Iterable file readers keep their state after a complete iteration: FolderBrowserTask, InputFolderBrowserTask, CsvReaderTask, LineReaderTask #208

Description

FolderBrowserTask, CsvReaderTask and LineReaderTask (and their Input* variants) do not reset their internal iterator or file once an iteration is finished. If the task runs again, for example because it receives a second input from an upstream iterable task, it does not start over:

• FolderBrowserTask / InputFolderBrowserTask, same folder twice: the second pass logs a false "No item found" and outputs nothing.
• InputFolderBrowserTask, two different folders: throws LogicException: Folder path '…/dB' already initialized with a different value …/dA, the folder path being only cleared on flush().
• CsvReaderTask / InputCsvReaderTask, same path twice: the second pass outputs nothing (skipped).
• LineReaderTask / InputLineReaderTask, same path twice: the second pass outputs false.

JsonStreamReaderTask, CsvSplitterTask and AbstractIterableOutputTask already release their state at the end of the iteration.

Reproduction

clever_age_process:
    configurations:
        demo.readers_reset.folder:
            tasks:
                folders:
                    service: '@CleverAge\ProcessBundle\Task\ConstantIterableOutputTask'
                    options:
                        output: ['/var/www/var/readers-reset/dA', '/var/www/var/readers-reset/dA', '/var/www/var/readers-reset/dB']
                    outputs: [browse]
                browse:
                    service: '@CleverAge\ProcessBundle\Task\File\InputFolderBrowserTask'
                    outputs: [debug_folder]
                debug_folder:
                    service: '@CleverAge\ProcessBundle\Task\Debug\DebugTask'
        demo.readers_reset.csv:
            tasks:
                csv_files:
                    service: '@CleverAge\ProcessBundle\Task\ConstantIterableOutputTask'
                    options:
                        output: ['/var/www/var/readers-reset/a.csv', '/var/www/var/readers-reset/a.csv']
                    outputs: [read_csv]
                read_csv:
                    service: '@CleverAge\ProcessBundle\Task\File\Csv\InputCsvReaderTask'
                    outputs: [debug_csv]
                debug_csv:
                    service: '@CleverAge\ProcessBundle\Task\Debug\DebugTask'
        demo.readers_reset.line:
            tasks:
                text_files:
                    service: '@CleverAge\ProcessBundle\Task\ConstantIterableOutputTask'
                    options:
                        output: ['/var/www/var/readers-reset/a.txt', '/var/www/var/readers-reset/a.txt']
                    outputs: [read_lines]
                read_lines:
                    service: '@CleverAge\ProcessBundle\Task\File\InputLineReaderTask'
                    outputs: [debug_line]
                debug_line:
                    service: '@CleverAge\ProcessBundle\Task\Debug\DebugTask'

Data: dA/f1.txt, dB/f2.txt, a.csv = id;name\n1;foo\n2;bar\n, a.txt = l1\nl2\n.

Output on main (c1a2804, PHP 8.5, Symfony 7.4):

Starting process 'demo.readers_reset.folder'...
"/var/www/var/readers-reset/dA/f1.txt"
WARNING   [cleverage_process_task] No item found in path /var/www/var/readers-reset/dA
CRITICAL  [cleverage_process_task] Folder path '/var/www/var/readers-reset/dB' already initialized with a different value /var/www/var/readers-reset/dA

Starting process 'demo.readers_reset.csv'...
array:2 [ "id" => "1"  "name" => "foo" ]
array:2 [ "id" => "2"  "name" => "bar" ]
Process 'demo.readers_reset.csv' executed successfully        <- second input produced nothing

Starting process 'demo.readers_reset.line'...
"l1\n"
"l2\n"
false                                                          <- second input
Process 'demo.readers_reset.line' executed successfully

Expected: dA/f1.txt, dA/f1.txt, dB/f2.txt; the two CSV rows twice; "l1\n", "l2\n" twice.

Cause

• FolderBrowserTask::next() returns $this->files->valid() but never resets the exhausted iterator: the next execute() finds $files !== null, skips the re-initialization and logs "No item found".
• InputFolderBrowserTask::getOptions() keeps $folderPath until flush(), so a different folder received afterwards throws.
• CsvReaderTask::execute() only re-opens the file when the path changes, and CsvReaderTask::next() never releases it at end of file.
• Same for LineReaderTask::execute() and LineReaderTask::next().

Proposed fix

Release the state in next() once the iteration is over, as JsonStreamReaderTask::next() already does:

• FolderBrowserTask: reset $files to null;
• InputFolderBrowserTask: override next() to also reset $folderPath (a different path received mid-iteration still throws);
• CsvReaderTask: close the CSV file and reset it to null;
• LineReaderTask: reset the SplFileObject to null.

The Input* variants inherit the fix: each new input, same path or not, is read from its beginning. Update the reference documentation, add unit tests.

Requirements

• Documentation updates
  • Reference
  • Cookbooks
  • Changelog
• Unit tests

Breaking changes

None.