Correl Roush
4244ed6d8b
git-svn-id: file:///srv/svn/scanner/trunk@1 a0501263-5b7a-4423-a8ba-1edf086583e7
330 lines
10 KiB
PHP
330 lines
10 KiB
PHP
<?php
|
|
define( 'PHPPARSER_FETCH_CLASSES', 1 );
|
|
define( 'PHPPARSER_FETCH_FUNCTIONS', 2 );
|
|
define( 'PHPPARSER_FETCH_CALLS', 4 );
|
|
define( 'PHPPARSER_FETCH_INCLUDES', 8 );
|
|
define( 'PHPPARSER_FETCH_INTERNAL', 16 );
|
|
define( 'PHPPARSER_FETCH_CONSTRUCTS', 32 );
|
|
define( 'PHPPARSER_FETCH_EXPRESSIONS', 64 );
|
|
|
|
define( 'PHPPARSER_FETCH_ALL', 65535 );
|
|
|
|
define( 'PHPPARSER_CLASS_DEF', 1 );
|
|
define( 'PHPPARSER_FUNCTION_DEF', 2 );
|
|
define( 'PHPPARSER_FUNCTION_CALL', 3 );
|
|
define( 'PHPPARSER_INCLUDE', 4 );
|
|
define( 'PHPPARSER_EXPRESSION', 5 );
|
|
define( 'PHPPARSER_LANGUAGE_CONSTRUCT', 6 );
|
|
|
|
|
|
|
|
class PHPParser {
|
|
|
|
var $fetch_mode;
|
|
var $file_name;
|
|
var $internal_functions;
|
|
var $defined_functions;
|
|
var $parsed_objects;
|
|
|
|
function PHPParser( $fetch_mode = PHPPARSER_FETCH_ALL, $functionlist = array() ) {
|
|
$this->reset( $fetch_mode );
|
|
$this->internal_functions = get_defined_functions(); $this->internal_functions = $this->internal_functions['internal'];
|
|
$this->addFunctionDefinitions( $functionlist );
|
|
}
|
|
|
|
function reset( $fetch_mode = null ) {
|
|
if( $fetch_mode > 0 ) {
|
|
$this->fetch_mode = $fetch_mode;
|
|
}
|
|
$this->file_name = '';
|
|
$this->defined_functions = array();
|
|
$this->parsed_objects = array();
|
|
}
|
|
|
|
function addFunctionDefinitions( $functionlist ) {
|
|
if( is_array( $functionlist ) ) {
|
|
array_merge( $this->defined_functions, $functionlist );
|
|
}
|
|
}
|
|
|
|
|
|
function parseFile( $file ) {
|
|
if( !file_exists( $file ) || !is_readable( $file ) ) { return false; }
|
|
$this->file_name = $file;
|
|
$this->parse( file_get_contents( $file ) );
|
|
}
|
|
|
|
function parse( $content ) {
|
|
$tokens = token_get_all( $content );
|
|
|
|
$line = 1;
|
|
$depth = 0;
|
|
$block = 0;
|
|
$block_count = 0;
|
|
|
|
$class = $classname = null;
|
|
$function = $functionname = null;
|
|
$switch = array();
|
|
$expression = '';
|
|
$line_text = '';
|
|
|
|
$internal_functions = get_defined_functions(); $internal_functions = $internal_functions['internal'];
|
|
$local_functions = array();
|
|
$local_classes = array();
|
|
$open_blocks = array( 0 );
|
|
$in_string = false;
|
|
foreach( $tokens as $token ) {
|
|
//echo ( is_string( $token ) ? 'CHAR: ' . $token : token_name( $token[0] ) . ': ' . $token[1] ) . "\n";
|
|
if( !in_array( 0, $open_blocks ) ) {
|
|
echo "PARSER ERROR: LOST ZERO BLOCK AT FILE {$this->file_name} LINE $line\n";
|
|
$open_blocks = array_merge( array(0), $open_blocks );
|
|
}
|
|
$classname = isset( $class['name'] ) ? $class['name'] : '';
|
|
$functionname = isset( $function['name'] ) ? $function['name'] : '';
|
|
if( $token == '"' ) {
|
|
$in_string = !$in_string;
|
|
}
|
|
if( $in_string ) {
|
|
$expression .= is_string( $token ) ? $token : $token[1];
|
|
$count = preg_match_all( '/\r?(\n|\r)/', is_string( $token ) ? $token : $token[1], $m );
|
|
$line += $count;
|
|
continue;
|
|
}
|
|
if( is_string( $token ) ) {
|
|
// Single character token
|
|
$text = $token;
|
|
switch( $token ) {
|
|
case '{':
|
|
$block_count++;
|
|
$block = $block_count;
|
|
array_push( $open_blocks, $block );
|
|
$depth = count( $open_blocks ) - 1;
|
|
break;
|
|
case '}':
|
|
array_pop( $open_blocks );
|
|
$depth = count( $open_blocks ) - 1;
|
|
$block = $depth == 0 ? 0 : $open_blocks[$depth];
|
|
if( !empty( $class ) && $class['block'] == $block ) { $class = $functionname = null; }
|
|
if( !empty( $function ) && $function['block'] == $block ) { $function = $functionname = null; }
|
|
if( in_array( $block, $switch ) ) {
|
|
array_pop( $open_blocks );
|
|
$depth = count( $open_blocks ) - 1;
|
|
$block = $depth == 0 ? 0 : $open_blocks[$depth-1];
|
|
}
|
|
break;
|
|
case '(':
|
|
if(
|
|
(bool)($this->fetch_mode & PHPPARSER_FETCH_CALLS)
|
|
&& !empty( $string )
|
|
&& $last_token != T_FUNCTION
|
|
&& $last_token != T_OBJECT_OPERATOR
|
|
&& $last_token != T_NEW
|
|
//&& ( (bool)($this->fetch_mode & PHPPARSER_FETCH_INTERNAL) && !in_array( $string, $this->internal_functions ) )
|
|
) {
|
|
$this->parsed_objects[] = array(
|
|
'type' => PHPPARSER_FUNCTION_CALL,
|
|
'name' => strtolower( $string ),
|
|
'file' => $this->file_name,
|
|
'line' => $line,
|
|
'block' => $block,
|
|
'depth' => $depth,
|
|
'in_class' => $classname,
|
|
'in_function' => $functionname,
|
|
'open_blocks' => $open_blocks
|
|
);
|
|
}
|
|
break;
|
|
}
|
|
if( !in_array( $token, array( '(', ')' ) ) ) { $last_token = null; }
|
|
if( (bool)($this->fetch_mode & PHPPARSER_FETCH_EXPRESSIONS) && in_array( $token, array( '{', '}', ';', '=', '?', ':' ) ) && strlen( trim( $expression ) ) > 0 ) {
|
|
$this->parsed_objects[] = array(
|
|
'type' => PHPPARSER_EXPRESSION,
|
|
'name' => trim( $expression ),
|
|
'file' => $this->file_name,
|
|
'line' => $line,
|
|
'block' => $block,
|
|
'depth' => $depth,
|
|
'in_class' => $classname,
|
|
'in_function' => $functionname,
|
|
'open_blocks' => $open_blocks
|
|
);
|
|
$expression = '';
|
|
} else { $expression .= $token; }
|
|
} else {
|
|
list($id, $text) = $token;
|
|
switch( $id ) {
|
|
case T_CURLY_OPEN:
|
|
case T_DOLLAR_OPEN_CURLY_BRACES:
|
|
$block_count++;
|
|
$block = $block_count;
|
|
array_push( $open_blocks, $block );
|
|
$depth = count( $open_blocks ) - 1;
|
|
break;
|
|
case T_STRING:
|
|
$string = $text;
|
|
switch( $last_token ) {
|
|
case T_CLASS:
|
|
if( (bool)($this->fetch_mode & PHPPARSER_FETCH_CLASSES) ) {
|
|
$this->parsed_objects[] = array(
|
|
'type' => PHPPARSER_CLASS_DEF,
|
|
'name' => strtolower( $string ),
|
|
'file' => $this->file_name,
|
|
'line' => $line,
|
|
'block' => $block,
|
|
'depth' => $depth,
|
|
'in_class' => $classname,
|
|
'in_function' => $functionname
|
|
);
|
|
}
|
|
$class = array( 'name' => $text, 'block' => $block );
|
|
break;
|
|
case T_FUNCTION:
|
|
if( (bool)($this->fetch_mode & PHPPARSER_FETCH_FUNCTIONS) && empty( $classname ) ) { // Not interested in member function definitions
|
|
$this->parsed_objects[] = array(
|
|
'type' => PHPPARSER_FUNCTION_DEF,
|
|
'name' => strtolower( $string ),
|
|
'file' => $this->file_name,
|
|
'line' => $line,
|
|
'block' => $block,
|
|
'depth' => $depth,
|
|
'in_class' => $classname,
|
|
'in_function' => $functionname
|
|
);
|
|
}
|
|
$function = array( 'name' => $text, 'block' => $block );
|
|
break;
|
|
}
|
|
break;
|
|
case T_CONSTANT_ENCAPSED_STRING:
|
|
switch( $last_token ) {
|
|
case T_INCLUDE:
|
|
case T_INCLUDE_ONCE:
|
|
case T_REQUIRE:
|
|
case T_REQUIRE_ONCE:
|
|
if( (bool)($this->fetch_mode & PHPPARSER_FETCH_INCLUDES) ) {
|
|
$this->parsed_objects[] = array(
|
|
'type' => PHPPARSER_INCLUDE,
|
|
'name' => trim( str_replace( '\'', '', $text ) ),
|
|
'file' => $this->file_name,
|
|
'line' => $line,
|
|
'block' => $block,
|
|
'depth' => $depth,
|
|
'in_class' => $classname,
|
|
'in_function' => $functionname
|
|
);
|
|
}
|
|
break;
|
|
}
|
|
break;
|
|
///*
|
|
case T_SWITCH:
|
|
$block_count++;
|
|
$block = $block_count;
|
|
array_push( $open_blocks, $block );
|
|
$depth = count( $open_blocks ) - 1;
|
|
$switch[$block] = $block;
|
|
break;
|
|
case T_CASE:
|
|
case T_DEFAULT:
|
|
// Each conditional of a switch statement is its own block
|
|
array_pop( $open_blocks );
|
|
$block_count++;
|
|
$block = $block_count;
|
|
array_push( $open_blocks, $block );
|
|
$depth = count( $open_blocks ) - 1;
|
|
break;
|
|
//*/
|
|
case T_EVAL:
|
|
if( (bool)($this->fetch_mode & PHPPARSER_FETCH_INTERNAL) ) {
|
|
$this->parsed_objects[] = array(
|
|
'type' => PHPPARSER_FUNCTION_CALL,
|
|
'name' => $text,
|
|
'file' => $this->file_name,
|
|
'line' => $line,
|
|
'block' => $block,
|
|
'depth' => $depth,
|
|
'in_class' => $classname,
|
|
'in_function' => $functionname
|
|
);
|
|
}
|
|
break;
|
|
}
|
|
if( !in_array( $id, array( T_WHITESPACE, /*T_COMMENT, T_DOC_COMMENT,*/ T_STRING ) ) ) { $string = null; $last_token = $id; }
|
|
if( (bool)($this->fetch_mode & PHPPARSER_FETCH_CONSTRUCTS) && in_array( $id, array(
|
|
T_ARRAY,
|
|
T_ECHO,
|
|
T_EMPTY,
|
|
T_EVAL,
|
|
T_EXIT,
|
|
T_HALT_COMPILER,
|
|
T_INCLUDE,
|
|
T_INCLUDE_ONCE,
|
|
T_ISSET,
|
|
T_LIST,
|
|
T_PRINT,
|
|
T_REQUIRE,
|
|
T_REQUIRE_ONCE,
|
|
T_UNSET
|
|
) ) ) {
|
|
$this->parsed_objects[] = array(
|
|
'type' => PHPPARSER_LANGUAGE_CONSTRUCT,
|
|
'name' => $text,
|
|
'file' => $this->file_name,
|
|
'line' => $line,
|
|
'block' => $block,
|
|
'depth' => $depth,
|
|
'in_class' => $classname,
|
|
'in_function' => $functionname
|
|
);
|
|
}
|
|
$expression .= ( in_array( $id, array(
|
|
/* Values */
|
|
T_STRING,
|
|
T_CHARACTER,
|
|
T_CONSTANT_ENCAPSED_STRING,
|
|
T_ENCAPSED_AND_WHITESPACE,
|
|
T_WHITESPACE,
|
|
T_DNUMBER,
|
|
T_LNUMBER,
|
|
T_NUM_STRING,
|
|
T_VARIABLE,
|
|
//T_ARRAY,
|
|
T_STRING_VARNAME,
|
|
|
|
/* Operators */
|
|
T_BOOLEAN_AND,
|
|
T_BOOLEAN_OR,
|
|
T_DEC,
|
|
T_INC,
|
|
T_IS_EQUAL,
|
|
T_IS_GREATER_OR_EQUAL,
|
|
T_IS_IDENTICAL,
|
|
T_IS_NOT_EQUAL,
|
|
T_IS_NOT_IDENTICAL,
|
|
T_IS_SMALLER_OR_EQUAL,
|
|
T_LOGICAL_AND,
|
|
T_LOGICAL_OR,
|
|
T_LOGICAL_XOR,
|
|
T_OBJECT_OPERATOR,
|
|
T_DOUBLE_COLON,
|
|
T_SL,
|
|
T_SR,
|
|
|
|
/* Casts */
|
|
T_DOUBLE_CAST,
|
|
T_INT_CAST,
|
|
T_OBJECT_CAST,
|
|
T_STRING_CAST,
|
|
T_UNSET_CAST,
|
|
|
|
/* Constructs */
|
|
T_ECHO,
|
|
T_PRINT
|
|
) ) ) ? $text : ' ';
|
|
$count = preg_match_all( '/\r?(\n|\r)/', $text, $m );
|
|
$line += $count;
|
|
}
|
|
}
|
|
}
|
|
}
|
|
?>
|