Mapfile configuration

As mentioned on the index page (Administrator Guide), the application administrator manages the application through the database and via the application’s MapServer mapfile.

The application’s mapfile is where WMS and WFS layers are defined. WMS is used for the map (WMS GetMap), and for the point query feature (WMS GetFeatureInfo). WFS is used for the query (box and point), the filters, and the WFS permalink features.

The application’s mapfile is located in the mapserver directory, it is commonly named mapserver.map.tmpl.

Note

The application’s mapfile is a .tmpl file because it contains variables. These variables are substituted at container start.

This section describes how to make layers printable and/or queryable and/or private (a.k.a restricted).

Print

MapFish Print performs single tile requests to the WMS server. For that reason, we need to use a relatively large value for the MAXSIZE parameter (of the MAP section); 5000 for example.

MapFish Print also supports map rotations. This implies specific requirements:

  • The MAP and all the LAYER definitions should have a PROJECTION. For example:

    PROJECTION
        "init=epsg:21781"
    END
    
  • When rotating the map (with a non-zero value for ANGLE) there are important things to be aware of. Make sure to read the notes for the ANGLE parameter on https://mapserver.org/mapfile/map.html.

MapFish Print uses a resolution of 254 dpi (instead of 72 dpi as used for the web application on the screen). Using LAYER/SYMBOLSCALEDENOM is therefore not recommended. LABEL/MINSIZE and LABEL/MAXSIZE should be used when necessary only, as these parameters do not take the MAP resolution into account.

WFS GetFeature

Layers that you want to use in a query (box and point), in filters, or for WFS permalink features must support WFS GetFeature. To support WFS GetFeature a LAYER should define a template:

TEMPLATE "fooOnlyForWFSGetFeature"

This is a fake template, but this is required by the querier and for the filter.

The LAYER should also define metadata, with a METADATA section. For example:

"gml_include_items" "all"
"gml_types" "auto"
"gml_featureid" "id"
"wfs_enable_request" "*"

gml_include_items

This is a comma-delimited list of attribute names, it specifies the list of attributes that should be returned in GML responses. all means that all the attributes of the layer should be returned.

gml_types

Set this to auto. This means that the layer’s field type information is obtained from the input feature driver (OGC). If the type is not well interpreted (notably if you have a wrong operator within your filter), you can define manually your attribute type with "gml_<attribute>_type" "<type>" where <type> is one of these: Integer, Long, Real, Character, Date or Boolean.

gml_featureid

References the name of layer’s primary key column. Setting this is mandatory for GetFeature request including ogc:FeatureId filters. Always set it. (This is required for the JavaScript API.)

wfs_enable_request

Space separated list of requests to enable. Use *.

Warning

The geometry columns of layers involved in query should have the same name. By default the filters assumes the name is geom.

In contrast to WMS GetFeatureInfo, WFS GetFeature supports point query as well as box query. However, it is to be noted that WFS GetFeature may return features that are not visible at the current resolution of the map. This is because of a limitation in MapServer, where MINSCALE/MAXSCALE values defined in the layer’s classes (CLASS) have no effect.

WMS GetFeatureInfo

To support WMS GetFeatureInfo, a LAYER should define a template:

TEMPLATE fooOnlyForWMSGetFeatureInfo

Similar to WFS GetFeature, this is a fake template, but it is required.

The gml_include_items, gml_<geometry name>_type and gml_geometries METADATA variables should also be defined in the LAYER. For example:

"gml_include_items" "all"
"gml_geometries" "geom"
"gml_geom_type" "polygon"

gml_include_items

See above.

gml_geometries

This is a string specifying the name used for geometry elements in GetFeatureInfo (GML) responses. This property, and gml_<name>_type, should be set for the GetFeatureInfo responses to include the features’ geometries instead of bboxes.

gml_<geometry name>_type

This specifies the type of a geometry column. Specifying this property is necessary if geometries, instead of bboxes, should be returned in GetFeatureInfo (GML) responses. <geometry name> should be replaced by the string set with the gml_geometries. For example, if geom_geometries is set to the_geom, then gml_the_geom_type should be used. The possible values are point, multipoint, line, multiline, polygon, multipolygon. If you do not set the right type for multi geometries, only the first will be visible on the map. See also gml_<geometry name>_type.

See the WMS Server MapFile Documentation for more detail.

Restricted layer

The restricted layers work only with PostgreSQL data. All layers defined as restricted in the mapfile should be defined as well in the administration interface and vice versa.

With a RestrictionArea area

A RestrictionArea is used to restrict the layer displaying to a given area. This area is specified in the administration interface while defining the RestrictionArea element.

Warning

Using a restriction area on a big layer or defining a very complex area may slow down the application.

To define a restricted layer in the Mapfile, the DATA property of the LAYER should look like this:

DATA "<the_geom> FROM (
    SELECT *
    FROM <schema>.<table>
    WHERE ST_Contains(
       (${MAPSERVER_DATA_SUBSELECT} '<layername>'),
       ST_SetSRID(<the_geom>, <projection>)
     )
) as foo USING UNIQUE gid USING srid=<projection>"

<schema>, <table>, <layername>, <projection> and <the_geom> need to be replaced as appropriate. <table> is the name of the PostGIS table including the geographic data for this layer. <the_geom> is the name of the table’s geometry column. <schema> is the name of the schema including the table. <layer_name> can be either the layer NAME or the layer GROUP, depending on what is configured in the administration interface for the layer.

The ${MAPSERVER_DATA_SUBSELECT} is defined as follows:

SELECT
    rra.role_id
FROM
    <main_schema>.restrictionarea AS ra,
    <main_schema>.role_restrictionarea AS rra,
    <main_schema>.layer_restrictionarea AS lra,
    <main_schema>.treeitem AS la
WHERE
    rra.restrictionarea_id = ra.id
AND
    lra.restrictionarea_id = ra.id
AND
    lra.layer_id = la.id AND la.name =

Warning

In some cases you can have geometries that overlap the restriction area. Theses features will not be displayed as they are not in the area (ie not contained). st_intersects or another operator could be used instead of the st_contains operator.

The ${mapfile_data_subselect} variable is defined in the CONST_vars.yaml configuration file. Its goal is to simplify the writing of the mapfile.

Without restriction on the RestrictionArea area

If we do not need to restrict on an area, we can use the following DATA property of the LAYER:

DATA "the_geom FROM (
    SELECT
        geo.*
    FROM
        <schema>.<table> AS geo
    WHERE (
        ARRAY[%role_ids%] && ARRAY(
            ${MAPSERVER_DATA_NOAREA_SUBSELECT} '<layername>'
        )
    )
) AS foo USING UNIQUE id USING srid=21781"

Then you do not need to define an area in the admin interface.

The ${MAPSERVER_DATA_NOAREA_SUBSELECT} is defined as follows:

SELECT
    rra.role_id
FROM
    <main_schema>.restrictionarea AS ra,
    <main_schema>.role_restrictionarea AS rra,
    <main_schema>.layer_restrictionarea AS lra,
    <main_schema>.treeitem AS la
WHERE
    rra.restrictionarea_id = ra.id
AND
    lra.restrictionarea_id = ra.id
AND
    lra.layer_id = la.id
AND
    la.name =

Metadata and filename

It is required to have the following in the VALIDATION section of the LAYER:

${mapserver_layer_validation}

This variable is defined in the CONST_vars.yaml configuration file as follows:

mapserver_layer_validation =
    "default_role_ids" "-1"
    "role_ids" "^-?[0-9]*$$"

The mapfile should be a .map.tmpl file, for the variable to be substituted at container start.

Variable Substitution

It is possible to adapt some values in the mapfile according to the user’s role by using variable substitution, for instance to hide some layer objects attributes. The list of parameters that support variable substitution is available here.

To define variables, edit the matching MAP/LAYER/VALIDATION section in the MapFile and add:

"default_s_<variable>" "<default_value>"
"s_<variable>" "<validation_pattern>"

The validation_pattern is a regular expression used to validate the argument. For example, if you only want lowercase characters and commas, use ^[a-z,]*$.

Now in LAYER place %s_<variable>% where you want to insert the variable value, but not at the start of a line (to avoid escape issues).

Then in the administration interface, create a functionality named mapserver_substitution with the value: <variable>=<value>.

Please note that we cannot use substitution in the METADATA values. As a result, if you would like to adapt the list of attributes returned in a WFS GetFeature or WMS GetFeatureInfo request, you have to adapt the columns listed in the DATA section. For instance:

LAYER
    ...
    DATA "geom FROM (SELECT t.geom, t.type, t.gid, %s_columns% FROM geodata.table as t)  AS foo using unique gid using SRID=21781"
    METADATA
        ...
        "gml_exclude_items" "type,gid"
        "gml_include_items" "all"
    END
    VALIDATION
        "default_s_columns" "t.name"
        "s_columns" "^[a-z,._]*$$"
    END
    CLASS
        EXPRESSION ([type]=1)
        ...
    END
    ...
END

Then add a mapserver_substitution functionality in the administration interface with for instance the following value for the given role: columns=t.private.

Note

You can also use variable substitution for the role_id and user_id, but beware that these attributes are not available for cached queries like: GetCapabilities, GetLegendGraphic, DescribeFeatureType.

MapServer documentation

Performance improvement

Adding an EXTENT parameter to the LAYER section may significantly improve performance because it saves MapServer from computing the extent of all layer features.

Prepare raster files

To achieve good performance, you should have tiled files with overviews, and ideally a tileindex. You can achieve this with these steps:

Convert your rasters in tiled GeoTIFF:

gdal_translate -of GTiff -co "TILED=YES" -co "TFW=YES" \
  <filename_in.tif> <filename_out.tif>

Then build overviews for your rasters:

gdaladdo -r average filename.tif 2 4 8 16

You can generate a shapefile indexing all your rasters:

gdaltindex filename_index.shp *.tif

Note about ECW

In general using ECW is not recommended, as MapServer often generates broken images and has memory leaks with ECW. See this MapServer ticket for example.

If you still want to use it, then replace SetHandler fcgid-script by SetHandler cgi-script in the apache/mapserver.conf.tmpl file. But note that this affects performance.