Spaces:
Running
Running
| <html> | |
| <head> | |
| <title>Document Query</title> | |
| </head> | |
| <body> | |
| <div x-data="{ | |
| defaults: { | |
| fetch_timeout: 30, | |
| fetch_retries: 3, | |
| fetch_retry_backoff: 1.0, | |
| per_document_timeout: 60, | |
| gather_timeout: 120, | |
| parser_concurrency: 1, | |
| context_intro_chunks: 2, | |
| chunk_size: 1000, | |
| chunk_overlap: 100, | |
| max_index_chunks: 1200, | |
| search_threshold: 0.5, | |
| search_limit: 100, | |
| max_remote_bytes: 52428800, | |
| liteparse_enabled: true, | |
| liteparse_ocr_enabled: true, | |
| liteparse_ocr_language: 'eng', | |
| liteparse_ocr_server_url: '', | |
| liteparse_tessdata_path: '', | |
| liteparse_max_pages: 1000, | |
| liteparse_target_pages: '', | |
| liteparse_dpi: 150, | |
| liteparse_preserve_very_small_text: false, | |
| liteparse_output_format: 'text', | |
| liteparse_num_workers: 2, | |
| pdf_ocr_fallback: true, | |
| thread_offload: true, | |
| }, | |
| initDefaults() { | |
| for (const [key, value] of Object.entries(this.defaults)) { | |
| if (config[key] === undefined || config[key] === null) config[key] = value; | |
| } | |
| this.ensureInt('parser_concurrency', 1, 1); | |
| this.ensureInt('context_intro_chunks', 2, 0); | |
| this.ensureInt('chunk_size', 1000, 100); | |
| this.ensureInt('chunk_overlap', 100, 0); | |
| this.ensureInt('max_index_chunks', 1200, 0); | |
| this.ensureInt('search_limit', 100, 1); | |
| this.ensureInt('max_remote_bytes', 52428800, 1); | |
| this.ensureNumber('search_threshold', 0.5, 0, 1); | |
| this.ensureNumber('fetch_timeout', 30, 1); | |
| this.ensureInt('fetch_retries', 3, 1); | |
| this.ensureNumber('fetch_retry_backoff', 1.0, 0); | |
| this.ensureNumber('per_document_timeout', 60, 1); | |
| this.ensureNumber('gather_timeout', 120, 1); | |
| this.ensureInt('liteparse_max_pages', 1000, 1); | |
| this.ensureNumber('liteparse_dpi', 150, 72); | |
| this.ensureInt('liteparse_num_workers', 2, 1); | |
| this.syncChunkOverlap(); | |
| }, | |
| ensureInt(key, fallback, min = null, max = null) { | |
| let value = parseInt(config[key], 10); | |
| if (!Number.isFinite(value)) value = fallback; | |
| if (min !== null) value = Math.max(min, value); | |
| if (max !== null) value = Math.min(max, value); | |
| config[key] = value; | |
| }, | |
| ensureNumber(key, fallback, min = null, max = null) { | |
| let value = parseFloat(config[key]); | |
| if (!Number.isFinite(value)) value = fallback; | |
| if (min !== null) value = Math.max(min, value); | |
| if (max !== null) value = Math.min(max, value); | |
| config[key] = value; | |
| }, | |
| syncChunkOverlap() { | |
| this.ensureInt('chunk_size', 1000, 100); | |
| this.ensureInt('chunk_overlap', 100, 0); | |
| if (config.chunk_overlap >= config.chunk_size) { | |
| config.chunk_overlap = Math.max(0, config.chunk_size - 1); | |
| } | |
| }, | |
| }"> | |
| <template x-if="config"> | |
| <div x-init="initDefaults()"> | |
| <div class="section-title">Document Query</div> | |
| <div class="section-description"> | |
| Settings for document parsing and retrieval. | |
| </div> | |
| <div class="field"> | |
| <div class="field-label"> | |
| <div class="field-title">Max parser concurrency</div> | |
| <div class="field-description"> | |
| Maximum document parser jobs allowed to run at the same time in this Agent Zero process. | |
| </div> | |
| </div> | |
| <div class="field-control"> | |
| <input type="number" min="1" step="1" | |
| @change="ensureInt('parser_concurrency', 1, 1)" | |
| x-model.number="config.parser_concurrency" /> | |
| </div> | |
| </div> | |
| <div class="field"> | |
| <div class="field-label"> | |
| <div class="field-title">Per-document timeout</div> | |
| <div class="field-description"> | |
| Seconds allowed for a single document parse before trying the next parser or returning an error. | |
| </div> | |
| </div> | |
| <div class="field-control"> | |
| <input type="number" min="1" step="1" | |
| @change="ensureNumber('per_document_timeout', 60, 1)" | |
| x-model.number="config.per_document_timeout" /> | |
| </div> | |
| </div> | |
| <div class="field"> | |
| <div class="field-label"> | |
| <div class="field-title">Batch timeout</div> | |
| <div class="field-description"> | |
| Seconds allowed for a document-query call that processes multiple files. | |
| </div> | |
| </div> | |
| <div class="field-control"> | |
| <input type="number" min="1" step="1" | |
| @change="ensureNumber('gather_timeout', 120, 1)" | |
| x-model.number="config.gather_timeout" /> | |
| </div> | |
| </div> | |
| <div class="section-title">Retrieval</div> | |
| <div class="section-description"> | |
| Controls how parsed text is split, indexed, and selected for Q&A. | |
| </div> | |
| <div class="field"> | |
| <div class="field-label"> | |
| <div class="field-title">Chunk size</div> | |
| <div class="field-description"> | |
| Target character count for indexed chunks. | |
| </div> | |
| </div> | |
| <div class="field-control"> | |
| <input type="number" min="100" step="50" | |
| @change="syncChunkOverlap()" | |
| x-model.number="config.chunk_size" /> | |
| </div> | |
| </div> | |
| <div class="field"> | |
| <div class="field-label"> | |
| <div class="field-title">Chunk overlap</div> | |
| <div class="field-description"> | |
| Characters repeated between neighboring chunks. | |
| </div> | |
| </div> | |
| <div class="field-control"> | |
| <input type="number" min="0" step="25" | |
| :max="Math.max(0, config.chunk_size - 1)" | |
| @change="syncChunkOverlap()" | |
| x-model.number="config.chunk_overlap" /> | |
| </div> | |
| </div> | |
| <div class="field"> | |
| <div class="field-label"> | |
| <div class="field-title">Search threshold</div> | |
| <div class="field-description"> | |
| Minimum vector similarity score for retrieved chunks. | |
| </div> | |
| </div> | |
| <div class="field-control"> | |
| <input type="number" min="0" max="1" step="0.05" | |
| @change="ensureNumber('search_threshold', 0.5, 0, 1)" | |
| x-model.number="config.search_threshold" /> | |
| </div> | |
| </div> | |
| <div class="field"> | |
| <div class="field-label"> | |
| <div class="field-title">Max index chunks</div> | |
| <div class="field-description"> | |
| Adapt chunk size when a parsed document would exceed this many indexed chunks. Use 0 for no cap. | |
| </div> | |
| </div> | |
| <div class="field-control"> | |
| <input type="number" min="0" step="50" | |
| @change="ensureInt('max_index_chunks', 1200, 0)" | |
| x-model.number="config.max_index_chunks" /> | |
| </div> | |
| </div> | |
| <div class="field"> | |
| <div class="field-label"> | |
| <div class="field-title">Search limit</div> | |
| <div class="field-description"> | |
| Maximum matching chunks considered during document Q&A. | |
| </div> | |
| </div> | |
| <div class="field-control"> | |
| <input type="number" min="1" step="1" | |
| @change="ensureInt('search_limit', 100, 1)" | |
| x-model.number="config.search_limit" /> | |
| </div> | |
| </div> | |
| <div class="field"> | |
| <div class="field-label"> | |
| <div class="field-title">Intro chunks</div> | |
| <div class="field-description"> | |
| Leading chunks always included per document to preserve titles, abstracts, and setup context. | |
| </div> | |
| </div> | |
| <div class="field-control"> | |
| <input type="number" min="0" step="1" | |
| @change="ensureInt('context_intro_chunks', 2, 0)" | |
| x-model.number="config.context_intro_chunks" /> | |
| </div> | |
| </div> | |
| <div class="section-title">Remote Files</div> | |
| <div class="section-description"> | |
| Limits for documents loaded from HTTP or HTTPS URLs. | |
| </div> | |
| <div class="field"> | |
| <div class="field-label"> | |
| <div class="field-title">Fetch timeout</div> | |
| <div class="field-description"> | |
| Seconds allowed for each remote fetch attempt. | |
| </div> | |
| </div> | |
| <div class="field-control"> | |
| <input type="number" min="1" step="1" | |
| @change="ensureNumber('fetch_timeout', 30, 1)" | |
| x-model.number="config.fetch_timeout" /> | |
| </div> | |
| </div> | |
| <div class="field"> | |
| <div class="field-label"> | |
| <div class="field-title">Fetch retries</div> | |
| <div class="field-description"> | |
| Attempts made before giving up on a remote document. | |
| </div> | |
| </div> | |
| <div class="field-control"> | |
| <input type="number" min="1" step="1" | |
| @change="ensureInt('fetch_retries', 3, 1)" | |
| x-model.number="config.fetch_retries" /> | |
| </div> | |
| </div> | |
| <div class="field"> | |
| <div class="field-label"> | |
| <div class="field-title">Retry backoff</div> | |
| <div class="field-description"> | |
| Delay in seconds between remote fetch retries. | |
| </div> | |
| </div> | |
| <div class="field-control"> | |
| <input type="number" min="0" step="0.1" | |
| @change="ensureNumber('fetch_retry_backoff', 1.0, 0)" | |
| x-model.number="config.fetch_retry_backoff" /> | |
| </div> | |
| </div> | |
| <div class="field"> | |
| <div class="field-label"> | |
| <div class="field-title">Max remote bytes</div> | |
| <div class="field-description"> | |
| Maximum size accepted for a remote document. | |
| </div> | |
| </div> | |
| <div class="field-control"> | |
| <input type="number" min="1" step="1048576" | |
| @change="ensureInt('max_remote_bytes', 52428800, 1)" | |
| x-model.number="config.max_remote_bytes" /> | |
| </div> | |
| </div> | |
| <div class="section-title">LiteParse and OCR</div> | |
| <div class="section-description"> | |
| Parser preference and OCR controls for PDFs, document images, and mixed-format inputs. | |
| </div> | |
| <div class="field"> | |
| <div class="field-label"> | |
| <div class="field-title">Use LiteParse first</div> | |
| <div class="field-description"> | |
| Prefer LiteParse before legacy parser fallbacks. | |
| </div> | |
| </div> | |
| <div class="field-control"> | |
| <label class="toggle"> | |
| <input type="checkbox" x-model="config.liteparse_enabled" /> | |
| <span class="toggler"></span> | |
| </label> | |
| </div> | |
| </div> | |
| <template x-if="config.liteparse_enabled !== false"> | |
| <div> | |
| <div class="field"> | |
| <div class="field-label"> | |
| <div class="field-title">LiteParse workers</div> | |
| <div class="field-description"> | |
| Worker count passed to LiteParse for a single parser job. | |
| </div> | |
| </div> | |
| <div class="field-control"> | |
| <input type="number" min="1" step="1" | |
| @change="ensureInt('liteparse_num_workers', 2, 1)" | |
| x-model.number="config.liteparse_num_workers" /> | |
| </div> | |
| </div> | |
| <div class="field"> | |
| <div class="field-label"> | |
| <div class="field-title">Max pages</div> | |
| <div class="field-description"> | |
| Maximum pages LiteParse should parse from one document. | |
| </div> | |
| </div> | |
| <div class="field-control"> | |
| <input type="number" min="1" step="1" | |
| @change="ensureInt('liteparse_max_pages', 1000, 1)" | |
| x-model.number="config.liteparse_max_pages" /> | |
| </div> | |
| </div> | |
| <div class="field"> | |
| <div class="field-label"> | |
| <div class="field-title">Target pages</div> | |
| <div class="field-description"> | |
| Optional page range expression passed to LiteParse. | |
| </div> | |
| </div> | |
| <div class="field-control"> | |
| <input type="text" | |
| placeholder="all pages" | |
| x-model="config.liteparse_target_pages" /> | |
| </div> | |
| </div> | |
| <div class="field"> | |
| <div class="field-label"> | |
| <div class="field-title">Enable OCR</div> | |
| <div class="field-description"> | |
| Allow OCR for scanned PDFs and document images. | |
| </div> | |
| </div> | |
| <div class="field-control"> | |
| <label class="toggle"> | |
| <input type="checkbox" x-model="config.liteparse_ocr_enabled" /> | |
| <span class="toggler"></span> | |
| </label> | |
| </div> | |
| </div> | |
| <template x-if="config.liteparse_ocr_enabled !== false"> | |
| <div> | |
| <div class="field"> | |
| <div class="field-label"> | |
| <div class="field-title">OCR language</div> | |
| <div class="field-description"> | |
| Tesseract language code used by LiteParse OCR. | |
| </div> | |
| </div> | |
| <div class="field-control"> | |
| <input type="text" | |
| placeholder="eng" | |
| x-model="config.liteparse_ocr_language" /> | |
| </div> | |
| </div> | |
| <div class="field"> | |
| <div class="field-label"> | |
| <div class="field-title">OCR DPI</div> | |
| <div class="field-description"> | |
| Render density used before OCR. | |
| </div> | |
| </div> | |
| <div class="field-control"> | |
| <input type="number" min="72" step="1" | |
| @change="ensureNumber('liteparse_dpi', 150, 72)" | |
| x-model.number="config.liteparse_dpi" /> | |
| </div> | |
| </div> | |
| <div class="field"> | |
| <div class="field-label"> | |
| <div class="field-title">OCR server URL</div> | |
| <div class="field-description"> | |
| Optional external OCR service URL. | |
| </div> | |
| </div> | |
| <div class="field-control"> | |
| <input type="text" | |
| placeholder="none" | |
| x-model="config.liteparse_ocr_server_url" /> | |
| </div> | |
| </div> | |
| <div class="field"> | |
| <div class="field-label"> | |
| <div class="field-title">Tessdata path</div> | |
| <div class="field-description"> | |
| Optional path to Tesseract language data. | |
| </div> | |
| </div> | |
| <div class="field-control"> | |
| <input type="text" | |
| placeholder="auto-detect" | |
| x-model="config.liteparse_tessdata_path" /> | |
| </div> | |
| </div> | |
| </div> | |
| </template> | |
| <div class="field"> | |
| <div class="field-label"> | |
| <div class="field-title">Output format</div> | |
| <div class="field-description"> | |
| LiteParse output format passed to the parser runtime. | |
| </div> | |
| </div> | |
| <div class="field-control"> | |
| <input type="text" | |
| placeholder="text" | |
| x-model="config.liteparse_output_format" /> | |
| </div> | |
| </div> | |
| <div class="field"> | |
| <div class="field-label"> | |
| <div class="field-title">Preserve very small text</div> | |
| <div class="field-description"> | |
| Keep tiny text regions that OCR or PDF extraction might otherwise discard. | |
| </div> | |
| </div> | |
| <div class="field-control"> | |
| <label class="toggle"> | |
| <input type="checkbox" x-model="config.liteparse_preserve_very_small_text" /> | |
| <span class="toggler"></span> | |
| </label> | |
| </div> | |
| </div> | |
| </div> | |
| </template> | |
| <div class="section-title">Fallbacks</div> | |
| <div class="section-description"> | |
| Legacy parser behavior used when the preferred parser cannot produce text. | |
| </div> | |
| <div class="field"> | |
| <div class="field-label"> | |
| <div class="field-title">PDF OCR fallback</div> | |
| <div class="field-description"> | |
| Try legacy Tesseract OCR when direct PDF text extraction is empty. | |
| </div> | |
| </div> | |
| <div class="field-control"> | |
| <label class="toggle"> | |
| <input type="checkbox" x-model="config.pdf_ocr_fallback" /> | |
| <span class="toggler"></span> | |
| </label> | |
| </div> | |
| </div> | |
| <div class="field"> | |
| <div class="field-label"> | |
| <div class="field-title">Thread offload</div> | |
| <div class="field-description"> | |
| Run synchronous parser fallbacks in a worker thread. | |
| </div> | |
| </div> | |
| <div class="field-control"> | |
| <label class="toggle"> | |
| <input type="checkbox" x-model="config.thread_offload" /> | |
| <span class="toggler"></span> | |
| </label> | |
| </div> | |
| </div> | |
| </div> | |
| </template> | |
| </div> | |
| </body> | |
| </html> | |