Convert ESRI data (shp, gdb, tables, etc) to .arches, used to upload shapefiles to the Arches platform.
Switch branches/tags
Nothing to show
Clone or download


Convert ESRI spatial data formats to the .arches format for upload to an Arches v3 deployment. Arc2arches includes an ESRI toolbox (.tbx) for use in ArcMap or ArcCatalog with tools that will help to configure and ultimately convert input datasets (any dataset read by arcpy, such as a file geodatabase table or shapefile) to the .arches format. Once a .arches file has been created, you can upload it to arches usings the built-in arches command line operations (python packages -o load_resources -s path/to/.arches/file).

These tools require the cohesive set of resource graphs and authority documents that were used during database installation. The choice was made to require these files because it makes things so much easier, and it allows for great quality control and error handling.


The arc2arches tools require two non-standard python packages: unicodecsv and pyshp. One way to fill these dependencies is install these packages with pip or easy_install. This is the recommended route, because it's good to understand pip/easy_install.

However, for ease of use, the unicodecsv and pyshp packages have been included in the arc2arches package, so if you do not want to deal with pip/easy_install (or don't have permission), you can just skip this step and the tools will use the packages that have already been included.

Also, at this point the geometry is written using tokens only made available at ArcGIS v10.1 SP 1. A more robust way of writing geometry could be implemented in the future.

simple use

Download and unzip (or clone or fork) this repository. Using ArcMap or ArcCatalog, navigate to the .tbx that is in the archestools directory (don't use the archestools_testing.tbx right now).

  1. Create a "conflig" file to accompany the dataset you plan to convert. This file holds the path to the dataset and some other basic metadata. It's also where you define all field mapping (next step).
  2. Add groups to the conflig file to correctly map specific fields in the dataset to existing arches nodes. Error handling has been added that relies completely on your existing authority documents. NOTE: The current tool interface only allows for the addition of 10 entity:field entries into a new group. However, you can open the .conflig file at any time in any text editor and copy/paste/write as many entries into a group as you need.
  3. Convert the dataset(s) to .arches format. All input datasets will be merged into the same .arches file. A (required) .relations file will be created as well. The handling of relationships is described below, and is ready for improvement...

Check out the official Arches v3.0 documentation for direction on how to upload the .arches file to your Arches installation.

relationships between resources

At present, you are able to automate relationships between uploaded resources in a useful but limited manor. When using the convert to .arches tool, you are able to choose a field from each input dataset whose value will be matched with values in other selected fields in other selected datasets. At this point, all relationship types default to RELATIONSHIP_TYPE:1. The following two examples will illustrate the good and bad qualities of the way that relationships are handled currently.

relationship example 1, cemetery data

There are two datasets: One with a polygon for every grave in a cemetery, and each polygon has a plot number stored in field "plot_id". The other is a geodatabase table of all the names of individuals that are interred in these graves, with the interment grave number stored in a field also named "plot_id". Not all graves have (indentified) individuals in them, and some graves hold more than on individual.

While running the Convert to .arches tool, the "plot_id" fields are selected as the relate fields. When the tool has finished, relationships will have been entered in the .relations file that will connect all resources that have the same plot_id value.

relationship example 2, building survey data

A windshield survey was carried out, and the information to make a single Activity resource to represent it in Arches is entered in a geodatabase table. A field is added named "surv" and its value is 'windshield'. The survey data itself is stored in a shapefile which is actually an aggregate of many building datasets. A field is added named "survname" and all building features that were part of the windshield survey are given the value "windshield" (all others left blank).

When running the tool, the "surv" and "survname" fields are chosen as relate fields for the survey table and the building shapefile, respectively. When the conversion is finished, the .relations file will have created a relationship between EVERY resource that had the same value in the relate fields. This means that in addition to each building being associated with the windshield survey activity resource, it is also related to every other building that was part of the survey. In this case, editing was done directly to the resulting .relations file in excel to remove all unwanted relationships between buildings. (Open in excel, choose "|" as the delimiter character, and then save as tab delimited csv and use notepad++ to replace all tabs with "|"...)

standalone script

This script is intended to be used in a command-line, preferably within the package root directory so the authority documents paths can be imported from It is in very rough shape.

planned improvements

The current intent is to greatly improve the relationship handling. At this point, a new interface has been created for the "3" tool, which you can see in the archestools_testing.tbx toolbox. The idea is to define all datasets, and then allow the user to create specific types of relationships between any two datasets, using matching source/target fields.

Another recent change is to add the DATASET_PATH as a property of the conflig file, so users only have to enter a conflig file, and the path to it's accompanying dataset will be automatically found. All in the name of reducing the amount of user input.

In tool "2", it would be good to allow the user to type in a value instead of choosing a field. So Entry A would have 3 parameters: entity name, field name, and single value. You would not be able to choose a field name and a single value, it would have to be one or the other. If the entity chosen uses an E32 authority document, that document should be referenced to create a dropdown list in the new single value parameter. This would eliminate the need to make a new field in the dataset named "name_type" (for example), and populate it with "Primary" for each row -- i.e. it would reduce the amount of work needed to prepare any given dataset for conversion. However, this would require support in the "3" tool, because that tool expects field names in the .conflig file, not actual values.