Added consistency check for numpy data arrays to ensure they match ch…

…annels when writing.

geosoft/gxpy/gdb.py

@@ -1090,6 +1090,8 @@ def new_channel(self, name, dtype=np.float64, array=1, dup=None, details=None):
         """
 
         symb = self._db.find_symb(name, gxapi.DB_SYMB_CHAN)
+        if array < 1:
+            array = 1
         if symb == gxapi.NULLSYMB:
             if dup:
                 symb = self._db.dup_symb_no_lock(self.channel_name_symb(dup)[1], name)
@@ -1694,10 +1696,14 @@ def write_channel(self, line, channel, data, fid=(0.0, 1.0), unit_of_measure=Non
         if not isinstance(data, np.ndarray):
             data = np.array(data)
 
+        if _va_width(data) == 0:
+            # no data to write
+            return
+
         w = self.channel_width(cs)
         if w != _va_width(data):
             raise GdbException(
-                _t("Array data width {} does not fit into VA channel '{}' with width {}").
+                _t("Array data width {} does not fit into channel '{}' with width {}").
                 format(_va_width(data), cn, w))
 
         # 1D channel
@@ -1757,7 +1763,8 @@ def write_line(self, line, data, channels=None, fid=(0.0, 1.0)):
         :param line:        line to write to, name or symbol
         :param data:        numpy array shape (records,channels).  If single dimension, one channel
         :param channels:    channel name or symbol list, or a single name/symbol.  If a single name is specified
-                            for multi-column data, a VA channel is assumed.
+                            for multi-column data, a VA channel is assumed. If None, a sorted list of all channels
+                            is assumed.
         :param fid:         option fid tuple (start, increment), default (0.0,1.0)
 
         .. versionadded:: 9.1
@@ -1767,9 +1774,28 @@ def write_line(self, line, data, channels=None, fid=(0.0, 1.0)):
             self.write_channel(line, channels, data, fid=fid)
 
         else:
+
             if channels is None:
                 channels = self._sorted_chan_list()
 
+            if data.ndim == 1:
+                data = data.reshape((-1, 1))
+
+            # ensure data matches channels
+            np_data = 0
+            for chan in channels:
+                try:
+                    ch, cs = self.channel_name_symb(chan)
+                    w = self.channel_width(cs)
+                except GdbException:
+                    w = 1
+                np_data += w
+
+            # channel - data mismatch
+            if data.shape[1] != np_data:
+                raise GdbException(_t('Data dimension ({}) does not match data required by channels ({}).').format(data.shape, channels))
+
+            # all good, write the data
             np_index = 0
             for chan in channels:
                 try:
@@ -1781,10 +1807,6 @@ def write_line(self, line, data, channels=None, fid=(0.0, 1.0)):
                 self.write_channel(line, cs, data[:, np_index: np_index + w], fid=fid)
                 np_index += w
 
-            # error if there is any data left
-            if np_index - data.shape[1] != 0:
-                raise GdbException(_t('More data than channels, but data up to channels was written out.'))
-
     def list_values(self, chan, max=1000, selected=True, dupl=50, progress=None, stop=None):
         """
         Build a list of unique values in a channel.  Uniqueness depends on the current display format for

geosoft/gxpy/tests/test_gdb.py

@@ -777,7 +777,8 @@ def test_newline_GDB(self):
                 pass
 
             ch = ['a','b','c','d','e']
-            gdb.write_line('testline',npd,channels=ch)
+            self.assertRaises(gxdb.GdbException, gdb.write_line, 'testline', npd, ['xx', 'yy'])
+            gdb.write_line('testline', npd, channels=ch)
             npd2, ch2, fid2 = gdb.read_line('testline',channels=ch)
             self.assertEqual(npd.shape,npd2.shape)
             self.assertEqual(ch2, ch)
@@ -1185,32 +1186,44 @@ def test_dup(self):
             finally:
                 gdb.discard()
 
-    @unittest.skip('This test shows throws error now, skipping to let fixture pass')
-    def test_large_va(self):
+    @unittest.skip('skipping to let fixture pass')
+    def test_large_stress(self):
         self.start()
 
         try:
             name = None
-            with gxdb.Geosoft_gdb.new('new', overwrite=True) as gdb:
+            with gxdb.Geosoft_gdb.new('new', overwrite=True, comp=gxdb.COMP_NONE, pageSize=64) as gdb:
                 name = gdb.file_name
-                npd = np.empty((2000000, 3))  # TODO - jacques, this works - see next one
-                npd[:, :] = np.nan
+                npd = np.zeros(1000000) #TODO, this is 8 meg of data. This should not fit in 4 meg.
                 line = gdb.new_line('test')
-                gxdb.Channel.new(gdb, 'xx', array=3)
-                gdb.write_line(line, npd, ['xx[0]', 'xx[1]', 'xx[2]'])
+                gdb.write_line(line, npd, ['xx'])
                 npd2, ch, fid = gdb.read_line(line)
-                self.assertEqual(len(ch), 3, npd.shape)
+                self.assertEqual(len(ch), 1)
+
+            with gxdb.Geosoft_gdb.new('new', overwrite=True, comp=gxdb.COMP_NONE, pageSize=64) as gdb:
+                name = gdb.file_name
+                npd = np.zeros(4000000) #TODO, this is 32 meg of data, reported as 4.11 meg in the error?
+                line = gdb.new_line('test')
+                gdb.write_line(line, npd, ['xx'])
+                npd2, ch, fid = gdb.read_line(line)
+                self.assertEqual(len(ch), 1)
 
+        finally:
+            gxdb.delete_files(name)
+
+    @unittest.skip('skipping to let fixture pass')
+    def test_very_large_stress(self):
+        self.start()
+
+        try:
             name = None
-            with gxdb.Geosoft_gdb.new('new', overwrite=True) as gdb:
+            with gxdb.Geosoft_gdb.new('new', overwrite=True, comp=gxdb.COMP_NONE, pageSize=64) as gdb:
                 name = gdb.file_name
-                npd = np.empty((20000000, 3))  # TODO - this bigger one fails, but not nicely (see above, which works), same as problem reported in forum
-                npd[:, :] = np.nan
+                npd = np.zeros(1000000000)  # TODO, this should NOT work. On one test I got an indexing error
                 line = gdb.new_line('test')
-                gxdb.Channel.new(gdb, 'xx', array=3) #TODO failure is when we attempt to write. Symbols are OK, but core reports invalid symbol (-1)
-                gdb.write_line(line, npd, ['xx[0]', 'xx[1]', 'xx[2]'])
+                gdb.write_line(line, npd, ['xx'])
                 npd2, ch, fid = gdb.read_line(line)
-                self.assertEqual(len(ch), 3, npd.shape)
+                self.assertEqual(len(ch), 1)
 
         finally:
             gxdb.delete_files(name)