function coder_format_string in Coder 7
Same name and namespace in other branches
- 5.2 scripts/coder_format/coder_format.inc \coder_format_string()
- 5 scripts/coder_format/coder_format.inc \coder_format_string()
- 6.2 scripts/coder_format/coder_format.inc \coder_format_string()
- 6 scripts/coder_format/coder_format.inc \coder_format_string()
- 7.2 scripts/coder_format/coder_format.inc \coder_format_string()
Format the source code according to Drupal coding style guidelines.
This function uses PHP's tokenizer functions.
To achieve the desired coding style, we have to take some special cases into account. These are:
Indent-related: $_coder_indent int Indent level The number of indents for the next line. This is
- increased after {, : (after case and default).
- decreased after }, break, case and default (after a previous case).
$in_case bool Is true after case and default. Is false after break and return, if $braces_in_case is not greater than 0. $switches int Switch level Nested switches need to have extra indents added to them. $braces_in_case array Count of braces The number of currently opened curly braces in a case. This is needed to support arbitrary function exits inside of a switch control strucure. This is an array to allow for nested switches. $parenthesis int Parenthesis level The number of currently opened parenthesis. This
- prevents line feeds in brackets (f.e. in arguments of for()).
- is the base for formatting of multiline arrays. Note: If the last ');' is not formatted to the correct indent level then there is no ',' (comma) behind the last array value.
$in_brace bool Is true after left curly braces if they are in quotes, an object or after a dollar sign. Prevents line breaks around such variable statements. $in_heredoc bool Is true after heredoc output method and false after heredoc delimiter. Prevents line breaks in heredocs. $first_php_tag bool Is false after the first PHP tag. Allows inserting a line break after the first one. $in_do_while bool Is true after a do {} statement and set to false in the next while statement. Prevents a line break in the do {...} while() construct.
Whitespace-related: $in_object bool Prevents whitespace after ->. Is true after ->. Is reset to false after the next string or variable. $in_at bool Prevents whitespace after @. Is true after @. Is reset to false after the next string or variable. $in_quote bool Prevents
- removal of whitespace in double quotes.
- injection of new line feeds after brackets in double quotes.
$inline_if bool Controls formatting of ? and : for inline ifs until a ; (semicolon) is processed. $in_function_declaration Prevents whitespace after & for function declarations, e.g. function &foo(). Is true after function token but before first parenthesis. $in_array Array of parenthesis level to whether or not the structure is for an array. $in_multiline Array of parenthesis level to whether or not the structure is multiline.
Context flags: These variables give information about what tokens have just been processed so that operations can change their behavior depending on the preceding context without having to scan backwards on the fully formed result. Most of these are ad hoc and have a very specific purpose in the program. It would probably be a good idea to generalize this facility.
$after_semicolon Is the token being processed on the same line as a semicolon? This allows for the semicolon processor to unconditionally add a newline while allowing things like inline comments on the same line to be bubbled up. $after_case Is the token being processed on the same line as a case? This is a specific override for comment movement behavior that places inline comments after a case before the case declaration. $after_comment Is the line being processed preceded by an inline comment? This is used to preserve newlines after comments. $after_initial_comment Is the line being processed preceded by the // $Id (ending dollar sign omitted) comment? This is a workaround to prevent the usual double-newline before docblocks for the very first docblock. $after_visibility_modifier Is the token being processed immediately preceded by a visibility modifier like public/protected/private? This prevents extra newlines added by T_FUNCTION. $after_return_in_case Whether or not the token is after a return statement in a case. This prevents the extra indent after case statements from being terminated prematurely for multiline return lines.
Parameters
$code: The source code to format.
Return value
The formatted code or false if it fails.
See also
http://www.php.net/manual/en/ref.tokenizer.php
1 call to coder_format_string()
- coder_format_string_all in scripts/
coder_format/ coder_format.inc - Formats source code according to Drupal conventions, also using post and pre-processors.
File
- scripts/
coder_format/ coder_format.inc, line 258 - Coder format helper functions.
Code
function coder_format_string($code = '') {
global $_coder_indent;
// Indent controls:
$_coder_indent = 0;
$in_case = FALSE;
$switches = 0;
$parenthesis = 0;
$braces_in_case = array();
$in_brace = FALSE;
$in_heredoc = FALSE;
$first_php_tag = TRUE;
$in_do_while = FALSE;
// Whitespace controls:
$in_object = FALSE;
$in_at = FALSE;
$in_php = FALSE;
$in_quote = FALSE;
$inline_if = FALSE;
$in_array = array();
$in_multiline = array();
// Context flags:
$after_semicolon = FALSE;
$after_case = FALSE;
$after_comment = FALSE;
$after_initial_comment = FALSE;
$after_visibility_modifier = FALSE;
$after_return_in_case = FALSE;
$after_php = FALSE;
// Whether or not a function token was encountered:
$in_function_declaration = FALSE;
// The position of the last character of the last non-whitespace
// non-comment token, e.g. it would be:
// function foo() { // bar
// ^ this character
$position_last_significant_token = 0;
$result = '';
$lasttoken = array(
0,
);
$tokens = token_get_all($code);
// Mask T_ML_COMMENT (PHP4) as T_COMMENT (PHP5).
if (!defined('T_ML_COMMENT')) {
define('T_ML_COMMENT', T_COMMENT);
}
elseif (!defined('T_DOC_COMMENT')) {
define('T_DOC_COMMENT', T_ML_COMMENT);
}
foreach ($tokens as $token) {
if (is_string($token)) {
// Simple 1-character token.
$text = trim($token);
switch ($text) {
case '{':
// Add a space before and behind a curly brace, if we are in inline
// PHP, e.g. <?php if ($foo) { print $foo }
if ($after_php) {
$text = " {$text} ";
}
// Write curly braces at the end of lines followed by a line break if
// not in quotes (""), object ($foo->{$bar}) or in variables (${foo}).
// (T_DOLLAR_OPEN_CURLY_BRACES exists but is never assigned.)
$c = substr(rtrim($result), -1);
if (!$after_php && !$in_quote && (!$in_variable && !$in_object && $c != '$' || $c == ')')) {
if ($in_case) {
++$braces_in_case[$switches];
$_coder_indent += $switches - 1;
}
++$_coder_indent;
$result = rtrim($result) . ' ' . $text;
coder_br($result);
}
else {
$in_brace = TRUE;
$result .= $text;
}
break;
case '}':
if (!$in_quote && !$in_brace && !$in_heredoc) {
if ($switches) {
--$braces_in_case[$switches];
}
--$_coder_indent;
if ($braces_in_case[$switches] < 0 && $in_case) {
// Decrease indent if last case in a switch is not terminated.
--$_coder_indent;
$in_case = FALSE;
}
if ($braces_in_case[$switches] < 0) {
$braces_in_case[$switches] = 0;
$switches--;
}
if ($switches > 0) {
$in_case = TRUE;
}
if (!$after_php) {
$result = rtrim($result);
if (substr($result, -1) != '{') {
// Avoid line break in empty curly braces.
coder_br($result);
}
$result .= $text;
coder_br($result);
}
else {
// Add a space before a curly brace, if we are in inline PHP, e.g.
// <?php if ($foo) { print $foo }
$result = rtrim($result, ' ');
if (substr($result, -1) !== "\n") {
$result .= ' ';
}
$result .= $text;
}
}
else {
$in_brace = FALSE;
$result .= $text;
}
break;
case ';':
$result = rtrim($result) . $text;
// Check if we had deferred reduction of indent because we were in
// a case statement. Now we can decrease the indent.
if ($after_return_in_case) {
--$_coder_indent;
$after_return_in_case = FALSE;
}
if (!$parenthesis && !$in_heredoc && !$after_php) {
coder_br($result);
$after_semicolon = TRUE;
}
else {
$result .= ' ';
}
if ($inline_if) {
$inline_if = FALSE;
}
break;
case '?':
$inline_if = TRUE;
$result .= ' ' . $text . ' ';
break;
case ':':
if ($inline_if) {
$result .= ' ' . $text . ' ';
}
elseif ($after_php) {
$result .= $text;
}
else {
if ($in_case) {
++$_coder_indent;
}
$result = rtrim($result) . $text;
coder_br($result);
}
break;
case '(':
$result .= $text;
++$parenthesis;
// Not multiline until proven so by whitespace.
$in_multiline[$parenthesis] = FALSE;
// If the $in_array flag for this parenthesis level was not
// set previously, set it to FALSE.
if (!isset($in_array[$parenthesis])) {
$in_array[$parenthesis] = FALSE;
}
// Terminate function declaration, as a parenthesis indicates
// the beginning of the arguments. This will catch all other
// instances of parentheses, but in this case it's not a problem.
$in_function_declaration = FALSE;
break;
case ')':
if ($in_array[$parenthesis] && $in_multiline[$parenthesis]) {
// Check if a comma insertion is necessary:
$c = $position_last_significant_token;
if ($result[$c] !== ',') {
// We need to add a comma at $c:
$result = substr($result, 0, $c + 1) . ',' . substr($result, $c + 1);
}
}
if (!$in_quote && !$in_heredoc && (substr(rtrim($result), -1) == ',' || $in_multiline[$parenthesis])) {
// Fix indent of right parenthesis in multiline structures by
// increasing indent for each parenthesis and decreasing one level.
$result = rtrim($result);
coder_br($result, $parenthesis - 1);
$result .= $text;
}
else {
$result .= $text;
}
if ($parenthesis) {
// Current parenthesis level is not an array anymore.
$in_array[$parenthesis] = FALSE;
--$parenthesis;
}
break;
case '@':
$in_at = TRUE;
$result .= $text;
break;
case ',':
$result .= $text . ' ';
break;
case '.':
// Starting from 7.x, string concatenations follow PEAR's standard.
$result = rtrim($result) . ' ' . $text . ' ';
break;
case '=':
case '<':
case '>':
case '+':
case '*':
case '/':
case '|':
case '^':
case '%':
$result = rtrim($result) . ' ' . $text . ' ';
break;
case '&':
if (substr(rtrim($result), -1) == '=' || substr(rtrim($result), -1) == '(' || substr(rtrim($result), -1) == ',') {
$result .= $text;
}
else {
$result = rtrim($result) . ' ' . $text;
// Ampersands used to declare reference return value for
// functions should not have trailing space.
if (!$in_function_declaration) {
$result .= ' ';
}
}
break;
case '-':
$result = rtrim($result);
// Do not add a space before negative numbers or variables.
$c = substr($result, -1);
// Do not add a space between closing parenthesis and negative arithmetic operators.
if ($c == '(') {
$result .= ltrim($text);
}
elseif ($c == '>' || $c == '=' || $c == ',' || $c == ':' || $c == '?') {
$result .= ' ' . $text;
}
else {
$result .= ' ' . $text . ' ';
}
break;
case '"':
// Toggle quote if the char is not escaped.
if (rtrim($result) != "\\") {
$in_quote = $in_quote ? FALSE : TRUE;
}
$result .= $text;
break;
default:
$result .= $text;
break;
}
// All text possibilities are significant:
$position_last_significant_token = strlen(rtrim($result)) - 1;
// Because they are all significant, we cannot possibly be after
// a comment now.
$after_comment = FALSE;
$after_initial_comment = FALSE;
// TODO: Make resetting context flags easier to do.
}
else {
// If we get here, then we have found not a single char, but a token.
// See <http://www.php.net/manual/en/tokens.php> for a reference.
// Fetch token array.
list($id, $text) = $token;
// Debugging:
/*
if ($lasttoken[0] == T_WHITESPACE) {
$result .= token_name($id);
}
*/
switch ($id) {
case T_ARRAY:
// Write array in lowercase.
$result .= strtolower(trim($text));
// Mark the next parenthesis level (we haven't consumed that token
// yet) as an array.
$in_array[$parenthesis + 1] = TRUE;
break;
case T_OPEN_TAG:
case T_OPEN_TAG_WITH_ECHO:
$in_php = TRUE;
// Add a line break between two PHP tags.
if (substr(rtrim($result), -2) == '?>' && !$after_php) {
coder_br($result);
}
$after_php = TRUE;
$nl = substr_count($text, "\n");
$result .= trim($text);
if ($first_php_tag) {
coder_br($result);
$first_php_tag = FALSE;
}
else {
if ($nl) {
coder_br($result, $parenthesis);
}
else {
$result .= ' ';
}
}
break;
case T_CLOSE_TAG:
$in_php = FALSE;
if ($after_php) {
$result = rtrim($result, ' ') . ' ';
$text = ltrim($text, ' ');
}
// Do not alter a closing PHP tag ($text includes trailing white-space)
// at all. Should allow to apply coder_format on phptemplate files.
$result .= $text;
break;
case T_OBJECT_OPERATOR:
$in_object = TRUE;
$result .= trim($text);
break;
case T_CONSTANT_ENCAPSED_STRING:
case T_STRING:
case T_VARIABLE:
// Boolean constants (TRUE, FALSE, NULL) are T_STRINGs, but must be
// written uppercase.
$text = trim($text);
if ($text == 'true' || $text == 'false' || $text == 'null') {
$text = strtoupper($text);
}
// No space after object operator ($foo->bar) and error suppression (@function()).
if ($in_object || $in_at) {
$result = rtrim($result) . $text;
$in_object = FALSE;
$in_at = FALSE;
}
else {
// Insert a space after right parenthesis, but not after type casts.
if (!in_array($lasttoken[0], array(
T_ARRAY_CAST,
T_BOOL_CAST,
T_DOUBLE_CAST,
T_INT_CAST,
T_OBJECT_CAST,
T_STRING_CAST,
T_UNSET_CAST,
))) {
coder_add_space($result);
}
$result .= $text;
}
$in_variable = TRUE;
break;
case T_CONST:
// Constants are written uppercase.
$result = rtrim($result) . strtoupper(trim($text));
break;
case T_ENCAPSED_AND_WHITESPACE:
$result .= $text;
break;
case T_WHITESPACE:
// Avoid duplicate line feeds outside arrays.
$c = $parenthesis || $after_comment ? 0 : 1;
for ($c, $cc = substr_count($text, "\n"); $c < $cc; ++$c) {
// Newlines were added; not after semicolon anymore
coder_br($result, $parenthesis);
}
// If there were newlines present inside a parenthesis,
// turn on multiline mode.
if ($cc && $parenthesis) {
$in_multiline[$parenthesis] = TRUE;
}
// If there were newlines present, move inline comments above.
if ($cc) {
$after_semicolon = FALSE;
$after_case = FALSE;
$after_php = FALSE;
}
$in_variable = FALSE;
break;
case T_SWITCH:
++$switches;
// Purposely fall through.
case T_FOR:
case T_FOREACH:
case T_GLOBAL:
case T_STATIC:
case T_ECHO:
case T_PRINT:
case T_NEW:
case T_REQUIRE:
case T_REQUIRE_ONCE:
case T_INCLUDE:
case T_INCLUDE_ONCE:
case T_VAR:
coder_add_space($result);
// Append a space.
$result .= trim($text) . ' ';
break;
case T_DO:
$result .= trim($text);
$in_do_while = TRUE;
break;
case T_WHILE:
if ($in_do_while && substr(rtrim($result), -1) === '}') {
// Write while after right parenthesis for do {...} while().
$result = rtrim($result) . ' ';
$in_do_while = FALSE;
}
// Append a space.
$result .= trim($text) . ' ';
break;
case T_IF:
// Use "elseif" instead of "else if".
if (substr(rtrim($result), -4) == 'else') {
$result = rtrim($result);
}
coder_add_space($result);
// Append a space.
$result .= trim($text) . ' ';
break;
case T_ELSE:
case T_ELSEIF:
// Write else and elseif to a new line.
$result = rtrim($result);
coder_br($result);
$result .= trim($text) . ' ';
break;
case T_CASE:
case T_DEFAULT:
$braces_in_case[$switches] = 0;
$result = rtrim($result);
$after_case = TRUE;
if (!$in_case) {
$in_case = TRUE;
// Add a line break between cases.
if (substr($result, -1) != '{') {
coder_br($result);
}
}
else {
// Decrease current indent to align multiple cases.
--$_coder_indent;
}
coder_br($result);
$result .= trim($text) . ' ';
break;
case T_BREAK:
// Write break to a new line.
$result = rtrim($result);
coder_br($result);
// Trailing space needed for 'break 3;'.
$result .= trim($text) . ' ';
if ($in_case && !$braces_in_case[$switches]) {
--$_coder_indent;
$in_case = FALSE;
}
break;
case T_RETURN:
if ($in_case && !$braces_in_case[$switches]) {
// Defer reduction of indent for later.
++$_coder_indent;
$after_return_in_case = TRUE;
}
case T_CONTINUE:
coder_add_space($result);
$result .= trim($text) . ' ';
// Decrease indent only if we're not in a control structure inside a case.
if ($in_case && !$braces_in_case[$switches]) {
--$_coder_indent;
$in_case = FALSE;
}
break;
case T_ABSTRACT:
case T_PRIVATE:
case T_PUBLIC:
case T_PROTECTED:
// Class member function properties must be treated similar to
// T_FUNCTION, but without line-break after the token. Because more
// than one of these tokens can appear in front of a function token,
// we need another white-space control variable.
$result .= trim($text) . ' ';
$after_visibility_modifier = TRUE;
break;
case T_FUNCTION:
$in_function_declaration = TRUE;
// Fall through.
case T_CLASS:
// Write function and class to new lines.
$result = rtrim($result);
if (substr($result, -1) == '}') {
coder_br($result);
}
if (!$after_visibility_modifier) {
coder_br($result);
}
else {
// This code only applies to T_FUNCTION; do not add a newline
// after public/protected/private/abstract.
$after_visibility_modifier = FALSE;
$result .= ' ';
}
$result .= trim($text) . ' ';
break;
case T_EXTENDS:
case T_INSTANCEOF:
// Add space before and after 'extends' and 'instanceof'.
$result = rtrim($result);
$result .= ' ' . trim($text) . ' ';
break;
case T_AND_EQUAL:
case T_AS:
case T_BOOLEAN_AND:
case T_BOOLEAN_OR:
case T_CONCAT_EQUAL:
case T_DIV_EQUAL:
case T_DOUBLE_ARROW:
case T_IS_EQUAL:
case T_IS_NOT_EQUAL:
case T_IS_IDENTICAL:
case T_IS_NOT_IDENTICAL:
case T_IS_GREATER_OR_EQUAL:
case T_IS_SMALLER_OR_EQUAL:
case T_LOGICAL_AND:
case T_LOGICAL_OR:
case T_LOGICAL_XOR:
case T_MINUS_EQUAL:
case T_MOD_EQUAL:
case T_MUL_EQUAL:
case T_OR_EQUAL:
case T_PLUS_EQUAL:
case T_SL:
case T_SL_EQUAL:
case T_SR:
case T_SR_EQUAL:
case T_XOR_EQUAL:
// Surround operators with spaces.
if (substr($result, -1) != ' ') {
// $result must not be trimmed to allow multi-line if-clauses.
$result .= ' ';
}
$result .= trim($text) . ' ';
break;
case T_COMMENT:
case T_ML_COMMENT:
case T_DOC_COMMENT:
if (substr($text, 0, 3) == '/**') {
// Prepend a new line.
$result = rtrim($result);
if (!$after_initial_comment) {
coder_br($result);
}
else {
// This probably will get set below, but it's good to
// explicitly turn it off after the initial comment has
// influenced behavior and now is not necessary.
$after_initial_comment = FALSE;
}
coder_br($result);
// Remove carriage returns.
$text = str_replace("\r", '', $text);
$lines = explode("\n", $text);
$params_fixed = FALSE;
for ($l = 0; $l < count($lines); ++$l) {
$lines[$l] = trim($lines[$l]);
// Add a new line between function description and first parameter description.
if (!$params_fixed && substr($lines[$l], 0, 8) == '* @param' && $lines[$l - 1] != '*') {
$result .= ' *';
coder_br($result);
$params_fixed = TRUE;
}
elseif (!$params_fixed && substr($lines[$l], 0, 8) == '* @param') {
// Do nothing if parameter description is properly formatted.
$params_fixed = TRUE;
}
// Add a new line between function params and return.
if (substr($lines[$l], 0, 9) == '* @return' && $lines[$l - 1] != '*') {
$result .= ' *';
coder_br($result);
}
// Add one space indent to get ' *[...]'.
if ($l > 0) {
$result .= ' ';
}
$result .= $lines[$l];
if ($l < count($lines)) {
coder_br($result);
}
}
}
else {
// Move the comment above if it's embedded.
$statement = FALSE;
// Some PHP versions throw a warning about wrong parameter count for
// substr_count().
$cc = substr_count(substr($result, $position_last_significant_token), "\n");
if ((!$cc || $after_semicolon) && !$after_case) {
$nl_position = strrpos(rtrim($result, " \n"), "\n");
$statement = substr($result, $nl_position);
$result = substr($result, 0, $nl_position);
$after_semicolon = FALSE;
coder_br($result, $parenthesis);
}
$result .= trim($text);
coder_br($result, $parenthesis);
if ($statement) {
// Newlines are automatically added, so remove these.
$result = rtrim($result, "\n ");
$result .= rtrim($statement, "\n ");
coder_br($result, $parenthesis);
// Need to update this, as our comment trickery has just
// reshuffled the index.
$position_last_significant_token = strlen(rtrim($result, " \n")) - 1;
}
else {
if (strpos($text, '$' . 'Id$') === FALSE) {
$after_comment = TRUE;
}
else {
// Is the number two so that our bottom code doesn't override
// our flag immediately.
$after_initial_comment = 2;
}
}
}
break;
case T_INLINE_HTML:
$result .= $text;
break;
case T_START_HEREDOC:
$result .= trim($text);
coder_br($result, FALSE, FALSE);
$in_heredoc = TRUE;
break;
case T_END_HEREDOC:
$result .= trim($text);
coder_br($result, FALSE, FALSE);
$in_heredoc = FALSE;
break;
default:
$result .= trim($text);
break;
}
// Store last token.
$lasttoken = $token;
// Excluding comments and whitespace, set the position of the
// last significant token's last character to the length of the
// string minus one.
switch ($id) {
case T_WHITESPACE:
case T_COMMENT:
case T_ML_COMMENT:
case T_DOC_COMMENT:
break;
default:
$position_last_significant_token = strlen(rtrim($result, " \n")) - 1;
break;
}
if ($id !== T_COMMENT && $id !== T_ML_COMMENT) {
$after_comment = FALSE;
}
if ($after_initial_comment && $id !== T_WHITESPACE) {
$after_initial_comment--;
}
}
}
return $result;
}