Struct PageText 

pub struct PageText { /* private fields */ }

A page’s extracted text: the runs in draw order, plus a whitespace-stripped, lowercased index for robust quote matching.

Implementations

impl PageText

pub fn is_empty(&self) -> bool

Whether the page has any extractable text (false for pure scans with no OCR layer — the host should then fall back to area markup).

Examples found in repository

examples/extract_text.rs (line 16)

fn main() {
    let mut args = std::env::args().skip(1);
    let path = args.next().expect("usage: extract_text <file.pdf> [page]");
    let page: usize = args.next().and_then(|s| s.parse().ok()).unwrap_or(0);

    let bytes = Arc::new(std::fs::read(&path).expect("read pdf"));
    let doc = gpui_pdf::parse(bytes).expect("parse pdf");

    match gpui_pdf::extract_page_text(&doc, page) {
        Some(pt) if !pt.is_empty() => {
            println!("--- page {page} text ---");
            println!("{}", pt.text());
        }
        Some(_) => eprintln!("page {page}: no extractable text (scanned image?)"),
        None => eprintln!("no page {page}"),
    }
}

pub fn text(&self) -> String

A readable reconstruction of the page text (spaces inserted on horizontal gaps, newlines on baseline changes). For search / display — locate uses the whitespace-insensitive index instead.

Examples found in repository

examples/extract_text.rs (line 18)

fn main() {
    let mut args = std::env::args().skip(1);
    let path = args.next().expect("usage: extract_text <file.pdf> [page]");
    let page: usize = args.next().and_then(|s| s.parse().ok()).unwrap_or(0);

    let bytes = Arc::new(std::fs::read(&path).expect("read pdf"));
    let doc = gpui_pdf::parse(bytes).expect("parse pdf");

    match gpui_pdf::extract_page_text(&doc, page) {
        Some(pt) if !pt.is_empty() => {
            println!("--- page {page} text ---");
            println!("{}", pt.text());
        }
        Some(_) => eprintln!("page {page}: no extractable text (scanned image?)"),
        None => eprintln!("no page {page}"),
    }
}

pub fn locate(&self, needle: &str, occurrence: usize) -> Vec<NormRect>

Locate the occurrence-th (0-based) case- and whitespace-insensitive match of needle on the page and return one normalized rect per line it spans (so a wrapped quote highlights as multiple line boxes). Empty if not found.

pub fn find_matches(&self, needle: &str) -> Vec<Vec<NormRect>>

Every (non-overlapping) case- and whitespace-insensitive match of needle on the page, each as one normalized rect per line it spans (for find-in-PDF). Matches are returned in reading order. (search feature.)

pub fn select(&self, from: NormPoint, to: NormPoint) -> Option<Selection>

Resolve a drag from from to to into a Selection: the run range between the nearest glyphs (draw order ≈ reading order), its one-line quote, the occurrence index, and the rects to draw. None if there’s no text or the selection is empty.