API: codes_extract_offsets_sizes_malloc

src/eccodes.h

@@ -1458,7 +1458,10 @@ int codes_bufr_header_get_string(codes_bufr_header* bh, const char* key, char* v
  * strict_mode  = If 1 means fail if any message is invalid.
  * returns 0 if OK, integer value on error.
  */
-int codes_extract_offsets_malloc(codes_context* c, const char* filename, ProductKind product, off_t** offsets, int* num_messages, int strict_mode);
+int codes_extract_offsets_malloc(codes_context* c, const char* filename, ProductKind product,
+                                 off_t** offsets, int* num_messages, int strict_mode);
+int codes_extract_offsets_sizes_malloc(codes_context* c, const char* filename, ProductKind product,
+                                       off_t** offsets, size_t** sizes, int* num_messages, int strict_mode);
 
 /* --------------------------------------- */
 #ifdef __cplusplus

src/eccodes_prototypes.h

@@ -1025,7 +1025,9 @@ int grib_read_any_from_memory_alloc(grib_context* ctx, unsigned char** data, siz
 int grib_read_any_from_memory(grib_context* ctx, unsigned char** data, size_t* data_length, void* buffer, size_t* len);
 int grib_count_in_file(grib_context* c, FILE* f, int* n);
 int grib_count_in_filename(grib_context* c, const char* filename, int* n);
-int codes_extract_offsets_malloc(grib_context* c, const char* filename, ProductKind product, off_t** offsets, int* length, int strict_mode);
+int codes_extract_offsets_malloc(grib_context* c, const char* filename, ProductKind product, off_t** offsets, int* num_messages, int strict_mode);
+int codes_extract_offsets_sizes_malloc(grib_context* c, const char* filename, ProductKind product,
+                                       off_t** offsets, size_t** sizes, int* num_messages, int strict_mode);
 
 
 /* grib_trie.cc*/

src/grib_io.cc

@@ -1815,7 +1815,11 @@ static int count_product_in_file(grib_context* c, FILE* f, ProductKind product,
     return err == GRIB_END_OF_FILE ? 0 : err;
 }
 
-int codes_extract_offsets_malloc(grib_context* c, const char* filename, ProductKind product, off_t** offsets, int* length, int strict_mode)
+static int codes_extract_offsets_malloc_internal(
+    grib_context* c, const char* filename, ProductKind product,
+    off_t** offsets, size_t** sizes,
+    int* number_of_elements,
+    int strict_mode)
 {
     int err      = 0;
     size_t size  = 0;
@@ -1848,7 +1852,7 @@ int codes_extract_offsets_malloc(grib_context* c, const char* filename, ProductK
         fclose(f);
         return err;
     }
-    *length = num_messages;
+    *number_of_elements = num_messages;
     if (num_messages == 0) {
         grib_context_log(c, GRIB_LOG_ERROR, "%s: No messages in file", __func__);
         fclose(f);
@@ -1859,6 +1863,13 @@ int codes_extract_offsets_malloc(grib_context* c, const char* filename, ProductK
         fclose(f);
         return GRIB_OUT_OF_MEMORY;
     }
+    if (sizes) {
+        *sizes = (size_t*)calloc(num_messages, sizeof(size_t));
+        if (!*sizes) {
+            fclose(f);
+            return GRIB_OUT_OF_MEMORY;
+        }
+    }
 
     i = 0;
     while (err != GRIB_END_OF_FILE) {
@@ -1868,6 +1879,9 @@ int codes_extract_offsets_malloc(grib_context* c, const char* filename, ProductK
         err = decoder(f, &size, &offset);
         if (!err) {
             (*offsets)[i] = offset;
+            if (sizes) {
+                (*sizes)[i] = size;
+            }
         }
         else {
             if (strict_mode && (err != GRIB_END_OF_FILE && err != GRIB_PREMATURE_END_OF_FILE)) {
@@ -1881,3 +1895,20 @@ int codes_extract_offsets_malloc(grib_context* c, const char* filename, ProductK
     fclose(f);
     return err;
 }
+
+// The lagacy version only did the offsets
+int codes_extract_offsets_malloc(
+    grib_context* c, const char* filename, ProductKind product,
+    off_t** offsets, int* number_of_elements, int strict_mode)
+{
+    // Call without doing the message sizes
+    return codes_extract_offsets_malloc_internal(c, filename, product, offsets, NULL, number_of_elements, strict_mode);
+}
+
+// New function does both message offsets and sizes
+int codes_extract_offsets_sizes_malloc(
+    grib_context* c, const char* filename, ProductKind product,
+    off_t** offsets, size_t** sizes, int* number_of_elements, int strict_mode)
+{
+    return codes_extract_offsets_malloc_internal(c, filename, product, offsets, sizes, number_of_elements, strict_mode);
+}

tests/extract_offsets.cc

@@ -15,23 +15,42 @@
 
 int main(int argc, char* argv[])
 {
-    char *filename;
+    char *filename = NULL;
+    char *what = NULL; // offsets or sizes
     int err = 0;
     int num_messages = 0, i =0;
     off_t* offsets = NULL;
+    size_t* sizes = NULL;
     codes_context* c = codes_context_get_default();
     const int strict_mode = 1;
 
-    /* Usage: prog file */
+    /* Usage: prog mode file */
-    assert(argc == 2);
+    assert(argc == 3);
 
-    filename = argv[1];
+    what = argv[1];
-    err = codes_extract_offsets_malloc(c, filename, PRODUCT_ANY, &offsets, &num_messages, strict_mode);
+    filename = argv[2];
-    if (err) return err;
 
-    for (i = 0; i < num_messages; ++i) {
+    if (strcmp(what, "-o")==0) {
-        printf("%lu\n", (unsigned long)offsets[i]);
+        err = codes_extract_offsets_malloc(c, filename, PRODUCT_ANY, &offsets, &num_messages, strict_mode);
+        if (err) return err;
+
+        for (i = 0; i < num_messages; ++i) {
+            printf("%lu\n", (unsigned long)offsets[i]);
+        }
+        free(offsets);
     }
-    free(offsets);
+
+    if (strcmp(what, "-s")==0) {
+        // Version getting offsets as well as sizes of messages
+        err = codes_extract_offsets_sizes_malloc(c, filename, PRODUCT_ANY, &offsets, &sizes, &num_messages, strict_mode);
+        if (err) return err;
+
+        for (i = 0; i < num_messages; ++i) {
+            printf("%zu\n", sizes[i]);
+        }
+        free(offsets);
+        free(sizes);
+    }
+
     return 0;
 }

tests/extract_offsets.sh

@@ -19,15 +19,20 @@ tempLog="temp.${label}.log"
 echo "Multi-message BUFR..."
 # ---------------------------
 input=${data_dir}/bufr/aeolus_wmo_26.bufr
-$EXEC ${test_dir}/extract_offsets  $input > $temp1
+$EXEC ${test_dir}/extract_offsets -o $input > $temp1
 ${tools_dir}/bufr_get -p offset:i  $input > $temp2
 diff $temp1 $temp2
 
+$EXEC ${test_dir}/extract_offsets -s $input > $temp1
+${tools_dir}/bufr_get -p totalLength  $input > $temp2
+diff $temp1 $temp2
+
+
 echo "Multi-message GRIBs..."
 # --------------------------
 inputs="${data_dir}/mixed.grib ${data_dir}/test.grib1  ${data_dir}/v.grib2"
 for input in $inputs; do
-    $EXEC ${test_dir}/extract_offsets  $input > $temp1
+    $EXEC ${test_dir}/extract_offsets -o $input > $temp1
     ${tools_dir}/grib_get -p offset:i  $input > $temp2
     diff $temp1 $temp2
 done
@@ -35,21 +40,21 @@ done
 echo "Test with invalid inputs..."
 # ---------------------------------
 set +e
-$EXEC ${test_dir}/extract_offsets ${data_dir} > $tempLog 2>&1
+$EXEC ${test_dir}/extract_offsets -o ${data_dir} > $tempLog 2>&1
 status=$?
 set -e
 [ $status -ne 0 ]
 grep -q "is a directory" $tempLog
 
 set +e
-$EXEC ${test_dir}/extract_offsets ${data_dir}/bad.grib > $tempLog 2>&1
+$EXEC ${test_dir}/extract_offsets -o ${data_dir}/bad.grib > $tempLog 2>&1
 status=$?
 set -e
 [ $status -ne 0 ]
 grep -q "Wrong message length" $tempLog
 
 set +e
-$EXEC ${test_dir}/extract_offsets nonexistentfile > $tempLog 2>&1
+$EXEC ${test_dir}/extract_offsets -o nonexistentfile > $tempLog 2>&1
 status=$?
 set -e
 [ $status -ne 0 ]