Python Libraries
The Standard Library
The libraries that tend to get all the attention are the ones heavily used for operating-system interaction, like sys, os, shutil, and to a slightly lesser extent, glob. This is understandable because most Python applications deal with input processing; however, the Python standard library is very rich and includes a bunch of additional functionality that many Python programmers take too long to discover. In this chapter we will mention a few libraries that every Python programmer should know very well.
collections
First up we have the collections module. If you’ve been working with Python for any length of time, it is very likely that you have made use of the this module; however, the batteries contained within are so important that we’ll go over them anyway, just in case.
collections.OrderedDict
collections.OrderedDict gives you a dict that will preserve the order in which items are added to it; note that this is not the same as a sorted order.1
The need for an ordered dict comes up surprisingly often. A common example is processing lines in a file where the lines (or something within them) maps to other data. A mapping is the right solution, and you often need to produce results in the same order in which the input data appeared. Here is a simple example of how the ordering changes with a normal dict:
See how the key "f" now appears before the "e" key in the sequence of keys? They no longer appear in the order of insertion, due to how the dict internals manage the assignment of hash entries.
The OrderedDict, however, retains the order in which items are inserted:
ORDEREDDICT: BEWARE CREATION WITH KEYWORD ARGUMENTS
There is an unfortunate catch with OrderedDict you need to be aware of: it doesn’t work when you create the OrderedDict with keyword arguments, a very common Python idiom:
This seems like a bug, but as explained in the documentation, it happens because the keyword arguments are first processed as a normal dict before they are passed on to the OrderedDict.
collections.defaultdict
collections.defaultdict is another special-case dictionary: it allows you to specify a default value for all new keys.
Here’s a common example:
You didn’t create this item yet? No problem! Key lookups automatically create values using the function provided when creating the defaultdict instance.
By setting up the default value as the list constructor in the preceding example, you can avoid wordy code that looks like this:
The setdefault() method of a dict can be used in a somewhat similar way to initialize items with defaults, but defaultdict generally results in clearer code.2
In the preceding examples, we’re saying that every new element, by default, will be an empty list. If, instead, you wanted every new element to contain a dictionary, you might say defaultdict(dict).
collections.namedtuple
The next tool, collections.namedtuple, is magic in a bottle! Instead of working with this:
You get to work with this:
The best thing about namedtuple is that you can add it to existing code and use it to progressively replace tuples: it can appear anywhere a tuple is currently being used, without breaking existing code, and without using any extra resources beyond what plain tuples require. Using namedtuple incurs no extra runtime cost, and can make code much easier to read. The most common situation where a namedtuple is recommended is when a function returns multiple results, which are then unpacked into a tuple. Let’s look at an example of code that uses plain tuples, to see why such code can be problematic:
Worse, the caller might access values inside the returned tuple by index.
The problem with this approach is that this code is fragile to future changes. If the function changes (perhaps by changing the order of the returned items, or adding more items), the unpacking of the returned value will be incorrect. Instead, you can modify existing code to return a namedtuple instance:
Even though our function now returns a namedtuple, the same calling code stills works.
You now also have the option of working with the returned namedtuple in the calling code:
Being able to use attributes to access data inside the tuple is much safer rather than relying on indexing alone; if future changes in the code added new fields to the namedtuple, the tup.count would continue to work.
The collections module has a few other tricks up its sleeve, and your time is well spent brushing up on the documentation. In addition to the classes shown here, there is also a Counter class for easily counting occurrences, a list-like container for efficiently appending and removing items from either end (deque), and several helper classes to make subclassing lists, dicts, and strings easier.
contextlib
A context manager is what you use with the with statement. A very common idiom in Python for working with file data demonstrates the context manager:
This is good syntax because it simplifies the cleanup step where the file handle is closed. Using the context manager means that you don’t have to remember to do f.close() yourself: this will happen automatically when the with block exits.
You can use the contextmanager decorator from the contextlib library to benefit from this language feature in your own nefarious schemes. Here’s a creative demonstration where we create a new context manager to print out performance (timing) data.
This might be useful for quickly testing the time cost of code snippets, as shown in the following example. The numbered notes are intentionally not in numerical order in the code. Follow the notes in numerical order as shown following the code snippet.
For fun, we’ll use our awesome, new context manager to also measure the total time.
On my computer, this code produces this output:
Quite surprisingly, the second method of producing a large array is around 100 times faster than the first. This means that it is much more efficient to create a small array, and then expand it, rather than to create an array entirely from a large list.
The point of this example is not to show the best way to create an array: rather, it is that the contextmanager decorator makes it exceptionally easy to create your own context manager, and context managers are a great way of providing a clean and safe means of managing before-and-after coding tasks.
concurrent.futures
The concurrent.futures module that was introduced in Python 3 provides a convenient way to manage pools of workers. If you have previously used the threading module in the Python standard library, you will have seen code like this before:
This code is very clean with only one thread, but with many threads it can become quite tricky to deal with sharing work between them. Also, in this example the result of the sum is not obtained from the work function, simply to avoid all the extra code that would be required to do so. There are various techniques for obtaining the result of a work function, such as passing a queue to the function, or subclassing threading.Thread, but we’re not going discuss them any further, because the multiprocessing package provides a better method for using pools, and the concurrent.futures module goes even further to simplify the interface. And, similar to multiprocessing, both thread-based pools and process-based pools have the same interface making it easy to switch between either thread-based or process-based approaches.
Here we have a trivial example using the ThreadPoolExecutor. We download the landing page of a plethora of popular social media sites, and, to keep the example simple, we print out the size of each. Note that in the results, we show only the first four to keep the output short.
This is a simple way of waiting for all the threads to return.
This produces the following output (I’ve shortened the number of results for brevity):
Even though one job is created for every URL, we limit the number of active threads to only four using max_workers and the results are all captured in the results list as they become available. If you wanted to use processes instead of threads, all that needs to change is the first line, from this:
To this:
Of course, for this kind of application, which is limited by network latency, a thread-based pool is fine. It is with CPU-bound tasks that Python threads are problematic because of how thread safety has been in implemented inside the CPython3 runtime interpreter, and in these situations it is best to use a process-based pool instead.
The primary problem with using processes for parallelism is that each process is confined to its own memory space, which makes it difficult for multiple workers to chew on the same large chunk of data. There are ways to get around this, but in such situations threads provide a much simpler programming model. However, as we shall see in “Cython”, there is a third-party package called Cython that makes it very easy to circumvent this problem with threads.
logging
The logging module is very well known in the web development community, but is far less used in other domains, such as the scientific one; this is unfortunate, because even for general use, the logging module is far superior to the print() function. It doesn’t seem that way at first, because the print() function is so simple; however, once you initialize logging, it can look very similar. For instance, compare these two:
The huge advantage of the latter is that, with a single change to a setting on the logger instance, you can either show or hide all your debugging messages. This means you no longer have to go through the process of commenting and uncommenting your print() statements in order to show or hide them. logging also gives you a few different levels so that you can adjust the verbosity of output in your programs. Here’s an example of different levels:
Another really neat trick is that when you use logging, writing messages during exception handling is a whole lot easier. You don’t have to deal with sys.exc_info() and the traceback module merely for printing out the exception message with a traceback. You can do this instead:
Just those four lines produces a full traceback in the output:
Earlier I said that logging requires some setup. The documentation for the logging module is extensive and might seem overwhelming; here is a quick recipe to get you started:
You need to call a config method; otherwise, calls to the logger will not log anything. The config step is necessary.
In the preceding basicConfig() line, by changing only logging.DEBUG to, say, logging.WARNING, you can affect which messages get processed and which don’t.
Finally, there’s a neat trick you can use for easily changing the logging level on the command line. Python scripts that are directly executable usually have the startup code in a conditional block beginning with if __name__ == '__main__'. This is where command-line parameters are handled, e.g., using the argparse library in the Python standard library. We can create command-line arguments specifically for the logging level:
With this setup, if you call your program with
it will run with the logging level set to DEBUG (so all logger messages will be shown), whereas if you run it with
it will run at the WARNING level for logging, so all INFO and DEBUG logger messages will be hidden.
There are many more features packed into the logging module, but I hope I’ve convinced you to consider using it instead of using print() for your next program. There is much more information about the logging module online, both in the official Python documentation and elsewhere in blogs and tutorials. The goal here is only to convince you that the logging module is worth investigating, and getting started with it is easy to do.
sched
There is increasing interest in the creation of bots4 and other monitoring and automation applications. For these applications, a common requirement is to perform actions at specified times or specified intervals. This functionality is provided by the sched module in the standard library. There are already similar tools provided by operating systems, such as cron on Linux and Windows Task Scheduler, but with Python’s own sched module you can ignore these platform differences, as well as incorporate scheduled tasks into a program that might have many other functions.
The documentation for sched is rather terse, but hopefully these examples will get you started. The easy way to begin is to schedule a function to be executed after a specified delay (this is a complete example to make sure you can run it successfully):
The scheduler is started with run(blocking=True), and the execution point remains here until the program is terminated or Ctrl-C is pressed.
There are a few annoying details about using sched: you have to pass timefunc=time.time as this isn’t set by default, and you have to supply a priority even when not required. However, overall, the sched module still provides a clean way to get cron-like behavior.
Working with delays can be frustrating if what you really want is for a task to execute at specific times. In addition to enter(), a sched instance also provides the enterabs() method with which you can trigger an event at a specific time. We can use that method to trigger a function, say, every whole minute:
The enterabs() method schedules the task.
This code produces the following output:
With the growing interest in “Internet of Things” applications, the built-in sched library provides a convenient way to manage repetitive tasks. The documentation provides further information about how to cancel future tasks.
In the Wild
From here on we’ll look at some third-party Python libraries that you might not yet have discovered. There are thousands of excellent packages described at the Python guide.
There are quite a few guides similar to this report that you can find online. There could be as many “favorite Python libraries” lists as there are Python developers, and so the selection presented here is necessarily subjective. I spent a lot of time finding and testing various third-party libraries in order to find these hidden gems.
My selection criteria were that a library should be:
easy to use
easy to install
cross-platform
applicable to more than one domain
not yet super-popular, but likely to become so
the X factor
The last two items in that list bear further explanation.
Popularity is difficult to define exactly, since different libraries tend to get used to varying degrees within different Python communities. For instance, Numpy and Scipy are much more heavily used within the scientific Python community, while Django and Flask enjoy more attention in the web development community. Furthermore, the popularity of these libraries is such that everybody already knows about them. A candidate for this list might have been something like Dask, which seems poised to become an eventual successor to Numpy, but in the medium term it is likely to be mostly applicable to the scientific community, thus failing my applicability test.
The X factor means that really cool things are likely to be built with that Python library. Such a criterion is of course strongly subjective, but I hope, in making these selections, to inspire you to experiment and create something new!
Each of the following chapters describes a library that met all the criteria on my list, and which I think you’ll find useful in your everyday Python activities, no matter your specialization.
Easier Python Packaging with flit
flit is a tool that dramatically simplifies the process of submitting a Python package to the Python Package Index (PyPI). The traditional process begins by creating a setup.py file; simply figuring out how to do that requires a considerable amount of work even to understand what to do. In contrast, flit will create its config file interactively, and for typical simple packages you’ll be ready to upload to PyPI almost immediately. Let’s have a look: consider this simple package structure:
The package’s init file is a great place to add package information like documentation, version numbers, and author information.
After installing flit into your environment with pip install flit, you can run the interactive initializer, which will create your package configuration. It asks only five questions, most of which will have applicable defaults once you have made your first package with flit:
The final line tells you that a flit.ini file was created. Let’s have a look at that:
It’s pretty much what we specified in the interactive flit init sequence. Before you can submit our package to the online PyPI, there are two more steps that you must complete. The first is to give your package a docstring. You add this to the _mypkg/__init__.py file at the top using triple quotes ("""). The second is that you must add a line for the version to the same file. Your finished __init__.py_ file might look like this:
Likewise, the version tag within your package will also be reused for PyPI when the package is uploaded. These automatic integrations help to simplify the packaging process. It might not seem like much is gained, but experience with Python packaging will show that too many steps (even when they’re simple), when combined, can lead to a complex packaging experience.5
After filling in the basic description and the version, you are ready to build a wheel and upload it to PyPI:
Note that flit automatically registers your package if the initial upload fails.
Figure 1-1. It’s alive! Our demo package on the PyPI.
flit allows you to specify more options, but for simple packages what you see here can get you pretty far.
Command-Line Applications
If you’ve spent any time whatsoever developing Python code, you will surely have used several command-line applications. Some command-line programs seem much friendlier than others, and in this chapter we show two fantastic libraries that will make it easy for you to offer the best experience for users of your own command-line applications. colorama allows you to use colors in your output, while begins makes it easy to provide a rich interface for specifying and processing command-line options.
colorama
Many of your desktop Python programs will only ever be used on the command line, so it makes sense to do whatever you can to improve the experience of your users as much as possible. The use of color can dramatically improve your user interface, and colorama makes it very easy to add splashes of color into your command-line applications.
Let’s begin with a simple example.
Figure 1-2. Add color to your output messages with simple string concatenation.
The great thing about colorama is that it also works on Windows, in addition to Linux and Mac OS X. In the preceding example, we used the init() function to enable automatic reset to default colors after each print(), but even when not required, init() should always be called (when your code runs on Windows, the init() call enables the mapping from ANSI color codes to the Windows color system).
The preceding example is clear enough to follow, but I would be a sorry author if I—after having emphasized the benefits of the logging module—told you that the only way to get colors into your console was to use print(). As usual with Python, it turns out that the hard work has already been done for us. After installing the colorlog package,6 you can use colors in your log messages immediately:
Set the message formatter to be the ColoredFormatter provided by the colorlog library.
This produces the output shown in Figure 1-3.
Figure 1-3. Beautiful and automatic colors for your logging messages.
begins
As far as user-interfaces go, most Python programs start out as command-line applications, and many remain so. It makes sense to offer your users the very best experience you can. For such programs, options are specified with command-line arguments, and the Python standard library offers the argparse library to help with that.
argparse is a robust, solid implementation for command-line processing, but it is somewhat verbose to use. For example, here we have an extremely simple script that will add two numbers passed on the command line:
As one would expect, help can be obtained by passing -h to this program:
In contrast, the begins library takes a machete to the API of argparse and maximally exploits features of the Python language to simplify setting up the same command-line interface:
There is so much happening in so few lines, yet everything is still explicit:
Each parameter in the main function becomes a command-line argument.
The function annotations are exploited to provide an inline help description of each parameter.
The default value of each parameter (here,
0.0) is used both as a default value, as well as to indicate the required datatype (in this case, afloatnumber value).The
auto_convert=Trueis used to enforce type coercion from a string to the target parameter type.The docstring for the function now becomes the help documentation of the program itself.
Also, you may have noticed that this example lacks the usual if __name__ == '__main__' boilerplate: that’s because begins is smart enough to work with Python’s stack frames so that your target function becomes the starting point.
For completeness, here is the help for the begins-version, produced with -h:
There are a bunch more tricks that begins makes available, and you are encouraged to read the documentation; for instance, if the parameters are words:
then both -d VALUE and --directory VALUE will work for specifying the value of the directory parameter on the command line. A sequence of positional arguments of unknown length is easily defined with unpacking:
When called with:
the filters argument would be the list ['tmp', 'temp', 'TEMP'].
If that were all that begins supported, it would already be sufficient for the vast majority of simple programs; however, begins also provides support for subcommands:
The main function is the same as before, but we’ve added two subcommands:
The first,
status, represents a subcommand that could be used to provide some kind of system status message (think “git status”).The second,
fetch, represents a subcommand for some kind of work function (think “git fetch”).
Each subcommand has its own set of parameters, and the rules work in the same way as before with the main function. For instance, observe the updated help (obtained with the -h parameter) for the program:
We still have the same documentation for the main program, but now additional help for the subcommands have been added. Note that the function docstrings for status and fetch have also been recycled into CLI help descriptions. Here is an example of how our program might be called:
You can also see how boolean parameters get some special treatment: they evaluate True if present and False when no- is prefixed to the parameter name, as shown for the parameter compact.
begins has even more tricks up its sleeve, such as automatic handling for environment variables, config files, error handling, and logging, and I once again urge you to check out the project documentation.
If begins seems too extreme for you, there are a bunch of other tools for easing the creation of command-line interfaces. One particular option that has been growing in popularity is click.
Graphical User Interfaces
Python offers a wealth of options for creating graphical user interfaces (GUIs) including PyQt, wxPython, and tkinter, which is also available directly in the standard library. In this chapter we will describe two significant, but largely undiscovered, additions to the lineup. The first, pyqtgraph, is much more than simply a chart-plotting library, while pywebview gives you a full-featured web-technology interface for your desktop Python applications.
pyqtgraph
The most popular chart-plotting library in Python is matplotlib, but you may not yet have heard of the wonderful alternative, pyqtgraph. Pyqtgraph is not a one-to-one replacement for matplotlib; rather, it offers a different selection of features and in particular is excellent for real-time and interactive visualization.
It is very easy to get an idea of what pyqtgraph offers: simply run the built-in examples application:
Figure 1-4. Interactive pyqtgraph window.
This example shows an array of interactive plots. You can’t tell from the screenshot, but when you run this example, the yellow chart on the right is a high-speed animation. Each plot, including the animated one (in yellow), can be panned and scaled in real-time with your cursor. If you’ve used Python GUI frameworks in the past, and in particular ones with free-form graphics, you would probably expect sluggish performance. This is not the case with pyqtgraph: because pyqtgraph uses, unsurprisingly, PyQt for the UI. This allows pyqtgraph itself to be a pure-python package, making installation and distribution easy.7
By default, the style configuration for graphs in pyqtgraph uses a black background with a white foreground. In Figure 1-4, I sneakily used a style configuration change to reverse the colors, since dark backgrounds look terrible in print. This information is also available in the documentation:
There is more on offer than simply plotting graphs: pyqtgraph also has features for 3D visualization, widget docking, and automatic data widgets with two-way binding. In Figure 1-5, the data-entry and manipulation widgets were generated automatically from a Numpy table array, and UI interaction with these widgets changes the graph automatically.
pyqtgraph is typically used as a direct visualizer, much like how matplotlib is used, but with better interactivity. However, it is also quite easy to embed pyqtgraph into another separate PyQt application and the documentation for this is easy to follow. pyqtgraph also provides a few useful extras, like edit widgets that are units-aware (kilograms, meters and so on), and tree widgets that can be built automatically from (nested!) standard Python data structures like lists, dictionaries, and arrays.
Figure 1-5. Another pyqtgraph example: the data fields and widgets on the left are automatically created from a Numpy table array.
pywebview
There are a huge number of ways to make desktop GUI applications with Python, but in recent years the idea of using a browser-like interface as a desktop client interface has become popular. This approach is based around tools like cefpython (using the Chrome-embedded framework) and Electron, which has been used to build many popular tools such as the Atom text editor and Slack social messaging application.
Usually, such an approach involves bundling a browser engine with your application, but Pywebview gives you a one-line command to create a GUI window that wraps a system native “web view” window. By combining this with a Python web abb like Flask or Bottle, it becomes very easy to create a local application with a GUI, while also taking advantage of the latest GUI technologies that have been developed in the browser space. The benefit to using the system native browser widget is that you don’t have to distribute a potentially large application bundle to your users.
Let’s begin with an example. Since our application will be built up as a web-page template in HTML, it might be interesting to use a Python tool to build the HTML rather than writing it out by hand.
I’ve embedded a few fun videos from YouTube.
Figure 1-6. A beautiful and pleasing interface with very little effort.
Figure 1-7. The same application, but using the Superhero theme from Bootswatch.
NOTE
In the preceding example we also used the dominate library, which is a handy utility for generating HTML procedurally. It’s not only for simple tasks: this demo shows how it can handle Bootstrap attributes and widgets quite successfully. dominate works with nested context managers to manage the scope of HTML tags, while HTML attributes are assigned with keyword arguments.
For the kind of application that Pywebview is aimed at, it would be very interesting to think about how dominate could be used to make an abstraction that completely hides browser technologies like HTML and JavaScript, and lets you create (e.g., button-click handlers) entirely in Python without worrying about the intermediate event processing within the JavaScript engine. This technique is already explored in existing tools like flexx.
System Tools
Python is heavily used to make tools that work closely with the operating system, and it should be no surprise to discover that there are excellent libraries waiting to be discovered. The psutil library gives you all the access to operating-system and process information you could possibly hope for, while the Watchdog library gives you hooks into system file-event notifications. Finally, we close out this chapter with a look at ptpython, which gives you a significantly enriched interactive Python prompt.
psutil
Psutil provides complete access to system information. Here’s a simple example of a basic report for CPU load sampled over 5 seconds:
Output:
It produces one value for each logical CPU: on my computer, eight. The rest of the psutil API is as simple and clean as this example shows, and the API also provides access to memory, disk, and network information. It is very extensive.
There is also detailed information about processes. To demonstrate, here is a program that monitors its own memory consumption and throws in the towel when a limit is reached:
This looks bad: an infinite loop and an ever-growing list!
Output:
The full path has been shortened here to improve the appearance of the snippet. There are many more methods besides name and exe available on psutil.Process() instances.
The type of memory shown here is the unique set size, which is the real memory released when that process terminates. The reporting of the unique set size is a new feature in version 4 of psutil, which also works on Windows.
There is an extensive set of process properties you can interrogate with psutil, and I encourage you to read the documentation. Once you begin using psutil, you will discover more and more scenarios in which it can make sense to capture process properties. For example, it might be useful, depending on your application, to capture process information inside exception handlers so that your error logging can include information about CPU and memory load.
Watchdog
Watchdog is a high-quality, cross-platform library for receiving notifications of changes in the file system. Such file-system notifications is a fundamental requirement in many automation applications, and Watchdog handles all the low-level and cross-platform details of notifications for system events. And, the great thing about Watchdog is that it doesn’t use polling.
The problem with polling is that having many such processes running can sometimes consume more resources than you may be willing to part with, particularly on lower-spec systems like the Raspberry Pi. By using the native notification systems on each platform, the operating system tells you immediately when something has changed, rather than you having to ask. On Linux, the inotify API is used; on Mac OS X, either kqueue or FSEvents are used; and on Windows, the ReadDirectoryChangesW API is used. Watchdog allows you to write cross-platform code and not worry too much about how the sausage is made.
Watchdog has a mature API, but one way to get immediate use out of it is to use the included watchmedo command-line utility to run a shell command when something changes. Here are a few ideas for inspiration:
Compiling template and markup languages:
Watchdog will substitute the name of the specific changed file using this template name.
Maintenance tasks, like backups or mirroring:
Convenience tasks, like automatically running formatters, linters, and tests:
Calling API endpoints on the Web, or even sending emails and SMS messages:
The API of Watchdog as a library is fairly similar to the command-line interface introduced earlier. There are the usual quirks with thread-based programming that you have to be aware of, but the typical idioms are sufficient:
Watchdog runs in a separate thread. By calling the join() method, you can force the program flow to block at this point.
With that code running, I cunningly executed a few of these:
This is the console output from the running Watchdog demo:
Here I terminated the program by pressing Ctrl-C.
As we can see, Watchdog saw all the modifications I made to a file in the directory being watched.
ptpython
ptpython is an alternative interpreter interface, offering a beautiful interactive Python experience. As such, ptpython is more like a tool than a library to include in your own projects, but what a tool! Figure 1-8 shows a screenshot showing the basic user interface.
When you need to work with blocks in the interpreter, like classes and functions, you will find that accessing the command history is much more convenient than the standard interpreter interface: when you scroll to a previous command, the entire block is shown, not only single lines, as shown in Figure 1-9.
Here, I retrieved the code for the function declaration by pressing the up arrow (or Ctrl-p) and the entire block is shown, rather than having to scroll through lines separately.
Figure 1-8. The code-completion suggestions pop up automatically as you type.
Figure 1-9. When you scroll to previous entries, entire blocks are suggested (e.g., function f()), not only single lines.
Figure 1-10. The settings menu accessible via F2.
On the command line, ptpython offers highly productive features that are sure to enhance your workflow. I have found that ptpython is the fastest way to quickly test some syntax or a library call, without having to open an editor or integrated development environment (IDE). And if you prefer IPython, you’ll be glad to hear that ptpython includes an integration which you can launch with ptipython. This makes available the shell integration that IPython offers, as well as its wealth of magic commands.
Web APIs with hug
Python is famously used in a large number of web frameworks, and an extension of this area is the web services domain in which APIs are exposed for other users’ programs to consume. In this domain, Django REST framework and Flask are very popular choices, but you may not yet have heard of hug.
hug is a library that provides an extremely simple way to create Internet APIs for web services. By exploiting some of the language features of Python, much of the usual boilerplate normally required when creating APIs for web services is removed.
Here is a small web service that converts between the hex value of a color and its CSS3 name:
NOTE
We are also using the webcolors library here: your one-stop shop for converting web-safe colors between various formats like name, hex, and rgb. Don’t forget that you can also convert between rgb and other formats like hsl with the colorsys library that’s already included in the Python standard library!
Figure 1-11. Hug goes a long way to make you feel welcome.
And that’s it! We can immediately test our API. You could use a web browser for this, but it’s easy enough to use a tool like cURL. You call the URL endpoint with the correct parameters, and hug returns the answer:
This API has a different endpoint function, nametohex, and its parameter is called name.
It’s quite impressive to get a live web API up with so little work. And the features don’t stop there: hug automatically generates API documentation, which is what you get by omitting an endpoint:
The types of the parameters we specified with hug.types.text are used not only for the documentation, but also for type conversions.
If these were all the features provided by hug, it would already be enough for many API tasks; however, one of hug’s best features is that it automatically handles versions. Consider the following example:
The second, newer version of the same API returns the result as calculated by inflect, without any alteration.
NOTE
We are also using the inflect library, which makes an enormous number of word transformations possible, including pluralization, gender transformation of pronouns, counting (“was no error” versus “were no errors”), correct article selection (“a,” “an,” and “the”), Roman numerals, and many more!
Imagine that we’ve created this web service to provide an API to transform words into their singular or plural version: one apple, many apples, and so on. For some reason that now escapes us, in the first release we lowercased the results before returning them:
Note: the URL now has a “v1” specifier for “version 1.”
It didn’t occur to us that some users may prefer to have the case of input words preserved: for example, if a word at the beginning of a sentence is being altered, it would be best to preserve the capitalization. Fortunately, the inflect library already does this by default, and hug provides versioning that allows us to provide a second version of the API (so as not to hurt existing users who may expect the lower-case transformation):
These API calls use version 2 of our web service. And finally, the documentation is also versioned, and for this example, you can see how the function docstrings are also incorporated as usage text:
The docstring of each function is reused as the usage text for that API call.
hug has extensive documentation in which there are even more features to discover. With the exploding interest in internet-of-things applications, it is likely that simple and powerful libraries like hug will enjoy widespread popularity.
Dates and Times
Many Python users cite two major pain points: the first is packaging, and for this we covered flit in an earlier chapter. The second is working with dates and times. In this chapter we cover two libraries that will make an enormous difference in how you deal with temporal matters. arrow is a reimagined library for working with datetime objects in which timezones are always present, which helps to minimize a large class of errors that new Python programmers encounter frequently. parsedatetime is a library that lets your code parse natural-language inputs for dates and times, which can make it vastly easier for your users to provide such information.
arrow
Though opinions differ, there are those who have found the standard library’s datetime module confusing to use. The module documentation itself in the first few paragraphs explains why: it provides both naive and aware objects to represent dates and times. The naive ones are the source of confusion, because application developers increasingly find themselves working on projects in which timezones are critically important. The dramatic rise of cloud infrastructure and software-as-a-service applications have contributed to the reality that your application will frequently run in a different timezone (e.g., on a server) than where developers are located, and different again to where your users are located.
The naive datetime objects are the ones you typically see in demos and tutorials:
This code could not be any simpler: we create two datetime objects, and calculate the difference. The problem is that we probably intended both now() and utcnow() to mean now as in “this moment in time,” but perhaps in different timezones. When the difference is calculated, we get what seems an absurdly large result:
The difference is, in fact, the timezone difference between my current location and UTC: +10 hours. This result is the worst kind of “incorrect” because it is, in fact, correct if timezones are not taken into account; and here lies the root of the problem: a misunderstanding of what the functions actually do. The now() and utcnow() functions do return the current time in local and UTC timezones, but unfortunately both results are naive datetime objects and therefore lack timezone data.
Note that it is possible to create datetime objects with timezone data attached, albeit in a subjectively cumbersome manner: you create an aware datetime by setting the tzinfo attribute to the correct timezone. In our example, we could (and should!) have created a current datetime object in the following way, by passing the timezone into the now() function:
Of course, if you make this change on line 3 and try to rerun the code, the following error appears:
This happens because even though the utcnow() function produces a datetime for the UTC timezone, the result is a naive datetime object, and Python prevents you from mixing naive and aware datetime objects. If you are suddenly scared to work with dates and times, that’s an understandable response, but I’d rather encourage you to instead be afraid of datetimes that lack timezone data. The TypeError just shown is the kind of error we really do want: it forces us to ensure that all our datetime instances are aware.
Naturally, to handle this in the general case requires some kind of database of timezones. Unfortunately, except for UTC, the Python standard library does not include timezone data; instead, the recommendation is to use a separately maintained library called pytz, and if you need to work with dates and times, I strongly encourage you to investigate that library in more detail.
Now we can finally delve into the main topic of this section, which is arrow: a third-party library that offers a much simpler API for working with dates and times. With arrow, everything has a timezone attached (and is thus “aware”):
Output:
As you can see, the now() function produces the current date and time with my local timezone attached (UTC+10), while the utcnow() function also produces the current date and time, but with the UTC timezone attached. As a consequence, the actual difference between the two times is as it should be: zero.
And the delightful arrow library just gets better from there. The objects themselves have convenient attributes and methods you would expect:
Note that the datetime produced from the same attribute correctly carries the essential timezone information.
There are several other features of the module that you can read more about in the documentation, but this last feature provides an appropriate segue to the next section:
The humanization even has built-in support for several locales!
ALTERNATIVE LIBRARIES FOR DATES AND TIMES
There are several other excellent libraries for dealing with dates and times in Python, and it would be worth checking out both Delorean and the very new Pendulum library.
parsedatetime
parsedatetime is a wonderful library with a dedicated focus: parsing text into dates and times. As you’d expect, it can be obtained with pip install parsedatetime. The official documentation is very API-like, which makes it harder than it should be to get a quick overview of what the library offers, but you can get a pretty good idea of what’s available by browsing the extensive test suite.
The minimum you should expect of a datetime-parsing library is to handle the more common formats, and the following code sample demonstrates this:
This produces the following, and unsurprising, output:
By default, if the year is given last, then month-day-year is assumed, and the library also conveniently handles the presence or absence of leading zeros, as well as whether hyphens (-) or slashes (/) are used as delimiters.
Significantly more impressive, however, is how parsedatetime handles more complicated, “natural language” inputs:
Incredibly, this all works just as you’d hope:
The urge to combine this with a speech-to-text package like SpeechRecognition or watson-word-watcher (which provides confidence values per word) is almost irresistible, but of course you don’t need complex projects to make use of parsedatetime: even allowing a user to type in a friendly and natural description of a date or time interval might be much more convenient than the usual but frequently clumsy DateTimePicker widgets we’ve become accustomed to.
NOTE
Another library featuring excellent datetime parsing abilities is Chronyk.
General-Purpose Libraries
In this chapter we take a look at a few batteries that have not yet been included in the Python standard library, but which would make excellent additions.
General-purpose libraries are quite rare in the Python world because the standard library covers most areas sufficiently well that library authors usually focus on very specific areas. Here we discuss boltons (a play on the word builtins), which provides a large number of useful additions to the standard library. We also cover the Cython library, which provides facilities for both massively speeding up Python code, as well as bypassing Python’s famous global interpreter lock (GIL) to enable true multi-CPU multi-threading.
boltons
The boltons library is a general-purpose collection of Python modules that covers a wide range of situations you may encounter. The library is well-maintained and high-quality; it’s well worth adding to your toolset.
As a general-purpose library, boltons does not have a specific focus. Instead, it contains several smaller libraries that focus on specific areas. In this section I will describe a few of these libraries that boltons offers.
boltons.cacheutils
boltons.cacheutils provides tools for using a cache inside your code. Caches are very useful for saving the results of expensive operations and reusing those previously calculated results.
The functools module in the standard library already provides a decorator called lru_cache, which can be used to memoize calls: this means that the function remembers the parameters from previous calls, and when the same parameter values appear in a new call, the previous answer is returned directly, bypassing any calculation.
boltons provides similar caching functionality, but with a few convenient tweaks. Consider the following sample, in which we attempt to rewrite some lyrics from Taylor Swift’s 1989 juggernaut record. We will use tools from boltons.cacheutils to speed up processing time:
Here we restrict which kinds of words will be substituted.
NOTE
The excellent vocabulary package is used here to provide access to synonyms and antonyms. Install it with pip install vocabulary.
For brevity, I’ve included only the first verse and chorus. The plan is staggeringly unsophisticated: we’re going to simply swap words with either a synonym or antonym, and which is decided randomly! Iteration over the words is straightforward, but we obtain synonyms and antonyms using the vocabulary package, which internally calls APIs on the Internet to fetch the data. Naturally, this can be slow since the lookup is going to be performed for every word, and this is why a cache will be used. In fact, in this code sample we use two different kinds of caching strategies.
boltons.cacheutils offers two kinds of caches: the least recently used (LRU) version, which is the same as functools.lru_cache, and a simpler least recently inserted (LRI) version, which expires entries based on their insertion order.
In our code, we use an LRU cache to keep a record of the parts of speech lookups, and we even save this cache to disk so that it can be reused in successive runs.
We also use an LRI cache to keep a record of word substitutions. For example, if a word is to be swapped with its antonym, the replacement will be stored in the LRI cache so that it can be reused. However, we apply a very small limit to the setting for maximum size on the LRI cache, so that words will fall out of the cache quite regularly. Using an LRI cache with a small maximum size means that the same word will be replaced with the same substitution only locally, say within the same verse; but if that same word appears later in the song (and that word has been dropped from the LRI cache), it might get a different substitution entirely.
The design of the caches in boltons.cacheutils is great in that it is easy to use the same cache for multiple functions, as we do here for the synonym() and antonym() functions. This means that once a word substitution appears in the cache, a call to either function returns the predetermined result from the same cache.
Here is an example of the output:
On second thought, perhaps the original was best after all! It is worth noting just how much functionality is possible with a tiny amount of code, as long as the abstractions available to you are powerful enough.
boltons has many features and we cannot cover everything here; however, we can do a whirlwind tour and pick out a few notable APIs that solve problems frequently encountered, e.g., in StackOverflow questions.
boltons.iterutils
boltons.iterutils.chunked_iter(src, size) returns pieces of the source iterable in size-sized chunks (this example was copied from the docs):
A similar requirement that often comes up is to have a moving window (of a particular size) slide over a sequence of data, and you can use boltons.iterutils.windowed_iter for that:
Note that both chunked_iter() and windowed_iter() can operate on iterables, which means that very large sequences of data can be processed while keeping memory requirements tolerable for your usage scenario.
boltons.fileutils
The copytree() function alleviates a particularly irritating behavior of the standard library’s shutil.copytree() function: Boltons’ copytree() will not complain if some or all of the destination file-system tree already exists.
The boltons.fileutils.AtomicSaver context manager helps to make sure that file-writes are protected against corruption. It achieves this by writing file data to temporary, or intermediate files, and then using an atomic renaming function to guarantee that the data is consistent. This is particularly valuable if there are multiple readers of a large file, and you want to ensure that the readers only ever see a consistent state, even though you have a (single!) writer changing the file data.
boltons.debugutils
If you’ve ever had a long-running python application, and wished that you could drop into an interactive debugger session to see what was happening, boltons.debugutils.pdb_on_signal() can make that happen. By default, a KeyboardInterrupt handler is automatically set up, which means that by pressing Ctrl-C you can drop immediately into the debugger prompt. This is a really great way to deal with infinite loops if your application is difficult to debug from a fresh start otherwise.
boltons.strutils
There are several functions in boltons.strutils that are enormously useful:
slugify(): modify a string to be suitable, e.g., for use as a filename, by removing characters and symbols that would be invalid in a filename.ordinalize(): given a numerical value, create a string referring to its position:
cardinalize: given a word and a count, change the word for plurality and preserve case:
singularizeandpluralize:
bytes2human: convert data sizes into friendler forms:
There are several other useful boltons libraries not mentioned here, and I encourage you to at least skim the documentation to learn about features you can use in your next project.
Cython
Cython is a magical tool! As with most magical devices, it is difficult to describe exactly what it is. Cython is a tool that converts Python source code into C source code; this new code is then compiled into a native binary that is linked to the CPython runtime.
That sounds complicated, but basically Cython lets you convert your Python modules into compiled extension modules. There are two main reasons you might need to do this:
You want to wrap a C/C++ library, and use it from Python.
You want to speed up Python code.
The second reason is the one I’m going to focus on. By adding a few type declarations to your Python source code, you can dramatically speed up certain kinds of functions.
Consider the following code, which is as simple as I could possibly make it for this example:
Print the modified data; here, we only show the first five entries.
This code represents the most basic computer processing: data comes in, is transformed, and goes out. The specific code we’re using is quite silly, but I hope it is clear enough so that it will be easy to understand how we implement this in Cython later.
We can run this program on the command line in the following way:
I’ve include the time command to get some performance measurements. Here we can see that this simple program takes around 30 seconds to run.
In order to use Cython, we need to modify the code slightly to take advantage of the Cython compiler’s features:
The work inside the loop is identical to before, except that we manipulate elements of the memory view rather than a itself.
Having modified our source code by adding information about native datatypes, we need to make three further departures from the normal Python workflow necessary before running our Cython code.
The first is that, by convention, we change the file extension of our source-code file to pyx instead of py, to reflect the fact that our source code is no longer normal Python.
The second is that we must use Cython to compile our source code into a native machine binary file. There are many ways to do this depending on your situation, but here we’re going to go with the simple option and use a command-line tool provided by Cython itself:
Running this command produces many lines of output messages from the compiler, but when the smoke clears you should find a new binary file in the same place as the .pyx file:
This is a native binary that Cython produced from our slightly modified Python source code! Now we need to run it, and this brings us to the third departure from the normal Python workflow: by default, Cython makes native extensions (as shared libraries), which means you have to import these in the same way you might import other Python extensions that use shared libraries.
With the first version of our example in ordinary Python, we could run the program easily with python cythondemoslow.py. We can run the code in our compiled Cython version simply by importing the native extension. As before, we include the time for measurement:
The Cython program gives us a speed-up over the plain Python program of almost 40 times! In larger numerical programs where the time cost of start-up and other initialization is a much smaller part of the overall execution time, the speed-up is usually more than 100 times!
In the example shown here, all our code was set out in the module itself, but usually you would write functions and after compiling with Cython, and then import these functions into your main Python program. Here’s how Cython can easily integrate into your normal Python workflow:
Begin your project with normal Python.
Benchmark your performance.
If you need more speed, profile your code to find the functions that consume most of the time.
Convert these functions to Cython functions, and compile the new
.pyxCython modules into native extensions.Import the new Cython functions into your main Python program.
Multithreading with Cython
It won’t take long for a newcomer to the Python world to hear about Python’s so-called GIL, a safety mechanism Python uses to decrease the possibility of problems when using threads.
Threading is a tool that lets you execute different sections of code in parallel, allowing the operating system to run each section on a separate CPU. The “GIL problem” is that the safety lock that prevents the interpreter state from being clobbered by parallel execution also has the unfortunate effect of limiting the ability of threads to actually run on different CPUs. The net effect is that Python threads do not achieve the parallel performance one would expect based on the availability of CPU resources.
Cython gives us a way out of this dilemma, and enables multithreading at full performance. This is because native extensions (which is what Cython makes) are allowed to tell the main Python interpreter that they will be well-behaved and don’t need to be protected with the global safety lock. This means that threads containing Cython code can run in a fully parallel way on multiple CPUs; we just need to ask Python for permission.
In the following code snippet, we demonstrate how to use normal Python threading to speed up the same nonsense calculation I used in previous examples:
We wait for the threads to complete.
I’ve also sneakily added a few small optimization options such as disabling bounds checking and enabling the faster “C division.” Cython is very configurable in how it generates C code behind the scenes and the documentation is well worth investigating.
As before, we must compile our program:
Then we can test the impact of our changes:
The use of threading has given us around 30% improvement over the previous, single-threaded version, and we’re about 50 times faster than the original Python version in this example. For a longer-running program the speedup factor would be even more significant because the startup time for the Python interpreter would account for a smaller portion of the time cost.
Executables with Cython
One final trick with Cython is creating executables. So far we’ve been compiling our Cython code for use as a native extension module, which we then import to run. However, Cython also makes it possible to create a native binary executable directly. The key is to invoke cython directly with the --embed option:
This produces a C source file that will compile to an executable rather than a shared library.
The next step depends on your platform because you must invoke the C compiler directly, but the main thing you need to provide is the path to the Python header file and linking library. This is how it looks on my Mac:
Here I’ve used a utility called python3.5-config that conveniently returns the path to the header file and the Python library, but you could also provide the paths directly.
The compilation step using gcc produces a native binary executable that can be run directly on the command line:
There is much more to learn about Cython, and I’ve made a comprehensive video series, Learning Cython (O’Reilly) that covers all the details. Cython’s online documentation is also an excellent reference.
awesome-python
Finally, we have awesome-python. It’s not a library, but rather a huge, curated list of a high-quality Python libraries covering a large number of domains. If you have not seen this list before, make sure to reserve some time before browsing because once you begin, you’ll have a hard time tearing yourself away!
Conclusion
There is much more to discover than what you’ve seen in this report. One of the best things about the Python world is its enormous repository of high-quality libraries.
You have seen a few of the very special features of the standard library like the collections module, contextlib, the concurrent.futures module, and the logging module. If you do not yet use these heavily, I sincerely hope you try them out in your next project.
In addition to those standard library modules, we also covered several excellent libraries that are also available to you on the PyPI. You’ve seen how:
flit makes it easy for you to create your own Python packages, and submit them to the PyPI.
libraries like colorama and begins improve your command-line applications.
tools like pyqtgraph and pywebview can save you lots of time when creating modern user interfaces, including hug, which can give your applications an easily created web API.
system libraries like psutil and watchdog can give you a clean integration with the host operating system.
temporal libraries like arrow and parsedatetime can simplify the tangled mess that working with dates and times often becomes.
general-purpose libraries like boltons and Cython can further enrich the already powerful facilities in the Python standard library.
Hopefully you will be able to use one or more of the great libraries in your next project, and I wish you the best of luck!
1 Looking for sorted container types? The excellent sorted containers package has high-performance sorted versions of the list, dict, and set datatypes.
2 For instance, this example with setdefault() looks like d.setdefault(k, []).append(...). The default value is always evaluated, whereas with defaultdict the default value generator is only evaluated when necessary. But there are still cases where you’ll need setdefault(), such as when using different default values depending on the key.
3 CPython means the specific implementation of the Python language that is written in the C language. There are other implementations of Python, created with various other languages and technologies such as .NET, Java and even subsets of Python itself.
4 Programs that automate some task, often communicating data across different network services like Twitter, IRC, and Slack.
5 The X.Y.Z versioning scheme shown here is known as semantic versioning (“semver”), but an alternative scheme worth investigating further is calendar versioning, which you can learn more about at calver.org.
6 pip install colorlog
7 Unfortunately, PyQt itself can be either trivial or very tricky to install, depending on your platform. At the time of writing, the stable release of pyqtgraph requires PyQt4 for which no prebuilt installer is available on PyPI; however, the development branch of pyqtgraph works with PyQt5, for which a prebuilt, pip-installable version does exist on PyPI. With any luck, by the time you read this a new version of pyqtgraph will have been released!
Last updated
