<tool id="geopandas_table2geojson" name="Table to GeoJSON" version="@TOOL_VERSION@+galaxy@VERSION_SUFFIX@" profile="@PROFILE@" license="MIT">
    <description></description>
    <macros>
        <token name="@TOOL_VERSION@">1.1.1</token>
        <token name="@VERSION_SUFFIX@">1</token>
        <token name="@PROFILE@">24.2</token>
    </macros>
    <requirements>
        <requirement type="package" version="@TOOL_VERSION@">geopandas</requirement>
    </requirements>
    <command detect_errors="exit_code"><![CDATA[
        python '$csv2geojson'
    ]]></command>
    <configfiles>
        <configfile name="csv2geojson">
import json
import pandas as pd
import geopandas as gpd

df = pd.read_csv('$infile', encoding="utf-8", sep=None, engine='python', header=None if '$has_header' == "false" else 0)

inc = [int(x)-1 for x in str('$include_columns').split(",") if x.strip().isdigit()] if '$include_columns' and str('$include_columns').lower() != "none" else []
exc = [int(x)-1 for x in str('$exclude_columns').split(",") if x.strip().isdigit()] if '$exclude_columns' and str('$exclude_columns').lower() != "none" else []

property_df = df

if inc:
    property_df = df.iloc[:, inc]
if exc:
    property_df = property_df.drop(columns=df.columns[exc])

gdf = gpd.GeoDataFrame(
    property_df,
    geometry=gpd.points_from_xy(
        df.iloc[:, int($lat_column)-1],   
        df.iloc[:, int($long_column)-1]),
    crs="$crs"
)
        
# drop_id: bool, default: False
data = json.loads(gdf.to_json())
with open('outfile.geojson', "w", encoding="utf8") as f:
    ## Minify output json
    f.write(json.dumps(data, indent=4))
        </configfile>
    </configfiles>
    <inputs>
        <param name="infile" type="data" format="csv,tabular,tsv" label="The file you want to convert"/>
        <param name="has_header" type="boolean" truevalue="true" falsevalue="false" label="First row contains column names (header)?" help="Does your input file consists of a header"/>
        <param name="lat_column" type="integer" value="1" label="Latitude column"/>
        <param name="long_column" type="integer" value="2" label="Longitude column"/>
        <param name="include_columns" type="data_column" data_ref="infile" use_header_names="true" multiple="true" optional="true"
            label="Include the following columns" help="Either include or exclude columns, by default, all columns will be taken over into the properties field"/>
        <param name="exclude_columns" type="data_column" data_ref="infile" use_header_names="true" multiple="true" optional="true"
            label="Exclude the following columns" help="Either include or exclude columns, by default, all columns will be taken over into the properties field"/>
        <param name="crs" type="select" label="Coordinate Reference System" help="Choose the Coordinate Reference Systems (CRS) your data is using">
            <option value="EPSG:3857">Web Mercator projection</option>
            <option value="EPSG:4326" selected="true">WGS84 projection (used by Google Earth)</option>
        </param>

    </inputs>
    <outputs>
        <data name="outfile" format="geojson" from_work_dir="outfile.geojson" label="${tool.name} on ${on_string}" />
    </outputs>
    <tests>
        <test expect_num_outputs="1">
            <param name="infile" value="geolocation.csv"/>
            <param name="has_header" value="true"/>
            <param name="lat_column" value="1"/>
            <param name="long_column" value="2"/>
            <param name="crs" value="EPSG:3857"/>
            <output name="outfile" file="geolocation_1.geojson">
                <assert_contents>
                    <has_json_property_with_text property="type" text="FeatureCollection"/>
                    <!-- all 5 columns are includes as properties, nothing excluded, nothing explicitly included -->
                    <has_text text='value": 10'/>
                    <has_text text='temperature": 23'/>
                    <has_text text='longitude'/>
                    <has_text text='latitude'/>
                    <has_text text='timestamp": "2023-01-01T12:05:00Z"'/>
                    <has_n_lines n="371"/>
                </assert_contents>
            </output>
        </test>
        <test expect_num_outputs="1">
            <param name="infile" value="geolocation_without_header.csv"/>
            <param name="has_header" value="false"/>
            <param name="lat_column" value="1"/>
            <param name="long_column" value="2"/>
            <param name="crs" value="EPSG:3857"/>
            <output name="outfile">
                <assert_contents>
                    <has_json_property_with_text property="type" text="FeatureCollection"/>
                    <has_n_lines n="371"/>
                </assert_contents>
            </output>
        </test>
        <test expect_num_outputs="1">
            <param name="infile" value="geolocation.csv"/>
            <param name="has_header" value="true"/>
            <param name="lat_column" value="1"/>
            <param name="long_column" value="2"/>
            <param name="include_columns" value="3,4"/>
            <param name="crs" value="EPSG:3857"/>
            <output name="outfile" ftype="geojson" file="geolocation_2.geojson">
                <assert_contents>
                    <!-- 2 columns are explicitly included as properties -->
                    <has_text text='timestamp": "2023-01-01T12:05:00Z"'/>
                    <has_text text='value": 10'/>
                    <has_text text="FeatureCollection"/>
                    <has_n_lines n="311"/>
                </assert_contents>
            </output>
        </test>
        <test expect_num_outputs="1">
            <param name="infile" value="geolocation.csv"/>
            <param name="has_header" value="true"/>
            <param name="lat_column" value="1"/>
            <param name="long_column" value="2"/>
            <param name="exclude_columns" value="5"/>
            <param name="crs" value="EPSG:3857"/>
            <output name="outfile" ftype="geojson" file="geolocation_3.geojson">
                <assert_contents>
                    <has_json_property_with_text property="type" text="FeatureCollection"/>
                    <!-- 4 columns are includes as properties ... one is excluded -->
                    <has_text text='value": 10'/>
                    <has_text text='longitude'/>
                    <has_text text='latitude'/>
                    <has_text text='timestamp": "2023-01-01T12:05:00Z"'/>
                    <has_n_lines n="351"/>
                </assert_contents>
            </output>
        </test>

    </tests>
    <help format="markdown"><![CDATA[

# 💡 Convert CSV or Tabular Data to GeoJSON Format

## Understanding Coordinate Reference Systems (CRS)

A Coordinate Reference System (CRS) defines how the coordinate values in your data relate to real-world locations on Earth. Without a CRS,
the coordinates in a GeoSeries or GeoDataFrame are simply numbers in an undefined space.
To find reference codes for widely used CRS projections, visit [spatialreference.org](https://spatialreference.org).

## What is GeoJSON?

GeoJSON is a widely-used format for encoding geographic data structures using JSON (JavaScript Object Notation).
A GeoJSON file contains an array of geographic features, where each feature includes:
* A geometry field describing the spatial shape (e.g., point, line, polygon),
* A properties field containing non-spatial attributes.

GeoJSON uses the WGS84 coordinate system by default (EPSG:4326), with coordinates expressed in decimal degrees.
To explore and edit GeoJSON interactively, visit [geojson.io](https://geojson.io), select a location or shape, and inspect the resulting GeoJSON output.

📌 Pro Tip, Galaxy can visualise GeoJSON files.

    ]]></help>
    <citations>
        <citation type="doi">10.5281/zenodo.3946761</citation>
    </citations>
</tool>