guard against large negative marks in grids

Author: gei-rusty

stompy/grid/unstructured_grid.py

@@ -2726,7 +2726,7 @@ def renumber_cells_ordering(self):
         Nactive = sum(~self.cells['deleted'])
         return np.argsort( self.cells['deleted'],kind='mergesort')[:Nactive]
 
-    def renumber_cells(self,order=None):
+    def renumber_cells(self,order=None,min_edge_mark=-2000):
         """
         Renumber cell indices, dropping deleted cells.
         Update edges['cells'], preserving negative values (e.g.
@@ -2735,12 +2735,16 @@ def renumber_cells(self,order=None):
         to get new cell indices.  Cell map is actually slightly
         larger than the number of old cells, to accomodate negative
         indexing.  For example, cell_map[-2]=-2
+        min_edge_mark: limits negative indexing, in case int.minValue
+        or something absurd comes in.
         """
         if order is None:
             csort = self.renumber_cells_ordering()
         else:
-            csort= order
-        Nneg=-min(-1,self.edges['cells'].min())
+            csort = order
+
+        min_mark = self.edges['cells'].min().clip(min_edge_mark,-1)
+        Nneg=-min_mark
         cell_map = np.zeros(self.Ncells()+Nneg,np.int32) # do this before truncating cells
         self.cells = self.cells[csort]
 
@@ -2751,14 +2755,14 @@ def renumber_cells(self,order=None):
         # or cell_map[csort[a]] = a
         # for all a, so
         # cell_map[csort[arange(Ncells)]] = arange(Ncells)
-        cell_map[:] = -999 # these should only remain for deleted cells, and never show up in the output
+        cell_map[:] = min_edge_mark # these should only remain for deleted cells, and never show up in the output
         cell_map[:-Nneg][csort] = np.arange(self.Ncells())
         # cell_map[-1] = -1 # land edges map cell -1 to -1
         # allow broader range of negatives:
         # map cell -1 to -1, -2 to -2, etc.
         cell_map[-Nneg:] = np.arange(-Nneg,0)
 
-        self.edges['cells'] = cell_map[self.edges['cells']]
+        self.edges['cells'] = cell_map[self.edges['cells'].clip(min_mark,None)]
         self._cell_center_index=None
         return cell_map
 
@@ -3566,7 +3570,7 @@ def cells_centroid_shapely(self,ids=None):
 
         for ci,c in enumerate(ids):
             if not self.cells['deleted'][c]:
-                centroids[ci]= np.array(self.cell_polygon(c).centroid)
+                centroids[ci]= np.array(self.cell_polygon(c).centroid.coords)
         return centroids
 
     def cells_centroid_py(self,ids=None):