Permalink
Fetching contributors…
Cannot retrieve contributors at this time
285 lines (217 sloc) 18.8 KB

Contributing to OpenAddresses

Build Status

Reporting Sources & Issues

We'd love to hear about a new address source, fixes to an old one, or make improvements.

OpenAddresses is a collection of authoritative data for address locations around the world. We collect address data from authoritative sources; we do not create our own data. A source is a location where authoritative address data can be found. Examples might be a downloadable CSV file or live ArcServer feature service hosted by a national postal service, a state GIS department, or a county property parcel database.

New Sources

Have a potential source? Fantastic! Follow these steps to help us get it into the project as fast as possible!

  • Check ./sources/ to make sure we don't have it already. Sources that overlap are ok as long as they are coming from different providers.
  • Check the wiki to make sure it isn't listed there.

Still a new source? Awesome!

  • If the source is raster data (images/webmap/not downloadable) please add it to the Raster Wiki
  • If the source costs money, please add it to the Commercial Wiki
  • If the source is parcel data, please add it to the Parcel Wiki
  • If you have an awesome link/contact but no data, add it to the Outreach Wiki
  • Finally if you have raw data, open a issue and we'll add it for you or add it yourself

Confused? Not sure where your source fits in? Open an issue, we’d rather a duplicate than miss it altogether!

Errors & Current Sources

If you are reporting an error, an improvement, or a suggestion, create a github issue here. We will do our best to review your issue and either fix or add the feature(s) to our plan.

Contributing Sources

Comfortable with JSON? Feel free to submit a pull request with the data instead of opening an issue. Before asking for a merge, please keep Travis CI happy by making sure you submit well-formed JSON: green is good!

For a first time contributor, getting the JSON right can be a bit of a challenge, check out other sources in ./sources/ to get an idea of what we are looking for. Still confused? Open an issue, we’ll be happy to help you out!

Naming Files

Although the file name is redundant (the same information is stored in JSON), using coherent file names makes it much easier for contributors to quickly evaluate whether a source already exists. Please use the following conventions.

/sources/{Country}/{Region}/{Source}

Directories

Coverage Code
Country ISO 3166-1 alpha-2 Country Code
Province ISO 3166-2 alpha-2 SubRegion Code

Source

Coverage Example
State us/md/statewide.json
County us/md/montgomery.json
City us/md/city_of_baltimore.json

JSON Tags

Sources use a standard set of attributes to allow for machine processing of each source. Use these tags where applicable. Check out other sources in the ./sources/ directory for examples.

Core Tags

Our testing platform, Travis CI, enforces required tags. These represent the minimum data required to process a source. If you are not able to provide the required tags, file an issue instead of a pull request. We’ll determine if the data is suitable for inclusion.

Tag Required? Note
data Yes A URL referencing the dataset. This should point to the raw data and not a web portal. If there isn't a good URL for the source, members of the OpenAddresses GitHub organization can upload files to https://results.openaddresses.io/upload-cache which provides a cached URL.
type Yes A string containing the protocol (One of: http, ftp, ESRI)
coverage Yes An object containing some combination of country, state, and either city or county. Each of which contain a String. See below for more details
conform Optional Object used to find address information in a source. See below for more details.
compression Optional string containing the compression type (usually zip). Omit if source is not compressed.

Conform Object

Although a conform Object is not a mandatory part of a source, it must be present for the source to be added into the end address file. This JSON Object tells the processing software how to handle the format as well as mapping fields to a standard set.

Conform contains tags for preparing data and tags for converting it. They are called Processing Tags and Attribute Tags.

Processing Tags
Tag Required? Note
type Yes The type properties stores the format. It can currently be one of gdb, shapefile, shapefile-polygon, csv, geojson, or xml (for GML).
srs Allows one to set a custom source srs. Currently only supported by type:shapefile, type:shapefile-polygon, and type:csv. Should be in the format of EPSG:#### and can be any code supported by ogr2ogr. Modern shapefiles typically store their projection in a .prj file. If this file exists, this tag should be omitted.
layer The gdb source type allows multiple layers of geodata in a single input file. Use the layer tag to specify which of those layers to use. It can either be the string name of the layer or an integer index of the layer.
file The majority of zips contain a single shapefile. Sometimes zips will contain multiple files, or the shapefile that is needed is located in a folder hierarchy in the zip. Since the program only supports determining single shapefiles not in a subfolder, file can be used to point the program to an exact file. The proper syntax would be "file": "addresspoints/address.shp" if the file was under a single subdirectory called addresspoints. Note there is no preceding forward slash.
encoding A character encoding from which an input file will first be converted (into utf-8). Must be recognizable by iconv.
csvsplit The character to delimit input CSV’s by. Defaults to comma.
headers (type 'csv' only) Some non-latin CSVs provide header lines in native script and in latin characters. If specified, this tag determines which line will be used to determine field names for other conform tags. If not specified, row 1 is assumed. Alternately, if a CSV file lacks headers, setting headers=-1 will add them. The autogenerated fields will be named COLUMN1, COLUMN2, COLUMN3 ... COLUMN10, COLUMN11 etc.
skiplines (type 'csv' only) May be used in conjunction with headers (see above). For example, if headers is 1 but a second header line exists and must be skipped.
accuracy The accuracy of the data source. See table below. Should never be 0, defaults to 5. If this is not set, address duplicates of higher accuracy will replace the addresses from this source when they are conflated.
Accuracy
ID Type
0 User Corrected
1 Rooftop
2 On Parcel
3 Driveway
4 Interpolation
5 Unknown
Attribute Tags

Attribute tags are functions or field names for mapping the source data into a given format.

Tag Required? Note
number Yes The name of the number field. This will either be the name of the field or a function.
street Yes The name of the street field. This will either be the name of the field, an ordered list of fields to merge together, or some other function.
unit The Unit number. This is an optional attribute and can be used for uniquely identifying addresses where the street number is the same but a "unit" number is specified, i.e. 300 Willow Valley Lakes Drive #3A where 3A can be found as a separate attribute in the source.
lon The longitude column. This is required for CSV sources and should be omitted for other types.
lat The latitude field. This is required for CSV sources and should be omitted for other types.
city Name of the City or Municipality in which the address falls
postcode Postcode or zip-code field in which the address falls
district District/County/Sub-Region in which the address falls
region State/Region/Province in which the address falls
id Unique identifier, such as a parcel APN.
addrtype Type of address. industrial, residential, etc.
notes Legal description of address or notes about the property.
Assembling Attributes

In many sources, attribute values are provided in either single fields or several fields to be merged together. Often times the number attribute is a single field in the source and the street attribute is merged together from several fields (see Alameda County, California).

Single Source Field

The most basic usage is simply mapping an attribute tag to a field in the source data.

Format

"{Attribute Tag}": "{Field Name}"

Example

"number": "SITUS_NUMBER",
"street": "SITUS_STREET"
Merged Fields

Often times there are multiple fields that should be merged together. An array ([]) of field names can be used with any attribute tag. The field names will joined with a space ().

Format

"{Attribute Tag}": ["{Field Name}"]

Example

"number": "SITUS_NUMBER",
"street": ["SITUS_STREET_PRE", "SITUS_STREET_NME", "SITUS_STREET_TYP", "SITUS_STREET_POST"]
Attribute Functions

Some sources do not offer data nicely separated into distinct fields so advanced techniques must be used extract and format values appropriately. Attribute functions allow basic text manipulation to be performed on any of the given attribute tags. This list gives a brief summary of what each function does. For more information and examples regarding attribute functions, click here.

Function Note
join Allow multiple fields to be joined with a given delimiter
format Allow multiple fields to be formatted into a single string
prefixed_number Allow number to be extracted from the beginning of a single field (extracts 102 from 102 East Maple Street)
postfixed_street Allow street to be extracted from the end of a single field (extracts East Maple Street from 102 East Maple Street)
postfixed_unit Allow unit to be extracted from the end of a single field (extracts Apt 4A from 102 East Maple Street Apt 4A)
remove_prefix Removes a field value from the beginning of another field value
remove_postfix Removes a field value from the end of another field value
regexp Allow regex find and/or replace on a given field. Useful to extract house number/street/city/region etc when the source has them in a single field

Sources vary in how they store data so several approaches to conforming attributes may apply.

The lon and lat attribute tags don't support attribute functions.

Coverage Object

Although a coverage Object is not a mandatory part of a source, its presence provides hints about the geographic extent of the address file and is used to render the map at data.openaddresses.io.

This object minimally contains some combination of country, state, and either city or county, all strings.

If one of the following tags are provided, it will be used to render the source to the map at data.openaddresses.io:

  1. US Census with geoid containing FIPS code
  2. ISO 3166 with alpha2 containing alphanumeric two-letter ISO-3166-1 country code or ISO-3166-2 subdivision code. See New Zealand, Victoria, Australia, or Dolnośląskie, Poland for examples.
  3. geometry with unprojected GeoJSON geometry object

Optional Tags

Although these tags are optional, their inclusion is very much appreciated. Additional metadata helps future proof the project!

Tag Note
website A URL referencing the data portal
license An object with license details for the dataset. Supported properties:

url: link to license terms, e.g. “http://creativecommons.org/licenses/by/4.0/”.
text: short description of license terms, e.g. “CC BY 4.0”.
attribution: Boolean value for required attribution to copyright holder. Defaults to true if other attribution details are present, otherwise false.
attribution name: name of data copyright holder requiring attribution, e.g. “United Federation of Planets”.
share-alike: Boolean value for requirement to license derivative works under the same terms or compatible terms as the original work. Defaults to false.
presumed: Boolean value. true if an OpenAddresses community member interpreted the license to derive the share-alike or attribution booleans above. false if the license explicitly includes share-alike or attribution requirements (like CC-BY or CC-BY-SA).
remarks: A fully qualified URL to a file (PDF or text document) of the full license of the source.

Deprecated value: a URL or string describing the license.
contact An object with contact information for the dataset or provider. Supported properties:

name: the name of a specific contact person, e.g. "Mary Smith".
title: the title for the contact person, e.g. "GIS Coordinator".
phone: A contact telephone number, e.g. "415-867-5309".
email: email address for contact person or organization, e.g. "mary@example.com".
address: A mailing address for the dataset provider, e.g. "500 Main Street, Washington, DC 10001".
note A String containing a human readable note.
attribution Deprecated: Where the license requires attribution, add it here. example CC-BY United Federation of Planets
email This email is used to send automated emails to the data provider if a user changes their data. Do not set unless the data provider wants to receive updates. Distinct from contact email in that the provider might want to send these notifications to a different mailing list than the contact email listed above. Note that this property is not used by any OpenAddresses software at this point.
language ISO 639-1, ISO 639-2, or ISO 639-3 code for the language of the data. For example: en, fr, de, or lld.

Example

{
    "coverage": {
        "country": "ca",
        "state": "nb"
    },
    "data": "http://geonb.snb.ca/downloads/gcadb/geonb_gcadb-bdavg_shp.zip",
    "website": "http://www.snb.ca/geonb1/e/DC/catalogue-E.asp",
    "license": {"url": "http://geonb.snb.ca/downloads/documents/geonb_license_e.pdf"},
    "type": "http",
    "compression": "zip",
    "language": "en",
    "conform": {
        "number": "civic_num",
        "street": "street_nam",
        "type": "shapefile",
        "accuracy": 3
    }
}

Language

Names of places, cities, addresses etc. are often translated to various languages. For example: Munich, Bavaria, Germany vs. München, Bayern, Deutschland. The optional language tag defines, which language is in use in the metadata entry. A data source usually includes addresses in a single language (e.g. Montreal in Canada contains only French names). Some data sources can define several translations of address components under appropriate csv column labels. For example, Brussels in Belgium has both Dutch and French names. Such a bilingual data source can be linked to OpenAddresses with two separate metadata entries, one for reading the French addresses and one for reading the Dutch addresses. An application,which uses OpenAddresses and wishes to generate multilingual address entries, can access the data via both metadata entries and merge the language versions by identifying the address items by their id unique identifier tag.

Formatting:

A few notes on formatting:

  • If a tag is empty please do not set it to null or an empty string. Omit it entirely.
  • No commas at the beginning of a line, only at the end.
  • Four spaces for indents (No tabs!)
  • No Blank lines

Although these are read by a machine, they are maintained by us mortals. Following the formatting guidelines keeps the rest of us sane!