Skip to main content
SearchCtrl K
Community

Source code

OfficeIMO.NET libraries, PowerShell module, CLI, Studio and browser tools 700 stars

Talk and help

DiscordQuestions, ideas and show and tell Report an issueBugs and feature requests SponsorFund ongoing development

Mode

Background

Search OfficeIMO

Enter a topic, API type, or PowerShell command.

Search the whole siteOpen search page →
Documentation Search all OfficeIMO/

Extract and Inspect Documents

Edit on GitHub

Extract Markdown or JSON from documents and folders with officeimo read, inspect files as JSON and read workbook schemas.

The extraction commands turn documents into Markdown or structured JSON for search indexes, review tools, diffs and language models. They use OfficeIMO.Reader, which handles Office, OpenDocument, PDF, email and mailboxes, web formats, EPUB, Visio, OneNote, CSV, YAML, TOML and more.

Extract one document

officeimo read and officeimo extract are the same command, a shortcut for officeimo reader read. Output goes to standard output unless you pass --output; an existing output file is refused unless you add --force:

officeimo read report.docx --format markdown
officeimo extract report.docx --format markdown --output report.md
officeimo extract handbook.pdf --format json --output handbook.json --assets handbook-assets

Read from standard input with -, and give the stream a name so the reader can pick the right handler:

cat report.docx | officeimo read - --name report.docx --format markdown

--max-input-bytes bounds the input size. --assets <directory> writes embedded images and other assets to a folder.

Extract a folder

officeimo reader folder ./contracts --output ./contracts-md --format markdown
officeimo reader folder ./archive --output ./archive-json --format json --concurrency 4 --max-files 5000

Each source becomes <file name>.md or <file name>.reader.json in the output folder, for example report.docx.md. Existing files in the output folder are refused unless you add --force. Use --no-recursive, --max-files, --max-total-bytes and --max-input-bytes to bound the run. Apple .pages, .numbers and .key directory packages are read as single documents.

See which formats are supported

officeimo reader capabilities
officeimo reader capabilities --format json

The text form prints one line per handler with its identifier and extensions. The command-line tool does not configure OCR or hosted providers, so scanned PDFs without a text layer return little text; see searchable scans.

Reader JSON

JSON output, from read --format json, reader folder --format json or convert <input> <output>.json for a source without a data route, uses one versioned envelope. For YAML, TOML, XML, CSV and other data files, convert writes converted data instead, so use read --format json for their Reader JSON:

{
  "schemaId": "officeimo.document.read-result",
  "schemaVersion": 15,
  "kind": "Word",
  "source": { "path": "...", "sourceId": "src:...", "sourceHash": "...", "lastWriteUtc": "...", "lengthBytes": 12478 },
  "markdown": "...",
  "chunks": [], "blocks": [], "tables": [], "pages": [], "metadata": [],
  "assets": [], "links": [], "forms": [], "visuals": [], "diagnostics": [], "nestedDocuments": []
}

Check schemaId and schemaVersion before parsing. chunks carry text, Markdown, a location, a token estimate and hashes for incremental indexing. tables carry columns, column profiles and rows, with totalRowCount and truncated when rows were limited. The Reader guide describes the model in detail.

Inspect a file

officeimo inspect returns one compact JSON object, sized for a script or a model context rather than a full extraction:

officeimo inspect deck.pptx
officeimo inspect report.docx --max-output-characters 4000

{
  "contentTrust": "untrusted",
  "contentSafety": { "status": "NotInspected" },
  "sourceId": "officeimo:f03752ed0b321dd92b95d277",
  "kind": "Word",
  "title": "", "author": "...",
  "preview": "Testing Tables on Load ...",
  "chunkCount": 4, "blockCount": 6, "pageCount": 0, "tableCount": 2,
  "metadata": [ { "name": "SectionCount", "value": "1" } ],
  "diagnostics": [],
  "truncated": false
}

inspect is a shortcut for officeimo agent inspect. The other agent commands search and fetch inside documents and mailboxes; see MCP and agents.

Workbook sheets and schemas

officeimo tabular sheets workbook.xlsx
officeimo tabular schema workbook.xlsx --sheet Data
officeimo tabular schema export.csv --delimiter ';'

sheets prints one sheet name per line. schema prints a tab-separated ordinal, name and type table for each column, with types such as string, integer, number, boolean and datetime. Names escape backslashes and control characters as \\, \t, \r, \n or \uXXXX, so each stays on one line and is safe to split in a script.

Mail data

officeimo agent inspect-email message.eml --max-output-characters 6000
officeimo agent search archive.pst --query "renewal date" --take 5

inspect-email reports body alternatives, charsets, attachment metadata, protection and signature status, and an HTML safety pass, without body previews. Mailbox search and fetch are covered in MCP and agents.

Example source