Specification of the tabular file format CSV (Comma Separated Values) with an optional geometry extension!
>> DRAFT Version 0.3 - Date of last modification: see bottom <<
Author: Stefan and contributors. (For notes and discussion see Diskussion:GeoCSV).
GeoCSV is an extension of the "human readable", tabular file format Comma-Separated Values (CSV) and/or Tab-Separated Values (TSV). CSV/TSV are well-known but spartanic format with possible information loss.
For exchanging geospatial data think about using more capable and elegant formats file exchange like e.g. GeoPackage. One the other hand it has some potential since it's quite more capable as e.g. a Shapefile. See also TheShapefileChallenge.
This format has following drawbacks:
- not suited for massive datasets (except when compressing/zipping)
- only one layer per file
- no layer name - except for the file name (which can be changed easily by others...).
- many other drawbacks like auxiliary cluttered accompanying files, like .csvt and .prj which it shares with Shapefiles.
GeoCSV file format specification
GeoCSV is based on the CSV specification (see following section) and comes with two variants: Option Point(X/Y) and Option WKT.
Option "WKT" (preferred):
- It's one single column of type String containing a constructor, like for example: "POINT (8.8249 47.2274)", meaning 8.8249 east and 47.2274 north (lon/lat).
- Note that WKT uses lon,lat notation.
- This option supports Point, LineString, Polygon, MultiPoint, MultiLineString, MultiPolygon and even GeometryCollection and ARCs!
- WKT ("Well Known Text") is defined by the Open Geospatial Consortium (OGC) and described in their "Simple Feature Access Specification" (also ISO SQL/MM). See e.g. .
- Geometry Point type as two columns, meaning an easting/northing coordinate pair (longitude/latitude, lon/lat, long/lat, sometimes also called x/y).
- Example for two lon and lat (Point) coordinate columns is "8.8249;47.2274".
- This option supports only Points.
- See "CSVT file format specification" below.
- There are more than one geometry columns allowed per sheet but only one column(-pair) can have either type Point(X/Y) or WKT.
- Coordinate system is WGS84 (EPSG:4326) by default. See section PRJ.
- All geometry values within one table are in the same coordinate reference system (CRS).
Optional auxiliary files (with same base filename but different file extensions) are:
- Contains field type information (schema).
- File extension is .CSVT (or .csvt).
- See section below.
- PRJ (to be clarified!):
- Contains Coordinate Reference System (CRS) information.
- File extension is .PRJ (or .prj).
- Default is EPSG:4326 (WGS84, lon/lat).
- File extension is .CSVZ (or .csvz)
- The CSV file can be accompanied with following files, having the same file base name: .csvt and .prj.
- Archiving and compressing in format .ZIP (or .zip) is also possible and encouraged.
CSV file format specification
- Contains the actual (geo-)data.
- File extension is .CSV (or .csv).
- Character Encoding and character set is UTF-8 (default) or ANSI/Windows-1252(?).
- End-of-lines are: CR, LF or CR/LF (unless embedded in parantheses).
- At the end of the file there may be an empty line.
- First row contains attribute names separated by a => delimiter.
- Following rows are contains values separated by a => delimiter.
- All rows have same number of attributes.
- Field delimiter default (and preferred) is semicolon (;) unless defined otherwise (Note: CSVT delimiter uses comma).
- Strings are enclosed by parantheses, to allow delimiters inside (e.g. "string").
- Data types (if supported from source or target system): See CSVT file format specification.
- Line End-of-lines (in String) fields are not recommended use e.g. HTML is needed); they are only allowed in strings within parantheses (see rfc).
- Calculations are not part of this spec.
See also CSV.
CSVT file format specification
CSVT means "CSV Types" and it describes the field types and eventually subtypes or properties separated by comma.
Field/column types, case insensitive, eventually in quotes ('"'):
- Integer or "Integer".
- Real or "Real".
- String or "String".
- Date (format "YYYY-MM-DD"), Time (format "HH:MM:SS+nn") and DateTime (format "YYYY-MM-DD HH:MM:SS+nn"), whereas nn is the timezone.
- "WKT" (preferred over Point(X/Y)). All WKT geometry types are allowed: Point, LineString, Polygon, Multipoint, MultiLinestring, MultiPolygon, GeometryCollection, Arcs, ... (see OGC WKT).
- "CoordX","CoordY" (preferred) or "Point(X)","Point(Y)". Two separate colums in either order and not necessary neighboring of type Integer or Float: one containing the easting coordinate, and another containing northing coordinate separated by a comma.
- All values of that WKT column MAY contain the same geometry (sub)type.
- Note that CSVT fields are separated by commas while GeoCSV fields are separated by semicolons.
- Types can be in quotes ('"') or not, e.g. <<"Integer";"Real">>.
- Types can have precision in parantheses, e.g. ('Real(20.2)')).
- There's only one geometry column per .csvt, namely either "Point(X)","Point(Y)" (or "CoordX","CoordY") or "WKT".
- Geometry types are a kind of subtypes: CoordX,CoordY values are stored as integer or float, a WKT field is stored in one single String column.
- See also http://www.gdal.org/drv_csv.html section with .csvt extension.
- (Enhancement issue: There could be more properties like "mandatory/optional" or, for strings field length, and for numbers precision etc.)
PRJ file format specification
- Default is EPSG:4326 (WGS84, geographic, geo-centered longitude/latitude).
- Contains a named CRS, i.e. the EPSG number "EPSG:nnnn" in OGR WKT format (the one natively spoken by OGR/GDAL, based on OGC 01-009).
- Desktop GIS:
- QGIS with Editable GeoCSV QGIS Plugin *** featured ***
- QGIS through usual "Add Vector Layer..." dialog
- OGR Version 2.x
- Desktop generic:
- CSV-to-GeoJSON: convertcsv.com, csv2geojson
- GeoJSON-to-CSV: convertcsv.com
CSV type file 'example1.csvt' - Option Point(X/Y):
CSV file 'example1.csv - Option Point(X/Y):
id;name;amount;city;lon;lat 1;Kevin;2.1;Rapperswil;8.8249;47.2274 2;Eva;2.2;Zürich;8.5435;47.3768 3;"Jimmy;Muff";2.3;;7.4397;46.9487
CSV type file 'example2.csvt' - Option WKT:
CSV file 'example2.csv - Option WKT:
id;name;amount;city;geom 1;Kevin;2.1;Rapperswil;POINT(8.8249 47.2274) 2;Eva;2.2;Zürich;POINT(8.5435 47.3768) 3;"Jimmy;Muff";2.3;;POINT(7.4397 46.9487)
...can be shown as following table:
Note the remarks string in row 2 and the empty string in row 3.
- GeoJSON Spec. at IETF: [http://tools.ietf.org/html/draft-butler-geojson-04
- rfc4180: http://tools.ietf.org/html/rfc4180
- TSV: http://www.cs.tut.fi/~jkorpela/TSV.html
- Super-CSV: http://super-csv.github.io/super-csv/csv_specification.html
- Tabular Data Package: http://dataprotocols.org/tabular-data-package/
- OGR CSV driver: http://www.gdal.org/drv_csv.html
- Google Earth: https://support.google.com/earth/answer/148104?hl=en