Skip to content

zifredder/IBdb-matlab

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

9 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

IBdb-matlab

API request handler for the IBdb (publication).

Getting started

Requirements
  • (Optional) Become a member of the IBdb community and request API usage permission.
  • (Required) MATLAB R2019a or newer
  • (Optional) Login credential input dialog with passfield
  • (Required for some functions) Python (tested with version 3.9) and the requests package.
Object construction and authentication
obj = IBdb; % create an IBdb object

During object construction, the user can authenticate for full API usage access in different ways. The easiest is providing a text file with the user's login credentials in the form username,password (single line, comma-delimited). If no file is provided, the user is prompted to type in their credentials. If this is aborted, no login is communicated with the data base; consequently, the user only has access to public API functions. This login process can be done without user interaction:

obj = IBdb('my_credentials_file.txt'); % provide the path to the credentials text file
obj = IBdb('Token asdf123'); % provide the TokenAuth string as acquired from the data base's Swagger UI
Make a request

Use a predefined function for a specific endpoint:

obj.neuron_get(111)
ans =
  struct with fields:

                        id: 111
        persistent_records: {["20.500.12158/NIN-0000111.1"]}
               uploaded_by: [1×1 struct]
                       nin: "NIN-0000111"
                 full_name: "sg-CPU1a-R2"
                      name: "CPU1a-R2"
                short_name: "CPU1a-R2"
                       sex: "UNKNOWN"
                       <...>

The same can be achieved by manually specifying request method and URL:

obj.request("get", "/neuron/111/");

Query parameters are (usually) passed as name value pairs to Matlab functions:

obj.neuron_query("short_name", "CPU1a-R2").id; % 111

% Add a neuron to the data base:
resp = obj.neuron_post("name", "test-asdf", "short_name", "test-asdf", "hemisphere", "u", "species", 9);
% And delete it:
obj.neuron_delete(resp.id);

For more code examples, see examples/.

Notes

Just as the API itself, this repository is a beta version. The API and this repository are subject to continuous change; they are not maintained by the same person, so if API endpoints are changed, parts of this code will break.

Pull requests are welcome!

Enumeration input is designed to be convenient and robust. E. g., for "hemisphere", accepted values are "LEFT", "RIGHT", "UNPAIRED" (case sensitive). However, you can pass "r" as value for the "hemisphere" parameter in request functions; it will automatically be parsed as "RIGHT".

Developer notes

  • Use strings where possible, not chars.
  • Use RESTful request method names in function signatures if possible. E. g., not neuron_add(), but neuron_post().
Function argument validation

Consistently use arguments with name value pairs. This is convenient because the name value pairs can be easily passed as parameter pairs for requests after converting them with namedargs2cell. Downside: Name value pairs are always optional, so it has to be manually validated which name value pairs are required. Check out IBdb.convertSchemaToArguments() to save some time when writing new functions. There may be exceptions from this rule in cases where using arguments is unnecessary overload, e. g., when a function requires only one or two inputs that are both not directly passed as parameters to the data base.

Use query parameter names for MATLAB functions literally like they are accepted by the data base (with reasonable exceptions). One exception would e. g. be POSTing data associated to an existing neuron: addDataToNeuron(neuron_id) makes more sense (to me) than addDataToNeuron('neuron', id, ...).

Concerning parameter validation: Validation in MATLAB functions could be skipped and handled by the data base. It is debatable which approach is better. Currently, there is quite some validation implemented in MATLAB, but this is subject to change.

Argument validation for parameters in requests: In the arguments block, use

  • arg {IBdb.mustBeEmptyOrScalarInteger} for scalar integer input
  • arg string {IBdb.mustBeEmptyOrScalarText} for string input
  • arg {IBdb.mustBeEmptyOrLogical} for scalar boolean input
  • arg {mustBeInteger} for integer array input If you are absolutely sure that a parameter is mandatory for the request and therefore must not be empty, specifying the size with arg (dim1,dim2) <class> works, too.
Enumerations

constant properties that are enumeration like strings are spelled in UPPERCASE. These properties need not be defined manually: See IBdb.convertSchemaEnumsToCode.

Use custom validation functions to parse function input. Especially, use validatestring for enumeration strings, because this enables fuzzy string matching. Enclose these validation function calls in if isfield(NameValue, '<parameter>') <...> end blocks if the input parameter is required to be non-empty.

Resources:

About

API request handler for the insectbrain data base (https://insectbraindb.org).

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages