Development Policies

From OpenEMR Project Wiki
Revision as of 03:02, 25 August 2016 by Matthew Vita (talk | contribs) (→‎Testing: new markdown test link)

Submitting Patches to Upstream

Overview of submitting code

  • Code patches for review can be submitted via two methods:
  • Please derive patches/code from the most current OpenEMR development version from either of below repositories:

Creating A Patch

(You should never need to do below and much better to instead use git/github)

Creating patches with diff

Using diff is simple whether you are working with single files or entire source directories. To create a patch for a single file, use the form:

   diff -u original.c new.c > original.patch

To create a patch for an entire source tree, make a copy of the tree:

   cp -R original new

Make any changes required in the directory new/. Then create a patch with the following command:

   diff -rupN original/ new/ > original.patch

That's all you need to get started with diff and patch. For more information use:

  man diff
  man patch

Notifying the Development Team

Place the patch or the github branch link in the one of the Source Forge trackers with an explanation. Use Bugs to log items that should work but don't; you can attach a patch to fix things if you have one. Use Patches for small changes that you consider complete. Use Code Review for items that need input from the community and possible re-work before being committed. Feature Requests is for items that you'd like OpenEMR to do, but it doesn't currently; developers might use these to guide development, but most of us are busy implementing stuff for our clients / employers; for new features it might be better to get OpenEMR Commercial Help.

Please also place an explanation of the patch or github branch in the developer forum so we know its in the tracker. SourceForge does not automatically link the tracker item to the forum thread, so make sure and post a (comment on the tracker item that includes the forum thread URL) and a (message in the forum thread that includes the tracker item URL).

We can't respond to everything immediately, and sometimes a tracker item is forgotten. If you think that may have happened to your item, feel free to post a "bump" to the forum thread.

Carriage Returns / Line Feeds

All text files for the project should have Unix-style line endings (i.e. no carriage returns). Developers who use a Windows desktop should also use a suitable text editor that respects this (last checked, EditPad Lite and Notepad++ are free examples). Git can also nicely automatically handle this for Windows users: http://help.github.com/line-endings/

General Development Best Practices

Copyright and Licensing

Each file in the source tree should begin with a copyright declaration, and information about what license the file is released under.

Testing

In order to verify that code changes pass certification tests, contributors must execute the relevant test cases via the test mapping document. This document is currently being built up (http://open-emr.org/wiki/index.php/Manual_Tests) and lives here: https://github.com/openemr/openemr/blob/tests/tests/certification/tests.md

Database Contents

When placing a string from the database inside of an html attribute, or placing it as the text content of an html object, one should call htmlspecialchars() to provide appropriate escaping.

PHP

Many of the practices at http://www.odi.ch/prog/design/php/guide.php appear to be good rules when working with the OpenEMR source.

When including a file, make sure to use 'require_once', or use 'include_once' and check the return!

All blocks of PHP code should start with '<?php', and end with '; ?>'. for example: <?php echo $testvar; ?>

Comments

Line comments (C++-style) should only be used for comments that (a) appear "after" (in reading order) the code the are documenting and (b) appear on the same line as the code they are documenting.

Block comments (C-style) should be used for comments that (a) require multiple lines OR (b) require their own line. Block comments should appear before the code they document. Block comments are also used for notices that are not code documentation, such as copyright and licensing notices.

Depending on the scope of your work, you may need one or both comment styles.

Comments should never be used to "preserve" old code or mark individual changes. This information should be available through our version control system, and is rarely that useful anyway. Comments shouldn't answer the "How?" of the code; that's the role of the source code. Comments should answer the "Why?" of the code, what issues are being address by this code.

Comments that are doxygen-ready are appreciated, but we currently do not use doxygen to generate documentation. Using doxygen with PHP appears to have some rough edges, but it might be useful in the future.

Please go to How to Document Your Code Properly to get a good idea on how to document properly.

Horizontal Spacing

The current OpenEMR source is inconsistent about indentation using spaces or tabs. Changes to existing code should preserve the indentation style of the changed lines. Changes to existing files should follow the indentation style of the rest of the file for new lines; if the indentation style is inconsistent you may use any style currently in the file. In new files, try to use a style that already exists in the source code.

Possible Future Guidelines

For most code:

  • Use a single tab for an indentation level. Outside of literal HTML, having code at more than 4 indentation levels is a good indicator the code should be re-factored.
  • Use spaces for alignment. Alignment is only done between two lines with the same indentation level. Tabs appear before the spaces. Alignment is rare, but may be used when breaking an overly long line.
  • Use a single space for a "half-indent". While not common in PHP code, the "half-indent" is often used for "case $n:" labels in C/C++ and for "public:"/"private:" labels in C++.
  • Tabs after spaces should be avoided. Tabs should be represented as "\t" in string literals.
  • Unnecessary whitespace at the end of lines should be avoided.

For columnar/tabular data in a fixed-width font, when the width of each column is known or can be determined.

  • Use spaces for left/right padding of data items and headers out to the fixed column width.
  • Use a single tab for the gap between columns.

The above guidelines allow both you can all other developers of the code to choose any tab stop width greater than the width of a single space character and have the code align itself nicely. The allows each developer to choose how much horizontal whitespace they need to efficiently read the code.

Vertical Spacing

Avoid adding unnecessary empty lines.

Below is WRONG

    echo "
\n"; ?>

Javascript

To ensure consistent interpretation of scripts in HTML and XHTML, any script which might contain the ampersand ('&') or left-angle-bracket ('<') characters must begin with

//<![CDATA[

on a line by itself and must end with

//]]>

on a line by itself. In addition, the script must not contain the strings "]]>" or "</", even within a Javascript string literal or comment. Currently, we do not have a function that sanitizes a php value for inclusion as a Javascript value.

When need to encode variables, use the encodeURIComponent() function instead of using the escape() function (to ensure compatibility with utf-8 characters).

HTML

Each page in OpenEMR should be valid HTML 4.0, XHTML 1.0, or XHTML 1.1. the validator at http://validator.w3.org/ is useful for ensuring compliance.

For textual data that should not be interpreted as markup, use htmlspecialchars(). Using a second ("quote_style") argument of ENT_NOQUOTES minimizes the length of the resulting string and is safe for non-markup.

For (parts of) attribute values, use htmlspecialchars() and provide a second ("quote_style") argument of ENT_QUOTES. XHTML requires either single- or double-quote characters around all attribute values, and using ENT_QUOTES allows either of those characters to be used.

To ensure (X)HTML validity, element names and attribute names but be drawn from the standards, instead of any outside data source. Because of this no special processing should be required -- just make sure they are spelled correctly in the PHP code.

Bad

<?php
$user_data = $_REQUEST['foo'];
$db_data = sqlQuery('some SQL statement')[0];
echo "<p class='$db_data'>$user_data</p>\n";
?>

Better

<?php
$user_data = $_REQUEST['foo'];
$db_data = sqlQuery('some SQL statement')[0];
$para_class = htmlspecialchars($db_data, ENT_QUOTES);
$para_content = htmlspecialchars($user_data, ENT_NOQUOTES);
echo "<p class='$para_class'>$para_content</p>\n";
?>

The basic rule of thumb is to use it on just code that is getting outputted to the screen (in html). So, not a good idea to do it to variables at the top of the file; need to do htmlspecialchars on them when you echo them onto the screen.

For example:

CORRECT:

 $date = $POST_['date']
 ...
 echo htmlspecialchars($date)

INCORRECT:

 $date = htmlspecialchars($POST_['date'])
 ...
 echo $date

Although sometimes you will build a variable just for output, then it's ok to hemlspecialchars that variable:

$string = "<b>".htmlspecialchars($string)."</b>"
echo $string

The key is to realize that htmlspecialchars() function is altering that string specifically for html output, thus using this altered string in other ways (such as php, mysql, pdf output, network exchange) can break things.

CSS

it is preferred that CSS stylesheets contain all style related contents of our html forms. css stylesheets should also validate.

OpenEMR Specific Development Best Practices

MySQL connections

All of your MySQL calls need to go through openemr/library/sql.inc or you will break the encoding (utf8). See that file for details and the large amount of example throughout code. NEVER, NEVER, NEVER use the native mysql_* calls.

PHP Sessions and Browser Windows

You must include a JavaScript call to top.restoreSession() wherever you invoke a PHP script that requires current session data (which is most of them). How to do this is discussed in more detail in the architecture discussion wiki page.

Internationalization

  • The main php function used for translation is xl(), basically all of labels and messages have to go through this function. To learn about this function definition, parameters, and general use, please read this wiki page, and ensure you understand it.
  • These things are what I consider the tenets of the xl() function:
For coding new xl functions:
1. No trailing or leading whitespace.
Below is WRONG
xl('Demographics ');
Below is CORRECT
xl('Demographics') . ' ';
2. No variables.
Below is WRONG
xl('please type $name here');
Below is CORRECT
xl('please type') . ' ' . $name . ' ' . xl('here');
For previously coded xl functions:
  1. To be safe, just leave them be (the above rules do not apply).
  • Don't forget about javascript strings. As long as the javascript is in a .php or .inc file it will be translated.
For example:
alert("Please type letters only");
Should be:
alert("<?php echo xl('Please type letters only'); ?>");
  • Do not use the text or values of your buttons or controls in your coding algorithm. For example, if you have a 'submit' button and use the 'submit' string(the 'value' of the button) in your algorithm, then it will not work if it's translated to another word.
  • xl() returns data that has been pulled from the database and then further mangled. However, it has not gone through the necessary HTML, Javascript, or other escaping that is required as part of input/output sanitizing; treat it like user-provided data pulled from the database.

For good examples, look through the code. If any questions don't hesitate to ask them on the sourceforge developer forums.

Input Collection

See here for new method for doing this here.

SQL Table Name Conventions

Creating new SQL tables names with uppercase letters is NOT allowed. This is because these table names are difficult to support in both Windows and Linux. There are times, however, when this needs to be supported in the the following settings:
  1. Older code (aka the form_CAMOS module)
  2. Code that is beyond OpenEMR's control (for example, the RxNORM tables that are imported into OpenEMR).
There is a method to support this (as of OpenEMR 4.1.2), so that these table are supported in both Linux and Windows. In order for this to work, you need to surround the table name with the mitigateSqlTableUpperCase() function. For example, it will look like this:
$query1 = "select id, category from ".mitigateSqlTableUpperCase("form_CAMOS_category");
(You may wonder, what the heck does this do; this function will actually look up the table in mysql in a case insensitive manner and return the table name from mysql)

Access Control Objects

If you add a new Access Control Object to the OpenEMR codebase, then also add it to the following three sites:

  1. Header notes of the library/acl.inc file
  2. acl_setup.php file
  3. acl_upgrade.php file

Email

We have a standardized method for sending emails. Please see here for details : Sending Email

Database Table Changes

Changes to the database schema are now versioned. Please increment the value of $v_database in version.php, so that we can know automatically when a database upgrade is required. This is checked in admin.php and if the versions differ, a manual upgrade is required.

Updates also need to be included in the database.sql and ####-to-####-upgrades.sql META files. See below for notes on using these files to support upgrades.

Upgrade SQL META Language

Comment Meta Language for sql upgrades (via the sql_upgrade.php script):

While these lines are as enigmatic as they are functional, there is a method to the madness. Let's take a moment to briefly go over proper comment meta language use. The best way to learn this is to look at examples in current/previous scripts.

 
--  The #If* sections have the behavior of functions and come complete with arguments supplied command-line style 
--
--  Your Comment meta language lines cannot contain any other comment styles such as the nefarious double dashes "--" lest your lines be skipped and the blocks automatcially executed with out regard to the existing database state.   
--
--  Comment Meta Language Constructs:
-- 
--  #IfNotTable
--    argument: table_name
--    behavior: if the table_name does not exist,  the block will be executed

--  #IfTable
--    argument: table_name
--    behavior: if the table_name does exist, the block will be executed

--  #IfMissingColumn
--    arguments: table_name colname
--    behavior:  if the colname in the table_name table does not exist,  the block will be executed

--  #IfNotColumnType
--    arguments: table_name colname value
--    behavior:  If the table table_name does not have a column colname with a data type equal to value, then the block will be executed

--  #IfNotRow
--    arguments: table_name colname value
--    behavior:  If the table table_name does not have a row where colname = value, the block will be executed.

--  #IfNotRow2D
--    arguments: table_name colname value colname2 value2
--    behavior:  If the table table_name does not have a row where colname = value AND colname2 = value2, the block will be executed.

--  #IfNotRow3D
--    arguments: table_name colname value colname2 value2 colname3 value3
--    behavior:  If the table table_name does not have a row where colname = value AND colname2 = value2 AND colname3 = value3, the block will be executed.

--  #IfNotRow4D
--    arguments: table_name colname value colname2 value2 colname3 value3 colname4 value4
--    behavior:  If the table table_name does not have a row where colname = value AND colname2 = value2 AND colname3 = value3 AND colname4 = value4, the block will be executed.

--  #IfNotRow2Dx2
--    desc:      This is a very specialized function to allow adding items to the list_options table to avoid both redundant option_id and title in each element.
--    arguments: table_name colname value colname2 value2 colname3 value3
--    behavior:  The block will be executed if both statements below are true:
--               1) The table table_name does not have a row where colname = value AND colname2 = value2.
--               2) The table table_name does not have a row where colname = value AND colname3 = value3.

--  #EndIf
--    all blocks are terminated with and #EndIF statement. 

Creating a list in list_options

Lists are a nice way to build customizable lists that support standardized output formatting and internationalization. See Administration->Lists for details and check out the list_options mysql table.


Example of placement of new list in database.sql:
INSERT INTO list_options ( list_id, option_id, title, seq, is_default ) VALUES ('lists','proc_specimen','Procedure Specimen Types', 1,0);
INSERT INTO list_options ( list_id, option_id, title, seq, is_default ) VALUES ('proc_specimen','blood' ,'Blood' ,10,0);
INSERT INTO list_options ( list_id, option_id, title, seq, is_default ) VALUES ('proc_specimen','saliva','Saliva',20,0);
INSERT INTO list_options ( list_id, option_id, title, seq, is_default ) VALUES ('proc_specimen','urine' ,'Urine' ,30,0);
INSERT INTO list_options ( list_id, option_id, title, seq, is_default ) VALUES ('proc_specimen','oth'   ,'Other' ,90,0);


Example of placement of new list in the upgrade sql file:
#IfNotRow2D list_options list_id lists option_id proc_specimen
INSERT INTO list_options ( list_id, option_id, title, seq, is_default ) VALUES ('lists','proc_specimen','Procedure Specimen Types', 1,0);
INSERT INTO list_options ( list_id, option_id, title, seq, is_default ) VALUES ('proc_specimen','blood' ,'Blood' ,10,0);
INSERT INTO list_options ( list_id, option_id, title, seq, is_default ) VALUES ('proc_specimen','saliva','Saliva',20,0);
INSERT INTO list_options ( list_id, option_id, title, seq, is_default ) VALUES ('proc_specimen','urine' ,'Urine' ,30,0);
INSERT INTO list_options ( list_id, option_id, title, seq, is_default ) VALUES ('proc_specimen','oth'   ,'Other' ,90,0);
#EndIf

Note as demonstrated above that the seq entries increment in 10's, and that the option_id entry should not contain any blanks or caps.


These lists can be used to configuration of layouts in Administration->layouts, and can also use the functions in options.inc.php within scripts.

Example of generating a item selector using the list(note the input variable will be $POST['form_proc_specimen']):
generate_form_field(array('data_type'=>1,'field_id'=>'proc_specimen','list_id'=>'proc_specimen'), $defaultProcSpecimen);
Example of generating the title of the list from the option_id:
generate_display_field(array('data_type'=>'1','list_id'=>'proc_specimen'),$chosen_proc_specimen);

Creating a global configuration setting

As of version 4.0, these can be added to library/globals.inc.php . Then the variable can be called in code as so : $GLOBALS['yourVariableName'] . It can also then be configured by the user in Adminstration->Globals screen.