Page Ranges

Page selection

Extractors, classifiers, and splitters let you restrict which pages of the source document are read and billed. Use either pageRanges or fixedPageLimit, preferring pageRanges, never both.

pageRanges

1
"advancedOptions": {
2
  "pageRanges": [
3
    { "start": 1,  "end": 3 },
4
    { "start": 10, "end": 10 }
5
  ]
6
}

• 1-based, inclusive page numbers
• Ranges can overlap or arrive out of order; the platform merges and sorts them automatically
• Omit the field or pass [] to process the full document (subject to global limits)

fixedPageLimit

1
"advancedOptions": {
2
  "fixedPageLimit": 25
3
}

Equivalent to pageRanges: [{ "start": 1, "end": 25 }]. All rules that apply to pageRanges also apply to the synthetic range generated from fixedPageLimit.

Chunking behavior

pageRanges (or the range derived from fixedPageLimit) is applied before chunking starts:

1. Selected pages are isolated and renumbered from 1
2. Chunking operates on this renumbered subset
3. Chunk-level options such as chunkingStrategy and pageChunkSize refer to the virtual page numbers

Example:

1
"pageRanges": [{ "start": 1, "end": 13 }, { "start": 20, "end": 28}],
2
"numPagesPerChunk": 5

Resulting chunks:

Best practices

• Prefer pageRanges over fixedPageLimit
• Use pageRanges for non-contiguous or trailing pages
• Remember that chunking operates on virtual page numbers after filtering