Merge pull request #1794 from bombipappoo/filename-encoding

Convert filename from ANSI to UTF-8 before calling HDF5.

Author: WardF

include/hdf5internal.h

@@ -197,4 +197,23 @@ extern int NC4_isnetcdf4(struct NC_FILE_INFO*); /*libsrc4/nc4hdf.c*/
 
 extern int nc4_find_default_chunksizes2(NC_GRP_INFO_T *grp, NC_VAR_INFO_T *var);
 
+#ifdef _WIN32
+
+/* Maxinum length of a typical path in UTF-8.
+ * When converting from ANSI to UTF-8, the length will be up to 3 times,
+ * so round up 260*3 to 1024. (260=MAX_PATH) */
+#define MAX_PATHBUF_SIZE 1024
+
+/* Struct for converting ANSI to UTF-8. */
+typedef struct pathbuf
+{
+    void *ptr;
+    char buffer[MAX_PATHBUF_SIZE];
+} pathbuf_t;
+
+const char *nc4_ndf5_ansi_to_utf8(pathbuf_t *pb, const char *path);
+void nc4_hdf5_free_pathbuf(pathbuf_t *pb);
+
+#endif /* _WIN32 */
+
 #endif /* _HDF5INTERNAL_ */

libhdf5/hdf5create.c

@@ -22,6 +22,12 @@ static const int ILLEGAL_CREATE_FLAGS = (NC_NOWRITE|NC_MMAP|NC_64BIT_OFFSET|NC_C
 /* From nc4mem.c */
 extern int NC4_create_image_file(NC_FILE_INFO_T* h5, size_t);
 
+#ifdef _WIN32
+static hid_t nc4_H5Fcreate(const char *filename, unsigned flags, hid_t fcpl_id, hid_t fapl_id);
+#else
+#define nc4_H5Fcreate  H5Fcreate
+#endif
+
 /**
  * @internal Create a netCDF-4/HDF5 file.
  *
@@ -217,13 +223,13 @@ nc4_create_file(const char *path, int cmode, size_t initialsz,
             /* Configure FAPL to use the core file driver */
             if (H5Pset_fapl_core(fapl_id, alloc_incr, (nc4_info->mem.persist?1:0)) < 0)
                 BAIL(NC_EHDFERR);
-            if ((hdf5_info->hdfid = H5Fcreate(path, flags, fcpl_id, fapl_id)) < 0)
+            if ((hdf5_info->hdfid = nc4_H5Fcreate(path, flags, fcpl_id, fapl_id)) < 0)
                 BAIL(EACCES);
         }
         else /* Normal file */
         {
             /* Create the HDF5 file. */
-            if ((hdf5_info->hdfid = H5Fcreate(path, flags, fcpl_id, fapl_id)) < 0)
+            if ((hdf5_info->hdfid = nc4_H5Fcreate(path, flags, fcpl_id, fapl_id)) < 0)
                 BAIL(EACCES);
         }
 
@@ -306,3 +312,31 @@ NC4_create(const char* path, int cmode, size_t initialsz, int basepe,
 
     return res;
 }
+
+#ifdef _WIN32
+
+/**
+ * Wrapper function for H5Fcreate.
+ * Converts the filename from ANSI to UTF-8 as needed before calling H5Fcreate.
+ *
+ * @param filename The filename encoded ANSI to access.
+ * @param flags File access flags.
+ * @param fcpl_id File creation property list identifier.
+ * @param fapl_id File access property list identifier.
+ * @return A file identifier if succeeded. A negative value if failed.
+ */
+static hid_t
+nc4_H5Fcreate(const char *filename, unsigned flags, hid_t fcpl_id, hid_t fapl_id)
+{
+    pathbuf_t pb;
+    hid_t hid;
+
+    filename = nc4_ndf5_ansi_to_utf8(&pb, filename);
+    if (!filename)
+        return H5I_INVALID_HID;
+    hid = H5Fcreate(filename, flags, fcpl_id, fapl_id);
+    nc4_hdf5_free_pathbuf(&pb);
+    return hid;
+}
+
+#endif /* _WIN32 */

libhdf5/hdf5internal.c

@@ -17,6 +17,9 @@
 #include "config.h"
 #include "hdf5internal.h"
 #include "ncfilter.h"
+#ifdef _WIN32
+#include <windows.h>
+#endif
 
 #undef DEBUGH5
 
@@ -1027,3 +1030,91 @@ hdf5_set_log_level()
     return NC_NOERR;
 }
 #endif /* LOGGING */
+
+
+#ifdef _WIN32
+
+/**
+ * Converts the filename from ANSI to UTF-8 if HDF5 >= 1.10.6. 
+ * nc4_hdf5_free_pathbuf must be called to free pb.
+ *
+ * @param pb Pointer that conversion information is stored.
+ * @param path The filename to be converted.
+ *
+ * @return The converted filename if succeeded. NULL if failed.
+ */
+const char *
+nc4_ndf5_ansi_to_utf8(pathbuf_t *pb, const char *path)
+{
+    const uint UTF8_MAJNUM = 1;
+    const uint UTF8_MINNUM = 10;
+    const uint UTF8_RELNUM = 6;
+    static enum {UNDEF, ANSI, UTF8} hdf5_encoding = UNDEF;
+    wchar_t wbuf[MAX_PATH];
+    wchar_t *ws = NULL;
+    char *ns = NULL;
+    int n;
+
+    if (hdf5_encoding == UNDEF) {
+        uint majnum, minnum, relnum;
+        H5get_libversion(&majnum, &minnum, &relnum);
+        hdf5_encoding = (((majnum == UTF8_MAJNUM && minnum == UTF8_MINNUM && relnum >= UTF8_RELNUM)
+                          || (majnum == UTF8_MAJNUM && minnum > UTF8_MINNUM)
+                          || majnum > UTF8_MAJNUM)
+                         ? UTF8 : ANSI);
+    }
+    if (hdf5_encoding == ANSI) {
+        pb->ptr = NULL;
+        return path;
+    }
+
+    n = MultiByteToWideChar(CP_ACP, 0, path, -1, NULL, 0);
+    if (!n) {
+        errno = EILSEQ;
+        goto done;
+    }
+    ws = n <= _countof(wbuf) ? wbuf : malloc(sizeof *ws * n);
+    if (!ws)
+        goto done;
+    if (!MultiByteToWideChar(CP_ACP, 0, path, -1, ws, n)) {
+        errno = EILSEQ;
+        goto done;
+    }
+
+    n = WideCharToMultiByte(CP_UTF8, 0, ws, -1, NULL, 0, NULL, NULL);
+    if (!n) {
+        errno = EILSEQ;
+        goto done;
+    }
+    ns = n <= sizeof pb->buffer ? pb->buffer : malloc(n);
+    if (!ns)
+        goto done;
+    if (!WideCharToMultiByte(CP_UTF8, 0, ws, -1, ns, n, NULL, NULL)) {
+        if (ns != pb->buffer)
+            free(ns);
+        ns = NULL;
+        errno = EILSEQ;
+        goto done;
+    }
+
+done:
+    if (ws != wbuf)
+        free (ws);
+
+    pb->ptr = ns;
+    return ns;
+}
+
+/**
+ * Free the conversion information used by nc4_ndf5_ansi_to_utf8.
+ *
+ * @param pb Pointer that hold conversion information to be freed.
+ */
+void
+nc4_hdf5_free_pathbuf(pathbuf_t *pb)
+{
+    if (pb->ptr && pb->ptr != pb->buffer)
+        free(pb->ptr);
+}
+
+#endif /* _WIN32 */

libhdf5/hdf5open.c

@@ -62,6 +62,12 @@ extern int NC4_open_image_file(NC_FILE_INFO_T* h5);
 /* Defined later in this file. */
 static int rec_read_metadata(NC_GRP_INFO_T *grp);
 
+#ifdef _WIN32
+static hid_t nc4_H5Fopen(const char *filename, unsigned flags, hid_t fapl_id);
+#else
+#define nc4_H5Fopen  H5Fopen
+#endif
+
 /**
  * @internal Struct to track HDF5 object info, for
  * rec_read_metadata(). We get this info for every object in the
@@ -833,7 +839,7 @@ nc4_open_file(const char *path, int mode, void* parameters, int ncid)
             if (H5Pset_fapl_core(fapl_id, min_incr, (nc4_info->mem.persist?1:0)) < 0)
                 BAIL(NC_EHDFERR);
             /* Open the HDF5 file. */
-            if ((h5->hdfid = H5Fopen(path, flags, fapl_id)) < 0)
+            if ((h5->hdfid = nc4_H5Fopen(path, flags, fapl_id)) < 0)
                 BAIL(NC_EHDFERR);
         }
 #ifdef ENABLE_BYTERANGE
@@ -843,14 +849,14 @@ nc4_open_file(const char *path, int mode, void* parameters, int ncid)
                 if (H5Pset_fapl_http(fapl_id) < 0)
                     BAIL(NC_EHDFERR);
                 /* Open the HDF5 file. */
-                if ((h5->hdfid = H5Fopen(path, flags, fapl_id)) < 0)
+                if ((h5->hdfid = nc4_H5Fopen(path, flags, fapl_id)) < 0)
                     BAIL(NC_EHDFERR);
             }
 #endif
             else
             {
                 /* Open the HDF5 file. */
-                if ((h5->hdfid = H5Fopen(path, flags, fapl_id)) < 0)
+                if ((h5->hdfid = nc4_H5Fopen(path, flags, fapl_id)) < 0)
                     BAIL(NC_EHDFERR);
             }
 
@@ -2707,3 +2713,30 @@ rec_read_metadata(NC_GRP_INFO_T *grp)
 
     return retval;
 }
+
+#ifdef _WIN32
+
+/**
+ * Wrapper function for H5Fopen.
+ * Converts the filename from ANSI to UTF-8 as needed before calling H5Fopen.
+ *
+ * @param filename The filename encoded ANSI to access.
+ * @param flags File access flags.
+ * @param fapl_id File access property list identifier.
+ * @return A file identifier if succeeded. A negative value if failed.
+ */
+static hid_t
+nc4_H5Fopen(const char *filename, unsigned flags, hid_t fapl_id)
+{
+    pathbuf_t pb;
+    hid_t hid;
+
+    filename = nc4_ndf5_ansi_to_utf8(&pb, filename);
+    if (!filename)
+        return H5I_INVALID_HID;
+    hid = H5Fopen(filename, flags, fapl_id);
+    nc4_hdf5_free_pathbuf(&pb);
+    return hid;
+}
+
+#endif /* _WIN32 */