Get Started

Reader and Extraction

Edit on GitHub

Overview of the OfficeIMO.Reader package for unified document extraction and AI-ready chunking workflows.

Reader and Extraction

OfficeIMO.Reader provides a single extraction surface for Office and adjacent document formats. Instead of maintaining separate parsing pipelines for .docx , .xlsx , .pptx , Markdown, PDF, or text-like files, you can normalize them into one chunk model and then feed that output into indexing, search, and AI workflows.

Best fit scenarios

  • Build ingestion pipelines for RAG, semantic search, or compliance review.
  • Normalize mixed document folders into one extraction and chunking model.
  • Preserve headings, citations, token estimates, and source hashes while preparing content for downstream tools.
  • Run extraction in background workers, containers, Azure Functions, or scheduled jobs.

Core workflow

  1. Extract a file into ReaderChunk instances with text, markdown, tables, visuals, and source information.
  2. Tune ReaderOptions so emitted slices stay deterministic and sized for search or AI prompts.
  3. Store chunks, citations, and source identifiers in your vector store, search index, or audit trail.

Quick start

using OfficeIMO.Reader;

var chunks = DocumentReader.Read("proposal.docx", new ReaderOptions
{
    MaxChars = 4_000,
    IncludeWordFootnotes = true,
    ComputeHashes = true
}).ToList();

foreach (var chunk in chunks)
{
    Console.WriteLine($"{chunk.Id} :: {chunk.Kind}");
    Console.WriteLine(chunk.Location.HeadingPath ?? chunk.Location.Path);
    Console.WriteLine(chunk.TokenEstimate);
}

Formats and behavior

InputTypical use
Word ( .docx )Rich business documents, reports, contracts, and templates
Excel ( .xlsx )Workbook content, tabular reports, and structured exports
PowerPoint ( .pptx )Slide decks, speaker notes, and presentation narratives
MarkdownDocumentation, changelogs, developer notes, and generated content
PDFPublished exports, archival documents, and third-party handoffs

Design goals

  • Deterministic chunking so repeated runs produce stable chunk boundaries.
  • Heading-aware extraction so downstream systems retain document structure.
  • Citation-friendly location data so search and AI responses can reference original sources.
  • Incremental indexing support through source IDs, hashes, and per-document chunk summaries.
  • Container-friendly execution with no Office installation requirements.