adejumotosin/africa-financials-extraction

Automated pipeline for extracting and structuring financial statements of African listed companies. # Africa Financials Extraction A financial data engineering project for turning African listed-company reports into structured, analysis-ready financial metrics. The current prototype combines digital PDF parsing, table extraction, OCR fallback, metric matching, balance-sheet validation, and CSV/Excel export. ## Problem Financial statements across African markets are frequently published as PDFs with inconsistent layouts, scanned pages, and exchange-specific reporting formats. This makes cross-company screening and analysis difficult to automate. This project explores a reusable extraction pipeline that converts those reports into structured data for analysts, researchers, and downstream financial applications. ## Current pipeline ```text Annual / quarterly report PDF | v PyMuPDF text blocks | +------> Camelot table extraction | +------> Tesseract OCR fallback | v Metric matching and normalization | v Balance-sheet consistency check | v CSV + multi-sheet Excel output ``` ## Metrics currently targeted - Total revenue - Net income / profit after tax - Total assets - Total liabilities - Shareholders' equity The extractor searches common reporting-language variants such as revenue, turnover, sales, net income, profit after tax, and equity. ## Implemented components - Digital PDF block extraction with PyMuPDF - Table extraction with Camelot - OCR fallback with Tesseract and pdfplumber - Regex-based financial metric matching - CSV output for extracted metrics - Multi-sheet Excel export containing metrics, text blocks, tables, and summary data - Basic accounting-equation validation for assets versus liabilities plus equity - Sample financial reports and structured output files for experimentation ## Run locally ```bash git clone github.com cd africa-financials-extraction python -m venv .venv ``` Activate the environment: ```bash # macOS / Linux source .venv/bin/activate # Windows .venv\Scripts\activate …