32_t codepoint;
     nc_utf8proc_ssize_t count;
 
     str = (const nc_utf8proc_uint8_t*)name;
     while(*str) {
-        count = nc_utf8proc_iterate(str,strlen,&codepoint);
+        count = nc_utf8proc_iterate(str,nchars,&codepoint);
 	if(count < 0) {
 	    switch (count) {
 	    case UTF8PROC_ERROR_NOMEM:
@@ -94,3 +100,73 @@ nc_utf8_normalize(const unsigned char* utf8, unsigned char** normalp)
 done:
     return ncstat;
 }
+
+/*
+ * Convert a normalized utf8 string to utf16. This is approximate
+ * because it just does the truncation version of conversion for
+ * each 32-bit codepoint to get the corresponding utf16.
+ * Return codes:
+ * NC_NOERR -- success
+ * NC_ENOMEM -- out of memory
+ * NC_EINVAL -- invalid argument or internal error
+ * NC_EBADNAME-- not valid utf16
+ */
+
+int nc_utf8_to_utf16(const unsigned char* s8, unsigned short** utf16p, size_t* len16p)
+{
+    int ncstat = NC_NOERR;
+    const nc_utf8proc_uint8_t *str;
+    nc_utf8proc_ssize_t nchars = -1;
+    nc_utf8proc_int32_t codepoint;
+    nc_utf8proc_ssize_t count;
+    size_t len8, len16;
+    unsigned short* utf16;
+    unsigned short* p16;
+
+    len8 = strlen((char*)s8);
+    utf16 = (unsigned short*)malloc(sizeof(unsigned short)*(len8+1));
+    if(utf16 == NULL) {
+	ncstat = NC_ENOMEM;
+	goto done;
+    }
+    str = (const nc_utf8proc_uint8_t*)s8;
+    /* Walk the string and convert each codepoint */
+    p16 = utf16;
+    len16 = 0;
+    while(*str) {
+        count = nc_utf8proc_iterate(str,nchars,&codepoint);
+	if(count < 0) {
+	    switch (count) {
+	    case UTF8PROC_ERROR_NOMEM:
+	    case UTF8PROC_ERROR_OVERFLOW:
+		ncstat = NC_ENOMEM;
+		break;
+	    case UTF8PROC_ERROR_INVALIDOPTS:
+		ncstat = NC_EINVAL;
+		break;
+	    case UTF8PROC_ERROR_INVALIDUTF8:
+	    case UTF8PROC_ERROR_NOTASSIGNED:
+	    default:
+		ncstat = NC_EBADNAME;
+		break;
+	    }
+	    goto done;
+	} else { /* move to next char */
+	    /* Complain if top 16 bits not zero */
+	    if((codepoint & 0x0000FFFF) != 0) {
+		ncstat = NC_EBADNAME;
+		goto done;
+	    }
+	    /* Truncate codepoint to 16 bits and store */
+	    *p16++ = (unsigned short)(codepoint & 0x0000FFFF);
+	    str += count;
+	    len16++;
+	}
+    }
+    *p16++ = (unsigned short)0;
+    if(utf16p) *utf16p = utf16;
+    if(len16p) *len16p = len16;
+done:
+    if(ncstat) free(utf16);
+    return ncstat;
+}

libdispatch/utf8proc.h

@@ -77,6 +77,7 @@
 /** @} */
 
 #include <stdlib.h>
+#include "ncexternl.h"
 
 #if defined(_MSC_VER) && _MSC_VER < 1800
 // MSVC prior to 2013 lacked stdbool.h and inttypes.h
@@ -120,20 +121,6 @@ typedef bool nc_utf8proc_bool;
 #endif
 #include <limits.h>
 
-#ifndef UTF8PROC_EXPORTS
-#ifdef _WIN32
-#  ifdef UTF8PROC_EXPORTS
-#    define UTF8PROC_DLLEXPORT __declspec(dllexport)
-#  else
-#    define UTF8PROC_DLLEXPORT __declspec(dllimport)
-#  endif
-#elif __GNUC__ >= 4
-#  define UTF8PROC_DLLEXPORT __attribute__ ((visibility("default")))
-#else
-#  define UTF8PROC_DLLEXPORT
-#endif
-#endif
-
 #ifdef __cplusplus
 extern "C" {
 #endif
@@ -418,7 +405,7 @@ const char *nc_utf8proc_errmsg(nc_utf8proc_ssize_t errcode);
  * In case of success, the number of bytes read is returned; otherwise, a
  * negative error code is returned.
  */
-UTF8PROC_DLLEXPORT nc_utf8proc_ssize_t nc_utf8proc_iterate(const nc_utf8proc_uint8_t *str, nc_utf8proc_ssize_t strlen, nc_utf8proc_int32_t *codepoint_ref);
+EXTERNL nc_utf8proc_ssize_t nc_utf8proc_iterate(const nc_utf8proc_uint8_t *str, nc_utf8proc_ssize_t strlen, nc_utf8proc_int32_t *codepoint_ref);
 
 /**
  * Check if a codepoint is valid (regardless of whether it has been
@@ -685,13 +672,13 @@ nc_utf8proc_ssize_t nc_utf8proc_map_custom(
  */
 /** @{ */
 /** NFD normalization (@ref UTF8PROC_DECOMPOSE). */
-UTF8PROC_DLLEXPORT nc_utf8proc_uint8_t *nc_utf8proc_NFD(const nc_utf8proc_uint8_t *str);
+EXTERNL nc_utf8proc_uint8_t *nc_utf8proc_NFD(const nc_utf8proc_uint8_t *str);
 /** NFC normalization (@ref UTF8PROC_COMPOSE). */
-UTF8PROC_DLLEXPORT nc_utf8proc_uint8_t *nc_utf8proc_NFC(const nc_utf8proc_uint8_t *str);
+EXTERNL nc_utf8proc_uint8_t *nc_utf8proc_NFC(const nc_utf8proc_uint8_t *str);
 /** NFKD normalization (@ref UTF8PROC_DECOMPOSE and @ref UTF8PROC_COMPAT). */
-UTF8PROC_DLLEXPORT nc_utf8proc_uint8_t *nc_utf8proc_NFKD(const nc_utf8proc_uint8_t *str);
+EXTERNL nc_utf8proc_uint8_t *nc_utf8proc_NFKD(const nc_utf8proc_uint8_t *str);
 /** NFKC normalization (@ref UTF8PROC_COMPOSE and @ref UTF8PROC_COMPAT). */
-UTF8PROC_DLLEXPORT nc_utf8proc_uint8_t *nc_utf8proc_NFKC(const nc_utf8proc_uint8_t *str);
+EXTERNL nc_utf8proc_uint8_t *nc_utf8proc_NFKC(const nc_utf8proc_uint8_t *str);
 /** @} */
 
 #ifdef __cplusplus

ncgen/CMakeLists.txt

@@ -3,7 +3,7 @@ IF(BUILD_SHARED_LIBS AND WIN32)
   remove_definitions(-DDLL_NETCDF)
 ENDIF()
 
-SET(ncgen_FILES generate.c main.c cdata.c bindata.c genchar.c cvt.c data.c debug.c escapes.c genc.c genbin.c generr.c genlib.c getfill.c odom.c semantics.c ncgeny.c dump.c util.c bytebuffer.c list.c genf77.c f77data.c genj.c jdata.c nc_iter.c ConvertUTF.c ncgen.h)
+SET(ncgen_FILES generate.c main.c cdata.c bindata.c genchar.c cvt.c data.c debug.c escapes.c genc.c genbin.c generr.c genlib.c getfill.c odom.c semantics.c ncgeny.c dump.c util.c bytebuffer.c list.c genf77.c f77data.c genj.c jdata.c nc_iter.c ncgen.h)
 
 # don't add the automatically determined parts of the RPATH
 # which point to directories outside the build tree to the install RPATH