1 # LayerMapping -- A Django Model/OGR Layer Mapping Utility
3 The LayerMapping class provides a way to map the contents of OGR
4 vector files (e.g. SHP files) to Geographic-enabled Django models.
6 For more information, please consult the GeoDjango documentation:
7 http://geodjango.org/docs/layermapping.html
10 from datetime
import date
, datetime
11 from decimal
import Decimal
12 from django
.core
.exceptions
import ObjectDoesNotExist
13 from django
.db
import connections
, DEFAULT_DB_ALIAS
14 from django
.contrib
.gis
.db
.models
import GeometryField
15 from django
.contrib
.gis
.gdal
import CoordTransform
, DataSource
, \
16 OGRException
, OGRGeometry
, OGRGeomType
, SpatialReference
17 from django
.contrib
.gis
.gdal
.field
import \
18 OFTDate
, OFTDateTime
, OFTInteger
, OFTReal
, OFTString
, OFTTime
19 from django
.db
import models
, transaction
20 from django
.contrib
.localflavor
.us
.models
import USStateField
22 # LayerMapping exceptions.
23 class LayerMapError(Exception): pass
24 class InvalidString(LayerMapError
): pass
25 class InvalidDecimal(LayerMapError
): pass
26 class InvalidInteger(LayerMapError
): pass
27 class MissingForeignKey(LayerMapError
): pass
29 class LayerMapping(object):
30 "A class that maps OGR Layers to GeoDjango Models."
32 # Acceptable 'base' types for a multi-geometry type.
33 MULTI_TYPES
= {1 : OGRGeomType('MultiPoint'),
34 2 : OGRGeomType('MultiLineString'),
35 3 : OGRGeomType('MultiPolygon'),
36 OGRGeomType('Point25D').num
: OGRGeomType('MultiPoint25D'),
37 OGRGeomType('LineString25D').num
: OGRGeomType('MultiLineString25D'),
38 OGRGeomType('Polygon25D').num
: OGRGeomType('MultiPolygon25D'),
41 # Acceptable Django field types and corresponding acceptable OGR
44 models
.AutoField
: OFTInteger
,
45 models
.IntegerField
: (OFTInteger
, OFTReal
, OFTString
),
46 models
.FloatField
: (OFTInteger
, OFTReal
),
47 models
.DateField
: OFTDate
,
48 models
.DateTimeField
: OFTDateTime
,
49 models
.EmailField
: OFTString
,
50 models
.TimeField
: OFTTime
,
51 models
.DecimalField
: (OFTInteger
, OFTReal
),
52 models
.CharField
: OFTString
,
53 models
.SlugField
: OFTString
,
54 models
.TextField
: OFTString
,
55 models
.URLField
: OFTString
,
56 USStateField
: OFTString
,
57 models
.XMLField
: OFTString
,
58 models
.SmallIntegerField
: (OFTInteger
, OFTReal
, OFTString
),
59 models
.PositiveSmallIntegerField
: (OFTInteger
, OFTReal
, OFTString
),
62 # The acceptable transaction modes.
63 TRANSACTION_MODES
= {'autocommit' : transaction
.autocommit
,
64 'commit_on_success' : transaction
.commit_on_success
,
67 def __init__(self
, model
, data
, mapping
, layer
=0,
68 source_srs
=None, encoding
=None,
69 transaction_mode
='commit_on_success',
70 transform
=True, unique
=None, using
=DEFAULT_DB_ALIAS
):
72 A LayerMapping object is initialized using the given Model (not an instance),
73 a DataSource (or string path to an OGR-supported data file), and a mapping
74 dictionary. See the module level docstring for more details and keyword
77 # Getting the DataSource and the associated Layer.
78 if isinstance(data
, basestring
):
79 self
.ds
= DataSource(data
)
82 self
.layer
= self
.ds
[layer
]
85 self
.spatial_backend
= connections
[using
].ops
87 # Setting the mapping & model attributes.
88 self
.mapping
= mapping
91 # Checking the layer -- intitialization of the object will fail if
92 # things don't check out before hand.
95 # Getting the geometry column associated with the model (an
96 # exception will be raised if there is no geometry column).
97 if self
.spatial_backend
.mysql
:
100 self
.geo_field
= self
.geometry_field()
102 # Checking the source spatial reference system, and getting
103 # the coordinate transformation object (unless the `transform`
104 # keyword is set to False)
106 self
.source_srs
= self
.check_srs(source_srs
)
107 self
.transform
= self
.coord_transform()
109 self
.transform
= transform
111 # Setting the encoding for OFTString fields, if specified.
113 # Making sure the encoding exists, if not a LookupError
114 # exception will be thrown.
115 from codecs
import lookup
117 self
.encoding
= encoding
122 self
.check_unique(unique
)
123 transaction_mode
= 'autocommit' # Has to be set to autocommit.
128 # Setting the transaction decorator with the function in the
129 # transaction modes dictionary.
130 if transaction_mode
in self
.TRANSACTION_MODES
:
131 self
.transaction_decorator
= self
.TRANSACTION_MODES
[transaction_mode
]
132 self
.transaction_mode
= transaction_mode
134 raise LayerMapError('Unrecognized transaction mode: %s' % transaction_mode
)
139 #### Checking routines used during initialization ####
140 def check_fid_range(self
, fid_range
):
141 "This checks the `fid_range` keyword."
143 if isinstance(fid_range
, (tuple, list)):
144 return slice(*fid_range
)
145 elif isinstance(fid_range
, slice):
152 def check_layer(self
):
154 This checks the Layer metadata, and ensures that it is compatible
155 with the mapping information and model. Unlike previous revisions,
156 there is no need to increment through each feature in the Layer.
158 # The geometry field of the model is set here.
159 # TODO: Support more than one geometry field / model. However, this
160 # depends on the GDAL Driver in use.
161 self
.geom_field
= False
164 # Getting lists of the field names and the field types available in
166 ogr_fields
= self
.layer
.fields
167 ogr_field_types
= self
.layer
.field_types
169 # Function for determining if the OGR mapping field is in the Layer.
170 def check_ogr_fld(ogr_map_fld
):
172 idx
= ogr_fields
.index(ogr_map_fld
)
174 raise LayerMapError('Given mapping OGR field "%s" not found in OGR Layer.' % ogr_map_fld
)
177 # No need to increment through each feature in the model, simply check
178 # the Layer metadata against what was given in the mapping dictionary.
179 for field_name
, ogr_name
in self
.mapping
.items():
180 # Ensuring that a corresponding field exists in the model
181 # for the given field name in the mapping.
183 model_field
= self
.model
._meta
.get_field(field_name
)
184 except models
.fields
.FieldDoesNotExist
:
185 raise LayerMapError('Given mapping field "%s" not in given Model fields.' % field_name
)
187 # Getting the string name for the Django field class (e.g., 'PointField').
188 fld_name
= model_field
.__class
__.__name
__
190 if isinstance(model_field
, GeometryField
):
192 raise LayerMapError('LayerMapping does not support more than one GeometryField per model.')
194 # Getting the coordinate dimension of the geometry field.
195 coord_dim
= model_field
.dim
199 gtype
= OGRGeomType(ogr_name
+ '25D')
201 gtype
= OGRGeomType(ogr_name
)
203 raise LayerMapError('Invalid mapping for GeometryField "%s".' % field_name
)
205 # Making sure that the OGR Layer's Geometry is compatible.
206 ltype
= self
.layer
.geom_type
207 if not (ltype
.name
.startswith(gtype
.name
) or self
.make_multi(ltype
, model_field
)):
208 raise LayerMapError('Invalid mapping geometry; model has %s%s, '
209 'layer geometry type is %s.' %
210 (fld_name
, (coord_dim
== 3 and '(dim=3)') or '', ltype
))
212 # Setting the `geom_field` attribute w/the name of the model field
213 # that is a Geometry. Also setting the coordinate dimension
215 self
.geom_field
= field_name
216 self
.coord_dim
= coord_dim
217 fields_val
= model_field
218 elif isinstance(model_field
, models
.ForeignKey
):
219 if isinstance(ogr_name
, dict):
220 # Is every given related model mapping field in the Layer?
221 rel_model
= model_field
.rel
.to
222 for rel_name
, ogr_field
in ogr_name
.items():
223 idx
= check_ogr_fld(ogr_field
)
225 rel_field
= rel_model
._meta
.get_field(rel_name
)
226 except models
.fields
.FieldDoesNotExist
:
227 raise LayerMapError('ForeignKey mapping field "%s" not in %s fields.' %
228 (rel_name
, rel_model
.__class
__.__name
__))
229 fields_val
= rel_model
231 raise TypeError('ForeignKey mapping must be of dictionary type.')
233 # Is the model field type supported by LayerMapping?
234 if not model_field
.__class
__ in self
.FIELD_TYPES
:
235 raise LayerMapError('Django field type "%s" has no OGR mapping (yet).' % fld_name
)
237 # Is the OGR field in the Layer?
238 idx
= check_ogr_fld(ogr_name
)
239 ogr_field
= ogr_field_types
[idx
]
241 # Can the OGR field type be mapped to the Django field type?
242 if not issubclass(ogr_field
, self
.FIELD_TYPES
[model_field
.__class
__]):
243 raise LayerMapError('OGR field "%s" (of type %s) cannot be mapped to Django %s.' %
244 (ogr_field
, ogr_field
.__name
__, fld_name
))
245 fields_val
= model_field
247 self
.fields
[field_name
] = fields_val
249 def check_srs(self
, source_srs
):
250 "Checks the compatibility of the given spatial reference object."
252 if isinstance(source_srs
, SpatialReference
):
254 elif isinstance(source_srs
, self
.spatial_backend
.spatial_ref_sys()):
256 elif isinstance(source_srs
, (int, basestring
)):
257 sr
= SpatialReference(source_srs
)
259 # Otherwise just pulling the SpatialReference from the layer
263 raise LayerMapError('No source reference system defined.')
267 def check_unique(self
, unique
):
268 "Checks the `unique` keyword parameter -- may be a sequence or string."
269 if isinstance(unique
, (list, tuple)):
270 # List of fields to determine uniqueness with
272 if not attr
in self
.mapping
: raise ValueError
273 elif isinstance(unique
, basestring
):
274 # Only a single field passed in.
275 if unique
not in self
.mapping
: raise ValueError
277 raise TypeError('Unique keyword argument must be set with a tuple, list, or string.')
279 #### Keyword argument retrieval routines ####
280 def feature_kwargs(self
, feat
):
282 Given an OGR Feature, this will return a dictionary of keyword arguments
283 for constructing the mapped model.
285 # The keyword arguments for model construction.
288 # Incrementing through each model field and OGR field in the
289 # dictionary mapping.
290 for field_name
, ogr_name
in self
.mapping
.items():
291 model_field
= self
.fields
[field_name
]
293 if isinstance(model_field
, GeometryField
):
294 # Verify OGR geometry.
295 val
= self
.verify_geom(feat
.geom
, model_field
)
296 elif isinstance(model_field
, models
.base
.ModelBase
):
297 # The related _model_, not a field was passed in -- indicating
298 # another mapping for the related Model.
299 val
= self
.verify_fk(feat
, model_field
, ogr_name
)
301 # Otherwise, verify OGR Field type.
302 val
= self
.verify_ogr_field(feat
[ogr_name
], model_field
)
304 # Setting the keyword arguments for the field name with the
305 # value obtained above.
306 kwargs
[field_name
] = val
310 def unique_kwargs(self
, kwargs
):
312 Given the feature keyword arguments (from `feature_kwargs`) this routine
313 will construct and return the uniqueness keyword arguments -- a subset
314 of the feature kwargs.
316 if isinstance(self
.unique
, basestring
):
317 return {self
.unique
: kwargs
[self
.unique
]}
319 return dict((fld
, kwargs
[fld
]) for fld
in self
.unique
)
321 #### Verification routines used in constructing model keyword arguments. ####
322 def verify_ogr_field(self
, ogr_field
, model_field
):
324 Verifies if the OGR Field contents are acceptable to the Django
325 model field. If they are, the verified value is returned,
326 otherwise the proper exception is raised.
328 if (isinstance(ogr_field
, OFTString
) and
329 isinstance(model_field
, (models
.CharField
, models
.TextField
))):
331 # The encoding for OGR data sources may be specified here
332 # (e.g., 'cp437' for Census Bureau boundary files).
333 val
= unicode(ogr_field
.value
, self
.encoding
)
335 val
= ogr_field
.value
336 if len(val
) > model_field
.max_length
:
337 raise InvalidString('%s model field maximum string length is %s, given %s characters.' %
338 (model_field
.name
, model_field
.max_length
, len(val
)))
339 elif isinstance(ogr_field
, OFTReal
) and isinstance(model_field
, models
.DecimalField
):
341 # Creating an instance of the Decimal value to use.
342 d
= Decimal(str(ogr_field
.value
))
344 raise InvalidDecimal('Could not construct decimal from: %s' % ogr_field
.value
)
346 # Getting the decimal value as a tuple.
349 d_idx
= dtup
[2] # index where the decimal is
351 # Maximum amount of precision, or digits to the left of the decimal.
352 max_prec
= model_field
.max_digits
- model_field
.decimal_places
354 # Getting the digits to the left of the decimal place for the
357 n_prec
= len(digits
[:d_idx
])
359 n_prec
= len(digits
) + d_idx
361 # If we have more than the maximum digits allowed, then throw an
362 # InvalidDecimal exception.
363 if n_prec
> max_prec
:
364 raise InvalidDecimal('A DecimalField with max_digits %d, decimal_places %d must round to an absolute value less than 10^%d.' %
365 (model_field
.max_digits
, model_field
.decimal_places
, max_prec
))
367 elif isinstance(ogr_field
, (OFTReal
, OFTString
)) and isinstance(model_field
, models
.IntegerField
):
368 # Attempt to convert any OFTReal and OFTString value to an OFTInteger.
370 val
= int(ogr_field
.value
)
372 raise InvalidInteger('Could not construct integer from: %s' % ogr_field
.value
)
374 val
= ogr_field
.value
377 def verify_fk(self
, feat
, rel_model
, rel_mapping
):
379 Given an OGR Feature, the related model and its dictionary mapping,
380 this routine will retrieve the related model for the ForeignKey
383 # TODO: It is expensive to retrieve a model for every record --
384 # explore if an efficient mechanism exists for caching related
387 # Constructing and verifying the related model keyword arguments.
389 for field_name
, ogr_name
in rel_mapping
.items():
390 fk_kwargs
[field_name
] = self
.verify_ogr_field(feat
[ogr_name
], rel_model
._meta
.get_field(field_name
))
392 # Attempting to retrieve and return the related model.
394 return rel_model
.objects
.get(**fk_kwargs
)
395 except ObjectDoesNotExist
:
396 raise MissingForeignKey('No ForeignKey %s model found with keyword arguments: %s' % (rel_model
.__name
__, fk_kwargs
))
398 def verify_geom(self
, geom
, model_field
):
400 Verifies the geometry -- will construct and return a GeometryCollection
401 if necessary (for example if the model field is MultiPolygonField while
402 the mapped shapefile only contains Polygons).
404 # Downgrade a 3D geom to a 2D one, if necessary.
405 if self
.coord_dim
!= geom
.coord_dim
:
406 geom
.coord_dim
= self
.coord_dim
408 if self
.make_multi(geom
.geom_type
, model_field
):
409 # Constructing a multi-geometry type to contain the single geometry
410 multi_type
= self
.MULTI_TYPES
[geom
.geom_type
.num
]
411 g
= OGRGeometry(multi_type
)
416 # Transforming the geometry with our Coordinate Transformation object,
417 # but only if the class variable `transform` is set w/a CoordTransform
419 if self
.transform
: g
.transform(self
.transform
)
421 # Returning the WKT of the geometry.
424 #### Other model methods ####
425 def coord_transform(self
):
426 "Returns the coordinate transformation object."
427 SpatialRefSys
= self
.spatial_backend
.spatial_ref_sys()
429 # Getting the target spatial reference system
430 target_srs
= SpatialRefSys
.objects
.get(srid
=self
.geo_field
.srid
).srs
432 # Creating the CoordTransform object
433 return CoordTransform(self
.source_srs
, target_srs
)
434 except Exception, msg
:
435 raise LayerMapError('Could not translate between the data source and model geometry: %s' % msg
)
437 def geometry_field(self
):
438 "Returns the GeometryField instance associated with the geographic column."
439 # Use the `get_field_by_name` on the model's options so that we
440 # get the correct field instance if there's model inheritance.
441 opts
= self
.model
._meta
442 fld
, model
, direct
, m2m
= opts
.get_field_by_name(self
.geom_field
)
445 def make_multi(self
, geom_type
, model_field
):
447 Given the OGRGeomType for a geometry and its associated GeometryField,
448 determine whether the geometry should be turned into a GeometryCollection.
450 return (geom_type
.num
in self
.MULTI_TYPES
and
451 model_field
.__class
__.__name
__ == 'Multi%s' % geom_type
.django
)
453 def save(self
, verbose
=False, fid_range
=False, step
=False,
454 progress
=False, silent
=False, stream
=sys
.stdout
, strict
=False):
456 Saves the contents from the OGR DataSource Layer into the database
457 according to the mapping dictionary given at initialization.
461 If set, information will be printed subsequent to each model save
462 executed on the database.
465 May be set with a slice or tuple of (begin, end) feature ID's to map
466 from the data source. In other words, this keyword enables the user
467 to selectively import a subset range of features in the geographic
471 If set with an integer, transactions will occur at every step
472 interval. For example, if step=1000, a commit would occur after
473 the 1,000th feature, the 2,000th feature etc.
476 When this keyword is set, status information will be printed giving
477 the number of features processed and sucessfully saved. By default,
478 progress information will pe printed every 1000 features processed,
479 however, this default may be overridden by setting this keyword with an
480 integer for the desired interval.
483 Status information will be written to this file handle. Defaults to
484 using `sys.stdout`, but any object with a `write` method is supported.
487 By default, non-fatal error notifications are printed to stdout, but
488 this keyword may be set to disable these notifications.
491 Execution of the model mapping will cease upon the first error
492 encountered. The default behavior is to attempt to continue.
494 # Getting the default Feature ID range.
495 default_range
= self
.check_fid_range(fid_range
)
497 # Setting the progress interval, if requested.
499 if progress
is True or not isinstance(progress
, int):
500 progress_interval
= 1000
502 progress_interval
= progress
504 # Defining the 'real' save method, utilizing the transaction
505 # decorator created during initialization.
506 @self.transaction_decorator
507 def _save(feat_range
=default_range
, num_feat
=0, num_saved
=0):
509 layer_iter
= self
.layer
[feat_range
]
511 layer_iter
= self
.layer
513 for feat
in layer_iter
:
515 # Getting the keyword arguments
517 kwargs
= self
.feature_kwargs(feat
)
518 except LayerMapError
, msg
:
519 # Something borked the validation
522 stream
.write('Ignoring Feature ID %s because: %s\n' % (feat
.fid
, msg
))
524 # Constructing the model using the keyword args
527 # If we want unique models on a particular field, handle the
528 # geometry appropriately.
530 # Getting the keyword arguments and retrieving
532 u_kwargs
= self
.unique_kwargs(kwargs
)
533 m
= self
.model
.objects
.using(self
.using
).get(**u_kwargs
)
536 # Getting the geometry (in OGR form), creating
537 # one from the kwargs WKT, adding in additional
538 # geometries, and update the attribute with the
539 # just-updated geometry WKT.
540 geom
= getattr(m
, self
.geom_field
).ogr
541 new
= OGRGeometry(kwargs
[self
.geom_field
])
542 for g
in new
: geom
.add(g
)
543 setattr(m
, self
.geom_field
, geom
.wkt
)
544 except ObjectDoesNotExist
:
545 # No unique model exists yet, create.
546 m
= self
.model(**kwargs
)
548 m
= self
.model(**kwargs
)
551 # Attempting to save.
552 m
.save(using
=self
.using
)
554 if verbose
: stream
.write('%s: %s\n' % (is_update
and 'Updated' or 'Saved', m
))
557 except Exception, msg
:
558 if self
.transaction_mode
== 'autocommit':
559 # Rolling back the transaction so that other model saves
561 transaction
.rollback_unless_managed()
563 # Bailing out if the `strict` keyword is set.
565 stream
.write('Failed to save the feature (id: %s) into the model with the keyword arguments:\n' % feat
.fid
)
566 stream
.write('%s\n' % kwargs
)
569 stream
.write('Failed to save %s:\n %s\nContinuing\n' % (kwargs
, msg
))
571 # Printing progress information, if requested.
572 if progress
and num_feat
% progress_interval
== 0:
573 stream
.write('Processed %d features, saved %d ...\n' % (num_feat
, num_saved
))
575 # Only used for status output purposes -- incremental saving uses the
576 # values returned here.
577 return num_saved
, num_feat
579 nfeat
= self
.layer
.num_feat
580 if step
and isinstance(step
, int) and step
< nfeat
:
581 # Incremental saving is requested at the given interval (step)
583 raise LayerMapError('The `step` keyword may not be used in conjunction with the `fid_range` keyword.')
584 beg
, num_feat
, num_saved
= (0, 0, 0)
585 indices
= range(step
, nfeat
, step
)
588 for i
, end
in enumerate(indices
):
589 # Constructing the slice to use for this step; the last slice is
590 # special (e.g, [100:] instead of [90:100]).
591 if i
+1 == n_i
: step_slice
= slice(beg
, None)
592 else: step_slice
= slice(beg
, end
)
595 num_feat
, num_saved
= _save(step_slice
, num_feat
, num_saved
)
598 stream
.write('%s\nFailed to save slice: %s\n' % ('=-' * 20, step_slice
))
601 # Otherwise, just calling the previously defined _save() function.