thuban 1.2.2-5 / usr / lib / thuban / Extensions / ogr / ogrshapes.py

# Copyright (C) 2004, 2007 by Intevation GmbH     vim:encoding=latin-1:
# Authors:
# Nina Hüffmeyer <nhueffme@intevation.de>
#
# This program is free software under the GPL (>=v2)
# Read the file COPYING coming with the software for details.

from __future__ import generators

__version__ = "$Revision: 2770 $"
# $Source$
# $Id: ogrshapes.py 2770 2007-05-22 08:54:41Z bh $

try:
    from osgeo import ogr
except ImportError:
    ogr = None

import os

from Thuban import _
from Thuban.Model import table
from Thuban.Model import transientdb
from Thuban.Model.transientdb import  TransientDatabase

from Thuban.Model.data import SHAPETYPE_POLYGON, SHAPETYPE_ARC, SHAPETYPE_POINT
from Thuban.Model.data import RAW_PYTHON, RAW_SHAPEFILE, RAW_WKT
from Thuban.Model.data import FileShapeStore

def has_ogr_support():
    """Return whether this Thuban instance supports ogr file formats

    Having OGR support means that the ogr module can be
    imported.
    """
    return ogr is not None

if ogr is not None:
    SHAPETYPE_UNKNOWN = ogr.wkbUnknown
    SHAPETYPE_GEOMCOLL = ogr.wkbGeometryCollection
    SHAPETYPE_NONE = ogr.wkbNone

    # mapping from ogr-lib shapetypes and table constants to our constants
    ogrlib_shapetypes = {ogr.wkbPolygon: SHAPETYPE_POLYGON,
                ogr.wkbLineString: SHAPETYPE_ARC,
                ogr.wkbPoint: SHAPETYPE_POINT,
                ogr.wkbUnknown: SHAPETYPE_UNKNOWN,
                ogr.wkbNone: SHAPETYPE_NONE,
                ogr.wkbGeometryCollection: SHAPETYPE_GEOMCOLL}

    fieldtype_map = {ogr.OFTString: table.FIELDTYPE_STRING,
                ogr.OFTInteger: table.FIELDTYPE_INT,
                ogr.OFTReal: table.FIELDTYPE_DOUBLE}

else:
    ogrlib_shapetypes = {}
    fieldtype_map = {}
    SHAPETYPE_UNKNOWN = 0
    SHAPETYPE_GEOMCOLL = 7
    SHAPETYPE_NONE = 100


class OGRShape:
    """Represent one shape of an OGRShapeStore
    """

    def __init__(self, shapestore, shape):
        """Initialize the shape object.

        shapestore should be an instance of OGRShapeStore,
        shape should be an instance of an OGRFeature.
        """
        self.ogrlayer = shapestore.ogrlayer
        id_column = shapestore.Id_column()
        self.feature = shape
        if id_column is None:
            self.shapeid = self.feature.GetFID()
        else:
            self.shapeid = self.feature.GetField(id_column)
        self.geom = self.feature.GetGeometryRef()
        if self.geom:
            self.shapetype = self.geom.GetGeometryType()
            self.bbox = self._compute_bbox()
            self.points = self._points()
        else:
            self.shapetype = ogr.wkbNone
            self.bbox = None
            self.points = [[]]
        try:
            self.shapetype = ogrlib_shapetypes[self.shapetype]
        except:
            self.shapetype = ogrlib_shapetypes[ogr.wkbUnknown]

        self.geoms = self._geoms()

    def _geoms(self):
        """Return a list of geometry objects.

        If the shape is a geometry collection, all contained geometry
        objects are stored to the list as OGRGeometry objects.
        """
        geoms = []
        if self.shapetype == SHAPETYPE_GEOMCOLL:
            for i in range(self.geom.GetGeometryCount()):
                geoms.append(OGRGeometry(self, self.geom, i))
        return geoms

    def _compute_bbox(self):
        """
        Compute the bounding box of the shape as a tuple (minx,miny,maxx,maxy)
        """
        minx, maxx, miny, maxy = self.geom.GetEnvelope()
        return (minx, miny, maxx, maxy)

    def compute_bbox(self):
        """
        Return the bounding box of the shape as a tuple (minx,miny,maxx,maxy)
        """
        return self.bbox

    def ShapeID(self):
        """Return the feature id of this shape.
        """
        return self.shapeid

    def Points(self):
        """Return the coordinates of the shape as a list of lists of pairs
        """
        return self.points

    def _points(self):
        """Get the coordinates of the shape as a list of lists of pairs
        """
        shape = []

        if self.geom is None:
            return shape.append([])

        # if geometry object is of type point or line
        if self.geom.GetGeometryCount() == 0:
            points =[]
            for point in range(self.geom.GetPointCount()):
                x = self.geom.GetX(point)
                y = self.geom.GetY(point)
                points.append((x, y))
            return [points]
        # if geometry object is of type polygon or multipolygon
        for i in range(self.geom.GetGeometryCount()):
            points = []
            geometry = self.geom.GetGeometryRef(i)
            # if geometry object is polygon
            if geometry.GetGeometryCount() == 0:
                for point in range(geometry.GetPointCount()):
                    x = geometry.GetX(point)
                    y = geometry.GetY(point)
                    points.append((x, y))
                shape.append(points)
            # if geometry object is of type multipolygon or geometry collection
            else:
                for j in range(geometry.GetGeometryCount()):
                    points = []
                    subgeom = geometry.GetGeometryRef(j)
                    for point in range(subgeom.GetPointCount()):
                        x = subgeom.GetX(point)
                        y = subgeom.GetY(point)
                        points.append((x, y))
                    shape.append(points)
        return shape

    def RawData(self):
        """Return the shape id to use with the shapestore
        """
        return self.shapeid

    def OGRLayer(self):
        """Return the ogrlayer object
        """
        return self.ogrlayer

    def ShapeType(self):
        """Return the shapetype of this shape (may differ from the layer's
        shapetype)
        """
        return self.shapetype

    def GetGeoms(self):
        """Return the list of geometries of this feature.

        If this feature is a geometry collection, all contained geometries
        are given. Else the returned list is empty.
        """
        return self.geoms

    def GetGeom(self, index):
        """Return the OGRGeometry object at the specified index.

        This is not none only if the shape is a geometry collection.
        """
        if index < len(self.geoms):
            return self.geoms[index]
        else:
            return None


class OGRGeometry:
    """This class represents a geometry belonging to a specified feature.
    """

    def __init__(self, shape, geom, index):
        """Initialize the geometry object.

        shape should be an OGRShape, which this geometry belongs to.
        geom is the base geometry, index is the ReferenceID.
        """
        self.shape = shape
        self.index = index

        self.geom = geom.GetGeometryRef(index)
        try:
            self.shapetype = ogrlib_shapetypes[self.geom.GetGeometryType()]
        except:
            self.shapetype = ogrlib_shapetypes[ogr.wkbUnknown]


    def ShapeType(self):
        """Return the shapetype of this geometry object."""
        return self.shapetype


class OGRShapeStore:
    """Corresponds to an OGRLayer object, containing features/shapes and
       providing the same methods like ShapefileStore.

       In fact, for all file based shape stores, the class OGRFileShapeStore
       should be used. Only database shape stores should be
       used with OGRShapeStore. It is subject to re-factoring
       to end up with better class names and sensible base classes.
    """

    # TODO: re-factor this class to be not responsible for file-based
    # stores anymore.

    def __init__(self, session, filename, layername, id_column = None):
        """Initialize the shapestore.

        All required information is loaded from the datasource.
        """
        # if id_column is None, data is loaded from file, so we need path
        # if id_column is not None, data is loaded from database
        if id_column is None:
            self.filename = os.path.abspath(filename)
        else:
            self.filename = filename
        self.layername = layername

        self.ogrdatasource = ogr.Open(filename)
        self.ogrlayer = (self.ogrdatasource).GetLayerByName(layername)

        if id_column is not None:
            self.id_column = id_column
        else:
            self.id_column = None

        self.table = OGRTable(session, self.ogrdatasource, self.ogrlayer,
                                self.id_column)

        self._open_ogrlayer(layername)

    def _open_ogrlayer(self, layername):
        """Get all required information from the datasource.
        """
        self.numshapes = self.ogrlayer.GetFeatureCount()
        self.shapetype = self.ogrlayer.GetLayerDefn().GetGeomType()

        extent = self.ogrlayer.GetExtent()
        if extent:
            self.bbox = [extent[0], extent[2], extent[1], extent[3]]
        else:
            self.bbox = None

        try:
            self.shapetype = ogrlib_shapetypes[self.shapetype]
        except:
            # if shapetype is not contained in ogrlib_shapetypes
            # treat it like SHAPETYPE_UNKNOWN
            self.shapetype = ogrlib_shapetypes[ogr.wkbUnknown]

        self.shapes = self.shapes()

    def shapes(self):
        """Return a collection of all features as OGRShape objects.
        """
        shapes = {}
        self.ogrlayer.ResetReading()
        if self.id_column is None:
            nextFeature = self.ogrlayer.GetNextFeature()
            while nextFeature is not None:
                fid = nextFeature.GetFID()
                shape = OGRShape(self, nextFeature)
                shapes[shape.ShapeID()] = shape
                nextFeature = self.ogrlayer.GetNextFeature()
        else:
            lay = self.ogrdatasource.ExecuteSQL("SELECT %s, * from %s" 
                                    % (self.id_column, self.layername))
            if lay is not None:
                lay.ResetReading()
                nextFeature = lay.GetNextFeature()
                while nextFeature is not None:
                    fid = nextFeature.GetField(0)
                    shape = OGRShape(self, nextFeature)
                    shapes[shape.ShapeID()] = shape
                    nextFeature = lay.GetNextFeature()
                self.ogrdatasource.ReleaseResultSet(lay)
        return shapes

    def OGRLayer(self):
        """Return the OGRLayer object
        """
        return self.ogrlayer

    def FileName(self):
        """Return the filename used to open the file
        """
        return self.filename

    def FileType(self):
        """Return the filetype. This is depending on the driver used to open
        the file.
        """
        return self.ogrdatasource.GetDriver().GetName()

    def ShapeType(self):
        """Return the type of the shapes in the shapestore.

        This is either SHAPETYPE_POINT, SHAPETYPE_ARC, SHAPETYPE_POLYGON,
        SHAEPTYPE_GEOMCOLL, SHAPETYPE_NONE or SHAPETYPE_UNKNOWN.
        """
        return self.shapetype

    def RawShapeFormat(self):
        """Return the raw data format of the shape data, i.e. RAW_PYTHON
        """
        return RAW_PYTHON

    def NumShapes(self):
        """Return the number of shapes in the shape store
        """
        return self.numshapes

    def BoundingBox(self):
        """Return the bounding box of the shapes in the shapestore.
        """
        return self.bbox

    def ShapesInRegion(self, bbox):
        """Return an iterable over the shapes that overlap the bounding box.

        The bbox parameter should be the bounding box as a tuple in the
        form (minx, miny, maxx, maxy) in the coordinate system of the
        shape store.
        """
        left, bottom, right, top = bbox

        # create a geometry which can be passed to the layer as spatial filter
        bboxpolygon = ogr.CreateGeometryFromWkt(
                      ('Polygon((%s %s, %s %s, %s %s,%s %s, %s %s))'
                                 %(left, bottom, left, top, right, top,
                                  right, bottom, left, bottom)))

        if self.ogrlayer.GetSpatialRef():
            bboxpolygon.AssignSpatialReference(self.ogrlayer.GetSpatialRef())

        self.ogrlayer.ResetReading()
        #ogrlayer.SetSpatialFilterRect(left, bottom, right, top)
        self.ogrlayer.SetSpatialFilter(bboxpolygon)

        numFeatures = self.ogrlayer.GetFeatureCount()
        # if no features are in bbox, return all features as shapesInRegion
        # (PostGIS sometimes returns no features even if they are within 
        #  the bounding box)
        if numFeatures == 0:
            self.ogrlayer.SetSpatialFilter(None)
            numFeatures = self.ogrlayer.GetFeatureCount()
        for feature in range(numFeatures):
            nextFeature = self.ogrlayer.GetNextFeature()
            if self.id_column is None:
                yield self.shapes[nextFeature.GetFID()]
            else:
                yield self.shapes[nextFeature.GetField(self.id_column)]

        self.ogrlayer.SetSpatialFilter(None)
        bboxpolygon.Destroy()

    def AllShapes(self):
        """Return an iterable over the shapes in the shape store.
        """
        for id in range(len(self.shapes)):
            yield self.shapes[id]

    def Shape(self, fid):
        """Return the shape with fid = fid
        """
        if fid in self.table.ids.keys():
            return self.shapes[fid]
        else:
            return None

    def Table(self):
        """Return the table containing the attribute data
        """
        return self.table

    def Dependencies(self):
        """Return the empty tuple.
        """
        return ()

    def OrigShapeStore(self):
        """Return None."""
        return None

    def Id_column(self):
        """Return the id_column.
        """
        return self.id_column

class OGRFileShapeStore(FileShapeStore):
    """Corresponds to an OGRLayer object, containing features/shapes and
       providing the same methods like ShapefileStore.
    """

    def __init__(self, session, filename, layername, id_column = None):
        """Initialize the shapestore.

        All required information is loaded from the datasource.
        """
        self._bbox = None
        self.ogrdatasource = ogr.Open(filename)

        # filetype is depending on the driver used to open the file.
        self._filetype = self.ogrdatasource.GetDriver().GetName()
        if self._filetype == 'ESRI Shapefile':
            self._filetype = "shapefile"
        FileShapeStore.__init__(self, filename,
                                sublayer_name = layername)

        self.ogrlayer = (self.ogrdatasource).GetLayerByName(layername)

        self._table = OGRTable(session, self.ogrdatasource, self.ogrlayer,
                               id_column)

        self._open_ogrlayer(layername)

    def _open_ogrlayer(self, layername):
        """Get all required information from the datasource.
        """
        self.numshapes = self.ogrlayer.GetFeatureCount()
        self.shapetype = self.ogrlayer.GetLayerDefn().GetGeomType()

        extent = self.ogrlayer.GetExtent()
        if extent:
            self._bbox = [extent[0], extent[2], extent[1], extent[3]]
        else:
            self._bbox = None

        try:
            self.shapetype = ogrlib_shapetypes[self.shapetype]
        except:
            # if shapetype is not contained in ogrlib_shapetypes
            # treat it like SHAPETYPE_UNKNOWN
            self.shapetype = ogrlib_shapetypes[ogr.wkbUnknown]

        self.shapes = self.shapes()

    def FileType(self):
        """Return the filetype."""
        return self._filetype

    def BoundingBox(self):
        """Return the bounding box of the shapes in the shape file.

        The coordinate system used is whatever was used in the shape file.
        If the shape file is empty, return None.
        """
        return self._bbox

    def shapes(self):
        """Return a collection of all features as OGRShape objects.
        """
        shapes = {}
        self.ogrlayer.ResetReading()

        nextFeature = self.ogrlayer.GetNextFeature()
        while nextFeature is not None:
            fid = nextFeature.GetFID()
            shape = OGRShape(self, nextFeature)
            shapes[shape.ShapeID()] = shape
            nextFeature = self.ogrlayer.GetNextFeature()

        return shapes

    def OGRLayer(self):
        """Return the OGRLayer object
        """
        return self.ogrlayer

    def ShapeType(self):
        """Return the type of the shapes in the shapestore.

        This is either SHAPETYPE_POINT, SHAPETYPE_ARC, SHAPETYPE_POLYGON,
        SHAEPTYPE_GEOMCOLL, SHAPETYPE_NONE or SHAPETYPE_UNKNOWN.
        """
        return self.shapetype

    def RawShapeFormat(self):
        """Return the raw data format of the shape data, i.e. RAW_PYTHON
        """
        return RAW_PYTHON

    def NumShapes(self):
        """Return the number of shapes in the shape store
        """
        return self.numshapes

    def ShapesInRegion(self, bbox):
        """Return an iterable over the shapes that overlap the bounding box.

        The bbox parameter should be the bounding box as a tuple in the
        form (minx, miny, maxx, maxy) in the coordinate system of the
        shape store.
        """
        left, bottom, right, top = bbox

        # create a geometry which can be passed to the layer as spatial filter
        bboxpolygon = ogr.CreateGeometryFromWkt(
                      ('Polygon((%s %s, %s %s, %s %s,%s %s, %s %s))'
                                 %(left, bottom, left, top, right, top,
                                  right, bottom, left, bottom)))

        if self.ogrlayer.GetSpatialRef():
            bboxpolygon.AssignSpatialReference(self.ogrlayer.GetSpatialRef())

        self.ogrlayer.ResetReading()
        #ogrlayer.SetSpatialFilterRect(left, bottom, right, top)
        self.ogrlayer.SetSpatialFilter(bboxpolygon)

        numFeatures = self.ogrlayer.GetFeatureCount()
        # if no features are in bbox, return all features as shapesInRegion
        # (PostGIS sometimes returns no features even if they are within 
        #  the bounding box)
        if numFeatures == 0:
            self.ogrlayer.SetSpatialFilter(None)
            numFeatures = self.ogrlayer.GetFeatureCount()
        for feature in range(numFeatures):
            nextFeature = self.ogrlayer.GetNextFeature()
            yield self.shapes[nextFeature.GetFID()]

        self.ogrlayer.SetSpatialFilter(None)
        bboxpolygon.Destroy()

    def AllShapes(self):
        """Return an iterable over the shapes in the shape store.
        """
        for id in range(len(self.shapes)):
            yield self.shapes[id]

    def Shape(self, fid):
        """Return the shape with fid = fid
        """
        if fid in self.Table().ids.keys():
            return self.shapes[fid]
        else:
            return None

    def Table(self):
        """Return the table containing the attribute data."""
        return self._table

    def Dependencies(self):
        """Return the empty tuple.
        """
        return ()

    def OrigShapeStore(self):
        """Return None."""
        return None

    def Id_column(self):
        """Return the id_column.
        """
        return None

class OGRTable(transientdb.AutoTransientTable):
    """A Table for an ogr datasource.
    """

    def __init__(self, session, ds, layer, id_column):
        """Initialize the OGRTable.

        session   - should be the current session.
        ds        - should be an instance of OGRDatasource.
        layer     - should be an instance of OGRLayer.
        id_column - should be the name of the column used as ID column
        """
        self.datasource = ds
        self.layer = layer
        self.tablename = self.layer.GetName()
        self.id_column = id_column

        # Map column names and indices to column objects.
        self.column_map = {}

        # Map feature ids to ordinals.
        self._map_ords_and_ids()

        self._fetch_table_information()
        self._fetch_table_content()

        transientdb.AutoTransientTable.__init__(self, session.TransientDB(),
                                                self)

    def _fetch_table_information(self):
        """Internal: Update information about the table
        """
        self.columns = []

        layerdefn = self.layer.GetLayerDefn()
        # if FID column is of interest
        #col = OGRColumn("FID", table.FIELDTYPE_INT, layerdefn.GetFieldCount())
        #self.columns.append(col)
        for i in range(layerdefn.GetFieldCount()):
            fielddef = layerdefn.GetFieldDefn(i)
            fieldname = fielddef.GetName()
            fieldtype = fieldtype_map[fielddef.GetType()]
            fieldindex = layerdefn.GetFieldIndex(fieldname)
            col = OGRColumn(fieldname, fieldtype, fieldindex)
            if col is not None:
                self.columns.append(col)

        for col in self.columns:
            self.column_map[col.name] = col
            self.column_map[col.index] = col

    def _fetch_table_content(self):
        """Internal: Update information contained in the table
        """
        self.content = []
        layerdefn = self.layer.GetLayerDefn()

        self.layer.ResetReading()
        for i in range(self.layer.GetFeatureCount()):
            nextFeature = self.layer.GetNextFeature()
            row = []
            for j in range(layerdefn.GetFieldCount()):
                row.append(nextFeature.GetField(j))
            # if FID should be listed in the table
            #if self.id_column is None:
            #    row.append(nextFeature.GetFID())
            #else:
            #    row.append(nextFeature.GetField(self.id_column))
            self.content.append(row)

    def _map_ords_and_ids(self):
        """Create collections which map ordinals to ids and verse visa.
        """
        self.ordinals = {}
        self.ids = {}

        if self.id_column is not None:
            lay = self.datasource.ExecuteSQL("SELECT %s from %s"
                    %(self.id_column, self.tablename))
            lay.ResetReading()
            nextFeature = lay.GetNextFeature()
        else:
            self.layer.ResetReading()
            nextFeature = self.layer.GetNextFeature()

        ord = 0
        while nextFeature is not None:
            if self.id_column is not None:
                id = nextFeature.GetField(self.id_column)
                nextFeature = lay.GetNextFeature()
            else:
                id = nextFeature.GetFID()
                nextFeature = self.layer.GetNextFeature()
            self.ordinals[ord] = id
            self.ids[id] = ord
            ord = ord + 1
        if self.id_column is not None:
            self.datasource.ReleaseResultSet(lay)

    def TableName(self):
        """Return the name of the table, which is the name of the layer
        """
        return self.tablename

    def Title(self):
        """Return the title of the table.
        """
        return self.tablename

    def Dependencies(self):
        """Return an empty tuple.
        """
        return ()

    def NumColumns(self):
        """Return the number of columns.
        """
        return len(self.columns)

    def Columns(self):
        """Return all columns.
        """
        return self.columns

    def Column(self, col):
        """Return the column col. col can be either a string or an integer.
        """
        return self.column_map[col]

    def HasColumn(self, col):
        """Return if column col exists. col can be either a string or an
        integer.
        """
        return self.column_map.has_key(col)

    def NumRows(self):
        """Return the number of rows in the table, which equals the number of
        features in the layer.
        """
        return len(self.ids)

    def RowIdToOrdinal(self, gid):
        """Return the row ordinal given its id
        """
        if gid < 0:
            return gid
        else:
            ord = self.ids[gid]
            return ord

    def RowOrdinalToId(self, num):
        """Return the rowid for given its ordinal
        """
        if num >= 0:
            id = self.ordinals[num]
            return id
        else:
            return num

    def ReadRowAsDict(self, row, row_is_ordinal = 0):
        """Return a dictionary which contains all the fields.
        """
        if row_is_ordinal == 0:
            rowId = self.RowIdToOrdinal(row)
        else:
            rowId = row
        result = {}
        for i in range(self.NumColumns()):
            result[self.Column(i).name] = self.content[rowId][i]
        return result

    def ReadValue(self, row, col, row_is_ordinal = 0):
        """Return the requested value.
        """
        if row_is_ordinal == 0:
            rowId = self.RowIdToOrdinal(row)
        else:
            rowId = row
        colIndex = self.column_map[col].index
        return self.content[rowId][colIndex]

    def ValueRange(self, col):
        """Return the value range of the given column (given as string).
        """

        result = self.datasource.ExecuteSQL("SELECT min(%s), max(%s) FROM %s"
                %(col, col, self.layer.GetName()))
        result.ResetReading()
        feature = result.GetNextFeature()
        try:
            min = feature.GetField(0)
            max = feature.GetField(1)
        except:
            min = 0
            max = 0
        self.datasource.ReleaseResultSet(result)
        return (min, max)

    def UniqueValues(self, col):
        """Return all the values being found in the column (given as string).
        """
        result = self.datasource.ExecuteSQL((
                    "SELECT DISTINCT %s FROM %s ORDER BY %s"
                     %(col,self.layer.GetName(),col)))
        values = []
        while 1:
            feature = result.GetNextFeature()
            if feature is None:
                break
            values.append(feature.GetField(0))
        self.datasource.ReleaseResultSet(result)
        return values

    def SimpleQuery(self, left, comparison, right):
        """Return the FIDs resulting from the given query.
        """

        if comparison not in ("==", "!=", "<", "<=", ">=", ">"):
            raise ValueError("Comparison operator %r not allowed" %comparison)

        if comparison == "==":
            comparison = "="

        if isinstance(right, OGRColumn):
            right_template = right.name
        else:
            right_template = right

        if self.id_column is None:
            id = "FID"
        else:
            id = self.id_column
        query = ("SELECT %s FROM %s WHERE %s %s %s ORDER BY %s"
                % (id, self.tablename,left.name, comparison,
                   right_template, id))

        lay = self.datasource.ExecuteSQL(query)
        result = []
        while lay is not None:
            feature = lay.GetNextFeature()
            if feature is None:
                break
            result.append(feature.GetField(0))
        if lay is not None:
            self.datasource.ReleaseResultSet(lay)
        return result

    def Id_column(self):
        """Return the id_column.
        """
        return self.id_column


class OGRColumn:
    """Column description for a table for an ogr file
    """

    def __init__(self, name, type, index):
        self.name = name
        self.type = type
        self.index = index