Using Type4Py Rest API

Usage Example

Type4Py's Rest API provides a predict endpoint, which accepts a Python source file via a POST request. A minimal example is given below in Python.

import requests

# Replace `example.py` with your desired Python source file.
with open('example.py') as f:
    r = requests.post("https://type4py.com/api/predict?tc=0", f.read())
    print(r.json())

The response contains type information in JSON format. See here for the description of the response format. For other programming languages, it should be possible to write a similar code like the above example.

Note 1: The sent Python file should have valid syntax.

Note 2: Above example uses https://type4py.com, our public API. Use http://localhost:5001 instead if Type4Py's Docker image is running. See here for the usage of Type4Py's Docker images.

JSON Response

A dummy response is shown below to describe the JSON's fields:

{
    "error":"None",
    "response":{
      "classes":[
        {
          "cls_var_ln":{
            "cls_var_name":[[4,4],[4,13]]
          },
          "cls_var_occur":{
            "cls_var_name":[["token","var","name"]]
          },
          "funcs":[
            {
              "docstring":{"func":"None","long_descr":"None","ret":"None"},
              "fn_lc":[[5,4],[8,28]],
              "fn_var_ln":{"var_name":[[7,8],[7,16]],},
              "fn_var_occur":{"var_name":[["token","var","name"]]},
              "name":"__init__",
              "params":{"age":"int",},
              "params_descr":{"age":"comment"},
              "params_occur":{"age":[["self","age","age"]],},
              "params_p":{"age":[["int",0.9999999991180025],["str", 2.3463255785983247e-10]],},
              "q_name":"Person.__init__",
              "ret_exprs":[""],
              "ret_type":"None",
              "variables":{"age":""
              },
              "variables_p":{"age":[["int", 0.2801499039103035]],}
            },
            {
              "docstring":{
                "func":"None","long_descr":"None","ret":"None"
              },
              "fn_lc":[[10,4],[11,24]],
              "fn_var_ln":{
              },
              "fn_var_occur":{
              },
              "name":"get_name",
              "params":{
              },
              "params_descr":{
              },
              "params_occur":{
              },
              "params_p":{
              },
              "q_name":"Person.get_name",
              "ret_exprs":["return self.name"],
              "ret_type":"",
              "ret_type_p":[["str", 0.7073830464758581]],
              "variables":{
              },
              "variables_p":{
              }
            }
          ],
          "name":"Person",
          "q_name":"Person",
          "variables":{"person_id":""},
          "variables_p":{
            "person_id":[["str", 0.703074717210447],]
          }
        }
      ],
      "funcs":[
        {
          "docstring":{"func":"None","long_descr":"None","ret":"None"},
          "fn_lc":[[18,0],[25,18]],
          "fn_var_ln":{"leave_hours":[[19,4],[19,15]],},
          "fn_var_occur":{"leave_hours":[["no_hours","leave_hours"]],},
          "name":"work",
          "params":{"no_hours":""},
          "params_descr":{"no_hours":""},
          "params_occur":{"no_hours":[["no_hours","leave_hours"]],},
          "params_p":{"no_hours":[["Type", 0.0999]]},
          "q_name":"work",
          "ret_exprs":["return \"Done!\""],
          "ret_type":"",
          "ret_type_p":[["str", 0.287441260068372]],
          "variables":{"leave_hours":""},
          "variables_p":{"leave_hours":[["int",0.2]]}
        }
      ],
      "imports":["os"],
      "mod_var_ln":{"A_GLOBAL_VAR":[[1,0],[1,12]]},
      "mod_var_occur":{"A_GLOBAL_VAR":["token"]},
      "no_types_annot":{"D":1,"I":0,"U":14},
      "session_id":"a0bvkdCC8utA35r8JrOho07FrDpV9qaLr2lccFzoXB4",
      "set":"None",
      "tc":[false,"None"],
      "type_annot_cove":0.07,
      "typed_seq":"",
      "untyped_seq":"",
      "variables":{"A_GLOBAL_VAR":""},
      "variables_p":{"A_GLOBAL_VAR":[["str",0.41389838221497904]]}
    }
  }

Description

Note: Predicted types by Type4Py are stored in the fields variables_p, params_p, ret_type_p.

• error: An error message if the process of a request is unsuccessful.
• response: Contains a processed Python file with type information
  • classes: A JSON array of processed classes
    • cls_var_ln: Contains line no. info for class variables.
    • cls_var_occur: Occurences of class variables in the class itself (tokenized)
    • funcs: Contains JSON array of class-level functions. Its structure is described below.
    • name: Name of the class
    • q_name: Fully qualified name of the class
    • variable: List of class variables and their type annotation (if available)
    • variables_p: List of predicted types for the class variables
  • funcs: Contains JSON array of processed module-level functions
    • docstring: Contains comments about what the function does (if available)
    • fn_lc: Line numbers for the start and end of the function
    • fn_var_ln: Contains line numbers for the start and end of local variables in the function
    • fn_var_occur: Occurences of local variables in the function itself (tokenized)
    • name: Name of the function
    • params: List of function's parameters with their type annotation (if available)
    • params_descr: Comments about function's parameters (if available)
    • params_occur: Occurences of parameters in the function itself (tokenized)
    • params_p: List of predicted types for the function's parameters. Each parameter's predicted types are sorted by the model's confidence.
    • q_name: Fully qualified name for the function
    • ret_exprs: Return expressions in the function (tokenized)
    • ret_type: The return type of the function (if available)
    • ret_type_p: List of predicted types for the function's return value.
    • variables: List of local variables and their type annotation (if available)
    • variables_p: List of predicted types for the function's local variables.
  • imports: An array of imported names in the Python file
  • mod_var_ln: Contains line numbers for the start and end of global variables in the file
  • mod_var_occur: Occurences of global variables in the file (tokenized)
  • no_types_annot: No. of pre-existing type annotations in the file, D: developer-provided, I: Inferred by Pyre, U: No. of type slots with no type annotation.
  • session_id: Each request has a unique session ID (hashed string).
  • set: Its value is always None and should be ignored.
  • tc: Whether the file can be type-checked by Mypy (this feature is disabled as of this writing).
  • type_annot_cove: The file's pre-existing type annotation coverage
  • typed_seq, untyped_seq: Seq2Seq representation of the file. [Always empty]
  • variables: List of the file's global variables and their type annotation (if available)
  • variables_p: List of predicted types for the file's global variables.

Note: Line numbers fields contains line and column number for both the start and end of an identifier.