Struct InvoiceExtractor 

pub struct InvoiceExtractor { /* private fields */ }

Invoice data extractor with configurable pattern matching

This is the main entry point for invoice extraction. Use the builder pattern to configure language, confidence thresholds, and other options.

Examples

use oxidize_pdf::text::invoice::InvoiceExtractor;

// Spanish invoices with high confidence threshold and kerning-aware spacing
let extractor = InvoiceExtractor::builder()
    .with_language("es")
    .confidence_threshold(0.85)
    .use_kerning(true)  // Enables font-aware spacing in text reconstruction
    .build();

Thread Safety

InvoiceExtractor is immutable after construction and can be safely shared across threads. Consider creating one extractor per language and reusing it.

Implementations

impl InvoiceExtractor

pub fn builder() -> InvoiceExtractorBuilder

Create a new builder for configuring the extractor

This is the recommended way to create an InvoiceExtractor.

Examples

use oxidize_pdf::text::invoice::InvoiceExtractor;

let extractor = InvoiceExtractor::builder()
    .with_language("es")
    .confidence_threshold(0.8)
    .build();

pub fn extract(&self, text_fragments: &[TextFragment]) -> Result<InvoiceData>

Extract structured invoice data from text fragments

This is the main extraction method. It processes text fragments from a PDF page and returns structured invoice data with confidence scores.

Process

1. Text fragments are reconstructed into full text
2. Language-specific patterns are applied
3. Matches are converted to typed fields
4. Confidence scores are calculated
5. Low-confidence fields are filtered out

Arguments

• text_fragments - Text fragments extracted from PDF page (from TextExtractor)

Returns

Returns Ok(InvoiceData) with extracted fields, or Err if:

• No text fragments provided
• PDF page is empty
• Text extraction failed

Examples

use oxidize_pdf::text::extraction::{TextExtractor, ExtractionOptions};
use oxidize_pdf::text::invoice::InvoiceExtractor;
use oxidize_pdf::Document;

let doc = Document::open("invoice.pdf")?;
let page = doc.get_page(1)?;

// Extract text
let text_extractor = TextExtractor::new();
let extracted = text_extractor.extract_text(&doc, page, &ExtractionOptions::default())?;

// Extract invoice data
let extractor = InvoiceExtractor::builder()
    .with_language("es")
    .build();

let invoice = extractor.extract(&extracted.fragments)?;

// Access extracted fields
for field in &invoice.fields {
    println!("{}: {:?} (confidence: {:.2})",
        field.field_type.name(),
        field.field_type,
        field.confidence
    );
}

Performance

Extraction is CPU-bound and typically completes in <100ms for standard invoices. The extractor can be safely reused across multiple pages and threads.

pub fn extract_from_text(&self, text: &str) -> Result<InvoiceData>

Extract invoice data from plain text (convenience method for testing)

This is a convenience wrapper around extract() that creates synthetic TextFragment objects from plain text input. Primarily useful for testing and simple scenarios where you don’t have actual PDF text fragments.

Note: This method creates fragments without position information, so proximity-based scoring may be less accurate than with real PDF fragments.

Arguments

• text - Plain text string to extract invoice data from

Returns

Returns Ok(InvoiceData) with extracted fields, or Err if text is empty

Examples

use oxidize_pdf::text::invoice::InvoiceExtractor;

let extractor = InvoiceExtractor::builder()
    .with_language("en")
    .confidence_threshold(0.7)
    .build();

let invoice_text = "Invoice Number: INV-001\nTotal: £100.00";
let result = extractor.extract_from_text(invoice_text)?;

assert!(!result.fields.is_empty());