Chapter 2. PostGIS Installation

Chapter 2. PostGIS Installation

This chapter details the steps required to install PostGIS.

2.1. Short Version

To compile assuming you have all the dependencies in your search path:

tar -xvfz postgis-3.3.4.tar.gz
cd postgis-3.3.4
./configure
make
make install

Once PostGIS is installed, it needs to be enabled ( Section 3.3, “Creating spatial databases” ) or upgraded ( Section 3.4, “Upgrading spatial databases” ) in each individual database you want to use it in.

2.2. Compiling and Install from Source

[Note]

Many OS systems now include pre-built packages for PostgreSQL/PostGIS. In many cases compilation is only necessary if you want the most bleeding edge versions or you are a package maintainer.

This section includes general compilation instructions, if you are compiling for Windows etc or another OS, you may find additional more detailed help at PostGIS User contributed compile guides and PostGIS Dev Wiki .

Pre-Built Packages for various OS are listed in PostGIS Pre-built Packages

If you are a windows user, you can get stable builds via Stackbuilder or PostGIS Windows download site We also have very bleeding-edge windows experimental builds that are built usually once or twice a week or whenever anything exciting happens. You can use these to experiment with the in progress releases of PostGIS

The PostGIS module is an extension to the PostgreSQL backend server. As such, PostGIS 3.3.4 requires full PostgreSQL server headers access in order to compile. It can be built against PostgreSQL versions 11 - 15. Earlier versions of PostgreSQL are not supported.

Refer to the PostgreSQL installation guides if you haven't already installed PostgreSQL. http://www.postgresql.org .

[Note]

For GEOS functionality, when you install PostgresSQL you may need to explicitly link PostgreSQL against the standard C++ library:

LDFLAGS=-lstdc++ ./configure [YOUR OPTIONS HERE]

This is a workaround for bogus C++ exceptions interaction with older development tools. If you experience weird problems (backend unexpectedly closed or similar things) try this trick. This will require recompiling your PostgreSQL from scratch, of course.

The following steps outline the configuration and compilation of the PostGIS source. They are written for Linux users and will not work on Windows or Mac.

2.2.1. Getting the Source

Retrieve the PostGIS source archive from the downloads website http://download.osgeo.org/postgis/source/postgis-3.3.4.tar.gz

wget http://download.osgeo.org/postgis/source/postgis-3.3.4.tar.gz
tar -xvzf postgis-3.3.4.tar.gz
cd postgis-3.3.4

This will create a directory called postgis-3.3.4 in the current working directory.

Alternatively, checkout the source from the git repository https://git.osgeo.org/gitea/postgis/postgis/ .

git clone https://git.osgeo.org/gitea/postgis/postgis.git postgis
cd postgis
sh autogen.sh
    

Change into the newly created postgis directory to continue the installation.

./configure

2.2.2. Install Requirements

PostGIS has the following requirements for building and usage:

Required

  • PostgreSQL 11 - 15. A complete installation of PostgreSQL (including server headers) is required. PostgreSQL is available from http://www.postgresql.org .

    For a full PostgreSQL / PostGIS support matrix and PostGIS/GEOS support matrix refer to http://trac.osgeo.org/postgis/wiki/UsersWikiPostgreSQLPostGIS

  • GNU C compiler ( gcc ). Some other ANSI C compilers can be used to compile PostGIS, but we find far fewer problems when compiling with gcc .

  • GNU Make ( gmake or make ). For many systems, GNU make is the default version of make. Check the version by invoking make -v . Other versions of make may not process the PostGIS Makefile properly.

  • Proj reprojection library. Proj 4.9 or above is required. The Proj library is used to provide coordinate reprojection support within PostGIS. Proj is available for download from https://proj.org/ .

  • GEOS geometry library, version 3.6 or greater, but GEOS 3.11+ is required to take full advantage of all the new functions and features. GEOS is available for download from https://libgeos.org/ .

  • LibXML2, version 2.5.x or higher. LibXML2 is currently used in some imports functions (ST_GeomFromGML and ST_GeomFromKML). LibXML2 is available for download from https://gitlab.gnome.org/GNOME/libxml2/-/releases .

  • JSON-C, version 0.9 or higher. JSON-C is currently used to import GeoJSON via the function ST_GeomFromGeoJson. JSON-C is available for download from https://github.com/json-c/json-c/releases/ .

  • GDAL, version 2+ is required 3+ is preferred. This is required for raster support. https://gdal.org/download.html .

  • If compiling with PostgreSQL+JIT, LLVM version >=6 is required https://trac.osgeo.org/postgis/ticket/4125 .

Optional

  • GDAL (pseudo optional) only if you don't want raster you can leave it out. Also make sure to enable the drivers you want to use as described in Section 3.2, “Configuring raster support” .

  • GTK (requires GTK+2.0, 2.8+) to compile the shp2pgsql-gui shape file loader. http://www.gtk.org/ .

  • SFCGAL, version 1.3.1 (or higher), 1.4.1 or higher is recommended. SFCGAL can be used to provide additional 2D and 3D advanced analysis functions to PostGIS cf Section 8.20, “SFCGAL Functions” . And also allow to use SFCGAL rather than GEOS for some 2D functions provided by both backends (like ST_Intersection or ST_Area, for instance). A PostgreSQL configuration variable postgis.backend allow end user to control which backend he want to use if SFCGAL is installed (GEOS by default). Nota: SFCGAL 1.2 require at least CGAL 4.3 and Boost 1.54 (cf: https://oslandia.gitlab.io/SFCGAL/dev.html ) https://gitlab.com/Oslandia/SFCGAL/ .

  • In order to build the Section 14.1, “Address Standardizer” you will also need PCRE http://www.pcre.org (which generally is already installed on nix systems). Regex::Assemble perl CPAN package is only needed if you want to rebuild the data encoded in parseaddress-stcities.h . Section 14.1, “Address Standardizer” will automatically be built if it detects a PCRE library, or you pass in a valid --with-pcre-dir=/path/to/pcre during configure.

  • To enable ST_AsMVT protobuf-c library 1.1.0 or higher (for usage) and the protoc-c compiler (for building) are required. Also, pkg-config is required to verify the correct minimum version of protobuf-c. See protobuf-c . By default, Postgis will use Wagyu to validate MVT polygons faster which requires a c++11 compiler. It will use CXXFLAGS and the same compiler as the PostgreSQL installation. To disable this and use GEOS instead use the --without-wagyu during the configure step.

  • CUnit ( CUnit ). This is needed for regression testing. http://cunit.sourceforge.net/

  • DocBook ( xsltproc ) is required for building the documentation. Docbook is available from http://www.docbook.org/ .

  • DBLatex ( dblatex ) is required for building the documentation in PDF format. DBLatex is available from http://dblatex.sourceforge.net/ .

  • ImageMagick ( convert ) is required to generate the images used in the documentation. ImageMagick is available from http://www.imagemagick.org/ .

2.2.3. Build configuration

As with most linux installations, the first step is to generate the Makefile that will be used to build the source code. This is done by running the shell script

./configure

With no additional parameters, this command will attempt to automatically locate the required components and libraries needed to build the PostGIS source code on your system. Although this is the most common usage of ./configure , the script accepts several parameters for those who have the required libraries and programs in non-standard locations.

The following list shows only the most commonly used parameters. For a complete list, use the --help or --help=short parameters.

--with-library-minor-version

Starting with PostGIS 3.0, the library files generated by default will no longer have the minor version as part of the file name. This means all PostGIS 3 libs will end in postgis-3 . This was done to make pg_upgrade easier, with downside that you can only install one version PostGIS 3 series in your server. To get the old behavior of file including the minor version: e.g. postgis-3.0 add this switch to your configure statement.

--prefix=PREFIX

This is the location the PostGIS loader executables and shared libs will be installed. By default, this location is the same as the detected PostgreSQL installation.

[Caution]

This parameter is currently broken, as the package will only install into the PostgreSQL installation directory. Visit http://trac.osgeo.org/postgis/ticket/635 to track this bug.

--with-pgconfig=FILE

PostgreSQL provides a utility called pg_config to enable extensions like PostGIS to locate the PostgreSQL installation directory. Use this parameter ( --with-pgconfig=/path/to/pg_config ) to manually specify a particular PostgreSQL installation that PostGIS will build against.

--with-gdalconfig=FILE

GDAL, a required library, provides functionality needed for raster support gdal-config to enable software installations to locate the GDAL installation directory. Use this parameter ( --with-gdalconfig=/path/to/gdal-config ) to manually specify a particular GDAL installation that PostGIS will build against.

--with-geosconfig=FILE

GEOS, a required geometry library, provides a utility called geos-config to enable software installations to locate the GEOS installation directory. Use this parameter ( --with-geosconfig=/path/to/geos-config ) to manually specify a particular GEOS installation that PostGIS will build against.

--with-xml2config=FILE

LibXML is the library required for doing GeomFromKML/GML processes. It normally is found if you have libxml installed, but if not or you want a specific version used, you'll need to point PostGIS at a specific xml2-config confi file to enable software installations to locate the LibXML installation directory. Use this parameter ( >--with-xml2config=/path/to/xml2-config ) to manually specify a particular LibXML installation that PostGIS will build against.

--with-projdir=DIR

Proj is a reprojection library required by PostGIS. Use this parameter ( --with-projdir=/path/to/projdir ) to manually specify a particular Proj installation directory that PostGIS will build against.

--with-libiconv=DIR

Directory where iconv is installed.

--with-jsondir=DIR

JSON-C is an MIT-licensed JSON library required by PostGIS ST_GeomFromJSON support. Use this parameter ( --with-jsondir=/path/to/jsondir ) to manually specify a particular JSON-C installation directory that PostGIS will build against.

--with-pcredir=DIR

PCRE is an BSD-licensed Perl Compatible Regular Expression library required by address_standardizer extension. Use this parameter ( --with-pcredir=/path/to/pcredir ) to manually specify a particular PCRE installation directory that PostGIS will build against.

--with-gui

Compile the data import GUI (requires GTK+2.0). This will create shp2pgsql-gui graphical interface to shp2pgsql.

--without-raster

Compile without raster support.

--without-topology

Disable topology support. There is no corresponding library as all logic needed for topology is in postgis-3.3.4 library.

--with-gettext=no

By default PostGIS will try to detect gettext support and compile with it, however if you run into incompatibility issues that cause breakage of loader, you can disable it entirely with this command. Refer to ticket http://trac.osgeo.org/postgis/ticket/748 for an example issue solved by configuring with this. NOTE: that you aren't missing much by turning this off. This is used for international help/label support for the GUI loader which is not yet documented and still experimental.

--with-sfcgal=PATH

By default PostGIS will not install with sfcgal support without this switch. PATH is an optional argument that allows to specify an alternate PATH to sfcgal-config.

--without-phony-revision

Disable updating postgis_revision.h to match current HEAD of the git repository.

[Note]

If you obtained PostGIS from the code repository , the first step is really to run the script

./autogen.sh

This script will generate the configure script that in turn is used to customize the installation of PostGIS.

If you instead obtained PostGIS as a tarball, running ./autogen.sh is not necessary as configure has already been generated.

2.2.4. Building

Once the Makefile has been generated, building PostGIS is as simple as running

make

The last line of the output should be " PostGIS was built successfully. Ready to install. "

As of PostGIS v1.4.0, all the functions have comments generated from the documentation. If you wish to install these comments into your spatial databases later, run the command which requires docbook. The postgis_comments.sql and other package comments files raster_comments.sql, topology_comments.sql are also packaged in the tar.gz distribution in the doc folder so no need to make comments if installing from the tar ball. Comments are also included as part of the CREATE EXTENSION install.

make comments

Introduced in PostGIS 2.0. This generates html cheat sheets suitable for quick reference or for student handouts. This requires xsltproc to build and will generate 4 files in doc folder topology_cheatsheet.html , tiger_geocoder_cheatsheet.html , raster_cheatsheet.html , postgis_cheatsheet.html

You can download some pre-built ones available in html and pdf from PostGIS / PostgreSQL Study Guides

make cheatsheets

2.2.5. Building PostGIS Extensions and Deploying them

The PostGIS extensions are built and installed automatically if you are using PostgreSQL 9.1+.

If you are building from source repository, you need to build the function descriptions first. These get built if you have docbook installed. You can also manually build with the statement:

make comments

Building the comments is not necessary if you are building from a release tar ball since these are packaged pre-built with the tar ball already.

The extensions should automatically build as part of the make install process. You can if needed build from the extensions folders or copy files if you need them on a different server.

cd extensions
cd postgis
make clean
make
export PGUSER=postgres #overwrite psql variables
make check #to test before install
make install
# to test extensions
make check RUNTESTFLAGS=--extension
[Note]

make check uses psql to run tests and as such can use psql environment variables. Common ones useful to override are PGUSER , PGPORT , and PGHOST . Refer to psql environment variables

The extension files will always be the same for the same version of PostGIS and PostgreSQL regardless of OS, so it is fine to copy over the extension files from one OS to another as long as you have the PostGIS binaries already installed on your servers.

If you want to install the extensions manually on a separate server different from your development, You need to copy the following files from the extensions folder into the PostgreSQL / share / extension folder of your PostgreSQL install as well as the needed binaries for regular PostGIS if you don't have them already on the server.

  • These are the control files that denote information such as the version of the extension to install if not specified. postgis.control, postgis_topology.control .

  • All the files in the /sql folder of each extension. Note that these need to be copied to the root of the PostgreSQL share/extension folder extensions/postgis/sql/*.sql , extensions/postgis_topology/sql/*.sql

Once you do that, you should see postgis , postgis_topology as available extensions in PgAdmin -> extensions.

If you are using psql, you can verify that the extensions are installed by running this query:

SELECT name, default_version,installed_version
FROM pg_available_extensions WHERE name LIKE 'postgis%' or name LIKE 'address%';

             name             | default_version | installed_version
------------------------------+-----------------+-------------------
 address_standardizer         | 3.3.4         | 3.3.4
 address_standardizer_data_us | 3.3.4         | 3.3.4
 postgis                      | 3.3.4         | 3.3.4
 postgis_raster               | 3.3.4         | 3.3.4
 postgis_sfcgal               | 3.3.4         |
 postgis_tiger_geocoder       | 3.3.4         | 3.3.4
 postgis_topology             | 3.3.4         |
(6 rows)

If you have the extension installed in the database you are querying, you'll see mention in the installed_version column. If you get no records back, it means you don't have postgis extensions installed on the server at all. PgAdmin III 1.14+ will also provide this information in the extensions section of the database browser tree and will even allow upgrade or uninstall by right-clicking.

If you have the extensions available, you can install postgis extension in your database of choice by either using pgAdmin extension interface or running these sql commands:

CREATE EXTENSION postgis;
CREATE EXTENSION postgis_raster;
CREATE EXTENSION postgis_sfcgal;
CREATE EXTENSION fuzzystrmatch; --needed for postgis_tiger_geocoder
--optional used by postgis_tiger_geocoder, or can be used standalone
CREATE EXTENSION address_standardizer;
CREATE EXTENSION address_standardizer_data_us;
CREATE EXTENSION postgis_tiger_geocoder;
CREATE EXTENSION postgis_topology;

In psql you can use to see what versions you have installed and also what schema they are installed.

\connect mygisdb
\x
\dx postgis*
List of installed extensions
-[ RECORD 1 ]-------------------------------------------------
Name        | postgis
Version     | 3.3.4
Schema      | public
Description | PostGIS geometry, geography, and raster spat..
-[ RECORD 2 ]-------------------------------------------------
Name        | postgis_raster
Version     | 3.0.0dev
Schema      | public
Description | PostGIS raster types and functions
-[ RECORD 3 ]-------------------------------------------------
Name        | postgis_tiger_geocoder
Version     | 3.3.4
Schema      | tiger
Description | PostGIS tiger geocoder and reverse geocoder
-[ RECORD 4 ]-------------------------------------------------
Name        | postgis_topology
Version     | 3.3.4
Schema      | topology
Description | PostGIS topology spatial types and functions
[Warning]

Extension tables spatial_ref_sys , layer , topology can not be explicitly backed up. They can only be backed up when the respective postgis or postgis_topology extension is backed up, which only seems to happen when you backup the whole database. As of PostGIS 2.0.1, only srid records not packaged with PostGIS are backed up when the database is backed up so don't go around changing srids we package and expect your changes to be there. Put in a ticket if you find an issue. The structures of extension tables are never backed up since they are created with CREATE EXTENSION and assumed to be the same for a given version of an extension. These behaviors are built into the current PostgreSQL extension model, so nothing we can do about it.

If you installed 3.3.4, without using our wonderful extension system, you can change it to be extension based by running the below commands to package the functions in their respective extension. Installing using `unpackaged` was removed in PostgreSQL 13, so you are advised to switch to an extension build before upgrading to PostgreSQL 13.

CREATE EXTENSION postgis FROM unpackaged;
CREATE EXTENSION postgis_raster FROM unpackaged;
CREATE EXTENSION postgis_topology FROM unpackaged;
CREATE EXTENSION postgis_tiger_geocoder FROM unpackaged;

2.2.6. Testing

If you wish to test the PostGIS build, run

make check

The above command will run through various checks and regression tests using the generated library against an actual PostgreSQL database.

[Note]

If you configured PostGIS using non-standard PostgreSQL, GEOS, or Proj locations, you may need to add their library locations to the LD_LIBRARY_PATH environment variable.

[Caution]

Currently, the make check relies on the PATH and PGPORT environment variables when performing the checks - it does not use the PostgreSQL version that may have been specified using the configuration parameter --with-pgconfig . So make sure to modify your PATH to match the detected PostgreSQL installation during configuration or be prepared to deal with the impending headaches.

If successful, make check will produce the output of almost 500 tests. The results will look similar to the following (numerous lines omitted below):


     CUnit - A unit testing framework for C - Version 2.1-3
     http://cunit.sourceforge.net/

	.
	.
	.

Run Summary:    Type  Total    Ran Passed Failed Inactive
              suites     44     44    n/a      0        0
               tests    300    300    300      0        0
             asserts   4215   4215   4215      0      n/a
Elapsed time =    0.229 seconds

	.
	.
	.

Running tests

	.
	.
	.

Run tests: 134
Failed: 0


-- if you build with SFCGAL

	.
	.
	.

Running tests

	.
	.
	.

Run tests: 13
Failed: 0

-- if you built with raster support

	.
	.
	.

Run Summary:    Type  Total    Ran Passed Failed Inactive
              suites     12     12    n/a      0        0
               tests     65     65     65      0        0
             asserts  45896  45896  45896      0      n/a


	.
	.
	.

Running tests

	.
	.
	.

Run tests: 101
Failed: 0

-- topology regress

.
.
.

Running tests

	.
	.
	.

Run tests: 51
Failed: 0

-- if you built --with-gui, you should see this too

     CUnit - A unit testing framework for C - Version 2.1-2
     http://cunit.sourceforge.net/

	.
	.
	.

Run Summary:    Type  Total    Ran Passed Failed Inactive
              suites      2      2    n/a      0        0
               tests      4      4      4      0        0
             asserts      4      4      4      0      n/a

The postgis_tiger_geocoder and address_standardizer extensions, currently only support the standard PostgreSQL installcheck. To test these use the below. Note: the make install is not necessary if you already did make install at root of PostGIS code folder.

For address_standardizer:

cd extensions/address_standardizer
make install
make installcheck
	  

Output should look like:

============== dropping database "contrib_regression" ==============
DROP DATABASE
============== creating database "contrib_regression" ==============
CREATE DATABASE
ALTER DATABASE
============== running regression test queries        ==============
test test-init-extensions     ... ok
test test-parseaddress        ... ok
test test-standardize_address_1 ... ok
test test-standardize_address_2 ... ok

=====================
 All 4 tests passed.
=====================

For tiger geocoder, make sure you have postgis and fuzzystrmatch extensions available in your PostgreSQL instance. The address_standardizer tests will also kick in if you built postgis with address_standardizer support:

cd extensions/postgis_tiger_geocoder
make install
make installcheck
	  

output should look like:

============== dropping database "contrib_regression" ==============
DROP DATABASE
============== creating database "contrib_regression" ==============
CREATE DATABASE
ALTER DATABASE
============== installing fuzzystrmatch               ==============
CREATE EXTENSION
============== installing postgis                     ==============
CREATE EXTENSION
============== installing postgis_tiger_geocoder      ==============
CREATE EXTENSION
============== installing address_standardizer        ==============
CREATE EXTENSION
============== running regression test queries        ==============
test test-normalize_address   ... ok
test test-pagc_normalize_address ... ok

=====================
All 2 tests passed.
=====================

2.2.7. Installation

To install PostGIS, type

make install

This will copy the PostGIS installation files into their appropriate subdirectory specified by the --prefix configuration parameter. In particular:

  • The loader and dumper binaries are installed in [prefix]/bin .

  • The SQL files, such as postgis.sql , are installed in [prefix]/share/contrib .

  • The PostGIS libraries are installed in [prefix]/lib .

If you previously ran the make comments command to generate the postgis_comments.sql , raster_comments.sql file, install the sql file by running

make comments-install

[Note]

postgis_comments.sql , raster_comments.sql , topology_comments.sql was separated from the typical build and installation targets since with it comes the extra dependency of xsltproc .

2.3. Installing and Using the address standardizer

The address_standardizer extension used to be a separate package that required separate download. From PostGIS 2.2 on, it is now bundled in. For more information about the address_standardize, what it does, and how to configure it for your needs, refer to Section 14.1, “Address Standardizer” .

This standardizer can be used in conjunction with the PostGIS packaged tiger geocoder extension as a replacement for the Normalize_Address discussed. To use as replacement refer to Section 2.4.3, “Using Address Standardizer Extension with Tiger geocoder” . You can also use it as a building block for your own geocoder or use it to standardize your addresses for easier compare of addresses.

The address standardizer relies on PCRE which is usually already installed on many Nix systems, but you can download the latest at: http://www.pcre.org . If during Section 2.2.3, “Build configuration” , PCRE is found, then the address standardizer extension will automatically be built. If you have a custom pcre install you want to use instead, pass to configure --with-pcredir=/path/to/pcre where /path/to/pcre is the root folder for your pcre include and lib directories.

For Windows users, the PostGIS 2.1+ bundle is packaged with the address_standardizer already so no need to compile and can move straight to CREATE EXTENSION step.

Once you have installed, you can connect to your database and run the SQL:

CREATE EXTENSION address_standardizer;

The following test requires no rules, gaz, or lex tables

SELECT num, street, city, state, zip
 FROM parse_address('1 Devonshire Place PH301, Boston, MA 02109');

Output should be

 num |         street         |  city  | state |  zip
-----+------------------------+--------+-------+-------
 1   | Devonshire Place PH301 | Boston | MA    | 02109

2.3.1. Installing Regex::Assemble

Perl Regex:Assemble is no longer needed for compiling address_standardizer extension since the files it generates are part of the source tree. However if you need to edit the usps-st-city-orig.txt or usps-st-city-orig.txt usps-st-city-adds.tx , you need to rebuild parseaddress-stcities.h which does require Regex:Assemble.

cpan Regexp::Assemble

or if you are on Ubuntu / Debian you might need to do

sudo perl -MCPAN -e "install Regexp::Assemble"

2.4. Installing, Upgrading Tiger Geocoder and loading data

Extras like Tiger geocoder may not be packaged in your PostGIS distribution. If you are missing the tiger geocoder extension or want a newer version than what your install comes with, then use the share/extension/postgis_tiger_geocoder.* files from the packages in Windows Unreleased Versions section for your version of PostgreSQL. Although these packages are for windows, the postgis_tiger_geocoder extension files will work on any OS since the extension is an SQL/plpgsql only extension.

2.4.1. Tiger Geocoder Enabling your PostGIS database: Using Extension

If you are using PostgreSQL 9.1+ and PostGIS 2.1+, you can take advantage of the new extension model for installing tiger geocoder. To do so:

  1. First get binaries for PostGIS 2.1+ or compile and install as usual. This should install the necessary extension files as well for tiger geocoder.

  2. Connect to your database via psql or pgAdmin or some other tool and run the following SQL commands. Note that if you are installing in a database that already has postgis, you don't need to do the first step. If you have fuzzystrmatch extension already installed, you don't need to do the second step either.

    CREATE EXTENSION postgis;
    CREATE EXTENSION fuzzystrmatch;
    CREATE EXTENSION postgis_tiger_geocoder;
    --this one is optional if you want to use the rules based standardizer (pagc_normalize_address)
    CREATE EXTENSION address_standardizer;

    If you already have postgis_tiger_geocoder extension installed, and just want to update to the latest run:

    ALTER EXTENSION postgis UPDATE;
    ALTER EXTENSION postgis_tiger_geocoder UPDATE;

    If you made custom entries or changes to tiger.loader_platform and tiger.loader_variables you may need to update these.

  3. To confirm your install is working correctly, run this sql in your database:

    SELECT na.address, na.streetname,na.streettypeabbrev, na.zip
    	FROM normalize_address('1 Devonshire Place, Boston, MA 02109') AS na;

    Which should output

     address | streetname | streettypeabbrev |  zip
    ---------+------------+------------------+-------
    	   1 | Devonshire | Pl               | 02109
  4. Create a new record in tiger.loader_platform table with the paths of your executables and server.

    So for example to create a profile called debbie that follows sh convention. You would do:

    INSERT INTO tiger.loader_platform(os, declare_sect, pgbin, wget, unzip_command, psql, path_sep,
    		   loader, environ_set_command, county_process_command)
    SELECT 'debbie', declare_sect, pgbin, wget, unzip_command, psql, path_sep,
    	   loader, environ_set_command, county_process_command
      FROM tiger.loader_platform
      WHERE os = 'sh';

    And then edit the paths in the declare_sect column to those that fit Debbie's pg, unzip,shp2pgsql, psql, etc path locations.

    If you don't edit this loader_platform table, it will just contain common case locations of items and you'll have to edit the generated script after the script is generated.

  5. As of PostGIS 2.4.1 the Zip code-5 digit tabulation area zcta5 load step was revised to load current zcta5 data and is part of the Loader_Generate_Nation_Script when enabled. It is turned off by default because it takes quite a bit of time to load (20 to 60 minutes), takes up quite a bit of disk space, and is not used that often.

    To enable it, do the following:

    UPDATE tiger.loader_lookuptables SET load = true WHERE table_name = 'zcta520';

    If present the Geocode function can use it if a boundary filter is added to limit to just zips in that boundary. The Reverse_Geocode function uses it if the returned address is missing a zip, which often happens with highway reverse geocoding.

  6. Create a folder called gisdata on root of server or your local pc if you have a fast network connection to the server. This folder is where the tiger files will be downloaded to and processed. If you are not happy with having the folder on the root of the server, or simply want to change to a different folder for staging, then edit the field staging_fold in the tiger.loader_variables table.

  7. Create a folder called temp in the gisdata folder or wherever you designated the staging_fold to be. This will be the folder where the loader extracts the downloaded tiger data.

  8. Then run the Loader_Generate_Nation_Script SQL function make sure to use the name of your custom profile and copy the script to a .sh or .bat file. So for example to build the nation load:

    psql -c "SELECT Loader_Generate_Nation_Script('debbie')" -d geocoder -tA > /gisdata/nation_script_load.sh
  9. Run the generated nation load commandline scripts.

    cd /gisdata
    sh nation_script_load.sh
  10. After you are done running the nation script, you should have three tables in your tiger_data schema and they should be filled with data. Confirm you do by doing the following queries from psql or pgAdmin

    SELECT count(*) FROM tiger_data.county_all;
     count
    -------
      3233
    (1 row)
    SELECT count(*) FROM tiger_data.state_all;
     count
    -------
        56
    (1 row)
    
  11. By default the tables corresponding to bg , tract , tabblock are not loaded. These tables are not used by the geocoder but are used by folks for population statistics. If you wish to load them as part of your state loads, run the following statement to enable them.

    UPDATE tiger.loader_lookuptables SET load = true WHERE load = false AND lookup_name IN('tract', 'bg', 'tabblock');

    Alternatively you can load just these tables after loading state data using the Loader_Generate_Census_Script

  12. For each state you want to load data for, generate a state script Loader_Generate_Script .

    [Warning]

    DO NOT Generate the state script until you have already loaded the nation data, because the state script utilizes county list loaded by nation script.

  13. psql -c "SELECT Loader_Generate_Script(ARRAY['MA'], 'debbie')" -d geocoder -tA > /gisdata/ma_load.sh
  14. Run the generated commandline scripts.

    cd /gisdata
    sh ma_load.sh
  15. After you are done loading all data or at a stopping point, it's a good idea to analyze all the tiger tables to update the stats (include inherited stats)

    SELECT install_missing_indexes();
    vacuum (analyze, verbose) tiger.addr;
    vacuum (analyze, verbose) tiger.edges;
    vacuum (analyze, verbose) tiger.faces;
    vacuum (analyze, verbose) tiger.featnames;
    vacuum (analyze, verbose) tiger.place;
    vacuum (analyze, verbose) tiger.cousub;
    vacuum (analyze, verbose) tiger.county;
    vacuum (analyze, verbose) tiger.state;
    vacuum (analyze, verbose) tiger.zip_lookup_base;
    vacuum (analyze, verbose) tiger.zip_state;
    vacuum (analyze, verbose) tiger.zip_state_loc;

2.4.1.1. Converting a Tiger Geocoder Regular Install to Extension Model

If you installed the tiger geocoder without using the extension model, you can convert to the extension model as follows:

  1. Follow instructions in Section 2.4.5, “Upgrading your Tiger Geocoder Install” for the non-extension model upgrade.

  2. Connect to your database with psql or pgAdmin and run the following command:

    CREATE EXTENSION postgis_tiger_geocoder FROM unpackaged;

2.4.2. Tiger Geocoder Enabling your PostGIS database: Not Using Extensions

First install PostGIS using the prior instructions.

If you don't have an extras folder, download http://download.osgeo.org/postgis/source/postgis-3.3.4.tar.gz

tar xvfz postgis-3.3.4.tar.gz

cd postgis-3.3.4/extras/tiger_geocoder

Edit the tiger_loader_2015.sql (or latest loader file you find, unless you want to load different year) to the paths of your executables server etc or alternatively you can update the loader_platform table once installed. If you don't edit this file or the loader_platform table, it will just contain common case locations of items and you'll have to edit the generated script after the fact when you run the Loader_Generate_Nation_Script and Loader_Generate_Script SQL functions.

If you are installing Tiger geocoder for the first time edit either the create_geocode.bat script If you are on windows or the create_geocode.sh if you are on Linux/Unix/Mac OSX with your PostgreSQL specific settings and run the corresponding script from the commandline.

Verify that you now have a tiger schema in your database and that it is part of your database search_path. If it is not, add it with a command something along the line of:

ALTER DATABASE geocoder SET search_path=public, tiger;

The normalizing address functionality works more or less without any data except for tricky addresses. Run this test and verify things look like this:

SELECT pprint_addy(normalize_address('202 East Fremont Street, Las Vegas, Nevada 89101')) As pretty_address;
pretty_address
---------------------------------------
202 E Fremont St, Las Vegas, NV 89101
			

2.4.3. Using Address Standardizer Extension with Tiger geocoder

One of the many complaints of folks is the address normalizer function Normalize_Address function that normalizes an address for prepping before geocoding. The normalizer is far from perfect and trying to patch its imperfectness takes a vast amount of resources. As such we have integrated with another project that has a much better address standardizer engine. To use this new address_standardizer, you compile the extension as described in Section 2.3, “Installing and Using the address standardizer” and install as an extension in your database.

Once you install this extension in the same database as you have installed postgis_tiger_geocoder , then the Pagc_Normalize_Address can be used instead of Normalize_Address . This extension is tiger agnostic, so can be used with other data sources such as international addresses. The tiger geocoder extension does come packaged with its own custom versions of rules table ( tiger.pagc_rules ) , gaz table ( tiger.pagc_gaz ), and lex table ( tiger.pagc_lex ). These you can add and update to improve your standardizing experience for your own needs.

2.4.4. Loading Tiger Data

The instructions for loading data are available in a more detailed form in the extras/tiger_geocoder/tiger_2011/README . This just includes the general steps.

The load process downloads data from the census website for the respective nation files, states requested, extracts the files, and then loads each state into its own separate set of state tables. Each state table inherits from the tables defined in tiger schema so that its sufficient to just query those tables to access all the data and drop a set of state tables at any time using the Drop_State_Tables_Generate_Script if you need to reload a state or just don't need a state anymore.

In order to be able to load data you'll need the following tools:

  • A tool to unzip the zip files from census website.

    For Unix like systems: unzip executable which is usually already installed on most Unix like platforms.

    For Windows, 7-zip which is a free compress/uncompress tool you can download from http://www.7-zip.org/

  • shp2pgsql commandline which is installed by default when you install PostGIS.

  • wget which is a web grabber tool usually installed on most Unix/Linux systems.

    If you are on windows, you can get pre-compiled binaries from http://gnuwin32.sourceforge.net/packages/wget.htm

If you are upgrading from tiger_2010, you'll need to first generate and run Drop_Nation_Tables_Generate_Script . Before you load any state data, you need to load the nation wide data which you do with Loader_Generate_Nation_Script . Which will generate a loader script for you. Loader_Generate_Nation_Script is a one-time step that should be done for upgrading (from 2010) and for new installs.

To load state data refer to Loader_Generate_Script to generate a data load script for your platform for the states you desire. Note that you can install these piecemeal. You don't have to load all the states you want all at once. You can load them as you need them.

After the states you desire have been loaded, make sure to run the:

SELECT install_missing_indexes();

as described in Install_Missing_Indexes .

To test that things are working as they should, try to run a geocode on an address in your state using Geocode

2.4.5. Upgrading your Tiger Geocoder Install

If you have Tiger Geocoder packaged with 2.0+ already installed, you can upgrade the functions at any time even from an interim tar ball if there are fixes you badly need. This will only work for Tiger geocoder not installed with extensions.

If you don't have an extras folder, download http://download.osgeo.org/postgis/source/postgis-3.3.4.tar.gz

tar xvfz postgis-3.3.4.tar.gz

cd postgis-3.3.4/extras/tiger_geocoder/tiger_2011

Locate the upgrade_geocoder.bat script If you are on windows or the upgrade_geocoder.sh if you are on Linux/Unix/Mac OSX. Edit the file to have your postgis database credentials.

If you are upgrading from 2010 or 2011, make sure to unremark out the loader script line so you get the latest script for loading 2012 data.

Then run th corresponding script from the commandline.

Next drop all nation tables and load up the new ones. Generate a drop script with this SQL statement as detailed in Drop_Nation_Tables_Generate_Script

SELECT drop_nation_tables_generate_script();

Run the generated drop SQL statements.

Generate a nation load script with this SELECT statement as detailed in Loader_Generate_Nation_Script

For windows

SELECT loader_generate_nation_script('windows'); 

For unix/linux

SELECT loader_generate_nation_script('sh');

Refer to Section 2.4.4, “Loading Tiger Data” for instructions on how to run the generate script. This only needs to be done once.

[Note]

You can have a mix of 2010/2011 state tables and can upgrade each state separately. Before you upgrade a state to 2011, you first need to drop the 2010 tables for that state using Drop_State_Tables_Generate_Script .

2.5. Common Problems during installation

There are several things to check when your installation or upgrade doesn't go as you expected.

  1. Check that you have installed PostgreSQL 11 or newer, and that you are compiling against the same version of the PostgreSQL source as the version of PostgreSQL that is running. Mix-ups can occur when your (Linux) distribution has already installed PostgreSQL, or you have otherwise installed PostgreSQL before and forgotten about it. PostGIS will only work with PostgreSQL 11 or newer, and strange, unexpected error messages will result if you use an older version. To check the version of PostgreSQL which is running, connect to the database using psql and run this query:

    SELECT version();

    If you are running an RPM based distribution, you can check for the existence of pre-installed packages using the rpm command as follows: rpm -qa | grep postgresql

  2. If your upgrade fails, make sure you are restoring into a database that already has PostGIS installed.

    SELECT postgis_full_version();

Also check that configure has correctly detected the location and version of PostgreSQL, the Proj library and the GEOS library.

  1. The output from configure is used to generate the postgis_config.h file. Check that the POSTGIS_PGSQL_VERSION , POSTGIS_PROJ_VERSION and POSTGIS_GEOS_VERSION variables have been set correctly.