Add standalone Automators for file to formatted text and JSON outputs (#3573331) · Issues · project / document_loader

Add standalone Automators for file to formatted text and JSON outputs

>>> [!note] Migrated issue   Reported by: [marcus_johansson](https://www.drupal.org/user/385947) Related to !21 >>> [Tracker] Update Summary: [One-line status update for stakeholders] Short Description: Add standalone Automators for file → formatted text and JSON outputs Check-in Date: MM/DD/YYYY [/Tracker] <h3 id="summary-problem-motivation">Problem/Motivation</h3> The existing AI Automators in the AI module demonstrate how file‐based content can be transformed into structured formats such as JSON or binary representations. For example, the AI module includes JSON native/binary Automators such as LlmJsonNativeBinary and base classes like TextToJsonField that target JSON output formats. These work by sending prompt/context to an AI provider and mapping results back into fields. For non-AI and deterministic workflows, we need a new set of Automators that use the <code>document_loader</code> module to extract/normalize file content into useful field/value formats without requiring any AI provider. The functionality must not depend on the AI module at runtime; the AI module can be placed in <code>require-dev</code> for tests, but the Automators should function without it. We want Automators that read file input and convert it into: <ul> <li>Formatted text (HTML output)</li> <li>Formatted text with summary (HTML output)</li> <li>Long text output (any supported output type)</li> <li>JSON native output (JSON output) — JSON field module required</li> <li>JSON field output (JSON output) — JSON field module required</li> <li>JSON native binary output — JSON field module required</li> </ul> The last three (JSON ones) are optional and can be split into a separate follow-up issue. Callers should be able to configure what input types are accepted (e.g., PDF, DOCX, URL) and what output types they support, especially for the long text Automator. <h3 id="summary-proposed-resolution">Proposed resolution</h3> <ul> <li>Create new Automator plugin types under a suitable namespace (e.g., <code>Plugin/DocumentAutomatorType</code>).</li> <li>Define a base class for file-to-text transformation that accepts document loader input types and drives html output.</li> <li>For “File to Formatted Text (HTML)” and “File to Formatted Text with Summary (HTML)”, pull normalized content using document_loader and ensure output is valid HTML.</li> <li>Define a “File to Long Text” Automator that accepts any output type supported by document_loader and delegates to the transformation manager to produce final text output.</li> <li>For JSON based Automators (native, field, binary), define a shared base class similar in pattern to AI module’s TextToJsonField, since the conversion code will be largely identical except for output target.</li> <li>Ensure the JSON Automators depend on the JSON module and gracefully degrade (disabled/not available) when JSON module is not installed.</li> <li>Design configuration settings on each Automator so site builders can specify accepted input types and output types.</li> <li>Ensure the Automators do NOT require the AI module for runtime operation — AI should only be included in <code>require-dev</code> for tests that may reference examples from the AI module.</li> <li>Add tests covering valid file inputs, invalid inputs, missing dependencies (JSON module), and expected output format handling.</li> <li>Document how to configure these Automators on fields in content types and how they integrate with document_loader normalization.</li> <li>Optionally split JSON Automators into a follow-up issue if complexity warrants it.</li> </ul> <h3 id="summary-ai-usage">AI usage (if applicable)</h3> [x] AI Assisted Issue This issue was generated with AI assistance, but was reviewed and refined by the creator. [ ] AI Assisted Code This code was mainly generated by a human, with AI autocompleting or parts AI generated, but under full human supervision. [ ] AI Generated Code This code was mainly generated by an AI with human guidance, and reviewed, tested, and refined by a human. [ ] Vibe Coded This code was generated by an AI and has only been functionally tes ::contentReference[oaicite:1]{index=1} > Related issue: [Issue #3575412](https://www.drupal.org/node/3575412)

issue