mirror of
https://github.com/phabelio/PHP-Parser.git
synced 2024-12-02 09:27:58 +01:00
1030 lines
39 KiB
PHP
1030 lines
39 KiB
PHP
<?php
|
|
|
|
namespace PhpParser;
|
|
|
|
use PhpParser\Node\Expr;
|
|
use PhpParser\Node\Scalar;
|
|
use PhpParser\Node\Stmt;
|
|
|
|
abstract class PrettyPrinterAbstract
|
|
{
|
|
const FIXUP_PREC_LEFT = 0; // LHS operand affected by precedence
|
|
const FIXUP_PREC_RIGHT = 1; // RHS operand affected by precedence
|
|
const FIXUP_CALL_LHS = 2; // LHS of call
|
|
const FIXUP_DEREF_LHS = 3; // LHS of dereferencing operation
|
|
const FIXUP_BRACED_NAME = 4; // Name operand that may require bracing
|
|
const FIXUP_VAR_BRACED_NAME = 5; // Name operand that may require ${} bracing
|
|
const FIXUP_ENCAPSED = 6; // Encapsed string part
|
|
|
|
protected $precedenceMap = array(
|
|
// [precedence, associativity]
|
|
// where for precedence -1 is %left, 0 is %nonassoc and 1 is %right
|
|
'Expr_BinaryOp_Pow' => array( 0, 1),
|
|
'Expr_BitwiseNot' => array( 10, 1),
|
|
'Expr_PreInc' => array( 10, 1),
|
|
'Expr_PreDec' => array( 10, 1),
|
|
'Expr_PostInc' => array( 10, -1),
|
|
'Expr_PostDec' => array( 10, -1),
|
|
'Expr_UnaryPlus' => array( 10, 1),
|
|
'Expr_UnaryMinus' => array( 10, 1),
|
|
'Expr_Cast_Int' => array( 10, 1),
|
|
'Expr_Cast_Double' => array( 10, 1),
|
|
'Expr_Cast_String' => array( 10, 1),
|
|
'Expr_Cast_Array' => array( 10, 1),
|
|
'Expr_Cast_Object' => array( 10, 1),
|
|
'Expr_Cast_Bool' => array( 10, 1),
|
|
'Expr_Cast_Unset' => array( 10, 1),
|
|
'Expr_ErrorSuppress' => array( 10, 1),
|
|
'Expr_Instanceof' => array( 20, 0),
|
|
'Expr_BooleanNot' => array( 30, 1),
|
|
'Expr_BinaryOp_Mul' => array( 40, -1),
|
|
'Expr_BinaryOp_Div' => array( 40, -1),
|
|
'Expr_BinaryOp_Mod' => array( 40, -1),
|
|
'Expr_BinaryOp_Plus' => array( 50, -1),
|
|
'Expr_BinaryOp_Minus' => array( 50, -1),
|
|
'Expr_BinaryOp_Concat' => array( 50, -1),
|
|
'Expr_BinaryOp_ShiftLeft' => array( 60, -1),
|
|
'Expr_BinaryOp_ShiftRight' => array( 60, -1),
|
|
'Expr_BinaryOp_Smaller' => array( 70, 0),
|
|
'Expr_BinaryOp_SmallerOrEqual' => array( 70, 0),
|
|
'Expr_BinaryOp_Greater' => array( 70, 0),
|
|
'Expr_BinaryOp_GreaterOrEqual' => array( 70, 0),
|
|
'Expr_BinaryOp_Equal' => array( 80, 0),
|
|
'Expr_BinaryOp_NotEqual' => array( 80, 0),
|
|
'Expr_BinaryOp_Identical' => array( 80, 0),
|
|
'Expr_BinaryOp_NotIdentical' => array( 80, 0),
|
|
'Expr_BinaryOp_Spaceship' => array( 80, 0),
|
|
'Expr_BinaryOp_BitwiseAnd' => array( 90, -1),
|
|
'Expr_BinaryOp_BitwiseXor' => array(100, -1),
|
|
'Expr_BinaryOp_BitwiseOr' => array(110, -1),
|
|
'Expr_BinaryOp_BooleanAnd' => array(120, -1),
|
|
'Expr_BinaryOp_BooleanOr' => array(130, -1),
|
|
'Expr_BinaryOp_Coalesce' => array(140, 1),
|
|
'Expr_Ternary' => array(150, -1),
|
|
// parser uses %left for assignments, but they really behave as %right
|
|
'Expr_Assign' => array(160, 1),
|
|
'Expr_AssignRef' => array(160, 1),
|
|
'Expr_AssignOp_Plus' => array(160, 1),
|
|
'Expr_AssignOp_Minus' => array(160, 1),
|
|
'Expr_AssignOp_Mul' => array(160, 1),
|
|
'Expr_AssignOp_Div' => array(160, 1),
|
|
'Expr_AssignOp_Concat' => array(160, 1),
|
|
'Expr_AssignOp_Mod' => array(160, 1),
|
|
'Expr_AssignOp_BitwiseAnd' => array(160, 1),
|
|
'Expr_AssignOp_BitwiseOr' => array(160, 1),
|
|
'Expr_AssignOp_BitwiseXor' => array(160, 1),
|
|
'Expr_AssignOp_ShiftLeft' => array(160, 1),
|
|
'Expr_AssignOp_ShiftRight' => array(160, 1),
|
|
'Expr_AssignOp_Pow' => array(160, 1),
|
|
'Expr_YieldFrom' => array(165, 1),
|
|
'Expr_Print' => array(168, 1),
|
|
'Expr_BinaryOp_LogicalAnd' => array(170, -1),
|
|
'Expr_BinaryOp_LogicalXor' => array(180, -1),
|
|
'Expr_BinaryOp_LogicalOr' => array(190, -1),
|
|
'Expr_Include' => array(200, -1),
|
|
);
|
|
|
|
/** @var string Token placed before newline to ensure it is not indented. */
|
|
protected $noIndentToken;
|
|
/** @var string Token placed at end of doc string to ensure it is followed by a newline. */
|
|
protected $docStringEndToken;
|
|
/** @var bool Whether semicolon namespaces can be used (i.e. no global namespace is used) */
|
|
protected $canUseSemicolonNamespaces;
|
|
/** @var array Pretty printer options */
|
|
protected $options;
|
|
|
|
/** @var array Original tokens for use in format-preserving pretty print */
|
|
protected $origTokens;
|
|
/** @var int Current indentation level during format-preserving pretty pting */
|
|
protected $indentLevel;
|
|
/** @var bool[] Map determining whether a certain character is a label character */
|
|
protected $labelCharMap;
|
|
/**
|
|
* @var int[][] Map from token types and subnode names to FIXUP_* constants. This is used
|
|
* during format-preserving prints to place additional parens/braces if necessary.
|
|
*/
|
|
protected $fixupMap;
|
|
/**
|
|
* @var int[][] Map from "{$node->getType()}->{$subNode}" to ['left' => $l, 'right' => $r],
|
|
* where $l and $r specify the token type that needs to be stripped when removing
|
|
* this node.
|
|
*/
|
|
protected $removalMap;
|
|
|
|
/**
|
|
* Creates a pretty printer instance using the given options.
|
|
*
|
|
* Supported options:
|
|
* * bool $shortArraySyntax = false: Whether to use [] instead of array() as the default array
|
|
* syntax, if the node does not specify a format.
|
|
*
|
|
* @param array $options Dictionary of formatting options
|
|
*/
|
|
public function __construct(array $options = []) {
|
|
$this->noIndentToken = '_NO_INDENT_' . mt_rand();
|
|
$this->docStringEndToken = '_DOC_STRING_END_' . mt_rand();
|
|
|
|
$defaultOptions = ['shortArraySyntax' => false];
|
|
$this->options = $options + $defaultOptions;
|
|
}
|
|
|
|
/**
|
|
* Pretty prints an array of statements.
|
|
*
|
|
* @param Node[] $stmts Array of statements
|
|
*
|
|
* @return string Pretty printed statements
|
|
*/
|
|
public function prettyPrint(array $stmts) {
|
|
$this->preprocessNodes($stmts);
|
|
|
|
return ltrim($this->handleMagicTokens($this->pStmts($stmts, false)));
|
|
}
|
|
|
|
/**
|
|
* Pretty prints an expression.
|
|
*
|
|
* @param Expr $node Expression node
|
|
*
|
|
* @return string Pretty printed node
|
|
*/
|
|
public function prettyPrintExpr(Expr $node) {
|
|
return $this->handleMagicTokens($this->p($node));
|
|
}
|
|
|
|
/**
|
|
* Pretty prints a file of statements (includes the opening <?php tag if it is required).
|
|
*
|
|
* @param Node[] $stmts Array of statements
|
|
*
|
|
* @return string Pretty printed statements
|
|
*/
|
|
public function prettyPrintFile(array $stmts) {
|
|
if (!$stmts) {
|
|
return "<?php\n\n";
|
|
}
|
|
|
|
$p = "<?php\n\n" . $this->prettyPrint($stmts);
|
|
|
|
if ($stmts[0] instanceof Stmt\InlineHTML) {
|
|
$p = preg_replace('/^<\?php\s+\?>\n?/', '', $p);
|
|
}
|
|
if ($stmts[count($stmts) - 1] instanceof Stmt\InlineHTML) {
|
|
$p = preg_replace('/<\?php$/', '', rtrim($p));
|
|
}
|
|
|
|
return $p;
|
|
}
|
|
|
|
/**
|
|
* Preprocesses the top-level nodes to initialize pretty printer state.
|
|
*
|
|
* @param Node[] $nodes Array of nodes
|
|
*/
|
|
protected function preprocessNodes(array $nodes) {
|
|
/* We can use semicolon-namespaces unless there is a global namespace declaration */
|
|
$this->canUseSemicolonNamespaces = true;
|
|
foreach ($nodes as $node) {
|
|
if ($node instanceof Stmt\Namespace_ && null === $node->name) {
|
|
$this->canUseSemicolonNamespaces = false;
|
|
}
|
|
}
|
|
}
|
|
|
|
/**
|
|
* Handles (and removes) no-indent and doc-string-end tokens.
|
|
*
|
|
* @param string $str
|
|
* @return string
|
|
*/
|
|
protected function handleMagicTokens($str) {
|
|
// Drop no-indent tokens
|
|
$str = str_replace($this->noIndentToken, '', $str);
|
|
|
|
// Replace doc-string-end tokens with nothing or a newline
|
|
$str = str_replace($this->docStringEndToken . ";\n", ";\n", $str);
|
|
$str = str_replace($this->docStringEndToken, "\n", $str);
|
|
|
|
return $str;
|
|
}
|
|
|
|
/**
|
|
* Pretty prints an array of nodes (statements) and indents them optionally.
|
|
*
|
|
* @param Node[] $nodes Array of nodes
|
|
* @param bool $indent Whether to indent the printed nodes
|
|
*
|
|
* @return string Pretty printed statements
|
|
*/
|
|
protected function pStmts(array $nodes, $indent = true) {
|
|
$result = '';
|
|
foreach ($nodes as $node) {
|
|
$comments = $node->getAttribute('comments', array());
|
|
if ($comments) {
|
|
$result .= "\n" . $this->pComments($comments);
|
|
if ($node instanceof Stmt\Nop) {
|
|
continue;
|
|
}
|
|
}
|
|
|
|
$result .= "\n" . $this->p($node);
|
|
}
|
|
|
|
if ($indent) {
|
|
return preg_replace('~\n(?!$|' . $this->noIndentToken . ')~', "\n ", $result);
|
|
} else {
|
|
return $result;
|
|
}
|
|
}
|
|
|
|
protected function pInfixOp($type, Node $leftNode, $operatorString, Node $rightNode) {
|
|
list($precedence, $associativity) = $this->precedenceMap[$type];
|
|
|
|
return $this->pPrec($leftNode, $precedence, $associativity, -1)
|
|
. $operatorString
|
|
. $this->pPrec($rightNode, $precedence, $associativity, 1);
|
|
}
|
|
|
|
protected function pPrefixOp($type, $operatorString, Node $node) {
|
|
list($precedence, $associativity) = $this->precedenceMap[$type];
|
|
return $operatorString . $this->pPrec($node, $precedence, $associativity, 1);
|
|
}
|
|
|
|
protected function pPostfixOp($type, Node $node, $operatorString) {
|
|
list($precedence, $associativity) = $this->precedenceMap[$type];
|
|
return $this->pPrec($node, $precedence, $associativity, -1) . $operatorString;
|
|
}
|
|
|
|
/**
|
|
* Prints an expression node with the least amount of parentheses necessary to preserve the meaning.
|
|
*
|
|
* @param Node $node Node to pretty print
|
|
* @param int $parentPrecedence Precedence of the parent operator
|
|
* @param int $parentAssociativity Associativity of parent operator
|
|
* (-1 is left, 0 is nonassoc, 1 is right)
|
|
* @param int $childPosition Position of the node relative to the operator
|
|
* (-1 is left, 1 is right)
|
|
*
|
|
* @return string The pretty printed node
|
|
*/
|
|
protected function pPrec(Node $node, $parentPrecedence, $parentAssociativity, $childPosition) {
|
|
$type = $node->getType();
|
|
if (isset($this->precedenceMap[$type])) {
|
|
$childPrecedence = $this->precedenceMap[$type][0];
|
|
if ($childPrecedence > $parentPrecedence
|
|
|| ($parentPrecedence == $childPrecedence && $parentAssociativity != $childPosition)
|
|
) {
|
|
return '(' . $this->p($node) . ')';
|
|
}
|
|
}
|
|
|
|
return $this->p($node);
|
|
}
|
|
|
|
/**
|
|
* Pretty prints an array of nodes and implodes the printed values.
|
|
*
|
|
* @param Node[] $nodes Array of Nodes to be printed
|
|
* @param string $glue Character to implode with
|
|
*
|
|
* @return string Imploded pretty printed nodes
|
|
*/
|
|
protected function pImplode(array $nodes, $glue = '') {
|
|
$pNodes = array();
|
|
foreach ($nodes as $node) {
|
|
if (null === $node) {
|
|
$pNodes[] = '';
|
|
} else {
|
|
$pNodes[] = $this->p($node);
|
|
}
|
|
}
|
|
|
|
return implode($glue, $pNodes);
|
|
}
|
|
|
|
/**
|
|
* Pretty prints an array of nodes and implodes the printed values with commas.
|
|
*
|
|
* @param Node[] $nodes Array of Nodes to be printed
|
|
*
|
|
* @return string Comma separated pretty printed nodes
|
|
*/
|
|
protected function pCommaSeparated(array $nodes) {
|
|
return $this->pImplode($nodes, ', ');
|
|
}
|
|
|
|
/**
|
|
* Signals the pretty printer that a string shall not be indented.
|
|
*
|
|
* @param string $string Not to be indented string
|
|
*
|
|
* @return string String marked with $this->noIndentToken's.
|
|
*/
|
|
protected function pNoIndent($string) {
|
|
return str_replace("\n", "\n" . $this->noIndentToken, $string);
|
|
}
|
|
|
|
/**
|
|
* Prints reformatted text of the passed comments.
|
|
*
|
|
* @param Comment[] $comments List of comments
|
|
*
|
|
* @return string Reformatted text of comments
|
|
*/
|
|
protected function pComments(array $comments) {
|
|
$formattedComments = [];
|
|
|
|
foreach ($comments as $comment) {
|
|
$formattedComments[] = $comment->getReformattedText();
|
|
}
|
|
|
|
return implode("\n", $formattedComments);
|
|
}
|
|
|
|
/**
|
|
* Perform a format-preserving pretty print of an AST.
|
|
*
|
|
* The format preservation is best effort. For some changes to the AST the formatting will not
|
|
* be preserved (at least not locally).
|
|
*
|
|
* In order to use this method a number of prerequisites must be satisfied:
|
|
* * The startTokenPos and endTokenPos attributes in the lexer must be enabled.
|
|
* * The parser must be run with the option useIdentifierNodes ENABLED.
|
|
* * The CloningVisitor must be run on the AST prior to modification.
|
|
* * The original tokens must be provided, using the getTokens() method on the lexer.
|
|
*
|
|
* @param Node[] $stmts Modified AST with links to original AST
|
|
* @param Node[] $origStmts Original AST with token offset information
|
|
* @param array $origTokens Tokens of the original code
|
|
*
|
|
* @return string
|
|
*/
|
|
public function printFormatPreserving(array $stmts, array $origStmts, array $origTokens) {
|
|
$this->initializeLabelCharMap();
|
|
$this->initializeFixupMap();
|
|
$this->initializeRemovalMap();
|
|
|
|
$this->origTokens = $origTokens;
|
|
$this->indentLevel = 0;
|
|
|
|
$this->preprocessNodes($stmts);
|
|
|
|
$pos = 0;
|
|
$result = $this->pArray($stmts, $origStmts, $pos, 0, null);
|
|
if (null !== $result) {
|
|
$result .= $this->getTokenCode($pos, count($this->origTokens), 0);
|
|
} else {
|
|
// Fallback
|
|
// TODO Add <?php properly
|
|
$result = "<?php\n" . $this->pStmts($stmts, false);
|
|
}
|
|
|
|
return ltrim($this->handleMagicTokens($result));
|
|
}
|
|
|
|
protected function pFallback(Node $node) {
|
|
return $this->{'p' . $node->getType()}($node);
|
|
}
|
|
|
|
/**
|
|
* Pretty prints a node.
|
|
*
|
|
* This method also handles formatting preservation for nodes.
|
|
*
|
|
* @param Node $node Node to be pretty printed
|
|
*
|
|
* @return string Pretty printed node
|
|
*/
|
|
protected function p(Node $node) {
|
|
// No orig tokens means this is a normal pretty print without preservation of formatting
|
|
if (!$this->origTokens) {
|
|
return $this->{'p' . $node->getType()}($node);
|
|
}
|
|
|
|
/** @var Node $origNode */
|
|
$origNode = $node->getAttribute('origNode');
|
|
if (null === $origNode) {
|
|
return $this->pFallback($node);
|
|
}
|
|
|
|
if (get_class($node) != get_class($origNode)) {
|
|
// Shouldn't happen
|
|
return $this->pFallback($node);
|
|
}
|
|
|
|
$startPos = $origNode->getAttribute('startTokenPos', -1);
|
|
$endPos = $origNode->getAttribute('endTokenPos', -1);
|
|
if ($startPos < 0 || $endPos < 0) {
|
|
// Shouldn't happen
|
|
return $this->pFallback($node);
|
|
}
|
|
|
|
if ($node instanceof Expr\New_ && $node->class instanceof Stmt\Class_) {
|
|
// For anonymous classes the new and class nodes are intermixed, this would require
|
|
// special handling (or maybe a new node type just for this?)
|
|
return $this->pFallback($node);
|
|
}
|
|
|
|
$indentAdjustment = $this->indentLevel - $this->getIndentationBefore($startPos);
|
|
|
|
$type = $node->getType();
|
|
$fixupInfo = isset($this->fixupMap[$type]) ? $this->fixupMap[$type] : null;
|
|
|
|
$result = '';
|
|
$pos = $startPos;
|
|
foreach ($node->getSubNodeNames() as $i => $subNodeName) {
|
|
$subNode = $node->$subNodeName;
|
|
$origSubNode = $origNode->$subNodeName;
|
|
|
|
if (!$origSubNode instanceof Node || (!$subNode instanceof Node && $subNode !== null)) {
|
|
if ($subNode === $origSubNode) {
|
|
// Unchanged, can reuse old code
|
|
continue;
|
|
}
|
|
|
|
if (is_array($subNode) && is_array($origSubNode)) {
|
|
// Array subnode changed, we might be able to reconstruct it
|
|
$fixup = isset($fixupInfo[$subNodeName]) ? $fixupInfo[$subNodeName] : null;
|
|
$listResult = $this->pArray(
|
|
$subNode, $origSubNode, $pos, $indentAdjustment, $fixup
|
|
);
|
|
if (null === $listResult) {
|
|
return $this->pFallback($node);
|
|
}
|
|
|
|
$result .= $listResult;
|
|
continue;
|
|
}
|
|
|
|
// If a non-node, non-array subnode changed, we don't be able to do a partial
|
|
// reconstructions, as we don't have enough offset information. Pretty print the
|
|
// whole node instead.
|
|
return $this->pFallback($node);
|
|
}
|
|
|
|
$subStartPos = $origSubNode->getAttribute('startTokenPos', -1);
|
|
$subEndPos = $origSubNode->getAttribute('endTokenPos', -1);
|
|
if ($subStartPos < 0 || $subEndPos < 0) {
|
|
// Shouldn't happen
|
|
return $this->pFallback($node);
|
|
}
|
|
|
|
if (null === $subNode) {
|
|
// A node has been removed, check if we have removal information for it
|
|
$key = $type . '->' . $subNodeName;
|
|
if (!isset($this->removalMap[$key])) {
|
|
return $this->pFallback($node);
|
|
}
|
|
|
|
// Adjust positions to account for additional tokens that must be skipped
|
|
$removalInfo = $this->removalMap[$key];
|
|
if (isset($removalInfo['left'])) {
|
|
$subStartPos = $this->skipLeft($subStartPos - 1, $removalInfo['left']) + 1;
|
|
}
|
|
if (isset($removalInfo['right'])) {
|
|
$subEndPos = $this->skipRight($subEndPos + 1, $removalInfo['right']) - 1;
|
|
}
|
|
}
|
|
|
|
$result .= $this->getTokenCode($pos, $subStartPos, $indentAdjustment);
|
|
|
|
if (null !== $subNode) {
|
|
$origIndentLevel = $this->indentLevel;
|
|
$this->indentLevel = $this->getIndentationBefore($subStartPos) + $indentAdjustment;
|
|
|
|
// If it's the same node that was previously in this position, it certainly doesn't
|
|
// need fixup. It's important to check this here, because our fixup checks are more
|
|
// conservative than strictly necessary.
|
|
if (isset($fixupInfo[$subNodeName])
|
|
&& $subNode->getAttribute('origNode') !== $origSubNode
|
|
) {
|
|
$fixup = $fixupInfo[$subNodeName];
|
|
$res = $this->pFixup($fixup, $subNode, $type, $subStartPos, $subEndPos);
|
|
} else {
|
|
$res = $this->p($subNode);
|
|
}
|
|
|
|
$this->safeAppend($result, $res);
|
|
$this->indentLevel = $origIndentLevel;
|
|
}
|
|
|
|
$pos = $subEndPos + 1;
|
|
}
|
|
|
|
$result .= $this->getTokenCode($pos, $endPos + 1, $indentAdjustment);
|
|
return $result;
|
|
}
|
|
|
|
/**
|
|
* Perform a format-preserving pretty print of an array
|
|
*
|
|
* @param array $nodes New nodes
|
|
* @param array $origNodes Original nodes
|
|
* @param int $pos Current token position (updated by reference)
|
|
* @param int $indentAdjustment Adjustment for indentation
|
|
* @param null|int $fixup Fixup information for array item nodes
|
|
*
|
|
* @return null|string Result of pretty print or null if cannot preserve formatting
|
|
*/
|
|
protected function pArray(array $nodes, array $origNodes, &$pos, $indentAdjustment, $fixup) {
|
|
$len = count($nodes);
|
|
$origLen = count($origNodes);
|
|
if ($len !== $origLen) {
|
|
return null;
|
|
}
|
|
|
|
$result = '';
|
|
for ($i = 0; $i < $len; $i++) {
|
|
$arrItem = $nodes[$i];
|
|
$origArrItem = $origNodes[$i];
|
|
if ($arrItem === $origArrItem) {
|
|
// Unchanged, can reuse old code
|
|
continue;
|
|
}
|
|
|
|
if (!$arrItem instanceof Node || !$origArrItem instanceof Node) {
|
|
// We can only handle arrays of nodes meaningfully
|
|
return null;
|
|
}
|
|
|
|
$itemStartPos = $origArrItem->getAttribute('startTokenPos', -1);
|
|
$itemEndPos = $origArrItem->getAttribute('endTokenPos', -1);
|
|
if ($itemStartPos < 0 || $itemEndPos < 0) {
|
|
// Shouldn't happen
|
|
return null;
|
|
}
|
|
|
|
if ($itemEndPos < $itemStartPos) {
|
|
// End can be before start for Nop nodes, because offsets refer to non-whitespace
|
|
// locations, which for an "empty" node might result in an inverted order.
|
|
assert($origArrItem instanceof Stmt\Nop);
|
|
continue;
|
|
}
|
|
|
|
$result .= $this->getTokenCode($pos, $itemStartPos, $indentAdjustment);
|
|
|
|
$origIndentLevel = $this->indentLevel;
|
|
$this->indentLevel = $this->getIndentationBefore($itemStartPos) + $indentAdjustment;
|
|
|
|
if (null !== $fixup && $arrItem->getAttribute('origNode') !== $origArrItem) {
|
|
$res = $this->pFixup($fixup, $arrItem, null, $itemStartPos, $itemEndPos);
|
|
} else {
|
|
$res = $this->p($arrItem);
|
|
}
|
|
$this->safeAppend($result, $res);
|
|
|
|
$this->indentLevel = $origIndentLevel;
|
|
$pos = $itemEndPos + 1;
|
|
}
|
|
return $result;
|
|
}
|
|
|
|
/**
|
|
* Print node with fixups.
|
|
*
|
|
* Fixups here refer to the addition of extra parentheses, braces or other characters, that
|
|
* are required to preserve program semantics in a certain context (e.g. to maintain precedence
|
|
* or because only certain expressions are allowed in certain places).
|
|
*
|
|
* @param int $fixup Fixup type
|
|
* @param Node $subNode Subnode to print
|
|
* @param string $parentType Type of parent node
|
|
* @param int $subStartPos Original start pos of subnode
|
|
* @param int $subEndPos Original end pos of subnode
|
|
*
|
|
* @return string Result of fixed-up print of subnode
|
|
*/
|
|
protected function pFixup($fixup, Node $subNode, $parentType, $subStartPos, $subEndPos) {
|
|
switch ($fixup) {
|
|
case self::FIXUP_PREC_LEFT:
|
|
case self::FIXUP_PREC_RIGHT:
|
|
if (!$this->haveParens($subStartPos, $subEndPos)) {
|
|
list($precedence, $associativity) = $this->precedenceMap[$parentType];
|
|
return $this->pPrec($subNode, $precedence, $associativity,
|
|
$fixup === self::FIXUP_PREC_LEFT ? -1 : 1);
|
|
}
|
|
break;
|
|
case self::FIXUP_CALL_LHS:
|
|
if ($this->callLhsRequiresParens($subNode)
|
|
&& !$this->haveParens($subStartPos, $subEndPos)
|
|
) {
|
|
return '(' . $this->p($subNode) . ')';
|
|
}
|
|
break;
|
|
case self::FIXUP_DEREF_LHS:
|
|
if ($this->dereferenceLhsRequiresParens($subNode)
|
|
&& !$this->haveParens($subStartPos, $subEndPos)
|
|
) {
|
|
return '(' . $this->p($subNode) . ')';
|
|
}
|
|
break;
|
|
case self::FIXUP_BRACED_NAME:
|
|
case self::FIXUP_VAR_BRACED_NAME:
|
|
if ($subNode instanceof Expr
|
|
&& !$this->haveBraces($subStartPos, $subEndPos)
|
|
) {
|
|
return ($fixup === self::FIXUP_VAR_BRACED_NAME ? '$' : '')
|
|
. '{' . $this->p($subNode) . '}';
|
|
}
|
|
break;
|
|
case self::FIXUP_ENCAPSED:
|
|
if (!$subNode instanceof Scalar\EncapsedStringPart
|
|
&& !$this->haveBraces($subStartPos, $subEndPos)
|
|
) {
|
|
return '{' . $this->p($subNode) . '}';
|
|
}
|
|
break;
|
|
default:
|
|
throw new \Exception('Cannot happen');
|
|
}
|
|
|
|
// Nothing special to do
|
|
return $this->p($subNode);
|
|
}
|
|
|
|
/**
|
|
* Appends to a string, ensuring whitespace between label characters.
|
|
*
|
|
* Example: "echo" and "$x" result in "echo$x", but "echo" and "x" result in "echo x".
|
|
* Without safeAppend the result would be "echox", which does not preserve semantics.
|
|
*
|
|
* @param string $str
|
|
* @param string $append
|
|
*/
|
|
protected function safeAppend(&$str, $append) {
|
|
// $append must not be empty in this function
|
|
if ($str === "") {
|
|
$str = $append;
|
|
return;
|
|
}
|
|
|
|
if (!$this->labelCharMap[$append[0]]
|
|
|| !$this->labelCharMap[$str[\strlen($str) - 1]]) {
|
|
$str .= $append;
|
|
} else {
|
|
$str .= " " . $append;
|
|
}
|
|
}
|
|
|
|
/**
|
|
* Get indentation before token position
|
|
*
|
|
* @param int $pos Token position
|
|
*
|
|
* @return int Indentation depth (in spaces)
|
|
*/
|
|
protected function getIndentationBefore($pos) {
|
|
$tokens = $this->origTokens;
|
|
$indent = 0;
|
|
$pos--;
|
|
for (; $pos >= 0; $pos--) {
|
|
if ($tokens[$pos][0] !== T_WHITESPACE) {
|
|
$indent = 0;
|
|
continue;
|
|
}
|
|
$content = $tokens[$pos][1];
|
|
$newlinePos = \strrpos($content, "\n");
|
|
if (false !== $newlinePos) {
|
|
$indent += \strlen($content) - $newlinePos - 1;
|
|
return $indent;
|
|
}
|
|
|
|
$indent += \strlen($content);
|
|
}
|
|
return $indent;
|
|
}
|
|
|
|
protected function haveParens($startPos, $endPos) {
|
|
return $this->haveTokenImmediativelyBefore($startPos, '(')
|
|
&& $this->haveTokenImmediatelyAfter($endPos, ')');
|
|
}
|
|
|
|
protected function haveBraces($startPos, $endPos) {
|
|
return $this->haveTokenImmediativelyBefore($startPos, '{')
|
|
&& $this->haveTokenImmediatelyAfter($endPos, '}');
|
|
}
|
|
|
|
/**
|
|
* Check whether the position is directly preceded by a certain token type.
|
|
*
|
|
* During this check whitespace and comments are skipped.
|
|
*
|
|
* @param int $pos Position before which the token should occur
|
|
* @param int|string $expectedTokenType Token to check for
|
|
*
|
|
* @return bool Whether the expected token was found
|
|
*/
|
|
protected function haveTokenImmediativelyBefore($pos, $expectedTokenType) {
|
|
$tokens = $this->origTokens;
|
|
$pos--;
|
|
for (; $pos >= 0; $pos--) {
|
|
$tokenType = $tokens[$pos][0];
|
|
if ($tokenType === $expectedTokenType) {
|
|
return true;
|
|
}
|
|
if ($tokenType !== T_WHITESPACE
|
|
&& $tokenType !== T_COMMENT && $tokenType !== T_DOC_COMMENT) {
|
|
break;
|
|
}
|
|
}
|
|
return false;
|
|
}
|
|
|
|
/**
|
|
* Check whether the position is directly followed by a certain token type.
|
|
*
|
|
* During this check whitespace and comments are skipped.
|
|
*
|
|
* @param int $pos Position after which the token should occur
|
|
* @param int|string $expectedTokenType Token to check for
|
|
*
|
|
* @return bool Whether the expected token was found
|
|
*/
|
|
protected function haveTokenImmediatelyAfter($pos, $expectedTokenType) {
|
|
$tokens = $this->origTokens;
|
|
$pos++;
|
|
for (; $pos < \count($tokens); $pos++) {
|
|
$tokenType = $tokens[$pos][0];
|
|
if ($tokenType === $expectedTokenType) {
|
|
return true;
|
|
}
|
|
if ($tokenType !== T_WHITESPACE
|
|
&& $tokenType !== T_COMMENT && $tokenType !== T_DOC_COMMENT) {
|
|
break;
|
|
}
|
|
}
|
|
return false;
|
|
}
|
|
|
|
/**
|
|
* Get the code corresponding to a token offset range, optionally adjusted for indentation.
|
|
*
|
|
* @param int $from Token start position (inclusive)
|
|
* @param int $to Token end position (exclusive)
|
|
* @param int $indent By how much the code should be indented (can be negative as well)
|
|
*
|
|
* @return string Code corresponding to token range, adjusted for indentation
|
|
*/
|
|
protected function getTokenCode($from, $to, $indent) {
|
|
$tokens = $this->origTokens;
|
|
$result = '';
|
|
for ($pos = $from; $pos < $to; $pos++) {
|
|
$token = $tokens[$pos];
|
|
if (\is_array($token)) {
|
|
$type = $token[0];
|
|
$content = $token[1];
|
|
if ($type === T_CONSTANT_ENCAPSED_STRING || $type === T_ENCAPSED_AND_WHITESPACE) {
|
|
$result .= $this->pNoIndent($content);
|
|
} else {
|
|
// TODO Handle non-space indentation
|
|
if ($indent < 0) {
|
|
$result .= str_replace("\n" . str_repeat(" ", -$indent), "\n", $content);
|
|
} else if ($indent > 0) {
|
|
$result .= str_replace("\n", "\n" . str_repeat(" ", $indent), $content);
|
|
} else {
|
|
$result .= $content;
|
|
}
|
|
}
|
|
} else {
|
|
$result .= $token;
|
|
}
|
|
}
|
|
return $result;
|
|
}
|
|
|
|
protected function skipLeft($pos, $skipTokenType) {
|
|
$tokens = $this->origTokens;
|
|
|
|
$pos = $this->skipLeftWhitespace($pos);
|
|
if ($skipTokenType === T_WHITESPACE) {
|
|
return $pos;
|
|
}
|
|
|
|
if ($tokens[$pos][0] !== $skipTokenType) {
|
|
// Shouldn't happen. The skip token MUST be there
|
|
throw new \Exception('Encountered unexpected token');
|
|
}
|
|
$pos--;
|
|
|
|
return $this->skipLeftWhitespace($pos);
|
|
}
|
|
|
|
protected function skipRight($pos, $skipTokenType) {
|
|
$tokens = $this->origTokens;
|
|
|
|
$pos = $this->skipRightWhitespace($pos);
|
|
if ($skipTokenType === T_WHITESPACE) {
|
|
return $pos;
|
|
}
|
|
|
|
if ($tokens[$pos][0] !== $skipTokenType) {
|
|
// Shouldn't happen. The skip token MUST be there
|
|
throw new \Exception('Encountered unexpected token');
|
|
}
|
|
$pos++;
|
|
|
|
return $this->skipRightWhitespace($pos);
|
|
}
|
|
|
|
protected function skipLeftWhitespace($pos) {
|
|
$tokens = $this->origTokens;
|
|
for (; $pos >= 0; $pos--) {
|
|
$type = $tokens[$pos][0];
|
|
if ($type !== T_WHITESPACE && $type !== T_COMMENT && $type !== T_DOC_COMMENT) {
|
|
break;
|
|
}
|
|
}
|
|
return $pos;
|
|
}
|
|
|
|
protected function skipRightWhitespace($pos) {
|
|
$tokens = $this->origTokens;
|
|
for ($count = \count($tokens); $pos < $count; $pos++) {
|
|
$type = $tokens[$pos][0];
|
|
if ($type !== T_WHITESPACE && $type !== T_COMMENT && $type !== T_DOC_COMMENT) {
|
|
break;
|
|
}
|
|
}
|
|
return $pos;
|
|
}
|
|
|
|
/**
|
|
* Determines whether the LHS of a call must be wrapped in parenthesis.
|
|
*
|
|
* @param Node $node LHS of a call
|
|
*
|
|
* @return bool Whether parentheses are required
|
|
*/
|
|
protected function callLhsRequiresParens(Node $node) {
|
|
return !($node instanceof Node\Name
|
|
|| $node instanceof Expr\Variable
|
|
|| $node instanceof Expr\ArrayDimFetch
|
|
|| $node instanceof Expr\FuncCall
|
|
|| $node instanceof Expr\MethodCall
|
|
|| $node instanceof Expr\StaticCall
|
|
|| $node instanceof Expr\Array_);
|
|
}
|
|
|
|
/**
|
|
* Determines whether the LHS of a dereferencing operation must be wrapped in parenthesis.
|
|
*
|
|
* @param Node $node LHS of dereferencing operation
|
|
*
|
|
* @return bool Whether parentheses are required
|
|
*/
|
|
protected function dereferenceLhsRequiresParens(Node $node) {
|
|
return !($node instanceof Expr\Variable
|
|
|| $node instanceof Node\Name
|
|
|| $node instanceof Expr\ArrayDimFetch
|
|
|| $node instanceof Expr\PropertyFetch
|
|
|| $node instanceof Expr\StaticPropertyFetch
|
|
|| $node instanceof Expr\FuncCall
|
|
|| $node instanceof Expr\MethodCall
|
|
|| $node instanceof Expr\StaticCall
|
|
|| $node instanceof Expr\Array_
|
|
|| $node instanceof Scalar\String_
|
|
|| $node instanceof Expr\ConstFetch
|
|
|| $node instanceof Expr\ClassConstFetch);
|
|
}
|
|
|
|
/**
|
|
* Lazily initializes label char map.
|
|
*
|
|
* The label char map determines whether a certain character may occur in a label.
|
|
*/
|
|
protected function initializeLabelCharMap() {
|
|
if ($this->labelCharMap) return;
|
|
|
|
$this->labelCharMap = [];
|
|
for ($i = 0; $i < 256; $i++) {
|
|
// Since PHP 7.1 The lower range is 0x80. However, we also want to support code for
|
|
// older versions.
|
|
$this->labelCharMap[chr($i)] = $i >= 0x7f || ctype_alnum($i);
|
|
}
|
|
}
|
|
|
|
/**
|
|
* Lazily initializes fixup map.
|
|
*
|
|
* The fixup map is used to determine whether a certain subnode of a certain node may require
|
|
* some kind of "fixup" operation, e.g. the addition of parenthesis or braces.
|
|
*/
|
|
protected function initializeFixupMap() {
|
|
if ($this->fixupMap) return;
|
|
|
|
$this->fixupMap = [
|
|
'Expr_PreInc' => ['var' => self::FIXUP_PREC_RIGHT],
|
|
'Expr_PreDec' => ['var' => self::FIXUP_PREC_RIGHT],
|
|
'Expr_PostInc' => ['var' => self::FIXUP_PREC_LEFT],
|
|
'Expr_PostDec' => ['var' => self::FIXUP_PREC_LEFT],
|
|
'Expr_Instanceof' => [
|
|
'expr' => self::FIXUP_PREC_LEFT,
|
|
'class' => self::FIXUP_PREC_RIGHT,
|
|
],
|
|
'Expr_Ternary' => [
|
|
'cond' => self::FIXUP_PREC_LEFT,
|
|
'else' => self::FIXUP_PREC_RIGHT,
|
|
],
|
|
|
|
'Expr_FuncCall' => ['name' => self::FIXUP_CALL_LHS],
|
|
'Expr_StaticCall' => ['class' => self::FIXUP_DEREF_LHS],
|
|
'Expr_ArrayDimFetch' => ['var' => self::FIXUP_DEREF_LHS],
|
|
'Expr_MethodCall' => [
|
|
'var' => self::FIXUP_DEREF_LHS,
|
|
'name' => self::FIXUP_BRACED_NAME,
|
|
],
|
|
'Expr_StaticPropertyFetch' => [
|
|
'class' => self::FIXUP_DEREF_LHS,
|
|
'name' => self::FIXUP_VAR_BRACED_NAME,
|
|
],
|
|
'Expr_PropertyFetch' => [
|
|
'var' => self::FIXUP_DEREF_LHS,
|
|
'name' => self::FIXUP_BRACED_NAME,
|
|
],
|
|
'Scalar_Encapsed' => [
|
|
'parts' => self::FIXUP_ENCAPSED,
|
|
],
|
|
];
|
|
|
|
$binaryOps = [
|
|
'Expr_BinaryOp_Pow', 'Expr_BinaryOp_Mul', 'Expr_BinaryOp_Div', 'Expr_BinaryOp_Mod',
|
|
'Expr_BinaryOp_Plus', 'Expr_BinaryOp_Minus', 'Expr_BinaryOp_Concat',
|
|
'Expr_BinaryOp_ShiftLeft', 'Expr_BinaryOp_ShiftRight', 'Expr_BinaryOp_Smaller',
|
|
'Expr_BinaryOp_SmallerOrEqual', 'Expr_BinaryOp_Greater', 'Expr_BinaryOp_GreaterOrEqual',
|
|
'Expr_BinaryOp_Equal', 'Expr_BinaryOp_NotEqual', 'Expr_BinaryOp_Identical',
|
|
'Expr_BinaryOp_NotIdentical', 'Expr_BinaryOp_Spaceship', 'Expr_BinaryOp_BitwiseAnd',
|
|
'Expr_BinaryOp_BitwiseXor', 'Expr_BinaryOp_BitwiseOr', 'Expr_BinaryOp_BooleanAnd',
|
|
'Expr_BinaryOp_BooleanOr', 'Expr_BinaryOp_Coalesce', 'Expr_BinaryOp_LogicalAnd',
|
|
'Expr_BinaryOp_LogicalXor', 'Expr_BinaryOp_LogicalOr',
|
|
];
|
|
foreach ($binaryOps as $binaryOp) {
|
|
$this->fixupMap[$binaryOp] = [
|
|
'left' => self::FIXUP_PREC_LEFT,
|
|
'right' => self::FIXUP_PREC_RIGHT
|
|
];
|
|
}
|
|
|
|
$assignOps = [
|
|
'Expr_Assign', 'Expr_AssignRef', 'Expr_AssignOp_Plus', 'Expr_AssignOp_Minus',
|
|
'Expr_AssignOp_Mul', 'Expr_AssignOp_Div', 'Expr_AssignOp_Concat', 'Expr_AssignOp_Mod',
|
|
'Expr_AssignOp_BitwiseAnd', 'Expr_AssignOp_BitwiseOr', 'Expr_AssignOp_BitwiseXor',
|
|
'Expr_AssignOp_ShiftLeft', 'Expr_AssignOp_ShiftRight', 'Expr_AssignOp_Pow',
|
|
];
|
|
foreach ($assignOps as $assignOp) {
|
|
$this->fixupMap[$assignOp] = [
|
|
'var' => self::FIXUP_PREC_LEFT,
|
|
'expr' => self::FIXUP_PREC_RIGHT,
|
|
];
|
|
}
|
|
|
|
$prefixOps = [
|
|
'Expr_BitwiseNot', 'Expr_BooleanNot', 'Expr_UnaryPlus', 'Expr_UnaryMinus',
|
|
'Expr_Cast_Int', 'Expr_Cast_Double', 'Expr_Cast_String', 'Expr_Cast_Array',
|
|
'Expr_Cast_Object', 'Expr_Cast_Bool', 'Expr_Cast_Unset', 'Expr_ErrorSuppress',
|
|
'Expr_YieldFrom', 'Expr_Print', 'Expr_Include',
|
|
];
|
|
foreach ($prefixOps as $prefixOp) {
|
|
$this->fixupMap[$prefixOp] = ['expr' => self::FIXUP_PREC_RIGHT];
|
|
}
|
|
}
|
|
|
|
protected function initializeRemovalMap() {
|
|
if ($this->removalMap) return;
|
|
|
|
$stripBoth = ['left' => T_WHITESPACE, 'right' => T_WHITESPACE];
|
|
$stripLeft = ['left' => T_WHITESPACE];
|
|
$stripRight = ['right' => T_WHITESPACE];
|
|
$stripDoubleArrow = ['right' => T_DOUBLE_ARROW];
|
|
$stripColon = ['left' => ':'];
|
|
$stripEquals = ['left' => '='];
|
|
$this->removalMap = [
|
|
'Expr_ArrayDimFetch->dim' => $stripBoth,
|
|
'Expr_ArrayItem->key' => $stripDoubleArrow,
|
|
'Expr_Closure->returnType' => $stripColon,
|
|
'Expr_Exit->expr' => $stripBoth,
|
|
'Expr_Ternary->if' => $stripBoth,
|
|
'Expr_Yield->key' => $stripDoubleArrow,
|
|
'Expr_Yield->value' => $stripBoth,
|
|
'Param->type' => $stripRight,
|
|
'Param->default' => $stripEquals,
|
|
'Stmt_Break->num' => $stripBoth,
|
|
'Stmt_ClassMethod->returnType' => $stripColon,
|
|
'Stmt_Class->extends' => ['left' => T_EXTENDS],
|
|
'Stmt_Continue->num' => $stripBoth,
|
|
'Stmt_Foreach->keyVar' => $stripDoubleArrow,
|
|
'Stmt_Function->returnType' => $stripColon,
|
|
'Stmt_If->else' => $stripLeft,
|
|
'Stmt_Namespace->name' => $stripLeft,
|
|
'Stmt_PropertyProperty->default' => $stripEquals,
|
|
'Stmt_Return->expr' => $stripBoth,
|
|
'Stmt_StaticVar->default' => $stripEquals,
|
|
'Stmt_TraitUseAdaptation_Alias->newName' => $stripLeft,
|
|
'Stmt_TryCatch->finally' => $stripLeft,
|
|
// 'Stmt_Case->cond': Replace with "default"
|
|
// 'Stmt_Class->name': Unclear what to do
|
|
// 'Stmt_Declare->stmts': Not a plain node
|
|
// 'Stmt_TraitUseAdaptation_Alias->newModifier': Not a plain node
|
|
];
|
|
}
|
|
}
|