Quick, no-setup web Server
QuickServ makes creating web applications dangerously easy, no matter what programming language you use.
QuickServ is a dependency-free, statically-linked, single-file web server that:
- Has sensible defaults
- Prints helpful error messages directly to the console
- Runs on any modern computer, with no setup or installation
- Needs no configuration
- Knows which files to run server-side, and which to serve plain
- Works with any programming language that can
read
andwrite
- Doesn't require understanding the intricacies of HTTP
- Enables Cross Origin Request Sharing (CORS) by default
- Works with or without the command line
QuickServ brings the heady fun of the 1990s Internet to the 2020s. It is
inspired by the Common Gateway Interface
(CGI), but is much
easier to set up and use. Unlike CGI, it works out of the box with no searching
for obscure log files, no learning how HTTP headers work, no fiddling with
permission bits, no worrying about CORS, no wondering where to put your scripts,
and no struggling with Apache mod_cgi
configurations.
Unlike with CGI, you don't have to know what anything from the previous paragraph means to use QuickServ.
It is perfect for:
- Building hackathon projects without learning a web framework
- Creating internal tools
- Prototyping web applications using any language
- Attaching web interfaces to scripts
- Controlling hardware with Raspberry Pis on your local network
- Trying out web development without being overwhelmed
QuickServ should not be used on the public Internet. It should only be used on private networks.
Using QuickServ is as easy as downloading the program, dragging it to your project folder, and double clicking to run. It automatically detects which files to execute, and which to serve directly to the user.
Click to view details
-
Make a project folder and add files to it. For example, if Python is installed, create a file called
test.py
in the project folder containing:#!python # Put your code here. For example: import random print(random.randint(0, 420))
Since
test.py
starts with#!something
, wheresomething test.py
is the command to execute the file, QuickServ will know to run it. If QuickServ is not running your file, make sure to add this to the beginning.On Windows, QuickServ also knows to automatically run files that end in
.exe
and.bat
. Any other file type needs to start with#!something
if it should be run. -
Move the downloaded
quickserv_windows_x64.exe
file to the project folder. -
Double click
quickserv_windows_x64.exe
in the project folder to start QuickServ. Allow access through Windows Defender if prompted. -
Go to http://127.0.0.1:42069 (or the address shown by QuickServ) to connect to your web application. In the example, to run
test.py
, go to http://127.0.0.1:42069/test.py.
Click to view details
-
Download the right version for your computer. If necessary, check what type of processor your Mac has. You will have to unzip the files after you download them.
-
Make a project folder and add files to it. For example, if Python is installed, create a file called
test.py
in the project folder containing:#!python # Put your code here. For example: import random print(random.randint(0, 420))
If you are making the file with TextEdit, you will need to go into
Format > Make Plain Text
to save the file in the correct format.Since
test.py
starts with#!something
, wheresomething test.py
is the command to execute the file, QuickServ will know to run it. If QuickServ is not running your file, make sure to add this to the beginning.On Mac, QuickServ also knows to automatically run files that have been compiled. Any other file type needs to start with
#!something
if it should be run. -
Move the downloaded
quickserv_macos_x64
orquickserv_macos_arm64
file to the project folder. -
Right click
quickserv_macos_x64
orquickserv_macos_arm64
in the project folder and select "Open." Then, press "Open" in the confirmation dialog box. After running it this way once, you will be able to start QuickServ by simply double-clicking the file. -
Go to http://127.0.0.1:42069 (or the address shown by QuickServ) to connect to your web application. In the example, to run
test.py
, go to http://127.0.0.1:42069/test.py.
Click to view details
It's easiest to install and run via the command line. Open the Terminal.
Enter the following commands. A password may be required for the first commands.
# Download
sudo curl \
--location \
--output /usr/local/bin/quickserv \
https://github.com/jstrieb/quickserv/releases/latest/download/quickserv_raspi_arm
# Make executable
sudo chmod +x /usr/local/bin/quickserv
# Make a project folder
mkdir -p my/project/folder
# Go to project folder
cd my/project/folder
# Add a test file
cat <<EOF > test.py
#!python3
# Put your code here. For example:
import random
print(random.randint(0, 420))
EOF
# Run QuickServ
quickserv
Go to http://127.0.0.1:42069 (or the address shown by QuickServ) to connect to
your web application. For example, to run test.py
, go to
http://127.0.0.1:42069/test.py.
Click to view details
Clicking to run executables does not have consistent behavior across Linux
distros, so it's easiest to install and run via the command line. Depending
on your computer's architecture, it may be necessary to change the filename
at the end of the curl
HTTP request URL below.
See all download options on the releases page.
# Download
sudo curl \
--location \
--output /usr/local/bin/quickserv \
https://github.com/jstrieb/quickserv/releases/latest/download/quickserv_linux_x64
# Make executable
sudo chmod +x /usr/local/bin/quickserv
# Make a project folder
mkdir -p /my/project/folder
# Go to project folder
cd /my/project/folder
# Add a test file
cat <<EOF > test.py
#!python3
# Put your code here. For example:
import random
print(random.randint(0, 420))
EOF
# Run QuickServ
quickserv
Go to http://127.0.0.1:42069 (or the address shown by QuickServ) to connect to
your web application. For example, to run test.py
, go to
http://127.0.0.1:42069/test.py.
Alternatively, use the instructions below to compile from source.
Click to view details
Compile and install from source using the following command. A version of Go greater than 1.16 is required because of the dependency on embedded filesystems.
go install github.com/jstrieb/quickserv@latest
Then create your project folder, populate it, and run QuickServ.
# Make a project folder
mkdir -p /my/project/folder
# Go to project folder
cd /my/project/folder
# Add a test file
cat <<EOF > test.py
#!python3
# Put your code here. For example:
import random
print(random.randint(0, 420))
EOF
# Run QuickServ
quickserv
To demonstrate key features of QuickServ, we will build a simple web application to perform addition. The code will not follow best practices, but it will show how little is needed to get started building with QuickServ.
First, we create create a project folder and drag the QuickServ executable into the folder, as in the getting started steps.
Next, inside the folder, we save the following text as index.html
:
<form action="/calculate">
<input name="first" type="number"> + <input name="second" type="number"> = ???
<br>
<button>Calculate</button>
</form>
This code submits two variables to the /calculate
page. In the browser, it
looks like this:
Then, we create a folder called calculate
inside the project folder. Inside
the calculate
folder, we save the following code as index.py
. The name
index.whatever
tells QuickServ to run this file when a user visits
http://website/calculate
instead of needing them to visit
http://website/calculate/index.py
.
Pay special attention to the code comments. They highlight a number of important QuickServ features.
#!python3
# Each QuickServ script must begin with a line like the one above so that
# QuickServ knows how to run the file. This line tells QuickServ that I would
# type `python3 this_file.py` to run this file at the command prompt. For
# example, if you wanted to do `julia this_file.py` instead, then you would
# make the first line of `this_file.py` be `#!julia`.
#
# Since we just want QuickServ to show the HTML code to the user and not run
# it, index.html does not begin with this. The first line is only required when
# QuickServ has to run the code.
import argparse
# All HTML form values get turned into command line arguments. The names are
# formatted like "--name" and the value comes right after the name.
parser = argparse.ArgumentParser()
parser.add_argument("--first", type=int, required=True)
parser.add_argument("--second", type=int, required=True)
args = parser.parse_args()
# Print the result -- anything printed out goes right to the user. In this
# case, the output is text. But you can print anything and QuickServ will guess
# the file type. Even printing the contents of image and video files works.
print(args.first + args.second)
Now double click QuickServ in your project folder and try it out in your browser. That's it!
See the examples linked in the next section for more QuickServ demonstrations. Read more details in the How it Works section, and in the code itself. The Advanced section has additional information about the environment QuickServ sets up for executables it runs.
All examples are located in the examples
folder, which is a Git submodule
connected to the
jstrieb/quickserv-examples
repo. Go to that repo for more information on how to run the examples.
Some highlights:
Summary | Language | Windows | Mac | Linux |
---|---|---|---|---|
YouTube to GIF | Shell | ✗ |
✓ |
✓ |
Create Bloom filters | C | ✓ |
✓ |
✓ |
Synchronized movie list | TypeScript | ✓ |
✓ |
✓ |
Brainfuck interpreter | x86-64 Assembly | ✗ |
✗ |
✓ |
Mandelbrot zoom | Shell | ✓ |
✓ |
✓ |
Cryptocurrency dashboard | Go | ✓ |
✓ |
✓ |
PDF tools | Python | ✗ |
✓ |
✓ |
Typeset Markdown as PDF | Standard ML | ✓ |
✓ |
✓ |
All of the QuickServ code lives in
quickserv.go
.
This well-commented file is about 700 lines long, and should take an experienced
programmer with no Golang familiarity at most an hour to read.
Click to view details
QuickServ has two main parts. The first is an initialization procedure, run exactly once at startup. The second is a handler function, called every time a user makes an HTTP request to the server.
When QuickServ starts up, it checks for command-line configuration flags, opens
a log file if one is passed with --logfile
(otherwise it logs to the standard
output), and changes directories if a working directory is passed with --dir
.
Note that the log file path is relative to the current working directory, not
relative to the one passed with --dir
.
Next, QuickServ scans the working directory for files to run. It prints all of the files that will be executed. This behavior is useful for determining if QuickServ recognizes a script as executable. It also prints helpful information for the user such as the web address to visit to access the server, and what folder the server is running in, as well as how to stop it.
If any part of the initialization fails, an error is reported. In the event of a fatal error, QuickServ waits for user input before quitting. This way, a user who double-clicks the executable (as opposed to starting it from the command line) does not have a window appear and then immediately disappear, flashing too quickly for the error to be read.
Error messages are purposefully written with as little technical jargon as possible, though some is unavoidable. Likely causes for the errors are also included in error messages, so that they are easier for users to identify and fix.
As the last step in the initialization procedure, QuickServ starts a web server
with a single handler function for all requests. The server listens on the
default port of 42069
, or on a random port if a user specified the
--random-port
command-line flag. A random port would be desirable if the user
has to show a project built with QuickServ to someone humorless, for example.
When a user visits a web page, QuickServ handles the request by calling the lone handler function.
First, this function tries to open the file the user requested. If it cannot
find or open the file, it tries to serve a default version of the file. For
example, there is an embedded, default favicon.ico
that gets served. If there
is no default file matching the path, it lets the built-in Go fileserver handle
the error and respond with a 404 error code.
If the file the user requested is present, it checks whether it is a directory.
If it is a directory, QuickServ looks inside for a file named index.xxx
where
xxx
is any file extension. If an index file is found, the index is served (and
possibly executed) as if it were the original page requested. Otherwise, the
user must have requested a directory without a default index, so QuickServ
responds with a listing of the other files in the directory.
If the file the user requested is present and not a directory (i.e., it is a
regular file), QuickServ checks whether or not it is executable. If so, it
executes the file it found. If not, it returns the raw file contents to the
user. In both cases, QuickServ will guess what filetype (and therefore which
mimetype
) to use for the response.
The technique for determining if a file is executable depends on the runtime
operating system. On Windows, any file with a .bat
or .exe
extension is
considered executable. On non-Windows systems, any file with the executable
permission bit set is considered executable. On all systems, a file is
executable if it has a valid pseudo-shebang at the beginning. The shebang must
be on the very first line, must begin with #!
, and must be a valid command.
For example, both of the following are acceptable, assuming python3
is
installed and on the PATH
:
#!/usr/bin/python3
#!python3
To execute a file, QuickServ either runs the file itself (if it is an .exe
or
has the executable bit set), or it passes the file's path as the first argument
to the executable listed in its shebang. The request body is passed to the
program on standard input, and everything printed by the program on standard
output is used as the response body. Executed programs are neither responsible
for writing—nor able to write—HTTP response headers.
All parsed HTTP form variables (if the Content-Type
is
x-www-form-urlencoded
) are also passed as command line arguments when the
program is executed. This way, the user does not need to parse the variables
themselves.
Whatever the executed program prints on standard error is logged by QuickServ, which means it gets printed in the console window by default. This is handy for debugging. If the program terminates with a non-zero exit code, QuickServ responds with a 500 internal server error. Otherwise it returns with a 200.
If the request is a URL-encoded POST request with form data, QuickServ
URL-decodes all of the characters except for three symbols: %
, &
, and =
.
The user is responsible for substituting these. Note that it is important to
always URL-decode %
last in the program that processes the form data.
Do not run QuickServ on the public Internet. Only run it on private networks.
QuickServ is not designed for production use. It was not created to be fast or secure. Using QuickServ in production puts your users and yourself at risk, please do not do it.
QuickServ lets people build dangerously insecure things. It does not sanitize any inputs or outputs. It uses one process per request, and is susceptible to a denial of service attack. Its security model presumes web users are trustworthy. These characteristics make prototyping easier, but are not safe on the public Internet.
To deter using QuickServ in production, it runs on port 42069
. Hopefully that
makes everyone think twice before entering it into a reverse proxy or port
forward config. For a more professional demo, the command-line flag
--random-port
will instead use a random port, determined at runtime.
QuickServ is similar to the ancient CGI protocol. There are many well-articulated, well-established reasons that CGI is bad in production, and they all apply to QuickServ in production.
QuickServ has advanced options configured via command line flags. These change how and where QuickServ runs, as well as where it saves its output.
Usage:
quickserv [options]
Options:
--dir string
Folder to serve files from. (default ".")
--logfile string
Log file path. Stdout if unspecified. (default "-")
--no-pause
Don't pause before exiting after fatal error.
--random-port
Use a random port instead of 42069.
In imitation of CGI, HTTP headers are passed to the executed program as
environment variables. A header called Header-Name
will be set as the
environment variable HTTP_HEADER_NAME
.
There is also a REQUEST_TYPE
variable that specifies whether the request was
GET
, POST
, etc.
HTTP requests with a body pass the body to the executed program on standard input. In most cases, the request body is passed verbatim. This is not the case for HTML forms.
HTML form data can either be read from command line arguments, as in the tutorial, or parsed from standard input. Variables take the form
name=value&othername=othervalue
The simple addition example from the tutorial can be rewritten to parse HTTP form values from the standard input instead of from the command line arguments.
#!python3
import sys
# In the form input, "=" and "&" determine where variables start and end. So if
# they are literally included in the variable name or value, they must be
# specially decoded. This code replaces every instance of the text on the left
# with the text on the right to do the decoding:
# %3D -> =
# %26 -> &
# %25 -> %
#
# NOTE: Order matters! "%" must be decoded last. If not, it can mess with
# decoding the others, since their encoded version uses "%"
def decode_characters(text):
text = text.replace("%3D", "=")
text = text.replace("%26", "&")
text = text.replace("%25", "%")
return text
first = second = 0
# Read all of the input into a variable. We are expecting the raw data to look
# like:
# first=123&second=456
data = sys.stdin.read()
# The raw data looks like the above, so split it into pairs at each "&"
pairs = data.split("&")
for pair in pairs:
# Each pair looks like the following, so split at each "=":
# name=value
name, value = pair.split("=")
# Decode any special characters (=, &, %) now that we have split the
# variables up. This isn't necessary here since we're expecting numbers and
# not expecting any of those characters. But it matters a lot when a user
# could submit text with those characters
name = decode_characters(name)
value = decode_characters(value)
# If the name is what we're looking for, store the value for adding
if name == "first":
first = int(value)
elif name == "second":
second = int(value)
# Print the result -- anything printed out goes right to the user. In this
# case, the output is text. But you can print anything and QuickServ will try and
# guess the file type.
print(first + second)
This project is actively developed and maintained. If there are no recent commits, it means that everything is running smoothly!
Please open an issue with any bugs, suggestions, or questions. This especially includes discussions about how to make error messages as clear as possible, and how to make the default settings applicable to as many users as possible.
Pull requests without prior discussion will be ignored – don't waste time writing code before confirming that it will be merged in. As a busy, lone developer, it is easier to be responsive when all code contributions have context.
If you make a blog post, video, tutorial, hackathon project, or anything else using QuickServ, please open an issue or message me on my contact form so that I can link back to it!
There are a few ways to support the project:
- Star the repository and follow me on GitHub
- Share and upvote on sites like Twitter, Reddit, and Hacker News
- Report any bugs, glitches, or errors that you find
- Translate into other languages so everyone can use the project
- Build and share your own projects made with QuickServ
These things motivate me to to keep sharing what I build, and they provide validation that my work is appreciated! They also help me improve the project. Thanks in advance!
If you are insistent on spending money to show your support, I encourage you to instead make a generous donation to one of the following organizations. By advocating for Internet freedoms, organizations like these help me to feel comfortable releasing work publicly on the Web.
This project would not be possible without the help of:
- Logan Snow
- Amy Liu
- Hacker News user rchaves, who helpfully suggested passing parsed form values as command line arguments
- Everyone who supports the project