Skip to main content

Selecting what to parse

By default LlamaParse will extract all the visible content of every page of a document



Parsing only some pages

You can specify the pages you want to parse by passing specific page numbers as a comma-separated list in the target_pages argument. Pages are numbered starting at 0.

parser = LlamaParse(
  target_pages="0,1,2,22,33"
)

The range syntax is also supported target_pages=0-2,6-22,33.



Parsing only a targeted area of a document

You can specify an area of a document that you want to parse. This can be helpful to remove headers and footers.

To do so you need to provide the bounding box margin expressed as a ratio compare to the page size between 0 and 1 in bbox_left, bbox_right, bbox_top and bbox_bottom.

Examples:

  • To not parse the top 10% of a document: bbox_top=0.1
  • To not parse the top 10% and bottom 20% of a document: bbbox_top=0.1 and bbox_bottom=0.2,
parser = LlamaParse(
  bbox_left=0.2
)


Bounding box (legacy)

We support a deprecated way of doing so where it is possible to provide the bounding box margin in clockwise order from the top in a comma separated string in the bounding_box arguments. The margins are expressed as a ratio compare to the page size between 0 and 1.

Examples:

  • To not parse the top 10% of a document: bounding_box="0.1,0,0,0"
  • To not parse the top 10% and bottom 20% of a document: bounding_box="0.1,0,0.2,0"
parser = LlamaParse(
  bounding_box="0.1,0.4,0.2,0.3"
)


Limiting number of page to parse

If you want to limit the maximum amount of pages to parse you can use the parameter max_pages. LlamaParse will stop parsing the document after the specified pages.

parser = LlamaParse(
  max_pages=25
)