HDFFV-10128 merge from develop

6 files changed, 192 insertions, 18 deletions

diff --git a/tools/h5diff/CMakeTests.cmake b/tools/h5diff/CMakeTests.cmake
index a9ec683..93fdd55 100644
--- a/tools/h5diff/CMakeTests.cmake
+++ b/tools/h5diff/CMakeTests.cmake
@@ -286,7 +286,7 @@
 ##############################################################################
 ##############################################################################
 
-  MACRO (ADD_H5_TEST resultfile resultcode)
+  macro (ADD_H5_TEST resultfile resultcode)
     # If using memchecker add tests without using scripts
     if (HDF5_ENABLE_USING_MEMCHECKER)
       add_test (NAME H5DIFF-${resultfile} COMMAND $<TARGET_FILE:h5diff> ${ARGN})
@@ -314,9 +314,9 @@
     if (H5_HAVE_PARALLEL)
       ADD_PH5_TEST (${resultfile} ${resultcode} ${ARGN})
     endif ()
-  ENDMACRO ()
+  endmacro ()
 
-  MACRO (ADD_PH5_TEST resultfile resultcode)
+  macro (ADD_PH5_TEST resultfile resultcode)
     # If using memchecker add tests without using scripts
     if (HDF5_ENABLE_USING_MEMCHECKER)
       add_test (NAME PH5DIFF-${resultfile} COMMAND $<TARGET_FILE:ph5diff> ${MPIEXEC} ${MPIEXEC_PREFLAGS} ${MPIEXEC_NUMPROC_FLAG} ${MPIEXEC_MAX_NUMPROCS} ${MPIEXEC_POSTFLAGS} ${ARGN})
@@ -348,9 +348,9 @@
       endif ()
       set (last_test "PH5DIFF-${resultfile}")
     endif ()
-  ENDMACRO ()
+  endmacro ()
 
-  MACRO (ADD_H5_UD_TEST testname resultcode resultfile)
+  macro (ADD_H5_UD_TEST testname resultcode resultfile)
     if (NOT HDF5_ENABLE_USING_MEMCHECKER)
       # Remove any output file left over from previous test run
       add_test (
@@ -393,7 +393,7 @@
       endif ()
       set_tests_properties (H5DIFF_UD-${testname} PROPERTIES DEPENDS H5DIFF_UD-${testname}-clearall-objects)
     endif ()
-  ENDMACRO ()
+  endmacro ()
 
 ##############################################################################
 ##############################################################################
@@ -450,6 +450,9 @@
   # attrs with verbose option level
   set (ATTR_VERBOSE_LEVEL_FILE1 h5diff_attr_v_level1.h5)
   set (ATTR_VERBOSE_LEVEL_FILE2 h5diff_attr_v_level2.h5)
+  # strings
+  set (STRINGS1 diff_strings1.h5)
+  set (STRINGS2 diff_strings2.h5)
 
   if (HDF5_ENABLE_USING_MEMCHECKER)
     # Remove any output file left over from previous test run
@@ -1089,8 +1092,7 @@ ADD_H5_TEST (h5diff_103 1 -v --use-system-epsilon ${FILE1} ${FILE1} g1/d1
 g1/d2)
 
 # with --use-system-epsilon for float value. expect less differences
-ADD_H5_TEST (h5diff_104 1 -v --use-system-epsilon ${FILE1} ${FILE1} g1/fp1
-g1/fp2)
+ADD_H5_TEST (h5diff_104 1 -v --use-system-epsilon ${FILE1} ${FILE1} g1/fp1 g1/fp2)
 
 # not comparable -c flag
 ADD_H5_TEST (h5diff_200 0 ${FILE2} ${FILE2} g2/dset1  g2/dset2)
@@ -1386,3 +1388,7 @@ ADD_H5_TEST (h5diff_646 1 -v --use-system-epsilon -p 0.05 ${FILE1} ${FILE2} /g1/
 ##############################################################################
 ADD_H5_UD_TEST (h5diff_plugin_test 0 h5diff_ud -v tudfilter.h5 tudfilter2.h5)
 ADD_H5_UD_TEST (h5diff_plugin_fail 2 h5diff_udfail -v tudfilter.h5 tudfilter2.h5)
+
+# ##############################################################################
+# # END
+# ##############################################################################
diff --git a/tools/h5diff/h5diffgentest.c b/tools/h5diff/h5diffgentest.c
index cfaffe6..50b32aa 100644
--- a/tools/h5diff/h5diffgentest.c
+++ b/tools/h5diff/h5diffgentest.c
@@ -93,6 +93,9 @@ hsize_t H5TOOLS_MALLOCSIZE = (128 * 1024 * 1024);
 /* non-comparable dataset and attribute */
 #define NON_COMPARBLES1 "non_comparables1.h5"
 #define NON_COMPARBLES2 "non_comparables2.h5"
+/* string dataset and attribute */
+#define DIFF_STRINGS1 "diff_strings1.h5"
+#define DIFF_STRINGS2 "diff_strings2.h5"
 
 #define UIMAX    4294967295u /*Maximum value for a variable of type unsigned int */
 #define STR_SIZE 3
@@ -157,6 +160,7 @@ static void test_comps_array_vlen (const char *fname, const char *dset, const ch
 static void test_comps_vlen_arry (const char *fname, const char *dset,const char *attr, int diff, int is_file_new);
 static void test_data_nocomparables (const char *fname, int diff);
 static void test_objs_nocomparables (const char *fname1, const char *fname2);
+static void test_objs_strings (const char *fname, const char *fname2);
 
 /* called by test_attributes() and test_datasets() */
 static void write_attr_in(hid_t loc_id,const char* dset_name,hid_t fid,int make_diffs);
@@ -257,6 +261,9 @@ int main(void)
     /* common objects (same name) with different object types. HDFFV-7644 */
     test_objs_nocomparables(NON_COMPARBLES1, NON_COMPARBLES2);
 
+    /* string dataset and attribute. HDFFV-10028 */
+    test_objs_strings(DIFF_STRINGS1, DIFF_STRINGS2);
+
     return 0;
 }
 
@@ -5214,6 +5221,160 @@ out:
 
 }
 
+static hid_t mkstr(int size, H5T_str_t pad) {
+    hid_t type;
+
+    if((type=H5Tcopy(H5T_C_S1)) < 0) return -1;
+    if(H5Tset_size(type, (size_t)size) < 0) return -1;
+    if(H5Tset_strpad(type, pad) < 0) return -1;
+
+    return type;
+}
+
+/*-------------------------------------------------------------------------
+* Function: test_objs_strings
+*
+* Purpose:
+*   Create test files with common objects (same name) but different string
+*   types.
+*   h5diff should show differences output from these common objects.
+*-------------------------------------------------------------------------*/
+static void test_objs_strings(const char *fname1, const char *fname2)
+{
+    herr_t  status = SUCCEED;
+    hid_t   fid1=0;
+    hid_t   fid2=0;
+    hid_t   dataset=0;
+    hid_t   space=0;
+    hid_t   f_type=0;
+    hid_t   m_type=0;
+    hsize_t dims1[] = {3, 4};
+    char string1A[12][3] = {"s1","s2","s3","s4","s5","s6","s","s","s9",
+            "s0","s1","s2"};
+    char string1B[12][3] = {"s1","s2","s3","s4","s","s","s7","s8","s9",
+            "s0","s1","s2"};
+
+    hsize_t dims2[]={20};
+    char string2A[20][10] = {"ab cd ef1", "ab cd ef2", "ab cd ef3", "ab cd ef4",
+            "ab cd ef5", "ab cd ef6", "ab cd ef7", "ab cd ef8",
+            "ab cd 9", "ab cd 0", "ab cd 1", "ab cd 2",
+            "ab cd ef3", "ab cd ef4", "ab cd ef5", "ab cd ef6",
+            "ab cd ef7", "ab cd ef8", "ab cd ef9", "ab cd ef0"};
+    char string2B[20][10] = {"ab cd ef1", "ab cd ef2", "ab cd ef3", "ab cd ef4",
+            "ab cd ef5", "ab cd ef6", "ab cd ef7", "ab cd ef8",
+            "ab cd ef9", "ab cd ef0", "ab cd ef1", "ab cd ef2",
+            "ab cd 3", "ab cd 4", "ab cd 5", "ab cd 6",
+            "ab cd ef7", "ab cd ef8", "ab cd ef9", "ab cd ef0"};
+
+    hsize_t dims3[] = {27};
+    char string3A[27][6] = {"abcd0", "abcd1", "abcd2", "abcd3",
+            "abcd4", "abcd5", "abcd6", "abcd7",
+            "abcd8", "abcd9", "abcd0", "abcd1",
+            "abd2", "abc3", "bcd4", "acd5",
+            "abcd6", "abcd7", "abcd8", "abcd9",
+            "abcd0", "abcd1", "abcd2", "abcd3",
+            "abc4", "abc5", "abc6"};
+    char string3B[27][6] = {"abcd0", "abcd1", "abcd2", "abcd3",
+            "abcd4", "abcd5", "abcd6", "abcd7",
+            "abcd8", "abcd9", "abcd0", "abcd1",
+            "abcd2", "abcd3", "abcd4", "abcd5",
+            "abd6", "abc7", "bcd8", "acd9",
+            "abcd0", "abcd1", "abcd2", "abcd3",
+            "abd4", "abd5", "abd6"};
+
+    hsize_t dims4[] = {3};
+    char string4A[3][21] = { "s1234567890123456789", "s1234567890123456789",
+            "s12345678901234567"};
+    char string4B[3][21] = { "s1234567890123456789", "s12345678901234567",
+            "s1234567890123456789"};
+
+    /*-----------------------------------------------------------------------
+     * Create file(s)
+     *------------------------------------------------------------------------*/
+     /* file1 */
+     fid1 = H5Fcreate (fname1, H5F_ACC_TRUNC, H5P_DEFAULT, H5P_DEFAULT);
+     if (fid1 < 0)
+     {
+         fprintf(stderr, "Error: %s> H5Fcreate failed.\n", fname1);
+         status = FAIL;
+         goto out;
+     }
+
+     /* file2 */
+     fid2 = H5Fcreate (fname2, H5F_ACC_TRUNC, H5P_DEFAULT, H5P_DEFAULT);
+     if (fid2 < 0)
+     {
+         fprintf(stderr, "Error: %s> H5Fcreate failed.\n", fname2);
+         status = FAIL;
+         goto out;
+     }
+
+    /* string 1 : nullterm string */
+    space = H5Screate_simple(2, dims1, NULL);
+    f_type = mkstr(5, H5T_STR_NULLTERM);
+    m_type = mkstr(3, H5T_STR_NULLTERM);
+    dataset = H5Dcreate2(fid1, "/string1", f_type, space, H5P_DEFAULT, H5P_DEFAULT, H5P_DEFAULT);
+    H5Dwrite(dataset, m_type, H5S_ALL, H5S_ALL, H5P_DEFAULT, string1A);
+    H5Dclose(dataset);
+    dataset = H5Dcreate2(fid2, "/string1", f_type, space, H5P_DEFAULT, H5P_DEFAULT, H5P_DEFAULT);
+    H5Dwrite(dataset, m_type, H5S_ALL, H5S_ALL, H5P_DEFAULT, string1B);
+    H5Tclose(m_type);
+    H5Tclose(f_type);
+    H5Sclose(space);
+    H5Dclose(dataset);
+
+    /* string 2 : space pad string */
+    space = H5Screate_simple(1, dims2, NULL);
+    f_type = mkstr(11, H5T_STR_SPACEPAD);
+    m_type = mkstr(10, H5T_STR_NULLTERM);
+    dataset = H5Dcreate2(fid1, "/string2", f_type, space, H5P_DEFAULT, H5P_DEFAULT, H5P_DEFAULT);
+    H5Dwrite(dataset, m_type, H5S_ALL, H5S_ALL, H5P_DEFAULT, string2A);
+    H5Dclose(dataset);
+    dataset = H5Dcreate2(fid2, "/string2", f_type, space, H5P_DEFAULT, H5P_DEFAULT, H5P_DEFAULT);
+    H5Dwrite(dataset, m_type, H5S_ALL, H5S_ALL, H5P_DEFAULT, string2B);
+    H5Tclose(m_type);
+    H5Tclose(f_type);
+    H5Sclose(space);
+    H5Dclose(dataset);
+
+    /* string 3 : null pad string */
+    space = H5Screate_simple(1, dims3, NULL);
+    f_type = mkstr(8, H5T_STR_NULLPAD);
+    m_type = mkstr(6, H5T_STR_NULLTERM);
+    dataset = H5Dcreate2(fid1, "/string3", f_type, space, H5P_DEFAULT, H5P_DEFAULT, H5P_DEFAULT);
+    H5Dwrite(dataset, m_type, H5S_ALL, H5S_ALL, H5P_DEFAULT, string3A);
+    H5Dclose(dataset);
+    dataset = H5Dcreate2(fid2, "/string3", f_type, space, H5P_DEFAULT, H5P_DEFAULT, H5P_DEFAULT);
+    H5Dwrite(dataset, m_type, H5S_ALL, H5S_ALL, H5P_DEFAULT, string3B);
+    H5Tclose(m_type);
+    H5Tclose(f_type);
+    H5Sclose(space);
+    H5Dclose(dataset);
+
+    /* string 4 : space pad long string */
+    space = H5Screate_simple(1, dims4, NULL);
+    f_type = mkstr(168, H5T_STR_SPACEPAD);
+    m_type = mkstr(21, H5T_STR_NULLTERM);
+    dataset = H5Dcreate2(fid1, "/string4", f_type, space, H5P_DEFAULT, H5P_DEFAULT, H5P_DEFAULT);
+    H5Dwrite(dataset, m_type, H5S_ALL, H5S_ALL, H5P_DEFAULT, string4A);
+    H5Dclose(dataset);
+    dataset = H5Dcreate2(fid2, "/string4", f_type, space, H5P_DEFAULT, H5P_DEFAULT, H5P_DEFAULT);
+    H5Dwrite(dataset, m_type, H5S_ALL, H5S_ALL, H5P_DEFAULT, string4B);
+    H5Tclose(m_type);
+    H5Tclose(f_type);
+    H5Sclose(space);
+    H5Dclose(dataset);
+
+out:
+    /*-----------------------------------------------------------------------
+    * Close IDs
+    *-----------------------------------------------------------------------*/
+    if(fid1)
+        H5Fclose(fid1);
+    if(fid2)
+        H5Fclose(fid2);
+}
+
 /*-------------------------------------------------------------------------
 * Function: write_attr_in
 *
diff --git a/tools/h5diff/testfiles/diff_strings1.h5 b/tools/h5diff/testfiles/diff_strings1.h5
new file mode 100644
index 0000000..8839f5a
--- /dev/null
+++ b/tools/h5diff/testfiles/diff_strings1.h5
diff --git a/tools/h5diff/testfiles/diff_strings2.h5 b/tools/h5diff/testfiles/diff_strings2.h5
new file mode 100644
index 0000000..810278c
--- /dev/null
+++ b/tools/h5diff/testfiles/diff_strings2.h5
diff --git a/tools/h5diff/testh5diff.sh.in b/tools/h5diff/testh5diff.sh.in
index 5748128..dedeb15 100644
--- a/tools/h5diff/testh5diff.sh.in
+++ b/tools/h5diff/testh5diff.sh.in
@@ -120,6 +120,8 @@ $SRC_H5DIFF_TESTFILES/h5diff_attr_v_level2.h5
 $SRC_H5DIFF_TESTFILES/h5diff_enum_invalid_values.h5
 $SRC_H5DIFF_TESTFILES/non_comparables1.h5
 $SRC_H5DIFF_TESTFILES/non_comparables2.h5
+$SRC_H5DIFF_TESTFILES/diff_strings1.h5
+$SRC_H5DIFF_TESTFILES/diff_strings2.h5
 "
 
 LIST_OTHER_TEST_FILES="
@@ -609,8 +611,6 @@ TOOLTEST h5diff_28.txt -v h5diff_types.h5 h5diff_types.h5 l1 l2
 TOOLTEST h5diff_30.txt -v h5diff_enum_invalid_values.h5 h5diff_enum_invalid_values.h5 dset1 dset2
 
 
-
-
 # ##############################################################################
 # # Dataset datatypes
 # ##############################################################################
@@ -646,6 +646,13 @@ TOOLTEST h5diff_58.txt -v h5diff_dset1.h5 h5diff_dset2.h5 refreg
 # ( HDDFV-7942 )
 TOOLTEST h5diff_59.txt -v h5diff_dtypes.h5 h5diff_dtypes.h5 dset11a dset11b
 
+# Strings
+# ( HDFFV-10128 )
+TOOLTEST h5diff_60.txt -v diff_strings1 diff_strings2 string1 string1
+TOOLTEST h5diff_61.txt -v diff_strings1 diff_strings2 string2 string2
+TOOLTEST h5diff_62.txt -v diff_strings1 diff_strings2 string3 string3
+TOOLTEST h5diff_63.txt -v diff_strings1 diff_strings2 string4 string4
+
 # ##############################################################################
 # # Error messages
 # ##############################################################################
@@ -717,8 +724,6 @@ TOOLTEST h5diff_618.txt -p 2 h5diff_basic1.h5 h5diff_basic2.h5 g1/dset3 g1/dset4
 # 6.19: number smaller than smallest difference
 TOOLTEST h5diff_619.txt -p 0.005 h5diff_basic1.h5 h5diff_basic2.h5 g1/dset3 g1/dset4
 
-
-
 # ##############################################################################
 # # -n
 # ##############################################################################
@@ -771,7 +776,6 @@ TOOLTEST h5diff_tmp2.txt tmpSingleSiteBethe.output.h5 tmpSingleSiteBethe.referen
 # ##################################################
 #  attrs with verbose option level
 # ##################################################
-
 TOOLTEST h5diff_700.txt -v1 h5diff_attr1.h5 h5diff_attr2.h5
 TOOLTEST h5diff_701.txt -v2 h5diff_attr1.h5 h5diff_attr2.h5
 TOOLTEST h5diff_702.txt --verbose=1 h5diff_attr1.h5 h5diff_attr2.h5
@@ -822,7 +826,6 @@ TOOLTEST h5diff_103.txt -v --use-system-epsilon h5diff_basic1.h5 h5diff_basic1.h
 # with --use-system-epsilon for float value
 TOOLTEST h5diff_104.txt -v --use-system-epsilon h5diff_basic1.h5 h5diff_basic1.h5 g1/fp1 g1/fp2
 
-
 # not comparable -c flag
 TOOLTEST h5diff_200.txt h5diff_basic2.h5 h5diff_basic2.h5 g2/dset1  g2/dset2
 
@@ -950,7 +953,6 @@ TOOLTEST h5diff_424.txt --follow-symlinks -v h5diff_ext2softlink_trg.h5 h5diff_e
 # extlink_to_softlink_to_dset1 vs extlink_to_softlink_to_dset2"
 TOOLTEST h5diff_425.txt --follow-symlinks -v h5diff_ext2softlink_src.h5 h5diff_ext2softlink_src.h5 /ext_link_to_slink1 /ext_link_to_slink2
 
-
 # ##############################################################################
 # # Dangling links compare (--follow-symlinks and --no-dangling-links)
 # ##############################################################################
@@ -1101,6 +1103,7 @@ TOOLTEST h5diff_540.txt -v compounds_array_vlen1.h5 compounds_array_vlen2.h5
 # ##############################################################################
 # # Test mutually exclusive options
 # ##############################################################################
+#
 # Test with -d , -p and --use-system-epsilon.
 TOOLTEST h5diff_640.txt -v -d 5 -p 0.05 --use-system-epsilon h5diff_basic1.h5 h5diff_basic2.h5 /g1/dset3 /g1/dset4
 TOOLTEST h5diff_641.txt -v -d 5 -p 0.05 h5diff_basic1.h5 h5diff_basic2.h5 /g1/dset3 /g1/dset4
diff --git a/tools/lib/h5diff_array.c b/tools/lib/h5diff_array.c
index e4b166f..fc19bbc 100644
--- a/tools/lib/h5diff_array.c
+++ b/tools/lib/h5diff_array.c
@@ -544,7 +544,7 @@ hsize_t diff_array( void *_mem1,
  *  Recursively call this function for each element
  * H5T_STRING
  *  compare byte by byte in a cycle from 0 to type_size. this type_size is the
- *  value obtained by the get_size function but it is the string lenght for
+ *  value obtained by the get_size function but it is the string length for
  *  variable sized strings
  * H5T_OPAQUE
  *  compare byte by byte in a cycle from 0 to type_size
@@ -600,7 +600,7 @@ static hsize_t diff_datum(void       *_mem1,
 
     /* Fast comparison first for atomic type by memcmp().
      * It is OK not to list non-atomic type here because it will not be caught
-     * by the confition, but it gives more clarity for code planning
+     * by the condition, but it gives more clarity for code planning
      */
     if (type_class != H5T_REFERENCE &&
         type_class != H5T_COMPOUND &&
@@ -690,6 +690,10 @@ static hsize_t diff_datum(void       *_mem1,
              *       of length of strings.
              *       For now mimic the previous way.
              */
+            if(size1 != size2)
+            {
+                nfound++;
+            }
             if(size1 < size2)
             {
                 size = size1;
@@ -710,7 +714,7 @@ static hsize_t diff_datum(void       *_mem1,
 
                 pad = H5Tget_strpad(m_type);
 
-                for (u=0; u<size && (s[u] || pad!=H5T_STR_NULLTERM); u++)
+                for (u=0; u<size; u++)
                     nfound+=character_compare(
                         s1 + u,
                         s2 + u, /* offset */