From c5a55416965d1d5f1aa0a53f1a4d24ab0a4a3e0b Mon Sep 17 00:00:00 2001
From: Eivind Jahren <eivind.jahren@webstep.no>
Date: Wed, 4 Aug 2021 08:03:03 +0200
Subject: [PATCH] Fix bug with incorrect array length in read

---
 src/ecl_data_io/_unformatted/common.py | 16 ++++++++++++++
 src/ecl_data_io/_unformatted/read.py   | 20 ++++++++----------
 tests/test_formatted_common.py         | 29 +++++++++++++++++++++++++-
 3 files changed, 53 insertions(+), 12 deletions(-)

diff --git a/src/ecl_data_io/_unformatted/common.py b/src/ecl_data_io/_unformatted/common.py
index 53c61fb..fd95968 100644
--- a/src/ecl_data_io/_unformatted/common.py
+++ b/src/ecl_data_io/_unformatted/common.py
@@ -29,3 +29,19 @@ def item_size(type_keyword):
     if type_keyword[0:2] == b"C0":
         return int(type_keyword[2:4].decode("ascii"))
     return static_item_sizes.get(type_keyword, None)
+
+
+def bytes_in_array(array_length, item_type):
+    """
+    :param array_length: Number of items in the array
+    :param item_type: Type of items in the array
+    :returns: Number of bytes used to store an array of
+        given type and length
+    """
+    g_len = group_len(item_type)
+    full_groups = array_length // g_len
+
+    if array_length % g_len:
+        return (full_groups + 1) * 8 + array_length * item_size(item_type)
+    else:
+        return full_groups * 8 + array_length * item_size(item_type)
diff --git a/src/ecl_data_io/_unformatted/read.py b/src/ecl_data_io/_unformatted/read.py
index 04f4631..cc199ba 100644
--- a/src/ecl_data_io/_unformatted/read.py
+++ b/src/ecl_data_io/_unformatted/read.py
@@ -1,9 +1,8 @@
 import io
 
-import numpy as np
-
 import ecl_data_io.types as ecl_io_types
-from ecl_data_io._unformatted.common import group_len, item_size
+import numpy as np
+from ecl_data_io._unformatted.common import bytes_in_array, group_len, item_size
 from ecl_data_io.array_entry import EclArray
 from ecl_data_io.errors import EclParsingError
 
@@ -61,7 +60,9 @@ def _read_record_marker(self, expected_value):
         """
         value = int.from_bytes(self.stream.read(4), byteorder="big", signed=True)
         if value != expected_value:
-            raise EclParsingError(f"Unexpected size of record {value}")
+            raise EclParsingError(
+                f"Unexpected size of record {value} ({value.to_bytes(4, byteorder='big', signed=True)})"
+            )
 
     def _read_keyword(self):
         """
@@ -102,7 +103,9 @@ def _read(self):
             return
         start_marker_value = int.from_bytes(start_marker, byteorder="big", signed=True)
         if start_marker_value != 16:
-            raise EclParsingError(f"Unexpected size of record {start_marker_value}")
+            raise EclParsingError(
+                f"Unexpected size of record {start_marker_value} ({start_marker})"
+            )
         self._read_keyword()
         self._length = self._read_length()
         self._read_type()
@@ -133,12 +136,7 @@ def _read(self):
                 f"has item length {type_len} which requires 0 number of"
                 f"elements, but found {self._length} amount of elements."
             )
-
-        g_len = group_len(self.type)
-        num_groups = self._length // g_len + 1
-        if self._length == 0:
-            num_groups = 0
-        bytes_to_skip = num_groups * 8 + self._length * type_len
+        bytes_to_skip = bytes_in_array(self._length, self.type)
 
         self._data_start = self.stream.tell()
         self.stream.seek(bytes_to_skip, io.SEEK_CUR)
diff --git a/tests/test_formatted_common.py b/tests/test_formatted_common.py
index 5758477..fecce5b 100644
--- a/tests/test_formatted_common.py
+++ b/tests/test_formatted_common.py
@@ -1,4 +1,7 @@
-from ecl_data_io._unformatted.common import group_len, item_size
+import hypothesis.strategies as st
+import pytest
+from ecl_data_io._unformatted.common import bytes_in_array, group_len, item_size
+from hypothesis import given
 
 
 def test_group_len():
@@ -21,3 +24,27 @@ def test_item_size():
     assert item_size(b"DOUB") == 8
     assert item_size(b"MESS") == 0
     assert item_size(b"x231") is None
+
+
+def test_bytes_in_array():
+    assert bytes_in_array(1000, b"DOUB") == 8008
+    assert bytes_in_array(1000, b"REAL") == 4008
+    assert bytes_in_array(900, b"REAL") == 900 * 4 + 8
+    assert bytes_in_array(1100, b"REAL") == 1000 * 4 + 8 + 100 * 4 + 8
+    assert bytes_in_array(0, b"REAL") == 0
+
+
+@pytest.mark.parametrize(
+    "item_type",
+    [
+        b"C032",
+        b"CHAR",
+        b"INTE",
+        b"REAL",
+        b"LOGI",
+        b"DOUB",
+    ],
+)
+@given(length=st.integers(min_value=1, max_value=104))
+def test_bytes_in_array_one_group(item_type, length):
+    assert bytes_in_array(length, item_type) == 8 + length * item_size(item_type)