edgar-crawler

Why we included this project

Building financial NLP models or datasets usually means wrestling with SEC filings first. The modeling is rarely the bottleneck; getting clean, structured text out of 10-K, 10-Q, and 8-K reports is. This toolkit handles that whole pipeline: it pulls filings from EDGAR for any set of US public companies, lets you filter by year, quarter, and filing type, and parses each document into standardized JSON with the key item sections (business, risk factors, MD&A, financial statements) separated out. The output drops straight into a corpus, a retrieval pipeline, or a fine-tuning run, and the authors used it to build the EDGAR-CORPUS dataset on HuggingFace. Researchers and quant teams who want reproducible, machine-readable financial text without hand-scraping HTML will find the structured extraction a real time-saver. It is a focused data-acquisition and parsing tool rather than an analytics platform, so pair it with your own downstream analysis or model work.