darkmirage/cv-randomize

Using python-docx to generate random CVs for correspondence resume experiments

   ______     __  ____                 _                 _
  / ___\ \   / / |  _ \ __ _ _ __   __| | ___  _ __ ___ (_)_______
 | |    \ \ / /  | |_) / _` | '_ \ / _` |/ _ \| '_ ` _ \| |_  / _ \
 | |___  \ V /   |  _ < (_| | | | | (_| | (_) | | | | | | |/ /  __/
  \____|  \_/    |_| \_\__,_|_| |_|\__,_|\___/|_| |_| |_|_/___\___|

Raven Jiang raven@cs.stanford.edu

Windows Instruction

1. Install python-2.7.5.msi
2. Install lxml-3.2.1.win32-py2.7.exe
3. Place the cv-randomize folder in C:\ (optional)
4. Press Win + R to open the Windows Run dialog, type cmd and press Enter
5. Type cd C:\cv-randomize (assuming step 3) and press Enter
6. Type C:\Python27\python cvrand.py template\cv1.docx 10 and press Enter

Mac and Linux Instructions

Python generally come preinstalled, so try opening Terminal and going to the cv-randomize folder and run python cvrand.py template/cv1.docx. You may need to install the lxml Python module using pip.

Command Parameters Explained

C:\Python27\python cvrand.py template\cv1.docx 10

This is an example of a command used to execute the program

C:\Python27\python

Refers to the installed path of the Python interpreter

cvrand.py

Refers to the CV Randomize script in the active folder

template\cv1.docx

This is the location of the template document relative to the current folder

10

This is the number of CVs to generate using the specified template. Defaults to 5 if not specified.

Templating System

The script looks for tags enclosed by [] in the template document and replaces them with random values to produce a CV. The source data for the random generator are stored as JSON files in the data folder.

Filename

Any tags enclosed by [] in the name of the template are replaced by the corresponding value used in the file. If the tag is found in the template name but not in the template body, it will not be randomized.

Simple

In the simplest case, a random tag representing a discrete random value with no dependencies is defined through the following:

Sample tag: [tag_name]

Corresponding JSON: tag_name.json

Content of JSON:

["random value 1", "random value 2", "random value 3"]

Group Dependency

A random tag can belong to a group such that all tags belonging to the same group will be have dependent values. (e.g. for a given person name, we want to always have the same email address)

Sample tags: [person.name] [person.email]

Corresponding JSON: person.json

Content of JSON:

[
  { "name": "John", "email": "john@mailcatch.com" },
  { "name": "Mary", "email": "mary@mailcatch.com" }
]

Internal Subfield Randomization

The above definitions ensure that whenever "John" is used for the value of [person.name], "john@mailcatch.com" must be the value for [person.email] Furthermore, tags can be further randomized within a single group.

For Example:

[
  {
    "name": "John",
    "email": ["john@mailcatch.com", "john2@yahoo.com"]
  },
  {
    "name": "Mary",
    "email": "mary@mailcatch.com"
  }
]

The above example means that whenever [person.name] is "John", one of the two email addresses belonging to him will be used for [person.email]. Refer to person.json for an example of this. Note that whitespace and line breaks in JSON files are mostly for readability and have no syntactic meaning.

External Subfield Randomization

Sometimes the list of random values for a subfield is too long to include in every single group. Instead of defining an array in the JSON file of the group, we can provide a reference to an external file.

Content of Group JSON (person.json):

[
  { "name": "John", "email": "_emails_" },
  { "name": "Mary", "email": "mary@mailcatch.com" }
]

The underscore characters indicate that this is not an actual value but a reference to an external file. Refer to education.json for an example of this.

Corresponding JSON: person.emails.json

Content of JSON:

["john@mailcatch.com", "john2@yahoo.com"]