Skip to main content

chcko randomly parameterized exercises automatically checked (formerly mamchecker)

Project description


Educational problem server at

developed at

with example content package

chcko wants to collect problems, usable by students with or without learn coaches around the world. An infrastructure.

chcko makes it easy to compose a problem set for students, it checks the answers and gives the results.

The focus is on collecting science problems not about code or format.

The problem packages are python packages. the whole system can be installed locally too.


Pages are requested with this URL format:


Pages are:

  • org: add roles and choose a color
  • contents: specific content items or overview if no query
  • done: done problems
  • todo: assigned problems
  • some additional helper pages

Example URLs:

With && instead of one & as separator in a contents query, is called a course, here. In a course only one problem at a time is shown. &&& marks the current position. For example, .../en/contents? has r.a as current item in the course.

The names School, Field, Teacher, Class, Role form namespaces levels to describe a student role. What string is used in each level is up to the user.

The namespaces are of organizational nature. The namepace names are taken from usual organizational structures.

See queries for details.

Use Cases

No Login

Without login a random Role is generated. One can revisit the Role later via the URL, for example

With Org one can choose names. One can choose names via the URL, too. For example:

Without login these names, random or not, are not protected. Their ownership changes to a logged-in user when accessed, either via Org or via URL.

Step-by-step try in class without logging in:

Reserve a Name

Login in.


Create a Class

In the Org tab, the Role input box uses the first of ;, as a separator to create a whole class with no owner (independent of logged in or not). Then send an email to the students, with the link to be filled with their name:<name>

Or send each student separately the full link.

If the students log in, they take ownership of the role.


To assign to others, you need to be logged in.

In the contents tab choose the problems or use an URL:

At the end of the page you can choose classes or students to assign to. Assigning a course (with the &&) assigns the problems individually.

URLs without problems cannot be assigned.

The students

  • log in
  • go to Todo
  • solve the assigned problems

Check Done

To check the done problems of others, you need to be logged in.

Change to the teacher / class / Role.

Clicking on the names for class, teacher, … forwards to these URLs.

* can be replaced by %2A when sending the link, because some programs drop the *.

*&* means: don’t take the default but any student and problem. ?<school>&<period>&<teacher>&<class>&<student>&<problem> is defaulted to the left with the current role names if omitted.

See done for details on queries.

Remove an Assignment

The todo page has the same query format as the done page.*&*

Shows the given (and not yet done) assignments and allows to delete them selectively.

Assume Role

As a logged in user you can have more roles. These roles are listed by clicking in the role field on the ☰, not on the link to the current role. Click on an entry to assume another role.

Remove a Role

  • Assume the role
  • Go to the Org tab
  • Choose delete
  • Confirm

There should be no easier way, because you lose all the history of the role, by deleting it.

Change a Role

Same as Remove a Role, but choose change instead.

This moves all the history associated with a role to the new role and deletes the previous one.

change is a way to

  • leave a class (teacher, period, school)
  • and join another one

without loosing one’s history.

Content Packages

Example content layout:

  ├── chcko
  │   ├──
  │   ├── _images
  │   │   ├── r_dg_c1.png
  │   │   ├── ...
  │   └── r
  │       ├──
  │       ├──
  │       ├── a
  │       │   ├── de.html
  │       │   ├── en.html
  │       │   └──
  │       ├── b
  │       │   ├── _de.html
  │       │   ├── de.rst
  │       │   ├── _en.html
  │       │   ├── en.rst
  │       │   ├──
  │       │   └── vector_dot_cross.tex
  │       └── ...
  ├── ...
  ├── README.rst

Image file names in _images are either random or otherwise unique by encoding author ID, problem ID, content and possibly language. is always there. Altogether it is a Python package, with chcko namespace

For problems, given() in provides random numbers and calc() solves the problem. The parameters from given() can be overridden via the URL parameters. cheader URL parameter is text placed at the beginning of a page with problems.

Generated files start with _ (_<language_id>.html). <language_id>.rst can contain tikz images and are statically converted to _<language_id>.html with:

doit -kd. html

It is better to just stick to HTML, though. HTML files are actually stpl template snippets, for example r/a/en.html:

%path = "maths/trigonometry/sss"
%kind = 0 #problems (see
%level = 11 # school year starting from elementary
The sides of a triangle are
a={{ g.a }},
b={{ g.b }},
c={{ g.c }}.
How big are the angles (in degrees).
%champles=['e.g.'+e for e in ['23.3','100','56.7']]

chq (defined in chelper.html ) creates the input field or shows the result, all according the output of calc() if no idx specified. chq optionally uses (if defined):

  • chames: as input names (per idx a html/tex string, e.g. r”(alpha)”)
  • champles: input examples ( ” )
  • chadios: texts for radio buttons (a tuple per idx)
  • checkos: texts for check boxes (a tuble per idx)

g is returned by given() in

import random
import math as m
from chcko.chcko.hlp import Struct
def angle_deg(i, g):
    d = dict(zip('abc', ([g.a, g.b, g.c]*2)[i:]))
    return eval('180*acos((a*a+b*b-c*c)/2/a/b)/pi', {**d,'acos':m.acos,'pi':m.pi})
def given():
    a, b = random.sample(range(1, 10), 2)
    c = random.randrange(max(a - b + 1, b - a + 1), a + b)
    return Struct(a=a, b=b, c=c)
def calc(g):
    return [angle_deg(i, g) for i in range(3)]
names = [r'\(\alpha=\)', r'\(\beta=\)', r'\(\gamma=\)']

Non-problem texts are OK, too, but should context-free, as they are composed to a page via the URL query string:

Replace the & with && to make a course:

In the URL

  • content items are <author_id>.<content_id>
  • corresponding to the folder chcko/<author_id>/<content_id>/ fills the database with content items. It is generated using:

doit -kd. initdb

To add a new content package on will be updated timely.

You can also run a server locally with:


If chcko is not installed:

Not installed content packages must be parallel to the main chcko folder.

New Package

Create a new content package with:

runchcko --init chcko-<id>

You run this command also to fill a repo you started on github and cloned.

Add a new content item with:

doit -kd. new


doit -kd. rst

Edit the problem text in en.html using a text editor. See the example above.

Then from the root of the content package:

doit initdb


If your are familiar with Linux, use it, possibly on a virtual machine like virtualbox. But all the needed tools are also available for Windows and Mac.

On your PC you will need

Then in a CLI and folder of your choice:

git clone
cd chcko
pip install -r requirements_dev.txt
pip install chcko

installs the python packages for development.

Sphinx is only needed if you use RST. And Latex is only needed if you use Sphinx plugins (sphinxcontrib.tikz, sphinxcontrib.texfigure).



Chcko is yet another solution for computer aided instructions (CAI). The internet has a huge potential in teaching and learning.

The main purpose:

  • Automatically correct problems
  • Infrastructure to organize teaching (school, period, teacher, class, student)
  • allow teachers/coaches to quickly check the problems of students
  • The use is of course not confined to schools. Teachers, professors, tutors, coaches, students, autodidacts, … can add problems and check themselves or others.
  • Share content via separate content packages like chcko-r.
  • The numbers in problems are randomly generated. This way a problem can be reused. Students sitting next to each others in class will have different numbers and therefore cannot copy the results.

Chcko can be used remotely as well as in class.

In class students can use the browser on their smartphones to answer problems. Teachers can immediately see, who answered correctly or who has not yet answered. This way the teacher is faster to find those students who have not yet memorized something or have not yet understood a concept or a relationship.

Students can do problems immediately after the teacher’s explanation in class in the same lesson. This way the students

  • need to pay attention, because they will have to know immediately afterwards
  • cannot copy from others, because the numbers are different, even with problems only due in the next lesson
  • do not need to admit that they have not understood, because the teacher sees, if they are unable to do the problem. Some students are too shy to ask. And there are other reasons, why student’s incomprehension can stay unnoticed for too long.

The teacher cannot look at all the done problems of a class at the same time, but the software can. To do it sequentially in class would hold up the students. If the teacher takes the exercise books home, there is an unwanted delay in feedback for the students.

More parallelism in class is very important in order to make the time spent there worthwhile for the students.

The time spent by a teacher to correct exercise books is also better invested in a good preparation:

  • how to motivate the students
  • how to present the topic as easy as possible
  • which questions to ask to practice and verify that the students have understood


  • Every content has a unique ID = ID_author.ID_content. This way no ID coordination is necessary once the author has an ID.

  • Every ID is also a folder

    • ID_author
      • ID_content1
      • ID_content2
  • IDs shall be as short as possible. They are best numbered through using a-z

    • numbers would not make it a Python identifier
    • capital letters would collide with windows case insensitivity for file names
  • Every content folder contains Python code and language files

    • A Python part ( to randomly generate for problems. It is also needed for content without numbers: just keep it empty.
    • Language template files (en.html, de.html, it.html, fr.html,…) that will produce html. en.html should always be there as starting points for translations.
    • A static off-line step is possible, to create content from other formats, currently from restructured text files (.rst) using Sphinx. This allows to use Sphinx contributions like tikz and texfigure (tex, tikz, chemfig, …) to create graphics.
  • Human language context paths to problems and keywords are language dependent and are therefore in the language files.

  • More problems can be combined in one URL / http request (contents query) e.g. to make a larger assignment.

  • Problem/Content pages can reference other content or inline it via the template engine (% include(``) for html or or :inl:`` for RST).

  • Answers to problems are stored in a DB and combined with the language texts during loading.

  • A user role is identified by an ID path/hierarchy:

    school 1-n period 1-n teacher 1-n class 1-n student
  • Via this hierarchy a teacher has fast access to the done problems of his classes and students via an URL query.

  • Teachers can assign problems to their classes/students, which they access via a todo query

  • Teachers see what their classes/students have done so far (done query)

  • Users initially get a generated role (generated random strings for each), which they can change, though (org query). There users can choose a color to help then see in which role they are.

  • Registered users can have more roles. Registration can also be done via Google, Twitter, Facebook or LinkedIn.


The code tries to stay minimal: Python 3 with bottle and a DB for the roles and problems.


The data model is:

school 1-n period 1-n teacher 1-n class 1-n student 1-n problem

The first 5 are called a role. A user has more roles.

DB is there for answers to problems, not for the problem texts.

  • On GCP, the DB is DataStore using ndb
  • On other server the DB is a SQL database using SqlAlchemy

Environment Variables

CHCKOSECRET:a secret used to encode the user token cookie
CHCKOPORT:used to change port for local server
 for social login
 for social login


The URL format is:

URL = "https://"domain"/"lang"/"page"
domain = ""
lang = "en"|"de"|...
page = ["contents"]["?"{author"."problem["="count]"&"}]
       | "done"[rlinc]
       | "todo"
       | "org"
rlinc = [[[[[school&]period&]teacher&]class&]student&]("*"|query)
query = {field("~"|"="|"!"|"<"|">")value","}

If <lang> is dropped, the last language or the browser setting is used. See

<page> is one of contents, done, todo, org. contents is default, if dropped.

<query> starts after the ? and it is a &-separated list. <query> can contain School=<LLL>&Field=<DDD>&Teacher=<RRR>&Class=<SSS>&Role=<TTT> for all pages.


With ../<lang>/contents all current contents are listed. One can select more entries here.

../en/contents?r.a& (r.a is equivalent to r.a=1) would create an English content page with one r.a and two problems. ../en/?r.a& is the same, i.e. contents is the default page.

Use && instead of & to show one problem at a time (course).

For logged-in users it is possible to make assignments to class/students with the same School-Field-Teacher prefix. You must have created the teacher role, before the others.

Problems have more questions and every question has points associated (default 1). After checking the entered values at the top there will be a summary of achieved points/total points twice, once not counting fields left empty.

The contents index can be limited with:

  • link: the author id
  • level: corresponds to school year starting from elemntary (1, 2, …)
  • kind: problems texts courses examples summaries formal fragments remarks citations definitions theorems corollaries lemmas propositions axioms conjectures claims identities paradoxes meta
  • path: as given in the header of the content sources


../<lang>/done lists the done problems with date and time and whether they were correct. One can open every done problem or do it again. It is possible to delete the selected problems.

The query



  • a student to filter his problems
  • a teacher to see the problems of his classes or students

Omitted entries on the left will be filled by the corresponding current role IDs. Therefore a student only needs <problem>, if it should be filtered at all. <..> are placeholders for the actual strings.

For ‘no restriction’ * is used.

An entry has this format:

name|field op value[,field op value[,...]]
  • name is the name of the record

  • field is a field of the record

    All records have a name, userkey and created. School, Field, Teacher and Class have no other fields. In addition Role has color and Problem has query_string, lang, given, created, answered, collection, inputids, results, oks, points, answers, nr.

  • op consists of ~=!<>, where ~ means =. For the age of a problem (since created) these abbreviations can be used:

    d=days, H=hours, M=minutes, S=seconds

1DK&*&d>3,d<1 would show all problems younger than 3 days (d) and older than one day of students from class 1DK


Bookmark often used requests.

Registered user’s data is protected against queries from anonymous users or other registered users.


../<lang>/todo lists the assignments with date/time given and date/time due.


../<lang>/org allows to add, change or delete IDs for School, Field, Teacher, Class and Role. For fields left empty

  • - is used for logged in users
  • a random ID is generated non-logged-in users

Setting role IDs fails, if the role is owned already. Role prefixes of others are italic. These other users can query your done problems.

new will create a new role.

change will change the identification of the current role, i.e. all the problems done will be copied over.

delete will delete the role and all its done problems.

A color can be chosen to more easily see in which role one is.


One level of privacy is via the IDs you choose. How the IDs link to the real things is only know to you. You could use first or last letter of names, add some additional characters, or do some other obfuscation, without compromising an easy mapping to the real things or person for your purpose.

All unregistered users fall into one user category. Therefore every other unregistered user can query all other unregistered users’ problems (non-owned).

A logged-in user assumes ownership of non-owned roles.

If you register and create instances of school, period, teacher, class and student, then they are associated to you as a user (owned). Then you can query all instances below your instance in the hierarchy:

    n Fields
        n Teachers
            n Classes
                n Roles


  • If a teacher role belongs to you, then classes and students that use the same IDs up to and inclusive teacher as your IDs, then you will be able to query them in the done page, even if they belong to some other user.
  • A director in an educational institution could make a School ID. If all teachers use the same School ID, then the director will be able to query the whole hierarchy.

On the other hand, if you start your query above an instance that does not belong to you, you will not see anything below, even if you have instances somewhere in the deeper levels of the hierarchy.

In .../<lang>/done?<school>&<period>&<teacher>&<class>&<student>&<problem> you can drop instances from the left, immediately after the ?. .../<lang>/done?aclass&*&d>2 would query all problems of any student of class aclass not older than 2 days. For this to work aclass needs to belong to you. If it does not, but the teacher role above belongs to your, then you can still query by entering .../<lang>/done?ateacher&aclass&*&d>2.



As I was about to engage in a teaching job in the beginning of 2013 I was looking for a way adequate for our times

  • to follow the progress of my students
  • to automate certain activities

I did not find a finished solution fitting to my ideas, but I found Google AppEngine, which seemed to be a good basis for an own project.

During my teaching job it was still in a very unsophisticated state, but it was usable already. During that time I added mostly problems, some summaries or other texts that did fit into the topics in class.

The first name, mamchecker, came about from this school’s abbreviation of the subject mathematics as MAM.

Since summer 2013 I restructured the code and added user management and I translated the problems and texts into English.

As I did not continue teaching in autumn, my major motivation for the additional effort was to make my initial effort usable for others.


I was kept busy 5+ years by a employment. Now I revisited the project,

  • renamed it to chcko
  • updated it to Python 3 and
  • to the change at Google AppEngine (now part of GCP): ndb changes, no email any more
  • added support for SQL databases using sqlalchemy
  • made it a python package chcko
  • separated the content to a separate chcko-r package, as an example
  • made some fixes

Project details

Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Files for chcko, version 0.2.1
Filename, size File type Python version Upload date Hashes
Filename, size chcko-0.2.1-py3-none-any.whl (161.9 kB) File type Wheel Python version py3 Upload date Hashes View

Supported by

Pingdom Pingdom Monitoring Google Google Object Storage and Download Analytics Sentry Sentry Error logging AWS AWS Cloud computing DataDog DataDog Monitoring Fastly Fastly CDN DigiCert DigiCert EV certificate StatusPage StatusPage Status page