Adds platform-independent basename and dirname (#1951)

* Adds platform-independent basename and dirname

* Tidy up H5_dirname and H5_basename implementations and add tests

* Committing clang-format changes

* Fix misspelling

* Several fixes for H5_dirname/H5_basename from review

* Committing clang-format changes

* Add reason to VERIFY_STR macros in th5_system.c

* Use H5MM_free instead of HDfree in H5_dirname/H5_basename tests

Co-authored-by: Jordan Henderson <jhenderson@hdfgroup.org>
Co-authored-by: github-actions <41898282+github-actions[bot]@users.noreply.github.com>

5 files changed, 321 insertions, 0 deletions

diff --git a/src/H5MM.c b/src/H5MM.c
index 6943887..9c03ceb 100644
--- a/src/H5MM.c
+++ b/src/H5MM.c
@@ -504,6 +504,53 @@ done:
 } /* end H5MM_strdup() */
 
 /*-------------------------------------------------------------------------
+ * Function:    H5MM_strndup
+ *
+ * Purpose:     Duplicates a string, including memory allocation, but only
+ *              copies at most `n` bytes from the string to be duplicated.
+ *              If the string to be duplicated is longer than `n`, only `n`
+ *              bytes are copied and a terminating null byte is added.
+ *              NULL is NOT an acceptable value for the input string.
+ *
+ *              If the string to be duplicated is the NULL pointer, then
+ *              an error will be raised.
+ *
+ * Return:      Success:    Pointer to a new string
+ *              Failure:    NULL
+ *-------------------------------------------------------------------------
+ */
+char *
+H5MM_strndup(const char *s, size_t n)
+{
+#if defined H5_MEMORY_ALLOC_SANITY_CHECK
+    size_t len;
+#endif
+    char *ret_value = NULL;
+
+    FUNC_ENTER_NOAPI(NULL)
+
+    if (!s)
+        HGOTO_ERROR(H5E_ARGS, H5E_BADVALUE, NULL, "NULL string not allowed")
+
+#if defined H5_MEMORY_ALLOC_SANITY_CHECK
+    for (len = 0; len < n && s[len] != '\0'; len++)
+        ;
+
+    if (NULL == (ret_value = H5MM_malloc(len + 1)))
+        HGOTO_ERROR(H5E_RESOURCE, H5E_NOSPACE, NULL, "memory allocation failed")
+
+    H5MM_memcpy(ret_value, s, len);
+    ret_value[len] = '\0';
+#else
+    if (NULL == (ret_value = HDstrndup(s, n)))
+        HGOTO_ERROR(H5E_RESOURCE, H5E_NOSPACE, NULL, "string duplication failed")
+#endif
+
+done:
+    FUNC_LEAVE_NOAPI(ret_value)
+} /* end H5MM_strndup() */
+
+/*-------------------------------------------------------------------------
  * Function:    H5MM_xfree
  *
  * Purpose:     Just like free(3) except null pointers are allowed as
diff --git a/src/H5MMprivate.h b/src/H5MMprivate.h
index f05a3da..0a5a011 100644
--- a/src/H5MMprivate.h
+++ b/src/H5MMprivate.h
@@ -44,6 +44,7 @@ H5_DLL void  *H5MM_calloc(size_t size) H5_ATTR_MALLOC;
 H5_DLL void  *H5MM_realloc(void *mem, size_t size);
 H5_DLL char  *H5MM_xstrdup(const char *s);
 H5_DLL char  *H5MM_strdup(const char *s);
+H5_DLL char  *H5MM_strndup(const char *s, size_t n);
 H5_DLL void  *H5MM_xfree(void *mem);
 H5_DLL void  *H5MM_xfree_const(const void *mem);
 H5_DLL void  *H5MM_memcpy(void *dest, const void *src, size_t n);
diff --git a/src/H5private.h b/src/H5private.h
index f9e7aff..3130bb1 100644
--- a/src/H5private.h
+++ b/src/H5private.h
@@ -1411,6 +1411,9 @@ H5_DLL H5_ATTR_CONST int Nflock(int fd, int operation);
 #ifndef HDstrncpy
 #define HDstrncpy(X, Y, Z) strncpy(X, Y, Z)
 #endif
+#ifndef HDstrndup
+#define HDstrndup(S, N) strndup(S, N)
+#endif
 #ifndef HDstrpbrk
 #define HDstrpbrk(X, Y) strpbrk(X, Y)
 #endif
@@ -2544,6 +2547,8 @@ H5_DLL double H5_get_time(void);
 /* Functions for building paths, etc. */
 H5_DLL herr_t H5_build_extpath(const char *name, char **extpath /*out*/);
 H5_DLL herr_t H5_combine_path(const char *path1, const char *path2, char **full_name /*out*/);
+H5_DLL herr_t H5_dirname(const char *path, char **dirname /*out*/);
+H5_DLL herr_t H5_basename(const char *path, char **basename /*out*/);
 
 /* getopt(3) equivalent that papers over the lack of long options on BSD
  * and lack of Windows support.
diff --git a/src/H5system.c b/src/H5system.c
index 677452d..9a19860 100644
--- a/src/H5system.c
+++ b/src/H5system.c
@@ -919,8 +919,274 @@ done:
 
     FUNC_LEAVE_NOAPI(ret_value)
 } /* end H5_expand_windows_env_vars() */
+
+/*-------------------------------------------------------------------------
+ * Function:    H5_strndup
+ *
+ * Purpose:     Similar to strndup() for use on Windows. Allocates a new
+ *              string and copies at most `n` bytes from the original
+ *              string into the new string. If the original string is
+ *              longer than `n`, only `n` bytes are copied from the
+ *              original string. In either case, the string being returned
+ *              is guaranteed to be terminated with a null byte.
+ *
+ *              The returned pointer is allocated by H5MM_malloc in this
+ *              routine and must be freed by the caller with H5MM_free or
+ *              H5MM_xfree.
+ *
+ * Return:      Pointer to copied string on success
+ *              NULL on failure
+ *
+ *-------------------------------------------------------------------------
+ */
+char *
+H5_strndup(const char *s, size_t n)
+{
+    size_t len;
+    char  *ret_value = NULL;
+
+    FUNC_ENTER_NOAPI_NOINIT
+
+    if (!s)
+        HGOTO_ERROR(H5E_ARGS, H5E_BADVALUE, NULL, "string cannot be NULL")
+
+    for (len = 0; len < n && s[len] != '\0'; len++)
+        ;
+
+    if (NULL == (ret_value = H5MM_malloc(len + 1)))
+        HGOTO_ERROR(H5E_RESOURCE, H5E_CANTALLOC, NULL, "can't allocate buffer for string")
+
+    H5MM_memcpy(ret_value, s, len);
+    ret_value[len] = '\0';
+
+done:
+    FUNC_LEAVE_NOAPI(ret_value)
+}
 #endif /* H5_HAVE_WIN32_API */
 
+/* dirname() and basename() are not easily ported to Windows and basename
+ * behavior varies depending on if you get POSIX vs. GNU. As a more
+ * platform-indpendent work-around, we've implemented H5_ versions of
+ * dirname() and basename().
+ *
+ * - The input string is never modified.
+ *
+ * - The out parameter is a new string that was allocated with H5MM routines
+ *   and must be freed by the caller via H5MM_free()/H5MM_xfree().
+ *
+ * - NULL pointers are errors.
+ *
+ * - On errors, FAIL will be returned and the output parameter will be
+ *   undefined.
+ *
+ * - Assumes the file separator is \ on Win32 and / everywhere else,
+ *   including Cygwin.
+ */
+
+/*-------------------------------------------------------------------------
+ * Function:    H5_dirname
+ *
+ * Purpose:     Similar to dirname(3) but more portable across platforms.
+ *              Returns a pointer to the directory component of a specified
+ *              pathname. The returned pointer is allocated by this routine
+ *              and must be freed by the caller with H5MM_free or
+ *              H5MM_xfree.
+ *
+ * Return:      Non-negative on success/Negative on failure
+ *
+ *-------------------------------------------------------------------------
+ */
+herr_t
+H5_dirname(const char *path, char **dirname)
+{
+    char  *sep;
+    char  *out       = NULL;
+    herr_t ret_value = SUCCEED;
+
+    FUNC_ENTER_NOAPI_NOINIT
+
+    if (!path)
+        HGOTO_ERROR(H5E_ARGS, H5E_BADVALUE, FAIL, "path can't be NULL")
+    if (!dirname)
+        HGOTO_ERROR(H5E_ARGS, H5E_BADVALUE, FAIL, "dirname can't be NULL")
+
+    if (NULL == (sep = HDstrrchr(path, H5_DIR_SEPC))) {
+        /* Pathname with no file separator characters */
+        out = H5MM_strdup(".");
+    }
+    else if (sep == path) {
+        /* Pathname of form "/" or "/filename" */
+        out = H5MM_strdup(H5_DIR_SEPS);
+    }
+    else {
+        if (sep[1] == '\0') {
+            /*
+             * Last file separator character is last character in
+             * pathname. Skip this and any other preceding trailing
+             * file separator characters
+             */
+            while (sep != path && sep[-1] == H5_DIR_SEPC)
+                sep--;
+
+            if (sep == path) {
+                /* Contrived case: "//", "///" and similar */
+                out = H5MM_strdup(H5_DIR_SEPS);
+                sep = NULL;
+            }
+            else {
+                /*
+                 * Must have found the filename component. Search
+                 * backwards to a previous file separator character,
+                 * if any.
+                 */
+                while (sep != path && sep[-1] != H5_DIR_SEPC)
+                    sep--;
+
+                if (sep == path) {
+                    /* No directory component found, just return "." */
+                    out = H5MM_strdup(".");
+                    sep = NULL;
+                }
+            }
+        }
+
+        if (sep) {
+            ptrdiff_t len;
+
+            /* Skip a possible run of duplicate file separator characters */
+            while (sep != path && sep[-1] == H5_DIR_SEPC)
+                sep--;
+
+            if (sep == path)
+                /* Pathname of form "/usr/" */
+                out = H5MM_strdup(H5_DIR_SEPS);
+            else {
+                /* Pathname of form "dir/filename" */
+                len = sep - path;
+                HDassert(len >= 0);
+
+                out = H5MM_strndup(path, (size_t)len);
+            }
+        }
+    }
+
+    if (NULL == out)
+        HGOTO_ERROR(H5E_RESOURCE, H5E_CANTALLOC, FAIL, "can't allocate buffer for dirname");
+
+    *dirname = out;
+
+done:
+    if (FAIL == ret_value) {
+        H5MM_free(out);
+        if (dirname)
+            *dirname = NULL;
+    }
+
+    FUNC_LEAVE_NOAPI(ret_value)
+} /* end H5_dirname() */
+
+/*-------------------------------------------------------------------------
+ * Function:    H5_basename
+ *
+ * Purpose:     Similar to basename(3) but more portable across platforms.
+ *              Returns a pointer to the filename component of a specified
+ *              pathname. The returned pointer is allocated by this routine
+ *              and must be freed by the caller with H5MM_free or
+ *              H5MM_xfree.
+ *
+ *              NOTE: This routine follows the POSIX semantics for
+ *              basename(3). That is, passing the path string "/" ("\" on
+ *              Windows) returns the string "/" (again, "\" on Windows) and
+ *              passing a path string with trailing file separator
+ *              characters returns the filename component with the trailing
+ *              file separator characters being ignored.
+ *
+ * Return:      Non-negative on success/Negative on failure
+ *
+ *-------------------------------------------------------------------------
+ */
+herr_t
+H5_basename(const char *path, char **basename)
+{
+    const char *sep;
+    char       *out       = NULL;
+    herr_t      ret_value = SUCCEED;
+
+    FUNC_ENTER_NOAPI_NOINIT
+
+    if (!path)
+        HGOTO_ERROR(H5E_ARGS, H5E_BADVALUE, FAIL, "path can't be NULL")
+    if (!basename)
+        HGOTO_ERROR(H5E_ARGS, H5E_BADVALUE, FAIL, "basename can't be NULL")
+
+    if (NULL == (sep = HDstrrchr(path, H5_DIR_SEPC))) {
+        if (*path == '\0')
+            /* Empty pathname */
+            out = H5MM_strdup(".");
+        else
+            /* Pathname with no file separator characters */
+            out = H5MM_strdup(path);
+    }
+    else if (sep == path) {
+        if (sep[1] == '\0')
+            /* Pathname of form "/" */
+            out = H5MM_strdup(H5_DIR_SEPS);
+        else
+            /* Pathname of form "/filename" */
+            out = H5MM_strdup(sep + 1);
+    }
+    else {
+        if (sep[1] != '\0')
+            /* Pathname of form "dir/filename" */
+            out = H5MM_strdup(sep + 1);
+        else {
+            /* Pathname of form "filename/", "/dir/filename/", etc. */
+
+            /*
+             * Last file separator character is last character in
+             * pathname. Skip this and any other preceding trailing
+             * file separator characters
+             */
+            while (sep != path && sep[-1] == H5_DIR_SEPC)
+                sep--;
+
+            if (sep == path)
+                /* Contrived case: "//", "///" and similar */
+                out = H5MM_strdup(H5_DIR_SEPS);
+            else {
+                const char *c_ptr = sep;
+                ptrdiff_t   len;
+
+                /*
+                 * Skip back to a previous file separator character,
+                 * if any, and form final filename component
+                 */
+                while (c_ptr != path && c_ptr[-1] != H5_DIR_SEPC)
+                    c_ptr--;
+
+                len = sep - c_ptr;
+                HDassert(len >= 0);
+
+                out = H5MM_strndup(c_ptr, (size_t)len);
+            }
+        }
+    }
+
+    if (NULL == out)
+        HGOTO_ERROR(H5E_RESOURCE, H5E_CANTALLOC, FAIL, "can't allocate buffer for basename");
+
+    *basename = out;
+
+done:
+    if (FAIL == ret_value) {
+        H5MM_free(out);
+        if (basename)
+            *basename = NULL;
+    }
+
+    FUNC_LEAVE_NOAPI(ret_value)
+} /* end H5_basename() */
+
 /* Global variables */
 int         H5_opterr = 1; /* Get_option prints errors if this is on */
 int         H5_optind = 1; /* Token pointer                          */
diff --git a/src/H5win32defs.h b/src/H5win32defs.h
index 5f384bb..1039f23 100644
--- a/src/H5win32defs.h
+++ b/src/H5win32defs.h
@@ -82,6 +82,7 @@ struct timezone {
 #define HDsleep(S)            Sleep(S * 1000)
 #define HDstat(S, B)          _stati64(S, B)
 #define HDstrcasecmp(A, B)    _stricmp(A, B)
+#define HDstrndup(S, N)       H5_strndup(S, N)
 #define HDstrtok_r(X, Y, Z)   strtok_s(X, Y, Z)
 #define HDtzset()             _tzset()
 #define HDunlink(S)           _unlink(S)
@@ -104,6 +105,7 @@ H5_DLL wchar_t *H5_get_utf16_str(const char *s);
 H5_DLL int      Wopen_utf8(const char *path, int oflag, ...);
 H5_DLL int      Wremove_utf8(const char *path);
 H5_DLL int      H5_get_win32_times(H5_timevals_t *tvs);
+H5_DLL char    *H5_strndup(const char *s, size_t n);
 #ifdef __cplusplus
 }
 #endif /* __cplusplus */