Skip to content

arjun-menon/pypage

Folders and files

NameName
Last commit message
Last commit date

Latest commit

Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 

Repository files navigation

PyPage pypi docs test

PyPage is a document template engine for Python programmers with a short learning curve.

Why use PyPage?

  • Easy to pick up. Syntax similar to Python's.
  • You need an eval-based template engine.

PyPage supports Python 3.x and 2.7, and has been tested (using test_cmd) on CPython and PyPy.

What does it look like?

Some fruits with the character `r` in their name:
<ul>
  {% for fruit in ['Apple', 'Strawberry', 'Orange', 'Raspberry'] if 'r' in fruit %}
    <li>
      {{ fruit }}
    </li>
  {% endfor %}
</ul>

Installation

You can install PyPage easily with pip:

pip install pypage

Try running pypage -h to see the command-line options available.

Why another templating language?

PyPage is lets you embed Python code easily and flexibly in textual documents (such as Markdown, HTML, reStructuredText, plain text, etc). It lets you construct powerful programmatically-generated documents by embedding Python code in an elegant and flexible manner. Its syntax is similar to and partially inspired by the templating languages Jinja and Liquid.

While there are many templating engines out there, the primarily advantage of PyPage is the fact that its syntax is very close to Python's, and therefore the learning curve is very short for Python programmers.

Rather than create a new mini domain-specific language for constructs such as for and if, PyPage does a teeny tiny bit of obvious string manipulation, and passes your logical directives unaltered to the Python interpreter. As such, PyPage inherits Python's syntax for the most part. For example, for loops in PyPage get converted into Python's generator expressions. The for loop in a Python generator expression (or list comprehension) is far more powerful than its regular for loop. This means that PyPage for loops are richer and more expressive than you'd otherwise expect, while the learning curve is nearly non-existent.

The primary disadvantage of using PyPage instead of a templating engine like Liquid is that PyPage does not operate on a restricted non-Turing-complete subset of programming languages, as Liquid for instance does. Liquid allows untrusted users to write and upload their own templates, because the expressiveness of Liquid is limited such that there is an implicit guarantee that the template will be processed in a reasonable (probably linear) amount of time using a reasonable amount of system resources. As such, Liquid's templating language is rather limited -- it offers a limited number of pre-defined functions/filters, and the overall flexibility of the language has been constrained in order to guarantee termination in a reasonable amount of time.

PyPage, on the other hand gives the template writer full unfettered access to the Python interpreter. As such, PyPage is meant only for use in trusted contexts (or containers), and in some ways it's similar to PHP in that a you're mixing a full-blown programming language (Python) and text that could be HTML.

This brings us to another topic: mixing code and UI. It is generally frowned upon to mix logic/code and the UI (or "view"). So it is good practise to not do any intelligent processing within your PyPage template. Instead, you can do it in a separate program, and pass an environment containing the results, to PyPage. An environment is a dictionary of variables that is passed to Python's exec, and is therefore accessible from all of the code in the PyPage template. From within your template you can focus solely on how to transform these input variables into the HTML/rST/other page you're building.

A pleasant aspect of PyPage, in comparison to other templating languages is that you don't have to learn much new syntax. It's probably the easiest to learn and most flexible templating language out there. It is highly flexible because of the plethora of easy-to-use powerful constructs that PyPage offers.

In order to embed code in a document, you wrap Python code with {{ and }}. The {{ ... }} constructs are called code tags. There are two kinds of code tags: inline and multiline.

Inline code tags occur entirely on the same line, i.e. the closing }} appears on the same line as the opening {{. Here is an example of an inline code tag:

There are {{ 5 + 2 }} days in a week.

The above, when processed by PyPage, yields:

There are 7 days in a week.

The Python eval statement is used to execute the code in an inline code tag. The result of the expression evaluation is converted into a string (with str) and the code tag is replaced with it.

Multiline code tags span multiple lines. The presence of one or more newline (\n) characters between the {{ and }} distinguishes it from an inline code tag. Here's an example:

{{
    x = 5
    y = 2

    write("There are", x + y, "days in a week.")
}}

The Python exec function is used to execute the code in a multiline code tag.

Why have distinct inline code tags? It's easier to write {{x}} than to write {{ write(x) }}. Many a time, all we need to do is inject the value of a variable at a specific location in the document.

All code is executed in a shared common environment. I.e., the locals and globals passed into eval and exec is a single shared dictionary, for all code tags in the same file.

As such, a variable instantiated in a code tag at the beginning of the document, will be available to all other code tags in the document. When PyPage is invoked as library, an initial seed environment consisting of a Python dictionary mapping variable names to values, can be provided.

A write function similar to the Python 3's print function is accessible from both kinds of code tags. It writes text into the document that substitutes/replaces the code tag it's used in.

write(*object, sep=' ', end='\n')

Objects passed to it are stringified with str, concatenated together with sep, and terminated with end. The outputs of multiple calls to write in a code tag are concatenated together, and the resulting final output is injected in place of the code tag.

If write is called from an inline code tag, the result of evaluating the expression (a None, since write will return a None) is ignored, and the output of the write call is used instead.

Block tags simplify certain tasks that would otherwise be cumbersome and ugly if done exclusively with code tags. One of the things it lets you do is wrap part of your page in an if/else conditional, or a for/while loop.

Here's an example of the for block tag:

{% for i in range(10) %}
    The square of {{i}} is {{i*i}}.
{% %}

A block tag begins with {% tag_name ... %} and ends with {% %}. Optionally, the end {% %} can be of the form {% endtag_name %} (i.e. prepend the tag_name with end), which in the above example would be {% endfor %}).

It's best to explain this with an example:

Hey,
{{
  import random
  # Randomly pick a greeting
  greeting = random.randint(1,4)
}}
{% if greeting == 1 %}
  Hello
{% elif greeting == 2 %}
  Bonjour
{% elif greeting == 3 %}
  Hey
{% else %}
  Hi
{% %}

When the above template is run, the resulting page will contain a randomly chosen greeting. As is evident, PyPage syntax for if/elif/else conditions closely mirrors Python's. The terminal {% %} can be replaced with an {% endif %} with no change in meaning (as with any block tag).

Let's start with a simple example:

{% for vowel in ['a', 'e', 'i', 'o', 'u'] %}{{vowel}} {% %}

This will print out the vowels with a space after every character.

Now that's an ordinary for loop. PyPage permits for loops that are more expressive than traditional Python for loops, by leveraging Python's generator expressions.

Here's an example of something that would be impossible to do in Python (with a regular for loop):

{% for x in [1,2,3] for y in ['a','b','c'] %}
    {{x}} ~ {{y}}
{%%}

The above loop would result in:

1 ~ a
1 ~ b
1 ~ c
2 ~ a
2 ~ b
2 ~ c
3 ~ a
3 ~ b
3 ~ c

Internally, PyPage morphs the expression for x in [1,2,3] for y in ['a','b','c'] into the generator expression (x, y) for x in [1,2,3] for y in ['a','b','c']. It exposes the the loop variables x and y by injecting them into your namespace.

Note: Injected loop variables replace variables with the same name for the duration of the loop. After the loop, the old variables with the identical names are restored (PyPage backs them up).

A while loops looks like {{% while condition %}} ... {{% %}, where condition can be any Python expression. Here's an example:

{{
    i = 10
    j = 20
}}
Numbers from {{i}} to {{j}}:
{% while i <= j %}
{{
    write(str(i))
    i += 1
}}
{% %}

This would simply list the numbers from 10 to 20.

{% while dofirst False %}
That's all, folks!
{%%}

Adding a dofirst right after the while and before the expression ensures that the loop is run at least once, before the condition is evaluated.

If a loop runs for more than 2 seconds, PyPage stops executing it, and writes an error message to stdout saying that the loop had been terminated. As PyPage is mostly intended to be used as a templating language, loops generally shouldn't be running for longer than two seconds, and this timeout was added to make it easier to catch accidental infinite loops. If you actually need a loop to run for longer than 2 seconds, you can add the keyword slow right after the condition expression ({{% while condition slow %}}), and that would suppress this 2-second timeout.

You can capture the output of part of your page using the capture tag:

{% capture x %}
  hello {{"bob"}}
{% %}

The above tag will not yield any output, but rather a new variable x will be created that captures the output of everything enclosed by it (which in this case is "hello bob").

The approach taken by PyPage toward template inheritance is quite distinct from that of other templating engines (like Jinja's). It's a lot simpler. You call a PyPage-provided function inject with the path of a PyPage template you want to inject (i.e. "extend" in Jinja parlance), and PyPage will process that template under the current scope (with all previously defined variables being available to the injected template), and the inject function will return its output.

A base template could look like this:

<html>
<head>
    <title>
        {% if exists('title') %}
        {{ title }}
        {% else %}
        No title
        {% %}
    </title>
</head>
<body>
{{ body }}
</body>
</html>

A derived templates only needs to define body and optionally title, to "extend" the template above.

{% capture body %}
The HTML body content would go in here.
{% %}
{{ inject('...path to the base template...') }}

We didn't specify a title above, but if we wanted to, we'd just need to make sure it was defined before inject was called. The base template checks whether a title variable exists by calling the function exists. As is obvious, the exists function simply takes a variable name as a string, and returns a boolean indicating whether the variable exists in the scope.

This approach to inheritance is explicit and easy-to-grasp. Rather than have complex inheritance rules, with a default block definition that is optionally overridden by a derived template, we make things more explicit by using conditionals for cases where we want to provide a default/fallback definition. We error out if a definition is expected to be provided, and is not present. The output of the "dervied" template is clear and obvious, with this approach.

If you want to include (as in, substitute) a file directly without processing it with PyPage, you can use the include function. It functions like the inject function, taking the path to a file as argument, and returning the contents of the file unprocessed.

Anything bounded by {# and #} will be omitted from the output. For example:

<p>
  Lorem ipsum dolor sit amet
  {#
    <ul>
        Non sequitur
    </ul>
  #}
  consectetur adipisicing elit
</p>

You can also easily comment an existing block, by simply placing the word comment in front of it:

<p>
  Lorem ipsum dolor sit amet
    {% comment for i in range(10) %}
        N = {{i}}
    {% %}
  consectetur adipisicing elit
</p>

The comment keyword before the for above results in the entire block being commented out and omitted from the output.

If a block tag is on a line by itself, surrounded only by whitespace, then that whitespace is automatically excluded from the output. This allows you indent your block tags without worrying about excess whitespace in the generated document.

PyPage smartly handles indentation for you. In a multi-line code tag, if you consistently indent your Python code with a specific amount of whitespace, that indentation will be stripped off before executing the code block (as Python is indentation-sensitive), and the resulting output of that code block will be re-indented with same whitespace that the initial code block was.

The whitespace preceding the second line of code determines the peripheral indentation for the entiee block. All subsequent lines (after second) must begin with exact same whitespace that preceded the second line, or be an empty line.

For example:

<p>
  Lorem ipsum dolor sit amet
    <ul>
      {{
        def foo():
          write("Hello!")
        foo()
      }}
    </ul>
  consectetur adipisicing elit
</p>

would produce the following output:

<p>
  Lorem ipsum dolor sit amet
    <ul>
        Hello!
    </ul>
  consectetur adipisicing elit
</p>

Note that the Hello! was indented with same whitespace that the code in the code block was.

PyPage automatically intends the output of a multi-line tag to match the indentation level of the code tag. The number of whitespace characters at the beginning of the second line of the code block determines the indentation level for the whole block. All lines of code following the second line must at least have the same level of indentation as the second line (or else, a PypageSyntaxError exception will be thrown).

Apache License Version 2.0