Enhance exception context, fix PCRE issue #6

[TASK] Add more context to exceptions * enhances `UnrecognizedToken` exception, see hoaproject/Compiler#111 * enhances `UnexpectedToken` exception, see hoaproject/Compiler#112 [BUGFIX] Allow colons in PCRE part of default namespace See: hoaproject/Compiler#113
sanmai · Oct 7, 2020 · 7ca878c · 7ca878c
2 parents 55577fe + e93b372
commit 7ca878c
Show file tree

Hide file tree

Showing 3 changed files with 20 additions and 6 deletions.
diff --git a/atoum/Unit/Llk/Llk.php b/atoum/Unit/Llk/Llk.php
@@ -217,7 +217,8 @@ public function case_parse_tokens()
                     '%token  foobar1            bazqux1' . "\n" .
                     '%token  sourceNS1:foobar2  bazqux2' . "\n" .
                     '%token  sourceNS2:foobar3  bazqux3  -> destinationNS' . "\n" .
-                    '%token  foobar4            barqux4  -> destinationNS'
+                    '%token  foobar4            barqux4  -> destinationNS' . "\n" .
+                    '%token  foobar5            ns:bar   -> destinationNS'
             )
             ->when($result = SUT::parsePP($pp, $tokens, $rules, $pragmas, 'streamFoo'))
             ->then
@@ -227,7 +228,8 @@ public function case_parse_tokens()
                     ->isEqualTo([
                         'default' => [
                             'foobar1'               => 'bazqux1',
-                            'foobar4:destinationNS' => 'barqux4'
+                            'foobar4:destinationNS' => 'barqux4',
+                            'foobar5:destinationNS' => 'ns:bar',
                         ],
                         'sourceNS1' => [
                             'foobar2' => 'bazqux2'

diff --git a/src/Llk/Lexer.php b/src/Llk/Lexer.php
@@ -155,14 +155,25 @@ public function lexMe($text, array $tokens): \Generator
             $nextToken = $this->nextToken($offset);
 
             if (null === $nextToken) {
+                $line              = 1;
+                $column            = $offset;
+                $offsetText        = substr($text, 0, $offset);
+                $pointerSpacing    = mb_strlen($offsetText);
+                $previousLineBreak = strrpos($offsetText, "\n");
+                if ($previousLineBreak !== false) {
+                    $line           = substr_count($offsetText, "\n") + 1;
+                    $column         = mb_strlen($offsetText) - $previousLineBreak - 1;
+                    $pointerSpacing = $column;
+                }
                 throw new Compiler\Exception\UnrecognizedToken(
-                    'Unrecognized token "%s" at line 1 and column %d:' .
+                    'Unrecognized token "%s" at line %d and column %d:' .
                     "\n" . '%s' . "\n" .
-                    str_repeat(' ', mb_strlen(substr($text, 0, $offset))) . '↑',
+                    str_repeat(' ', $pointerSpacing) . '↑',
                     0,
                     [
                         mb_substr(substr($text, $offset), 0, 1),
-                        $offset + 1,
+                        $line,
+                        $column + 1,
                         $text
                     ],
                     1,

diff --git a/src/Llk/Parser.php b/src/Llk/Parser.php
@@ -215,11 +215,12 @@ final public function parse(string $text, string $rule = null, bool $tree = true
                 }
 
                 throw new Compiler\Exception\UnexpectedToken(
-                    'Unexpected token "%s" (%s) at line %d and column %d:' .
+                    'Unexpected token "%s" (%s:%s) at line %d and column %d:' .
                     "\n" . '%s' . "\n" . str_repeat(' ', $column - 1) . '↑',
                     0,
                     [
                         $token['value'],
+                        $token['namespace'],
                         $token['token'],
                         $line,
                         $column,