haseeb-140/bloblens

BlobLens is a self-hosted search tool for Azure Blob Storage, the file storage service in Microsoft's cloud. It lets you search not just file names, but the actual text inside PDF and Word documents stored across your storage containers. The project positions itself as a free alternative to Azure's own AI Search product, which the README says can cost from about 75 to 250 dollars or more per month, while BlobLens runs on a small server for no extra cost. Setting it up means cloning the project, adding your Azure storage connection string to a configuration file, and running a single Docker command. Once started, a background worker begins scanning your storage containers and extracting text from supported file types, including PDFs, Word documents, and many plain text and code file formats. That extracted text, along with details like file name, size, and last modified date, gets sent to Meilisearch, a fast open source search engine, which powers a search-as-you-type web interface at localhost port 8000. To keep repeated scans cheap, the worker only re-downloads files that changed since the last pass, tracking this using each container's last modified timestamps. Very large files skip text extraction and are indexed by metadata only. When you want to open a file, the app generates a temporary, time-limited download link rather than storing or proxying the file itself. The README is clear that BlobLens has no login system of its own and is meant to run inside a private network, behind a VPN or your own authentication layer. It also lists planned future features, including automatic detection and redaction of sensitive information like emails and ID numbers before indexing. The project is released under the MIT license.