Config option added to allow specification of source file encoding in case auto detection doesn't work. Fixed #212

Author: theseer

src/collector/Collector.php

@@ -67,15 +67,22 @@ class Collector {
          */
         private $backend;
 
+        /**
+         * @var string
+         */
+        private $encoding;
 
         /**
-         * @param ProgressLogger  $logger
-         * @param Project $project
+         * @param ProgressLogger   $logger
+         * @param Project          $project
+         * @param BackendInterface $backend
+         * @param                  $encoding
          */
-        public function __construct(ProgressLogger $logger, Project $project, BackendInterface $backend) {
+        public function __construct(ProgressLogger $logger, Project $project, BackendInterface $backend, $encoding) {
             $this->logger = $logger;
             $this->project = $project;
             $this->backend = $backend;
+            $this->encoding = $encoding;
         }
 
         /**
@@ -88,7 +95,7 @@ public function run(DirectoryScanner $scanner) {
             $srcDir = $this->project->getSourceDir();
             $this->logger->log("Scanning directory '{$srcDir}' for files to process\n");
 
-            $iterator = new SourceFileIterator($scanner($srcDir), $srcDir);
+            $iterator = new SourceFileIterator($scanner($srcDir), $srcDir, $this->encoding);
             foreach($iterator as $file) {
                 $needsProcessing = $this->project->addFile($file);
                 if (!$needsProcessing) {

src/collector/SourceFileIterator.php

@@ -7,17 +7,24 @@ class SourceFileIterator implements \Iterator {
 
         private $iterator;
         private $srcDir;
+        private $encoding;
 
-        public function __construct(\Iterator $iterator, $srcDir) {
+        /**
+         * @param \Iterator $iterator
+         * @param FileInfo  $srcDir
+         * @param string    $encoding
+         */
+        public function __construct(\Iterator $iterator, FileInfo $srcDir, $encoding) {
             $this->iterator = $iterator;
             $this->srcDir = $srcDir;
+            $this->encoding = $encoding;
         }
 
         /**
          * @return SourceFile
          */
         public function current() {
-            return new SourceFile($this->iterator->current()->getPathname(), $this->srcDir);
+            return new SourceFile($this->iterator->current()->getPathname(), $this->srcDir, $this->encoding);
         }
 
         /**

src/collector/project/SourceFile.php

@@ -23,9 +23,15 @@ class SourceFile extends FileInfo {
          */
         private $srcDir;
 
-        public function __construct($file_name, FileInfo $srcDir = NULL) {
+        /**
+         * @var
+         */
+        private $encoding;
+
+        public function __construct($file_name, FileInfo $srcDir = NULL, $encoding = 'auto') {
             parent::__construct($file_name);
             $this->srcDir = $srcDir;
+            $this->encoding = $encoding;
         }
 
         /**
@@ -44,10 +50,12 @@ public function getSource() {
                 return '';
             }
 
-            $info = new \finfo();
-            $encoding = $info->file((string)$this, FILEINFO_MIME_ENCODING);
+            if ($this->encoding == 'auto') {
+                $info = new \finfo();
+                $this->encoding = $info->file((string)$this, FILEINFO_MIME_ENCODING);
+            }
             try {
-                $source = iconv($encoding, 'UTF-8//TRANSLIT', $source);
+                $source = iconv($this->encoding, 'UTF-8//TRANSLIT', $source);
             } catch (\ErrorException $e) {
                 throw new SourceFileException('Encoding error - conversion to UTF-8 failed', SourceFileException::BadEncoding, $e);
             }

src/config/CollectorConfig.php

@@ -75,6 +75,13 @@ public function getSourceDirectory() {
             return $this->project->getSourceDirectory();
         }
 
+        /**
+         * @return string
+         */
+        public function getFileEncoding() {
+            return $this->ctx->getAttribute('encoding', 'auto');
+        }
+
         public function isPublicOnlyMode() {
             if ($this->ctx->hasAttribute('publiconly')) {
                 return $this->ctx->getAttribute('publiconly', 'false') === 'true';

src/config/skeleton.xml

@@ -35,9 +35,10 @@
          -->
 
         <!--  Additional configuration for the collecting process (parsing of php code, generation of xml data) -->
-        <collector publiconly="false" backend="parser">
+        <collector publiconly="false" backend="parser" encoding="auto">
             <!--  @publiconly - Flag to disable/enable processing of non public methods and members -->
             <!--  @backend    - The collector backend to use, currently only shipping with 'parser' -->
+            <!--  @encoding   - Charset encoding of source files (overwrite default 'auto' if detection fails) -->
 
             <!--  <include / exclude filter for filelist generator, mask must follow fnmatch() requirements  -->
             <include mask="*.php" />

src/shared/Factory.php

@@ -199,7 +199,8 @@ public function getCollector(CollectorConfig $config) {
                     $config->getSourceDirectory(),
                     $config->getWorkDirectory()
                 ),
-                $this->getBackendFactory()->getInstanceFor($config->getBackend())
+                $this->getBackendFactory()->getInstanceFor($config->getBackend()),
+                $config->getFileEncoding()
             );
         }