Skip to main content
Type: partition Subtype: unstructured_api For more information about how the various partitioning strategies balance output quality, speed, and cost, see Partitioning.

​
Usage guidance

Use this strategy for documents with tables, complex layouts, or scanned pages, and for any workflow that includes enrichment nodes (such as Image Description, Table to HTML, or Generative OCR), which require per-block image crops that only this strategy produces. Enable pdf_infer_table_structure to generate metadata.text_as_html for table elements, which is required by the Chunk by Table Merging chunker. Set extract_image_block_types to include only the element types your downstream enrichment nodes will process, which reduces output size.

​
Settings

​
strategy
string
required
Partitioning strategy to use. Must be set to hi_res.
​
include_page_breaks
boolean
If true, includes page breaks in the output where supported by the file type. Default: false.
​
pdf_infer_table_structure
boolean
If true, any Table elements extracted from a PDF include an additional text_as_html metadata field containing the table data as an HTML table. Default: false.
​
exclude_elements
array
List of Unstructured element types to exclude from the output. Default: none. Allowed values:
  • Address
  • EmailAddress
  • FigureCaption
  • Footer
  • Formula
  • Header
  • Image
  • ListItem
  • NarrativeText
  • PageBreak
  • Table
  • Title
  • UncategorizedText
​
xml_keep_tags
boolean
If true, retains XML tags in the output. If false, extracts only the text content from XML tags. Default: false.
​
encoding
string
Encoding method used to decode the text input. Default: utf-8.
​
ocr_languages
array
Languages present in the input, for use in partitioning, OCR, or both. Multiple values indicate the text could be in any of the specified languages. Default: ['eng']. See the language codes list.
​
extract_image_block_types
array
Unstructured element types for which image blocks are extracted as Base64-encoded data and stored in metadata fields. Default: none. Allowed values:
  • Abstract
  • BulletedText
  • Caption
  • CodeSnippet
  • CompositeElement
  • Figure
  • FigureCaption
  • Form
  • FormKeysValues
  • Formula
  • Header
  • Image
  • List
  • List-item
  • ListItem
  • NarrativeText
  • Paragraph
  • Picture
  • Table
  • Text
  • Threading
  • Title
  • UncategorizedText
​
infer_table_structure
boolean
If true, any table elements extracted from a PDF include an additional text_as_html metadata field containing an HTML <table> representation. Default: false.
​
coordinates
boolean
If true, each element extracted from a PDF includes position information relative to its page. Default: false.
high_res_paritioner_workflow_node = WorkflowNode(
    name="Partitioner",
    subtype="unstructured_api",
    type="partition",
    settings={
        "strategy": "hi_res",
        "include_page_breaks": <True|False>,
        "pdf_infer_table_structure": <True|False>,
        "exclude_elements": [
            "<element-name>",
            "<element-name>"
        ],
        "xml_keep_tags": <True|False>,
        "encoding": "<encoding>",
        "ocr_languages": [
            "<language>",
            "<language>"
        ],
        "extract_image_block_types": [
            "image",
            "table"
        ],
        "infer_table_structure": <True|False>,
        "coordinates": <True|False>
    }
)

{
    "name": "Partitioner",
    "type": "partition",
    "subtype": "unstructured_api",
    "settings": {
        "strategy": "hi_res",
        "include_page_breaks": <true|false>,
        "pdf_infer_table_structure": <true|false>,
        "exclude_elements": [
            "<element-name>",
            "<element-name>"
        ],
        "xml_keep_tags": <true|false>,
        "encoding": "<encoding>",
        "ocr_languages": [
            "<language>",
            "<language>"
        ],
        "extract_image_block_types": [
            "image",
            "table"
        ],
        "infer_table_structure": <true|false>,
        "coordinates": <true|false>
    }
}