more robust csv tests

tests/data/csv/empty_rows.csv

@@ -1,4 +1,4 @@
-x,y,text,date,integer,boolean,float,time,datetime,empty_column
+"x","y","text","date","integer","boolean","float","time","datetime","empty_column"
 0,0,a b,1971-01-01,40,True,1.0,04:14:00,1971-01-01T04:14:00,
 0,0,c d,1948-01-01,63,True,1.27,14:57:13,1948-01-01T14:57:13,
 0,0,e f,1920-01-01,164,False,41800000.01,00:00:00,1920-01-01T00:00:00,

tests/python_tests/csv_test.py

@@ -15,6 +15,9 @@ def setup():
 
 if 'csv' in mapnik2.DatasourceCache.instance().plugin_names():
 
+    def get_csv_ds(filename):
+        return mapnik2.Datasource(type='csv',file=os.path.join('../data/csv/',filename),quiet=True)
+
     def test_broken_files(visual=False):
         broken = glob.glob("../data/csv/fails/*.*")
         broken.extend(glob.glob("../data/csv/warns/*.*"))
@@ -43,6 +46,71 @@ if 'csv' in mapnik2.DatasourceCache.instance().plugin_names():
                 except Exception:
                     print '\x1b[33mfailed\x1b[0m',csv
     
+    def test_type_detection(**kwargs):
+        ds = get_csv_ds('nypd.csv')
+        eq_(ds.fields(),['Precinct','Phone','Address','City','geo_longitude','geo_latitude','geo_accuracy'])
+        eq_(ds.field_types(),['str','str','str','str','float','float','str'])
+        feat = ds.featureset().next()
+        attr = {'City': u'New York, NY', 'geo_accuracy': u'house', 'Phone': u'(212) 334-0711', 'Address': u'19 Elizabeth Street', 'Precinct': u'5th Precinct', 'geo_longitude': -70, 'geo_latitude': 40}
+        eq_(feat.attributes,attr)
+        eq_(len(ds.all_features()),2)
+
+    def test_skipping_blank_rows(**kwargs):
+        ds = get_csv_ds('blank_rows.csv')
+        eq_(ds.fields(),['x','y','name'])
+        eq_(ds.field_types(),['int','int','str'])
+        eq_(len(ds.all_features()),2)
+
+    def test_empty_rows(**kwargs):
+        ds = get_csv_ds('empty_rows.csv')
+        eq_(len(ds.fields()),10)
+        eq_(len(ds.field_types()),10)
+        eq_(ds.fields(),['x', 'y', 'text', 'date', 'integer', 'boolean', 'float', 'time', 'datetime', 'empty_column'])
+        eq_(ds.field_types(),['int', 'int', 'str', 'str', 'int', 'bool', 'float', 'str', 'str', 'str'])
+        fs = ds.featureset()
+        feat = fs.next()
+        attr = {'x': 0, 'empty_column': None, 'text': u'a b', 'float': 1.0, 'datetime': u'1971-01-01T04:14:00', 'y': 0, 'boolean': True, 'time': u'04:14:00', 'date': u'1971-01-01', 'integer': 40}
+        eq_(feat.attributes,attr)
+        while feat:
+            eq_(len(feat),10)
+            eq_(feat['empty_column'],None)
+            feat = fs.next()
+    
+    def test_slashes(**kwargs):
+        ds = get_csv_ds('has_attributes_with_slashes.csv')
+        eq_(len(ds.fields()),3)
+        fs = ds.all_features()
+        eq_(fs[0].attributes,{'x':0,'y':0,'name':u'a/a'})
+        eq_(fs[1].attributes,{'x':1,'y':4,'name':u'b/b'})
+        eq_(fs[2].attributes,{'x':10,'y':2.5,'name':u'c/c'})
+
+    def test_wkt_field(**kwargs):
+        ds = get_csv_ds('wkt.csv')
+        eq_(len(ds.fields()),2)
+        eq_(ds.fields(),['type','WKT'])
+        eq_(ds.field_types(),['str','str'])
+        fs = ds.all_features()
+        #import pdb;pdb.set_trace()
+        eq_(len(fs[0].geometries()),1)
+        eq_(fs[0].geometries()[0].type(),mapnik2.GeometryType.Point)
+        eq_(len(fs[1].geometries()),1)
+        eq_(fs[1].geometries()[0].type(),mapnik2.GeometryType.LineString)
+        eq_(len(fs[2].geometries()),1)
+        eq_(fs[2].geometries()[0].type(),mapnik2.GeometryType.Polygon)
+        eq_(len(fs[3].geometries()),1) # one geometry, two parts
+        eq_(fs[3].geometries()[0].type(),mapnik2.GeometryType.Polygon)
+        # tests assuming we want to flatten geometries
+        eq_(len(fs[4].geometries()),4)
+        eq_(fs[4].geometries()[0].type(),mapnik2.GeometryType.Point)
+        eq_(len(fs[5].geometries()),2)
+        eq_(fs[5].geometries()[0].type(),mapnik2.GeometryType.LineString)
+        eq_(len(fs[6].geometries()),2)
+        eq_(fs[6].geometries()[0].type(),mapnik2.GeometryType.Polygon)
+        eq_(len(fs[7].geometries()),2)
+        eq_(fs[7].geometries()[0].type(),mapnik2.GeometryType.Polygon)
+        
+        # ideally we should not have to:
+        # https://github.com/mapnik/mapnik/issues?labels=multigeom+robustness&sort=created&direction=desc&state=open&page=1
 
 if __name__ == "__main__":
     setup()