[GIS] Change layer datasource in qgis project with python

pythonqgis

I have different qgis project with some layer and their datasource like ../my_path/my_file.shp.
I need to insert this shp file to a remote db. For do it i use shapefile from PGeo module and it's work fine.
But i need change layer datasource in qgs project with dbname= , host etc where schema_name is a name of qgis project, table_name is a layer id, and i need to maintain all layer's property. I tried to use:

 uri = QgsDataSourceURI()
 uri.setConnection("localhost", "5432", "mydb", "postgres", "postgres")
 uri.setDataSource(str(schema_name),str(layer.id()),'the_geom')
 layer = QgsVectorLayer(uri.uri(), str(layer.id()), "postgres")
 QgsMapLayerRegistry.instance().addMapLayer(layer)
 layer.reload()
 print layer.source()
 if not layer.isValid():
    print "Layer is not valid"
 project.write(QFileInfo(str(dir)+'/'+project_name))

But i always got "Layer is not valid" and nothing is changed on my qgis project.
Is there another way to change layer datasource and save changes to the project? I see this solution but it doesn't work for me

Best Answer

It's seem that's uri is a key sensitive, so I solved a problem with:

table_name = table_name.lower()
schema_name = schema_name.lower()
uri.setDataSource(schema_name,table_name,"the_geom", '', 'gid')

After this change layer.isValid() return true

Related Solutions

[GIS] Using group by to set a PostGIS datasource using QGIS Python API

From my own further investigation it looks like such a group by clause is not useful because the primary key needs to be included as a field and so no grouping will actually be done.

As commented on another website where I asked this question, the data will need to be preprocessed and probably best made available as a view.

To do this dynamically (as I was initially trying to do), the answer would be to create a table (or view) on the fly in PostGIS and then set the datasource to show this data in QGIS. I don't like creating a lot of tables just for the sake of viewing data, but can't see another way of doing it.

QGIS Python – How to Change Attributes

You are right, that you don't need to call commit if you work on the dataProvider directly. However, this will also commit every update immediately, leading to more I/O overhead (often decreased performance) and poor transaction handling. This is to say, the method works, but use it carefully. I have added another method below which is better suited in most cases.

The problem in the code is, that you are only setting the attribute on the feature in memory and don't tell the provider, that this feature changed. Instead, try the following code. It buffers all updates in the updateMap and then executes it in one go at the end, therefore allowing batch updates (useful e.g. on slow networks). Another possibility would be to issue updateAttributeValue directly in the loop.

from qgis.utils import iface
from PyQt4.QtCore import *

layers = iface.legendInterface().layers()

for layer in layers:
    name = layer.name()
    if name.endswith('x'):
        provider = layer.dataProvider()
        updateMap = {}
        fieldIdx = provider.fields().indexFromName( 'attr' )
        features = provider.getFeatures()
        for feature in features:
            updateMap[feature.id()] = { fieldIdx: 'a' }

        provider.changeAttributeValues( updateMap )

The generally recommended way:

with edit(layer):
    for feature in layer.getFeatures():
        feature['fieldName'] = newValue
        layer.updateFeature(feature)

http://qgis.org/api/2.2/classQgsVectorDataProvider.html#a99a5d01c33c1cadb7ae9209b68cd8743

Related Question