Merge pull request #219 from cleverage/208

#208 - Fix FolderBrowserTask, InputFolderBrowserTask, CsvReaderTask and LineReaderTask (and their `Input*` variants)

Author: njoubert-cleverage

CHANGELOG.md

@@ -18,6 +18,7 @@ Latest
 * [#205](https://github.com/cleverage/process-bundle/issues/205) Fix FileSplitterTask: produced files lost the first line, doubled line breaks and ended with an empty file; each file now contains exactly `max_lines` lines of the source. Update documentation, add tests.
 * [#206](https://github.com/cleverage/process-bundle/issues/206) Fix AdvancedStatCounterTask, IterableBatchTask, ConditionTrait and ColumnAggregatorTask edge cases: log the first counted execution with correct counts, accept a `null` `batch_count` (only flush at the end), allow scalar inputs in conditions (`''` path on the whole value), aggregate `null` column values. Update documentation, add tests.
 * [#207](https://github.com/cleverage/process-bundle/issues/207) Fix JsonStreamReaderTask / JsonStreamWriterTask: throw an explicit `\UnexpectedValueException` when a line decodes to a scalar, create the missing parent directory when writing. Update documentation, add tests.
+* [#208](https://github.com/cleverage/process-bundle/issues/208) Fix FolderBrowserTask, InputFolderBrowserTask, CsvReaderTask and LineReaderTask (and their `Input*` variants): reset the state at the end of the iteration, so that a following input is read from its beginning. Update documentation, add tests.
 
 ## Deprecated
 * [#189](https://github.com/cleverage/process-bundle/issues/189) EventDispatcherTask: when `event_name` is set, listening to `CleverAge\ProcessBundle\Event\EventDispatcherTaskEvent` is deprecated (the event is still dispatched under its class name, with an `E_USER_DEPRECATED` error, if it has listeners). Listen to the configured `event_name` instead: the BC layer will be removed in v6.0.

docs/reference/tasks/csv_reader_task.md

@@ -72,4 +72,6 @@ Notes
 * Each line must contain exactly as many columns as there are headers, otherwise an `\UnexpectedValueException` is thrown.
 * A UTF-8 BOM is removed from the first header when headers are read from the file.
 * `csv_file` and `csv_line` are added to the error context of the process.
+* The file is closed once fully read: if the task is executed again (e.g. for a new input), the file is read again
+  from its beginning.
 * See also [InputCsvReaderTask](input_csv_reader_task.md) to read a file path given as input.

docs/reference/tasks/folder_browser_task.md

@@ -53,4 +53,6 @@ Notes
 -----
 
 * `current_file_path` is added to the error context of the process.
+* The iteration is reset once all files have been output: if the task is executed again (e.g. for a new input), the
+  folder is browsed again from the start.
 * See also [InputFolderBrowserTask](input_folder_browser_task.md) to browse a folder path given as input.

docs/reference/tasks/input_csv_reader_task.md

@@ -14,7 +14,8 @@ Accepted inputs
 ---------------
 
 `string`: path of the file to read, prefixed by the `base_path` option if set.
-When a different path is received, the previous file is dropped and the new one is opened.
+The file is closed once fully read, so each input (the same path again or a different one) is read from its
+beginning.
 
 Possible outputs
 ----------------

docs/reference/tasks/input_folder_browser_task.md

@@ -17,7 +17,9 @@ Accepted inputs
 `string`: path of the folder to browse, prefixed by the `base_folder_path` option. It must be an existing readable
 directory.
 
-Receiving a different folder path before the task has been flushed throws a `\LogicException`.
+The folder path is released once its files have all been iterated, so each input (the same path again or a different
+one) is browsed from the start. Receiving a different folder path while an iteration is in progress throws a
+`\LogicException`.
 
 Possible outputs
 ----------------

docs/reference/tasks/input_line_reader_task.md

@@ -14,7 +14,8 @@ Accepted inputs
 ---------------
 
 `string`: path of the file to read. An `\UnexpectedValueException` is thrown if the file does not exist or is not
-readable. When a different path is received, the previous file is dropped and the new one is opened.
+readable. The file is released once fully read, so each input (the same path again or a different one) is read from
+its beginning.
 
 Possible outputs
 ----------------

docs/reference/tasks/line_reader_task.md

@@ -51,4 +51,6 @@ trim:
 Notes
 -----
 
+* The file is released once fully read: if the task is executed again (e.g. for a new input), the file is read again
+  from its beginning.
 * See also [InputLineReaderTask](input_line_reader_task.md) to read a file path given as input.

src/Task/File/Csv/CsvReaderTask.php

@@ -74,7 +74,14 @@ public function next(ProcessState $state): bool
         $state->removeErrorContext('csv_file');
         $state->removeErrorContext('csv_line');
 
-        return !$this->csv->isEndOfFile();
+        $endOfFile = $this->csv->isEndOfFile();
+        if ($endOfFile) {
+            // Release the file to allow the following iteration to read it again
+            $this->csv->close();
+            $this->csv = null;
+        }
+
+        return !$endOfFile;
     }
 
     protected function getHeaders(ProcessState $state, array $options): ?array

src/Task/File/FolderBrowserTask.php

@@ -80,7 +80,14 @@ public function next(ProcessState $state): bool
         $this->files->next();
         $state->removeErrorContext('current_file_path');
 
-        return $this->files->valid();
+        if (!$this->files->valid()) {
+            // Reset the iterator to allow the following iteration
+            $this->files = null;
+
+            return false;
+        }
+
+        return true;
     }
 
     protected function configureOptions(OptionsResolver $resolver): void

src/Task/File/InputFolderBrowserTask.php

@@ -31,6 +31,18 @@ public function flush(ProcessState $state): void
         $state->setSkipped(true);
     }
 
+    #[\Override]
+    public function next(ProcessState $state): bool
+    {
+        $hasNext = parent::next($state);
+        if (!$hasNext) {
+            // Release the folder path to allow the following input to browse another folder
+            $this->folderPath = null;
+        }
+
+        return $hasNext;
+    }
+
     #[\Override]
     public function initialize(ProcessState $state): void
     {