spatial-data-lab / knime-geospatial-extension Goto Github PK

It seems that the Geospatial View and Spatial Heatmap do not escape special characters when creating the GeoJSON stringe e.g. the following input string creates an exception I guess because of the {} brakets:
I didn’t want to do it. but I did it anyway. I looked at the Town Halls. {{{what a mess}}} they are so different. #Biden is a gentle caterpillar while #Trump is a rabid fox 🥺 #TownHalls #trustYourInstinct #election2020 #holdontight #bumpyride

Example workflow can be found here: https://hub.knime.com/tobias.koetter/spaces/Geospatial/latest/bugs/ViewsWithUnescapedStringProblem~m87FqNLk_gM68CYx

To reproduce:

1. load a vector data set
2. append date&time information to each object in the table (as Date+Time type)
3. consume data in “Kepler.gl View”

Converting the date&time columns into a string makes it work for the node.

We should use 4.7.0 as minimum version for the org.knime.features.geospatial feature_dependencies

The results of the overlay node have none values which different from ArcGIS pro.
@UrbanGISer

The Python node API now supports adding static resources to a view which we should use to load the libraries from there instead of the web to prevent security issues.

Geoview, add numberic tooptips support

Depending on the query the output table can’t be viewed (it gets stuck at “Loading port content…”. In addition, when wanting to reset the node, the following error is shown: Error when attempting to close open closeable.WARN ResetAction Reset not allowed

Node config:

• State FIPS: 01
• County FIPS: 01 (same as State FIPS because I want all counties of state 01)
• TIGER/line data type: county20

The node execute successfully but I can’t use it for further processing (tried Joiner node and Geospatial View node)

See example workflow here: https://hub.knime.com/tobias.koetter/spaces/Geospatial/latest/bugs/Faulty%20exection%20Tiger%20Map%20node~p65jSa5GEUA_THX_

Can't install knime-python-base on macbook m1.

python 3.10

• nothing provides requested knime-python-base

When I configure a <geospatial view> node and want to use one of my columns for the 'Marker Color Column' field. If my selected column holds too little values the preview fails, which is ok. So far, I didn't find a way to deselect the column from that field, so that I basically select no column for the 'Marker Color Column'. Could we add an empty field for the drop down selection of the columns or any way to deselect the column again?

As a lay user I have no clue about which projection to use. That is why we should add the most common globally useable projections to the CRS description and what they are for e.g. EPSG:3857 for distances in meters.

When adding a <buffer> node to the workflow and connecting it to a table with a geometry column, it doesn't pre-select the geometry column (even though there is only one) and hence, the node is in an unconfigured state.

Add line with support for geoview node

When configuring the <geospatial view> node for the first time, the 'geometry' column is pre-selected. At the bottom of the (pre-)view we prompt a warning that this column is preset.
It seems to me, that this might be counter-intuitive, since the orange color might indicate that something is off.
I would think that we do not need to prompt this warning, since the user sees that the geometry column is selected at the top right. But this is just an idea.

The Spatial Join node right now uses HTML tags (e.g. Join mode section) in the node description which are not correct rendered in KNIME. Instead it should use markdown for lists etc.

Add visualization settings from the Geospatial View to the Heatmap view e.g. color and legend settings. Show the heatmap settings (weight, opacity, zoom, radius, blur) as first group in the dialog followed by the visualization settings from the Geospatial View.

Create a new settings class for the general options (base map, stroke, marker size, tooltip, popup) in the Geospatial View node to reuse it also in the Heatmap View along with the other existing settings groups e.g. Coloring Settings and Legend Settings.

Getting errors when trying to execute US2020 TIGER Map node. Two types of error messages depending on what I want to do:

Execute failed: HTTP Error 404: Not Found or

Execute failed: An error occurred while calling o15.reportBatchWritte

Node works fine for some states (e.g., configuring the node to extract all tract20 for Florida or Maryland works fine, but fails when trying for New York (36) or District of Columbia (11). Also tried different data type like tract, block group, etc., node just fails.

I checked archive (Index of /geo/tiger/TIGER2020PL/STATE ) and was able to download the zipped archive for New York manually.

If the input table contains more than one row the node fails with an exception that is not helpful. Instead of failing the node should only process the first input row and show the following warning: "Only first row processed. Multi Ring Buffer does not support multiple input rows."

The GeoJSON to Geometry node fails with
File "C:\DEV\GIT\EXTERNAL\knime-geospatial-extension\knime_extension\src\nodes\conversion.py", line 254, in execute raise RuntimeError( RuntimeError: Error converting 'GeoJSON' column to geometry: Unknown column geometry

Node description states that the field “geojson column” expects a GeoJSON column to convert. However, it does not recognize any JSON column but only STRING columns.

The input port description lacks in detail, as it states “input table to extract the geometry column from”. Actually, the geometry column is what we want to generate, so why should our input table have it already?

The embedded “From GeoJSON” link actually points to the from_wkt method.

I tried to read a GADM geopackage file with the GeoPackage Reader node but failed. With different error messages. What do I need to specify as the Input layer name for reading (Has also a typo).

After my recent pull (and nightly build update), the "ok" button of a configuration window (of any geospatial node) seems to be non-responsive. If I click the button nothing happens and the configuration window remains open.

As one needs to type in FIPS for state (and county) it might be helpful to include this additional link to the node description to let unexperienced users understand the states/counties and their corresponding FIPS. As one needs to type in FIPS for state (and county) it might be helpful to include this additional link to the node description to let unexperienced users understand the states/counties and their corresponding FIPS.

In addition include some of the most common counties (e.g. New York) as an example into the node description.

As a user I want to calculate the drive distance rather than the Euclidean between two places. To compute the distance I want to use a free service such as OSM but also want to use commercial providers such as here.com, Google API.

With the 2020 US Census Data node it is possible to enter “*” to select all states/counties. Would be nice to have the same option for the Tiger Map node

Add string column support for geoview static

The node actually takes all column information which is included inside the table. Therefore, it might be better to let the user know, if one wants to limit the amount of information which is passed onto the node (also stating what information is necessary).

All the spatial modeling nodes do not have a general description about what they are doing. Ideally each node has a short description with link(s) to other resources describing the methods in detail.

Double check that the extension also works with the default table backend in KNIME AP

As a user I want to get the geo coordinates of an address. To do so I want to use a free provider such as OSM but also commercial once such as here.com or Google.
The input of the node would be a table with the address information and the node would append a new column to each row with the geo coordinates of the corresponding address.

Move the dataverse nodes to their own extension to make them usable also for none Geospatial users.

Do not forget to remove the dataverse dependency from the geospatial_env.yml file

Ideally the different options of the view nodes (e.g. opacity etc. in the heatmap view) have some examples on how the settings will influence the graphic.

To reproduce create a KNIME table with a Geo column with name column1. Select this column to visualize it in the Kepler.gl view node and you will get the following exception:
WARN Kepler.gl Geoview 3:117 Traceback (most recent call last): File "C:\KNIME\NIGHTLY\knime_nightly\plugins\org.knime.python3.nodes_4.7.0.v202210281512\src\main\python\knime_node_backend.py", line 470, in execute outputs = self._node.execute(exec_context, *inputs) File "C:\DEV\GIT\EXTERNAL\knime-geospatial-extension\knime_extension\src\nodes\visualize.py", line 865, in execute map_1.add_data(data=gdf.copy(), name="state") File "C:\Users\Tobias\.conda\envs\geospatial_env\lib\site-packages\keplergl\keplergl.py", line 140, in add_data normalized = _normalize_data(data) File "C:\Users\Tobias\.conda\envs\geospatial_env\lib\site-packages\keplergl\keplergl.py", line 50, in _normalize_data return _gdf_to_dict(data) if isinstance(data, geopandas.GeoDataFrame) else _df_to_dict(data) File "C:\Users\Tobias\.conda\envs\geospatial_env\lib\site-packages\keplergl\keplergl.py", line 44, in _gdf_to_dict df[name] = df.geometry.apply(lambda x: shapely.wkt.dumps(x)) File "C:\Users\Tobias\.conda\envs\geospatial_env\lib\site-packages\pandas\core\generic.py", line 5907, in __getattr__ return object.__getattribute__(self, name) AttributeError: 'DataFrame' object has no attribute 'geometry'

Right now the nodes have the wrong color which is because of the wrong node type. Change type to SINK.

The H3 library from Uber provides a nice way of creating a world wide grid that can be used to aggregate information on different levels of granularity. The node would allow the user to create such a grid for a certain area.

This blog post shows a nice example on how to use the library: https://towardsdatascience.com/uber-h3-for-data-analysis-with-python-1e54acdcc908

The new node dialogs have a little icon next to each dialog option. To make it easier for people to find the right MIPS etc numbers it would be great if we could add a link to a lookup page to the corresponding dialog option so that the link is shown directly in the node dialog when clicking on the question mark of the option.

As a user I want to get the address of a geo coordinate. To do so I want to use a free provider such as OSM but also commercial once such as here.com or Google.
The input of the node would be a table with geo coordinate (point) and the node would append new columns to each row with the corresponding address.
This was requested during the summit and in the KNIME forum.

@UrbanGISer It would be nice to have options letting people choose CRSs from the dropdown list.

To reproduce use an input table with a single Geo column. Execute the node as it is predefined (e.g. Geometry column is the input column and Marker color column is empty). Do not select None in the Marker column select box. This results in the following error:
WARN Geospatial View Static 3:119 Traceback (most recent call last): File "C:\KNIME\NIGHTLY\knime_nightly\plugins\org.knime.python3.nodes_4.7.0.v202210281512\src\main\python\knime_node_backend.py", line 470, in execute outputs = self._node.execute(exec_context, *inputs) File "C:\DEV\GIT\EXTERNAL\knime-geospatial-extension\knime_extension\src\nodes\visualize.py", line 721, in execute self.legend_caption = self.color_col File "C:\KNIME\NIGHTLY\knime_nightly\plugins\org.knime.python3.nodes_4.7.0.v202210281512\src\main\python\knime_parameter.py", line 426, in __set__ self._validate(value) File "C:\KNIME\NIGHTLY\knime_nightly\plugins\org.knime.python3.nodes_4.7.0.v202210281512\src\main\python\knime_parameter.py", line 435, in _validate self._validator(value) File "C:\KNIME\NIGHTLY\knime_nightly\plugins\org.knime.python3.nodes_4.7.0.v202210281512\src\main\python\knime_parameter.py", line 652, in default_validator raise TypeError( TypeError: None is of type <class 'NoneType'>, but should be of type string.

Right now the order of list values is unpredictable maybe due to the usage of the toString() method. Instead the node should use a reproduceable way to generate the string from a list e.g. by sorting the elements prior converting to string.

I updated to the most recent nightly build and pulling the most recent contents from master branch this afternoon. After that, clicking the "save & execute" button of the configuration window of a view node (e.g. ) does not react and hence, does not load a visualization map.

As a user I want to extract the geo type and/or coordinate reference system and maybe other information into new data columns.

The resulting composite view is limited, bringing up a scroll bar at the right side and hence, the view is not shown at the full extent.

To reproduce:

1. load a vector data set
2. display it with a view node
3. put the view node into a component
4. consume the composite view

Whenever a configuration of a view node or a view output is opened, I see the screen below. It takes a moment until it disappears and the map is finally displayed.
What is the reason for this and how can we prevent this from being shown?

As a user I want to get the geo coordinates of an IP address. To do so I want to use a free provider such as OSM but also commercial once such as here.com or Google.
The input of the node would be a table with the IP address and the node would append a new column to each row with the geo coordinates of the corresponding address.

One can only type in a hard-coded path, but browsing the file to read or eve just the destination (for writer nodes) would improve the user experience.