Document Archive & Search (PDFs and Scans)

About the Document Archive & Search (PDFs and Scans)

Document Archive & Search builds a private, searchable archive of a folder on your own device. Open the folder — PDFs, Word, Excel and PowerPoint files, OpenDocument files, text, e-mails, scans and photos, in any subfolders, up to 20,000 files — and every file is read in your browser: the text of PDFs and Office files, scans and photos with OCR in English, Hindi and 11 more languages, a SHA-256 fingerprint that finds exact duplicates, and the details business papers carry, found by rules: the document date, amounts and total, invoice and purchase-order numbers, GSTINs (checked), PANs and e-mail addresses. Each file gets its kind (invoice, receipt, bank statement, payslip, contract, tax document and more).

Add your own fields — client, project, kind, status and a retain-until date — and tags, one file at a time or many together. Then search the words inside the files and the details together (“acme client:acme type:invoice fy:2025-26 amount>10000”), with the matching passages marked, save the searches you use as views, and list the files whose retention period is over, by the rules of the Data Retention Schedule Generator for India, the UK, the US and the EU. Nothing is uploaded: the archive stays in this browser, a new scan picks up new and changed files, the whole archive can be saved as one encrypted file, and the catalogue exports as CSV or Excel. Without a pass the free preview indexes and searches the first 50 files.

Common uses

• Finding an old invoice, contract or statement by a word, a number or an amount in seconds.
• Preparing for an audit or a GST or tax filing: every invoice of the financial year with its total.
• Cleaning up shared drives: duplicates found by fingerprint, and records past their retention date listed.
• Running a small office, a clinic or a practice paperless, with client and project fields instead of folders.
• Keeping a private, encrypted index of sensitive papers that never goes to a cloud service.

What is read, and how

• PDFs: the text layer of every page (pdf.js); pages without one are scans, read with OCR when you ask.
• Word, Excel and PowerPoint (.docx, .xlsx, .pptx) and OpenDocument (.odt, .ods, .odp): the document text, every sheet’s cells, the slides and their notes, and the title, author and dates in the file’s properties.
• Text, CSV, Markdown, JSON, XML, HTML, RTF and e-mails (.eml, with their subject, sender and date).
• Photos and scans (JPEG, PNG, WebP, TIFF, HEIC): OCR with Tesseract, on this device.
• Password-protected files and old .doc, .xls and .ppt files are listed by name, size and fingerprint.

Details found by rules

Dates in the common forms, with a numeric date such as 03/04/2025 read the way your country writes dates (day first in India, Europe and most of the world; month first in the United States), amounts in rupees, dollars, euros and pounds with Indian or Western grouping, the total on a “Total”, “Grand total” or “Amount due” line, invoice and PO numbers after their labels, GSTINs only when their check character is right (the GSTN’s Luhn mod 36), PANs of the right pattern and e-mail addresses. The kind of each file comes from weighted words in its name and text. Nothing is guessed by AI, and you can correct any of it.

Retention dates

For invoices, receipts and orders, bank statements and tax papers, payslips, contracts and CVs, the retain-until date follows the presets of the Data Retention Schedule Generator for the rules you choose, which start as your country’s — for India, books of account and their vouchers for 8 years from the end of the financial year (Companies Act, 2013, s.128(5)). Rules that count from the end of a financial year use the month you say your financial year ends in (March for India, December elsewhere, until you change it); UK payroll records count from the end of the tax year on 5 April.

A rule that counts from an event the file can’t show — the end of a contract or of an employment, the filing of a tax return — gives only the earliest date the file may go, counted from the document’s date, and such files are never listed as past their retention date: set their date by hand once you know it. A period the presets give without a legal source says it is a common choice, not a rule of law. A date you set by hand always wins. These are starting points, not legal advice.

Limitations

• OCR quality depends on the scan: clean pages at 200 to 300 dpi read well; photos taken at an angle or in poor light less so.
• The rules find details in the forms listed above; unusual layouts may need a correction by hand.
• Opening the folder again on a later visit and scanning it needs a Chromium browser on a computer; elsewhere, choose the folder again.
• The archive lives in this browser: clearing the site’s data removes it (save it as an encrypted file to keep a copy).

Privacy

On this device; its files download once

Everything happens in your browser. What you enter or open here is not uploaded or stored by MySmartCoPilot. Reading scans and photos (OCR) downloads the recognition engine and the data for the chosen language from this site the first time. Your files are never uploaded.

Open-source components

This page can load these open-source components, each only when a task needs it:

• It uses libheif and libde265 under the LGPL v3, through heic-to: licence and source code.

You may replace them with your own builds. Their licences and the credits of every library the site uses are on the open-source credits page.