The following are general guidelines to getting Giterary running. These instructions are biased towards the Linux side of things, but there is no reason an enterprising individual couldn’t get them running on Mac OS or Windows. They also assume you have a basic working knowledge of the components of the Internet, how to configure your systems to run “services” or “daemons,” or a certain quantity of patience and a willingness to Google something if you don’t understand it.
These are widely available from many different Linux distribution packaging mechanisms (Debian and Ubuntu’s apt-get, Centos and Fedora’s yum, etc.) Binary packages are also available from their respective sites for all operating systems.
Web server software that supports running PHP
Numerous pieces of web server software support the use of PHP. Popular choices include:
- Apache (Available on Linux, Mac OS, and Windows)
- Nginx (Available on Linux, Windows, probably Mac OS)
- IIS (Windows) (Available on Windows)
You may not feel up to installing and configuring these yourself. I know: I’ve had those days, where the weight of the world dictates that you won’t look at another config file. But don’t despair: these are so popular and ubiquitous that many people sell services, construct virtual machines appliances, and make a ton of tools so that you can get off the ground with a web server and PHP easily. So much so that they’ve even given these types of setups a clever acronym: LAMP or WAMP ("Linux/Windows, Apache, MySQL, and PHP", minus the MySQL in our case). Do a couple of Google queries for “LAMP server” or “LAMP appliance” and you’ll see what I mean: tons of step by step guides, virtual machine appliances, and hosting services, all waiting for your time and patience (or money, potentially).
Unfortunately, those steps are beyond the scope of this document. We provide verification steps later here to make sure things are working, but getting to that point has been documented and solved much better by smarter people than the author of this document. No matter if you’re installing on your daily-driver laptop, or a fully-redundant cloud server, I would invite you to learn about such things, as they are valuable things to be knowledgeable about, will teach you about the underpinnings of the Internet at large, and makes you a blast at parties).
OPTIONAL: An SSH server program (if you are doing git-over-SSH synchronization, which is recommended)
git has the ability to push and pull from multiple sources, and using multiple network protocols. The most reliable and conventionally accepted appears to be SSH, which provides encryption as well as authentication. If you plan on “synchronizing” another computer with your Giterary instance, you will want to have an SSH server or git-capable equivalent to provide access to your Giterary repository.
- OpenSSH is widely supported on Linux, Mac OS, and on Windows.
OPTIONAL: A git GUI client
You will need the command line git client installed. However, there are a number of GUI clients (graphical user interfaces) that are a lot more friendly to human eyes.
It is highly recommended that you keep a git client handy, just in case you don’t like how Giterary does something, or that you have to do something Giterary doesn’t. These tools let you interact with your Giterary repository in lots of nice ways. They also provide a more forgiving git learning tool than the command line, man pages, and frustrated Google searches. I recommend the following:
These requirements may seem out of control, but they’re not so bad, and in fact, are pretty common among web applications (though hidden by the fact that you only need a web browser to use them successfully). I can give a few recommendations, pointers, and places to get started with hosting your own Giterary instance.
(I’ll always advocate running your own servers before “going to the cloud.” It means that you own your hardware and your data, and you have to worry slightly less about how your novel will fare after the zombie apocalypse. However, starting from scratch, the learning curve for “rolling your own” is admittedly steep and having other services available to jumpstart your progress is too useful to leave it unmentioned. Plus, setting up and maintaining highly redundant, highly available server hardware is expensive and time consuming, and some people just don’t value geek cred as much as I do :) )
Here are a few things to look into if you’re a) installing your own server, b) exploring options for your own server, or c) curious about how to become a hit at parties:
- Ubuntu happens to be my favorite “I don’t care about the details, I just want to use something” flavor of Linux. Its package system tends to have everything, and if you’re wanting/needing it, it has a well-supported desktop environment, and a large user community.
- Amazon’s EC2 (Elastic Compute Cloud) is a cheap and quick way to get hosted in the “cloud,” and they can bill at cents per hour, if you’re wanting to be economical.
- TurnKey Linux is a collection of tested, pre-configured virtual machine appliances that can get you set up and running with what you want in just a few minutes. They have a pre-configured LAMP stack, and partner with Amazon’s EC2 to provide cheap virtual machine hosting on their Turnkey Hub service.
- prgrmr's tagline is “We don’t assume you are stupid.” Which is pretty great. However, they aren’t for the faint of heart. In fact, you may or may not have to send in a photo of your neckbeard for verification before you can be sold anything.
Another option is to call up or email that computer-savvy person you know, or, make friends with one you don’t. Honestly: we live for this stuff. Memorize the following phrase (phonetically, if you have to): “I want to set up an Apache server with git and SSH, and have it sync with my laptop.” If you work in a business that has an Information Technology department, you can walk into the middle of their work area and shout that above the server noise. Any geek worth their salt will venture out of their cube to ask further questions. Best case: they set it up for you. Worst case: you learn who to talk to next time something is really broken.
There are also Linux User Groups in pretty much every major city, state, and university campus. They like to be able to serve their community, and installing something like Giterary is an easy and straightforward task for them.
Grab the latest version of Giterary from:
TODO: Github? Bitbucket?
Verify PHP versus Your Web Server
Every web server has a directory into which you can put files, and it will serve those files, for better or for worse.
Much of the configuration headache for web servers is getting the names, access rules, and pointers to the right directories into place. We leave it up to the reader to maintain their own security, but take heed when we say that the intent of a web server is to put things on the Internet. Be sure that you check to see if things are exposed that should not be (for instance, if your Giterary instance is exposed to whoever happens to be sitting next to your at the airport terminal).
Most important to establishing this “hosted directory” is ensuring that PHP is configured correctly. Within your “hosted” directory, place a file called test.php, which has only these contents.
<? phpinfo(); ?>
Afterwards, using your web browser, browse to a URL referencing your host and the test.php file, for instance:
If you plan to be accessing Giterary from only your own machine:
If you are installing to a separate server:
If you are installing Giterary to a place other than the URL’s “root” (referring to the “/giterary/” portion of the URL not being at the “root” of the path after the myserver host specification.)
This page should contain an output from the phpinfo function that lists all of the compilation, configuration, and dependency information for your PHP installation. It will be purple.
If this is running successfully, then you have a working PHP configuration for your web server. Good job!
The Easy Way, or The Hard Way (install.sh)
There is Linux install script that tries to walk you through your various configuration necessities (names, paths, usernames, passwords, etc.). It isn’t particularly sophisticated, and isn’t particularly platform agnostic (has only been ran on Ubuntu), but it’s a lot easier than wading through all this mess.
It is located in the root directory of the source code, called install.sh.
If successful, it sets you up with a Giterary instance with:
- A working git repository
Reasonable directory structure for git repository, cache, temp, draft, and session information:
/var/lib/giterary /var/lib/giterary/repos/your_repo_name /var/lib/giterary/caches/your_repo_name /var/lib/giterary/temps/your_repo_name /var/lib/giterary/temps/your_repo_name/drafts /var/lib/giterary/temps/your_repo_name/sessions
Initializes your git repository, sets an anonymous user for the repository (required from operating from a web application like Giterary), and other small git repository configurations.
- A “passfile.csv” configuration, where a special CSV is added to Giterary that contains your username, git user information, and hashed password (pointed to with the config/auth.php config file). It is then configured as a “sensitive” file that only only be edited by the username your enter (with the config/permissions.php config file).
- Adds your DEFAULT_FILE to the repository with a welcoming message.
This is a fairly straightforward and secure configuration out of the box.
Verify Git Is Working
From your operating system’s command line (any terminal in Linux, Terminal in MacOS, the cmd MS-DOS prompt in Windows), attempt to execute the following command.
It should print out something like this:
git version 220.127.116.11
Alternatively, if your environment paths are somehow incorrect, you may need to be explicit in the path to your git executable.
To find your explicit path on Linux and MacOSX, you can issue the following command:
…which should return the explicit path. In my case, it returns this:
On Windows you may have to search the filesystem for git.exe and use the explicit path it returns.
With your explicit path, run the following command (but with your explicit path):
This should give you version information for your installation of git.
Verify the security configuration on your system for your git client. It needs to be executable by the user that is running your web server. This is critical for Giterary to function.
Note this path value, as this is the path you will need to use later in the Giterary configuration.
Create Giterary directories
Giterary needs a few directories to run:
The directory of the git repository.
This is a directory that will need to be readable/writable, and will store the git repository and its working directory. Recommended locations:
Linux or MacOSX:
The directory for Giterary to store “cache” information to make the application speedier.
This is a directory that will need to be readable/writable, and will store the Giterary cache files to help make browsing via Giterary speedier. Recommended locations:
Linux or MacOSX:
The directory for Giterary to store “session” information that allows you to be logged in to the application.
This is a directory that will need to be readable/writable, and will store the PHP session files that allow you to log in. Recommended locations:
Linux or MacOSX:
The directory for Giterary to store “temp” data, like draft files and file comparison inputs.
This is a directory that will need to be readable/writable, and will store the PHP session files that allow you to log in. Recommended locations:
Linux or MacOSX:
These locations are relatively arbitrary, and you could potentially combine the cache and session directories into one (but never the repository directory, that must remain apart from everything).
Oh, the config files…
There are a series of configuration files necessary to getting Giterary up and running. They are all PHP files, and can be edited using any text editor. They are located within the include/config/ directory of the Giterary source. They are situated such that one could have multiple instances of Giterary, but with different include/config directories, and the instances would be able to run simultaneously.
base.php is the largest configuration file, as it contains references to the most basic portions of Giterary that are necessary to run ("Where is my repository?", “What is my name?", etc.). Below is a line-by-line description of the configuration values within base.php.
The following are required fields.
SITE_NAME and SHORT_NAME are display values for showing on the Giterary interface which instance of Giterary is running. This is where you put a name for the “title” of Giterary.
define('SITE_NAME', "My New Story"); define('SHORT_NAME', "New Story");
The BASE_URL is the base HTTP URL that will be used to navigate to your Giterary instance. This is used in link generation.
The DEFAULT_FILE gives the name of the first file to be created within your Giterary instance, and the default file to be displayed when no file is specified (likely the first file that will be displayed with accessing your Giterary instance).
define('DEFAULT_FILE', 'Home' );
SRC_DIR is the filesystem location of the source files for Giterary.
GIT_PATH is the filesystem path to the command line git client on your server.
GIT_REPO_DIR is the path to the readable/writable directory that contains the git repository for this instance of Giterary.
CACHE_DIR is the readable/writable directory for maintaining caches of certain data within Giterary.
define( 'CACHE_DIR','/var/lib/giterary/caches/NAME_OF_YOUR_REPO/' );
CACHE_ENABLE is a value of 1 or 0, 1 being that use of the cache is enabled, 0 is that it is disabled.
define( 'CACHE_ENABLE', 0 );
TMP_DIR is the temporary directory for temporary files generated by Giterary. It can be under the same directory as the Giterary cache.
DRAFT_DIR is the directory for storing draft files within Giterary. It can be under the directory as the Giterary cache.
SESS_PATH is the directory for storing cookie information for interacting with your browser.
The following are optional fields.
STYLESHEET and CSS_DIR are relative URL references to the default CSS to be used by the Giterary application.
define('STYLESHEET', "simpler.css"); define('CSS_DIR', "css/");
The following are variables for configuring cookie information for communicating login session information with your web browser.
define('COOKIE_DOMAIN', 'www.YOUR_DOMAIN_HERE.com'); define('COOKIE_PATH', '/'); define('COOKIE_EXPR_TIME', 86400); define('SESS_NAME', 'GITERARYSESSION');
permissions.php and permissions.lib.php
permissions.php is the configuration file for how users are permitted to perform certain actions within Giterary. permissions.lib.php is a set of common permission settings, defined as named classes, and usable as objects to be “registered” within permissions.php.
A common, starting instance is “AllPlay", where all users can perform all functions without limit.
<? require_once( dirname( __FILE__ ) . '/permissions.lib.php' ); $registered_auth_calls = array(); register_auth_call( new AllPlay(), "can" ); ?>
For a more “private” Giterary experience, the MustBeLoggedIn class defines that in order to view any portions of the site, you must have successfully logged in to the site.
<? require_once( dirname( __FILE__ ) . '/permissions.lib.php' ); $registered_auth_calls = array(); register_auth_call( new MustBeLoggedIn(), "can" ); ?>
For a “mixed” Giterary experience, SensitiveFiles allows you to define certain files which are considered “sensitive,” in that you must belong to a set of defined users otherwise you can neither read nor write to the files.
<? require_once( dirname( __FILE__ ) . '/permissions.lib.php' ); $registered_auth_calls = array(); register_auth_call( new SensitiveFiles( array( "passfile.csv" => array( "jrhoades" ) ) ), "can" ); ?>
auth.php and auth.lib.php
Similar to permissions.php and permissions.lib.php, auth.php and auth.lib.php provide the configuration and the common case library for defining how people log in to Giterary. Essentially, how the user list is determined, and how passwords will be stored.
Currently there are two common configurations, StaticUserList and PasswordFile.
In a StaticUserList configuration, a class is defined in auth.lib.php that describes the user list, their properties, and their passwords. Editing the StaticUserList class in auth.lib.php is necessary for this configuration. This method is not necessarily recommended, though, is common due to the ease of maintenance.
<? require_once( dirname( __FILE__ ) . '/auth.lib.php' ); $registered_login_calls = array(); register_login_call( new StaticUserList(), "validate_login" ); ?>
In a PasswordFile configuration, a CSV is referenced that defines username, user git properties, and a hashed password. This file can exist within your a repository (and be editable from the Giterary interface), or be outside of your repository and be edited manually, it is up to your security requirements. A permissions.lib.php class is defined specifically for this type of file called “SensitiveFiles,” which requires specific user credentials to be able to edit a file.
<? require_once( dirname( __FILE__ ) . '/auth.lib.php' ); $registered_login_calls = array(); register_login_call( new PasswordFile( '/var/lib/giterary/repos/my_repo/passfile.csv' ), "validate_login" ); ?>
A “passfile” is a CSV formatted file that contains names, emails, and password information for users to log in with. An example file might look like this:
Username,GitUser,Password jimmy,"Jimmy <email@example.com>",6f1ed002ab5595859014ebf0951522d9
The file is a CSV, and as such, can be edited using the Giterary interface (and will render as a table when viewing). However, you don’t want just anyone on your Giterary instance viewing and/or editing your password file. In these cases, you want to use permissions.php to flag your passfile.csv as “Sensitive,” and only viewable/editable by a small set of users.
WARNING: Just because the Giterary interface restricts access to such a file doesn’t mean that people with access to your git repository won’t. Anyone who can read information from git can read your passfile. While efforts are taken to make sure your passwords are obfuscated, it’s important to know that git itself doesn’t do “security.” If you want to use a passfile, but not have the file be located in your git repository, simple move your passfile to somewhere that isn’t in your git repository. You will have to edit this file by hand.
Finally, you might be asking yourself, “Who wants to enter in ‘6f1ed002ab5595859014ebf0951522d9’ as a password?” The answer is: absolutely nobody. Nor do they. This is a “hash” of your password, a mathematical algorithm performed on your password to make it difficult to guess its original contents. When you log in to Giterary, the password you enter is ran through this same algorithm, and if the two “hashed” results are equal, you have successfully entered your password and you are allowed to log in. It’s a clean and useful way of storing passwords without having to store them at all.
However, the question becomes how to generate these password hashes. On Linux, the following commands are available:
md5sum sha256sum sha512sum
…And others, which let you pipe in arbitrary text (ie, your password) and return a hashed result. For example, for the password “blah” (which is a terrible password which you should never use):
echo -n "blah" | md5sum
…will return the following:
NOTE: You must include the “—n” portion of the echo command, otherwise echo will emit a newline character, which will invalidate your hash value.
The 6f1…22d9 value is what you paste into your Giterary passfile.csv. However, Giterary assumes that unless you specify your hashing algorithm, it will just use the MD5 algorithm. This is fine if you only ever generated hashes with MD5, but some people want a more cryptographically hardened hash value. If you want to do a SHA256 hash, you would have to put in the following:
Or, for SHA512:
"sha512$39ca2b1f97...(long, long result )...1c2fa54b7"
The important part being the “xxxx$” prefix. Anything before the “$” in the password fields is treated as a hashing hint, to suggest to Giterary which hashing algorithm is used. The “xxxx” hash hint can be anything that is supported by PHP’s hash() function.
The dict.php file defines the path to the “dictionary” file, as well as sets of configurable groupings of words for calculating document metrics (the “Statistics” page).
DICTIONARY_PATH is the path to a file that consists of 1 word per line. On Linux, this path is normally /usr/share/dict/words.
define( 'DICTIONARY_PATH', '/usr/share/dict/words' );
The following are the lists that define the groupings for which words are to be considered conjunctions, “non-counted” words (common words you don’t want counted as “words” for your stats), past tense verbs, and present tense verbs.
$conjunctions = array( /* ...for, and, not, nor... */ ); # http://en.wikipedia.org/wiki/Most_common_words_in_English $non_counted_words = array( /* ...the, be, to, of... */ ); $past_tense_verbs = array( /* ...was, wasn't, did, didn't... */ ); $present_tense_verbs = array( /* ...is, isn't, am, are... */ );
A flag for whether to calculate performance stats and include a “performance report” in the HTML comments of every Giterary page. Enabled by default.
<? define( 'PERF_STATS', 1 ); ?>
Defines the default timezone to be used when making changes from the Giterary interface.
<? date_default_timezone_set ( 'America/Anchorage' ); ?>
The conventions.php file defines the basic “conventional” assumptions Giterary makes when defining things. Examples of conventions are:
The suffix used for “dirified” directory names (dir by default).
define( 'DIRIFY_SUFFIX', "dir" );
Default character encoding for the Giterary instance (UTF-8 by default).
define( 'ENCODING', "UTF-8" );
Regular expressions to determine the valid filename and file path patterns used on the site.
$wikiname_pattern = '-_a-zA-Z0-9\.\s'; $wikifile_pattern = "@^([$wikiname_pattern]+)(\\/[$wikiname_pattern]+)*$@"; $wikilink_pattern = "@(\\\)?\[\[([$wikiname_pattern]+(\\/[$wikiname_pattern]+)*)(\|([\w\s\.\,\\/-]+))?\]\]@"; $functionlink_pattern = "@(\\\)?\[\[([a-zA-Z]+):(([^\]|,]+=[^\]|,]+)(,[^\]|,]+=[^\]|,]+)*)?(\|([\w\s\.\,\"/-]+))\]\]@";
html.php defines parameters for dealing with the HTML generation within Giterary.
The $allowed_tags configuration variable defines which tags are allowed to be used for final rendering within Giterary.
$allowed_tags = array( /* ...'<p>', '<h>', '<pre>', '<img>', '<table>'... */ );
themes.php defines the names and paths responsible for rendering the various pieces of Giterary. This file is not necessary to edit in an initial configuration, but is important if you want to add your own elements as “overrides” to the default theme.
DEFAULT_THEME is the name of the “default” theme to be used for your Giterary instance. This theme must be the name of a array key within the $themes configuration variable later on (and is by default).
The $renderables array is a list of name-to-path associations for ‘well-known’ renderable files, responsible for rendering anything from page layouts to reusable widgets or error messages withing Giterary.
$renderables = array( 'default_layout' => 'theme/default/renderable/layout/standard.php', 'edit_layout' => 'theme/default/renderable/layout/edit.php', 'show_layout' => 'theme/default/renderable/layout/show_layout.php', 'note' => 'theme/default/renderable/note.php', 'gen_error' => 'theme/default/renderable/gen_error.php', 'gen_header' => 'theme/default/renderable/gen_header.php', 'gen_history' => 'theme/default/renderable/gen_history.php', /* ... */ 'not_logged_in' => 'theme/default/renderable/not_logged_in.php', 'gen_csv' => 'theme/default/renderable/gen_csv.php' );
The $themes array defines as the keys the names of available themes, and as the values of those keys, arrays that describe “overrides” of the $renderables definitions for the named renderables. This allows a theme author to make as much use of the default theme as possible, and only define their theme as “exceptions” to the default theme as necessary.
$themes = array( 'default' => array(), );
Note again that editing this file is not required, but we’re just providing a description here for documentation purposes.
Initialize your git repository
Your git repository needs to be “primed,” so to speak, before being used by Giterary. First off, within your repository directory (perhaps /var/lib/giterary/repos/your_repo), do the following command:
This will initialize an empty directory for you to work with (and creates the “hidden” files that git uses alongside your own).
You also need to inform git “who you are,” though, this isn’t really your name, as the Giterary application provides your committing information for you. However, git still squawks when this is unavailable. To silence this, we set a temporary, “anonymous” user for the repository. The name doesn’t particularly matter, nor does the email address have to be valid.
git config user.name "Anonymous User" git config user.email "firstname.lastname@example.org"
(Once the Giterary authors figure out how to get around this, we will likely be able to skip this step.)
That was horrific. Am I done?
Well, I think so. At least the Giterary part anyway. You should browse to your Giterary instance with your web browser, loading up index.php to see if it displays the editing screen for the file indicated by the file name configured with DEFAULT_FILE in base.php.
If the page loads, and you’re able to save your file successfully, then you’ve succeeded in installing and configuring Giterary.
It didn’t work. I have shamed myself and my family.
No! Not at all. We are human, we must err, or, do things which cause error messages.
As we get more people performing installations we will include “common” errors and hopefully provide reasonable solutions.
I say “optional,” but really I mean “very cool, but if it’s already 2am and you’ve got stuff to do in the morning, it’s alright if you come back later.”
The following are the cool things around which Giterary was based: being able to use your own favorite editing tools, and being able to synchronize from any source, and in any direction.
Being able to push/pull from Giterary
I can’t recommend enough being able to push and pull from a setup external to the one running your Giterary instance. This is bread and butter of why you would use git for something like this in the first place. Unfortunately, unless you’ve worked with a version control system before, or have written programs collaboratively before, it’s unlikely that you’ve been a part of this kind of workflow.
The first part of being able to push/pull using git is probably to learn more about git itself. There are a ton of sources out there to learn how to use Git.
- gittutorial, git’s own tutorial (admittedly dense).
- git-scm’s documentation and videos, particularly their “Distributed Git” section
- Github’s 15 minute tutorial
Honestly, I learned git by way of Googling, man pages, and more furious Googling. The resources are out there. Stackoverflow is rife with questions on git and its usage, and usually has great answers for some of the more subtle use cases around git. They may not be obvious, and the verbs are sometimes a little odd, but there are some amazing things you can do with git. Pushing and pulling is just the start.
What you first want to be able to do is use a git client to be able to clone your Giterary repository, and use it as one of its “origins.” From there, you can see your Giterary files as they truly exist: in files and folders, named as close as could get to how they’re represented in Giterary.
From there, you can open any file in your git repository working directory in your favorite text editor. You may or may not have the live-preview of Markdown, or whichever file format you’re editing, but you also have a suite of tools available to you that are difficult or otherwise impossible to do from a web interface (multiple selections, column edits, search and replace, pattern recognition, intelligent spellchecking, etc.).
Save the file, and then use your git client to “commit.” This is an important step in the git workflow. Committing to your repository informs git that “Yes, indeed, I intended to make those changes, and I think I’m ready to save these.” Certain git clients smooth this over, informing you of changes and automatically adding them to your commit, which saves a bit of time. However, “true” git considers your working directory to be just that: a work in progress. It may be that you aren’t done yet, but you may have portions of your work you want to commit. Or, you can stage your changes and swap out your entire working directory to focus on something else entirely. git, like Giterary, tries to help you with the hard stuff, but also tries to get out of your way when you just want to work.
After committing, you can “push” back to your Giterary origin. Depending on your Giterary setup (and the presence of your post-receive git server hook (as described later), you should see your updates reflected in the Giterary interface. If not, be sure that you clear the cache of the edited documents in question.
Post-receive git server hook
git gives you the ability to push and pull from whatever repository you like. A recommended setup with Giterary is to synchronize a copy of your Giterary repository to a separate computer (your daily driver laptop, etc.), and synchronize via SSH.
This creates a problem, however. Potentially, you can push to your Giterary server repository, and while your push may succeed, the files in the working directory on the server may be out of sync.
To solve this, we use a feature of git called “hooks.” In particular, there is a hook called the post-receive hook, which allows you to perform arbitrary user action on the server after successfully receiving your commits.
Put the following into /path/to/your/repo/.git/hooks/post-receive to enable this feature:
#!/bin/bash echo "Refreshing server's working tree"; /usr/local/bin/git --git-dir=/path/to/your/repo/.git --work-tree=/path/to/your/repo/ reset --hard; exit 0;
Additionally, git takes issue with pushing into non-bare repositories. While normally this would be for good reason, Giterary needs to be able to push into a non-bare repository in order to synchronize its files for display on the Giterary interface.
git config receive.denyCurrentBranch ignore