Reading Structure

Pulling out just the outline

Lesson 3 iterated every item in reading order. Sometimes you don't want every item, you want the document's outline: what are the section headings, in order, with body text stripped out. Because doc.texts items carry a .label, this is a filter, not a parser:

from docling_core.types.doc import DocItemLabel

headings = [
    t for t in doc.texts
    if t.label in (DocItemLabel.TITLE, DocItemLabel.SECTION_HEADER)
]

DocItemLabel is the enum backing every .label value seen so far (title, section_header, text, and more used in later lessons). Importing it directly, instead of comparing against the raw string "section_header", keeps this code correct even if the underlying string values ever change across docling versions.

export_to_text()

export_to_markdown() produces Markdown syntax, # for headings, | for tables. export_to_text() produces the same reading-order content without any of that, just plain paragraphs and simple - bullets. Reach for it when the destination can't handle Markdown at all, an older full-text search index, a plain log file, a system that would choke on stray # characters.