r2108 + introduce $any_op autotag
[racktables] / inc / code.php
index 0cf0b0213188913135a5dcc7adaf0093b549c4a6..867bc66eb23f4dcb8f5a30bfd0631dfa839682bb 100644 (file)
  *
  */
 
-// Evaluation function table. The functions below process a lexical/syntax node
-// of given type and return boolean value. SYNT_EXPRESSION is missing from this
-// list, because the places, where eval_expression() should be called, are all
-// defined. Each function is passed three arguments: the evaluated item; list
-// of tags, against which we evaluate the whole code; predicates table.
-// The latter changes as the top-level processor meets new definitions on its
-// way down the code list.
-$evalfunc = array
-(
-       'LEX_BOOLCONST' => 'eval_boolconst',
-       'SYNT_NOTEXPR' => 'eval_notexpr',
-       'SYNT_BOOLOP' => 'eval_boolop',
-       'LEX_TAG' => 'eval_tag',
-       'LEX_PREDICATE' => 'eval_predicate',
-);
-
 // Complain about martian char.
-function abortLex1 ($state, $text, $pos)
+function lexError1 ($state, $text, $pos)
 {
-       echo "Error! Could not parse the following text (current state is '${state}'): ";
-       echo substr ($text, 0, $pos);
-       echo '<font color = red>-&gt;' . $text{$pos} . '&lt;-</font>';
-       echo substr ($text, $pos + 1);
-       die;
+       $message = "invalid char with code " . ord (mb_substr ($text, $pos, 1)) . ' (';
+       $message .= mb_substr ($text, $pos, 1) . ") at position ${pos} (FSM state is '${state}')";
+       return array ('result' => 'NAK', 'load' => $message);
 }
 
 // Complain about martian keyword.
-function abortLex2 ($state, $word)
+function lexError2 ($word)
 {
-       echo "Error! Could not parse word (current state is '${state}'): '${word}'.";
-       die;
+       return array
+       (
+               'result' => 'NAK',
+               'load' => "could not parse keyword '${word}'"
+       );
 }
 
 // Complain about wrong FSM state.
-function abortLex3 ($state)
+function lexError3 ($state)
 {
-       echo "Error! Lexical scanner final state is still '${state}' after scanning the last char.";
-       die;
+       return array
+       (
+               'result' => 'NAK',
+               'load' => "FSM state is still '${state}' at end of input"
+       );
 }
 
-function abortSynt ($lexname)
+function lexError4 ($s)
 {
-       echo "Error! Unknown lexeme '${lexname}'.";
-       die;
+       return array
+       (
+               'result' => 'NAK',
+               'load' => "Invalid tag or predicate name '${s}'"
+       );
 }
 
 // Produce a list of lexems from the given text. Possible lexems are:
 function getLexemsFromRackCode ($text)
 {
        $ret = array();
-       $textlen = strlen ($text);
+       $textlen = mb_strlen ($text);
        $state = "ESOTSM";
        for ($i = 0; $i < $textlen; $i++) :
-               $char = $text{$i};
+               $char = mb_substr ($text, $i, 1);
                $newstate = $state;
                switch ($state) :
                        case 'ESOTSM':
@@ -85,11 +76,11 @@ function getLexemsFromRackCode ($text)
                                        case ($char == '#'):
                                                $newstate = 'skipping comment';
                                                break;
-                                       case (preg_match ('/^[a-zA-Z]$/', $char)):
-                                               $newstate = 'reading word';
+                                       case (mb_ereg ('[[:alpha:]]', $char) > 0):
+                                               $newstate = 'reading keyword';
                                                $buffer = $char;
                                                break;
-                                       case (preg_match ('/^[ \t\n]$/', $char)):
+                                       case (preg_match ('/^[ \t\n\r]$/', $char)):
                                                // nom-nom...
                                                break;
                                        case ($char == '{'):
@@ -99,13 +90,13 @@ function getLexemsFromRackCode ($text)
                                                $newstate = 'reading predicate 1';
                                                break;
                                        default:
-                                               abortLex1 ($state, $text, $i);
+                                               return lexError1 ($state, $text, $i);
                                }
                                break;
-                       case 'reading word':
+                       case 'reading keyword':
                                switch (TRUE)
                                {
-                                       case (preg_match ('/^[a-zA-Z]$/', $char)):
+                                       case (mb_ereg ('[[:alpha:]]', $char) > 0):
                                                $buffer .= $char;
                                                break;
                                        case (preg_match ('/^[ \t\n]$/', $char)):
@@ -131,26 +122,26 @@ function getLexemsFromRackCode ($text)
                                                                $ret[] = array ('type' => 'LEX_BOOLCONST', 'load' => $buffer);
                                                                break;
                                                        default:
-                                                               abortLex2 ($state, $buffer);
+                                                               return lexError2 ($buffer);
                                                }
                                                $newstate = 'ESOTSM';
                                                break;
                                        default:
-                                               abortLex1 ($state, $text, $i);
+                                               return lexError1 ($state, $text, $i);
                                }
                                break;
                        case 'reading tag 1':
                                switch (TRUE)
                                {
-                                       case (preg_match ('/^[ \t\n]$/', $char)):
+                                       case (preg_match ('/^[ \t\n\r]$/', $char)):
                                                // nom-nom...
                                                break;
-                                       case (preg_match ('/^[a-zA-Z\$]$/', $char)):
+                                       case (mb_ereg ('[[:alnum:]\$]', $char) > 0):
                                                $buffer = $char;
                                                $newstate = 'reading tag 2';
                                                break;
                                        default:
-                                               abortLex1 ($state, $text, $i);
+                                               return lexError1 ($state, $text, $i);
                                }
                                break;
                        case 'reading tag 2':
@@ -158,30 +149,30 @@ function getLexemsFromRackCode ($text)
                                {
                                        case ($char == '}'):
                                                $buffer = rtrim ($buffer);
-                                               if (!preg_match ('/^[a-zA-Z0-9]$/', substr ($buffer, -1)))
-                                                       abortLex1 ($state, $text, $i);
+                                               if (!validTagName ($buffer, TRUE))
+                                                       return lexError4 ($buffer);
                                                $ret[] = array ('type' => 'LEX_TAG', 'load' => $buffer);
                                                $newstate = 'ESOTSM';
                                                break;
-                                       case (preg_match ('/^[a-zA-Z0-9 _-]$/', $char)):
+                                       case (mb_ereg ('[[:alnum:]\. _~-]', $char) > 0):
                                                $buffer .= $char;
                                                break;
                                        default:
-                                               abortLex1 ($state, $text, $i);
+                                               return lexError1 ($state, $text, $i);
                                }
                                break;
                        case 'reading predicate 1':
                                switch (TRUE)
                                {
-                                       case (preg_match ('/^[ \t\n]$/', $char)):
+                                       case (preg_match ('/^[ \t\n\r]$/', $char)):
                                                // nom-nom...
                                                break;
-                                       case (preg_match ('/^[a-zA-Z]$/', $char)):
+                                       case (mb_ereg ('[[:alnum:]]', $char) > 0):
                                                $buffer = $char;
                                                $newstate = 'reading predicate 2';
                                                break;
                                        default:
-                                               abortLex1 ($state, $text, $i);
+                                               return lexError1 ($state, $text, $i);
                                }
                                break;
                        case 'reading predicate 2':
@@ -189,20 +180,20 @@ function getLexemsFromRackCode ($text)
                                {
                                        case ($char == ']'):
                                                $buffer = rtrim ($buffer);
-                                               if (!preg_match ('/^[a-zA-Z0-9]$/', substr ($buffer, -1)))
-                                                       abortLex1 ($state, $text, $i);
+                                               if (!validTagName ($buffer))
+                                                       return lexError4 ($buffer);
                                                $ret[] = array ('type' => 'LEX_PREDICATE', 'load' => $buffer);
                                                $newstate = 'ESOTSM';
                                                break;
-                                       case (preg_match ('/^[a-zA-Z0-9 _-]$/', $char)):
+                                       case (mb_ereg ('[[:alnum:]\. _~-]', $char) > 0):
                                                $buffer .= $char;
                                                break;
                                        default:
-                                               abortLex1 ($state, $text, $i);
+                                               return lexError1 ($state, $text, $i);
                                }
                                break;
                        case 'skipping comment':
-                               switch ($text{$i})
+                               switch ($char)
                                {
                                        case "\n":
                                                $newstate = 'ESOTSM';
@@ -215,9 +206,9 @@ function getLexemsFromRackCode ($text)
                endswitch;
                $state = $newstate;
        endfor;
-       if ($state != 'ESOTSM')
-               abortLex3 ($state);
-       return $ret;
+       if ($state != 'ESOTSM' and $state != 'skipping comment')
+               return lexError3 ($state);
+       return array ('result' => 'ACK', 'load' => $ret);
 }
 
 // Parse the given lexems stream into a list of RackCode sentences. Each such
@@ -429,7 +420,7 @@ function getSentencesFromLexems ($lexems)
                                        (
                                                'type' => 'SYNT_DEFINITION',
                                                'term' => $stacksecondtop['load'],
-                                               'definition' => $stacktop
+                                               'definition' => $stacktop['load']
                                        )
                                );
                                continue;
@@ -483,79 +474,68 @@ function getSentencesFromLexems ($lexems)
                }
                // The moment of truth.
                if (count ($stack) == 1 and $stack[0]['type'] == 'SYNT_CODETEXT')
-                       return $stack[0]['load'];
-               return NULL;
+                       return array ('result' => 'ACK', 'load' => $stack[0]['load']);
+               return array ('result' => 'NAK', 'load' => 'Syntax error!');
        }
 }
 
-// Evaluate a boolean constant.
- function eval_boolconst ($const, $tagchain, $ptable)
+function eval_expression ($expr, $tagchain, $ptable)
 {
-       switch ($const['load'])
+       switch ($expr['type'])
        {
-               case 'true':
-                       return TRUE;
-               case 'false':
+               case 'LEX_TAG': // Return true, if given tag is present on the tag chain.
+                       foreach ($tagchain as $tagInfo)
+                               if ($expr['load'] == $tagInfo['tag'])
+                                       return TRUE;
                        return FALSE;
+               case 'LEX_PREDICATE': // Find given predicate in the symbol table and evaluate it.
+                       $pname = $expr['load'];
+                       if (!isset ($ptable[$pname]))
+                       {
+                               showError ("Predicate '${pname}' is referenced before declaration", __FUNCTION__);
+                               return;
+                       }
+                       return eval_expression ($ptable[$pname], $tagchain, $ptable);
+               case 'LEX_BOOLCONST': // Evaluate a boolean constant.
+                       switch ($expr['load'])
+                       {
+                               case 'true':
+                                       return TRUE;
+                               case 'false':
+                                       return FALSE;
+                               default:
+                                       showError ("Could not parse a boolean constant with value '${expr['load']}'", __FUNCTION__);
+                                       return; // should failure be harder?
+                       }
+               case 'SYNT_NOTEXPR':
+                       return !eval_expression ($expr['load'], $tagchain, $ptable);
+               case 'SYNT_BOOLOP':
+                       $leftresult = eval_expression ($expr['left'], $tagchain, $ptable);
+                       switch ($expr['subtype'])
+                       {
+                               case 'or':
+                                       if ($leftresult)
+                                               return TRUE; // early success
+                                       return eval_expression ($expr['right'], $tagchain, $ptable);
+                               case 'and':
+                                       if (!$leftresult)
+                                               return FALSE; // early failure
+                                       return eval_expression ($expr['right'], $tagchain, $ptable);
+                               default:
+                                       showError ("Cannot evaluate unknown boolean operation '${boolop['subtype']}'");
+                                       return;
+                       }
                default:
-                       showError ("Could not parse a boolean constant with value '${const['load']}'");
+                       showError ("Evaluation error, cannot process expression type '${expr['type']}'", __FUNCTION__);
                        break;
        }
 }
 
-function eval_expression ($expr, $tagchain, $ptable)
-{
-}
-
-function eval_notexpr ($notexpr, $tagchain, $ptable)
-{
-       return !eval_expression ($notexpr['load']);
-}
-
-function eval_boolop ($boolop, $tagchain, $ptable)
-{
-       $leftresult = eval_expression ($boolop['left']);
-       switch ($boolop['subtype'])
-       {
-               case 'or':
-                       if ($leftresult)
-                               return TRUE; // early success
-                       return eval_expression ($boolop['right']);
-               case 'and':
-                       if (!$leftresult)
-                               return FALSE; // early failure
-                       return eval_expression ($boolop['right']);
-               default:
-                       showError ("Cannot evaluate boolean operation '${boolop['subtype']}'");
-                       return;
-       }
-}
-
-// Return true, if given tag is present on the tag chain.
-function eval_tag ($tag, $tagchain, $ptable)
-{
-       foreach ($tagchain as $tagInfo)
-               if ($tag['load'] == $tagInfo['tag'])
-                       return TRUE;
-       return FALSE;
-}
-
-// Find given predicate in the symbol table and evaluate it.
-function eval_predicate ($tag, $tagchain, $ptable)
-{
-       $pname = $predicate['load'];
-       if (!isset ($ptable[$pname]))
-       {
-               showError ("Predicate '${pname}' is referenced before declaration");
-               return;
-       }
-       return eval_expression ($ptable[$pname]);
-}
-
-function gotClearanceForTagChain ($code, $tagchain)
+function gotClearanceForTagChain ($tagchain)
 {
+       global $rackCode;
        $ptable = array();
-       foreach ($code as $sentence)
+       foreach ($rackCode as $sentence)
        {
                switch ($sentence['type'])
                {
@@ -571,15 +551,89 @@ function gotClearanceForTagChain ($code, $tagchain)
                                                case 'deny':
                                                        return FALSE;
                                                default:
-                                                       showError ("Condition match for unknown grant decision '${sentence['decision']}'");
+                                                       showError ("Condition match for unknown grant decision '${sentence['decision']}'", __FUNCTION__);
                                                        break;
                                        }
+                               break;
                        default:
-                               showError ("Can't process sentence of unknown type '${sentence['type']}'");
+                               showError ("Can't process sentence of unknown type '${sentence['type']}'", __FUNCTION__);
                                break;
                }
        }
        return FALSE;
 }
 
+function getRackCode ($text)
+{
+       if (!mb_strlen ($text))
+               return array ('result' => 'NAK', 'load' => 'The RackCode text was found empty in ' . __FUNCTION__);
+       $text = str_replace ("\r", '', $text) . "\n";
+       $lex = getLexemsFromRackCode ($text);
+       if ($lex['result'] != 'ACK')
+               return $lex;
+       $synt = getSentencesFromLexems ($lex['load']);
+       if ($synt['result'] != 'ACK')
+               return $synt;
+       // An empty sentence list is semantically valid, yet senseless,
+       // so checking intermediate result once more won't hurt.
+       if (!count ($synt['load']))
+               return array ('result' => 'NAK', 'load' => 'Empty parse tree found in ' . __FUNCTION__);
+       return semanticFilter ($synt['load']);
+}
+
+// Return NULL, if the given expression can be evaluated against the given
+// predicate list. Return the name of the first show stopper otherwise.
+function firstUnrefPredicate ($plist, $expr)
+{
+       switch ($expr['type'])
+       {
+               case 'LEX_BOOLCONST':
+               case 'LEX_TAG':
+                       return NULL;
+               case 'LEX_PREDICATE':
+                       return in_array ($expr['load'], $plist) ? NULL : $expr['load'];
+               case 'SYNT_NOTEXPR':
+                       return firstUnrefPredicate ($plist, $expr['load']);
+               case 'SYNT_BOOLOP':
+                       if (($tmp = firstUnrefPredicate ($plist, $expr['left'])) !== NULL)
+                               return $tmp;
+                       if (($tmp = firstUnrefPredicate ($plist, $expr['right'])) !== NULL)
+                               return $tmp;
+                       return NULL;
+               default:
+                       return NULL;
+       }
+}
+
+function semanticFilter ($code)
+{
+       $predicatelist = array();
+       foreach ($code as $sentence)
+               switch ($sentence['type'])
+               {
+                       case 'SYNT_DEFINITION':
+                               $up = firstUnrefPredicate ($predicatelist, $sentence['definition']);
+                               if ($up !== NULL)
+                                       return array
+                                       (
+                                               'result' => 'NAK',
+                                               'load' => "definition [${sentence['term']}] uses unknown predicate [${up}]"
+                                       );
+                               $predicatelist[] = $sentence['term'];
+                               break;
+                       case 'SYNT_GRANT':
+                               $up = firstUnrefPredicate ($predicatelist, $sentence['condition']);
+                               if ($up !== NULL)
+                                       return array
+                                       (
+                                               'result' => 'NAK',
+                                               'load' => "grant sentence uses unknown predicate [${up}]"
+                                       );
+                               break;
+                       default:
+                               return array ('result' => 'NAK', 'load' => 'unknown sentence type');
+               }
+       return array ('result' => 'ACK', 'load' => $code);
+}
+
 ?>