| Index: trunk/tools/code-utils/find-long-functions.php |
| — | — | @@ -0,0 +1,186 @@ |
| | 2 | +<?php |
| | 3 | +/** |
| | 4 | + * T_WHITESPACE \t \r\n |
| | 5 | + * T_FUNCTION |
| | 6 | + * T_ENDDECLARE |
| | 7 | + * T_DECLARE |
| | 8 | + * 0 : token index |
| | 9 | + * 1 : content |
| | 10 | + * 2 : line number |
| | 11 | + * |
| | 12 | + * @param $token_array An array of tokens as returned by PHP token_get_all() |
| | 13 | + */ |
| | 14 | +function _analyze_tokens( $token_array ) { |
| | 15 | + $state = array( |
| | 16 | + 'function' => '', # function name |
| | 17 | + 'fdepth' => null, # function depth in the brace hierarchy |
| | 18 | + 'fstart_line' => 0, # function declaration start line |
| | 19 | + |
| | 20 | + 'class' => '', # class name |
| | 21 | + 'cdepth' => null, # class depth in the brace hierarchy |
| | 22 | + 'cstart_line' => 0, # class declaration start line |
| | 23 | + |
| | 24 | + 'cur_line' => 0, # current line analyzed |
| | 25 | + 'line' => 0, # line number |
| | 26 | + 'depth' => 0, # braces depth |
| | 27 | + ); |
| | 28 | + |
| | 29 | + $tokens = new TokenIterator( $token_array ); |
| | 30 | + $tokens->rewind(); |
| | 31 | + |
| | 32 | + // loop while we get tokens from the iterator |
| | 33 | + while( $tokens->valid() ) { |
| | 34 | + $token = $tokens->current(); |
| | 35 | + |
| | 36 | + # handles braces returned by token_get_all() as strings |
| | 37 | + if( is_string( $token ) ) { |
| | 38 | + if( $token == '{' ) { |
| | 39 | + $state['depth']++; |
| | 40 | + } elseif( $token == '}' ) { |
| | 41 | + $state['depth']--; |
| | 42 | + if( $state['depth'] === $state['cdepth'] ) { |
| | 43 | + $l = $state['cur_line'] - $state['cstart_line']; |
| | 44 | + print "Done with class {$state['class']}, $l lines long.\n\n"; |
| | 45 | + $state['class'] = ''; |
| | 46 | + $state['cdepth'] = null; |
| | 47 | + $state['cstart_line'] = null; |
| | 48 | + } elseif( $state['depth'] === $state['fdepth'] ) { |
| | 49 | + $l = $state['cur_line'] - $state['fstart_line']; |
| | 50 | + printf( "%5s lines for %s.\n", $l, $state['function'] ); |
| | 51 | + $state['function'] = ''; |
| | 52 | + $state['fdepth'] = null; |
| | 53 | + $state['fstart_line'] = null; |
| | 54 | + } |
| | 55 | + } else { print "Got unwanted string: $token\n"; } |
| | 56 | + |
| | 57 | + debug_state( $state, $token ); |
| | 58 | + |
| | 59 | + $tokens->next(); |
| | 60 | + continue; |
| | 61 | + } |
| | 62 | + |
| | 63 | + $state['cur_line'] = $token[2]; |
| | 64 | + |
| | 65 | + # handles CLASS and FUNCTION tokens |
| | 66 | + switch( $token[0] ) { |
| | 67 | + case T_CURLY_OPEN: |
| | 68 | + $state['depth']++; |
| | 69 | + break; |
| | 70 | + case T_CLASS: |
| | 71 | + case T_FUNCTION: |
| | 72 | + # find the token giving function or class name |
| | 73 | + $name_token = $tokens->nextOfKind( T_STRING ); |
| | 74 | + |
| | 75 | + if( $token[0] == T_CLASS ) { |
| | 76 | + $TYPE = 'class'; |
| | 77 | + $state['cdepth'] = $state['depth']; |
| | 78 | + $state['cstart_line'] = $state['cur_line']; |
| | 79 | + $state['class'] = $name_token[1]; |
| | 80 | + print "Analyzing class {$state['class']}\n"; |
| | 81 | + } else { |
| | 82 | + $TYPE = 'function'; |
| | 83 | + $state['fdepth'] = $state['depth']; |
| | 84 | + $state['fstart_line'] = $state['cur_line']; |
| | 85 | + if( $state['class'] ) { |
| | 86 | + $state['function'] = $state['class'] . '::' . $name_token[1]; |
| | 87 | + } else { |
| | 88 | + $state['function'] = $name_token[1]; |
| | 89 | + } |
| | 90 | + } |
| | 91 | + |
| | 92 | + default: |
| | 93 | + debug_state( $state, $token ); |
| | 94 | + } |
| | 95 | + |
| | 96 | + $tokens->next(); |
| | 97 | + } |
| | 98 | +} |
| | 99 | + |
| | 100 | +/** |
| | 101 | + * Debugging function used to output a token position in the brace hierarchy |
| | 102 | + */ |
| | 103 | +function debug_state( $state, $token = null ) { |
| | 104 | + return; |
| | 105 | + printf( "cdepth: %s, fdepth: %s, depth: %s | ", |
| | 106 | + $state['cdepth'], |
| | 107 | + $state['fdepth'], |
| | 108 | + $state['depth'] |
| | 109 | + ); |
| | 110 | + if( is_array( $token ) ) { |
| | 111 | + printf( "line %s - %s: %s\n", $token[2], token_name( $token[0] ), $token[1] ); |
| | 112 | + } else { |
| | 113 | + print "string $token\n"; |
| | 114 | + } |
| | 115 | +} |
| | 116 | + |
| | 117 | +/** |
| | 118 | + * A basic iterator extending the PHP ArrayIterator class. |
| | 119 | + */ |
| | 120 | +class TokenIterator extends ArrayIterator { |
| | 121 | + function __construct( $array ) { |
| | 122 | + parent::__construct( $array ); |
| | 123 | + } |
| | 124 | + |
| | 125 | + /** |
| | 126 | + * Skip tokens until we reach the wanted token. Return it. |
| | 127 | + */ |
| | 128 | + function nextOfKind( $wanted_token_index ) { |
| | 129 | + $found = false; |
| | 130 | + while( true ) { |
| | 131 | + $this->next(); |
| | 132 | + $token = $this->current(); |
| | 133 | + if( $token[0] == $wanted_token_index ) { |
| | 134 | + return $token; |
| | 135 | + } |
| | 136 | + } |
| | 137 | + } |
| | 138 | +} |
| | 139 | + |
| | 140 | +// We just keep blocks |
| | 141 | +$wanted_tokens = array( T_FUNCTION, T_STRING, T_CLASS, T_IF, T_WHILE, T_SWITCH, T_CURLY_OPEN ); |
| | 142 | +$wanted_strings = array( '{', '}' ); |
| | 143 | +$unwanted_tokens = array( T_WHITESPACE ); |
| | 144 | + |
| | 145 | +/** |
| | 146 | + * Callback used to get ride of unneeded tokens. |
| | 147 | + * @param mixed A PHP Token |
| | 148 | + * @return Boolean Whether to keep the token |
| | 149 | + */ |
| | 150 | +function _filter( $token ) { |
| | 151 | + global $wanted_tokens, $wanted_strings, $unwanted_tokens ; |
| | 152 | + if( false && is_array( $token ) ) { |
| | 153 | + return in_array( $token[0] , $wanted_tokens ); |
| | 154 | + } |
| | 155 | + if( is_array( $token ) ) { |
| | 156 | + return !in_array( $token[0] , $unwanted_tokens ); |
| | 157 | + } |
| | 158 | + if( is_string( $token ) ) { |
| | 159 | + return in_array( $token, $wanted_strings ); |
| | 160 | + } |
| | 161 | + return true; |
| | 162 | +} |
| | 163 | + |
| | 164 | +function analyze_file( $file ) { |
| | 165 | + $content = file_get_contents( $file ); |
| | 166 | + if( $content === false ) { |
| | 167 | + print "Could not open file $file\n"; |
| | 168 | + return null; |
| | 169 | + } |
| | 170 | + |
| | 171 | + $tokens = token_get_all( $content ); |
| | 172 | + $tokens = array_filter( $tokens, '_filter' ); |
| | 173 | + _analyze_tokens( $tokens ); |
| | 174 | +} |
| | 175 | + |
| | 176 | + |
| | 177 | +# Print usage when no source file given |
| | 178 | +if( $argc == 1 ) { |
| | 179 | + die( "Usage: $argv[0] <PHP_source_file>\n" ); |
| | 180 | +} |
| | 181 | +array_shift( $argv ); // skip script name |
| | 182 | + |
| | 183 | +# Parse each file given as an argument, one after the other... |
| | 184 | +foreach( $argv as $arg ) { |
| | 185 | + print "Trying file $arg...\n"; |
| | 186 | + analyze_file( $arg ); |
| | 187 | +} |