Python for CAs

Real, runnable Python code for real CA problems. File automation, Excel processing, PDF extraction, bulk email, GST validation.

Everything in this section is real code. Not pseudocode. Not simplified examples. You can run every script here right now if you have Python installed.

For each example, we show you:

The problem it solves (the CA scenario)
The code Claude would generate
The prompt you would give Claude Code to get it
How to install any required libraries

Before You Start: Installing Libraries

Python libraries are pre-built toolkits that someone else has already written. You install them once with pip and then they are available to use in any script.

Open the command prompt

Press Windows key + R, type cmd, press Enter.

Install the core CA libraries

Run each of these commands one at a time, pressing Enter after each one and waiting for it to finish before running the next:

pip install openpyxl

pip install pandas

pip install pdfplumber

pip install pypdf

Verify the installs

After all four complete, run pip list and scroll through the output. You should see openpyxl, pandas, pdfplumber, and pypdf listed.

What each library does: openpyxl and pandas work with Excel files. pdfplumber extracts text from PDFs where you can select and copy text. pypdf merges and splits PDF files. These four libraries cover the majority of CA automation tasks.

Install libraries one at a time. If you run all four in one long command and one fails, it's hard to know which one caused the error. One at a time is slower but much easier to debug.

Category 1: File and Folder Automation

Problem: Rename 500 PDF invoices to a standard format

The scenario: Your firm receives invoices from vendors in whatever format they use. Every month you spend hours renaming them to your firm's standard: YYYY-MM_VendorName_InvoiceNum.pdf. One script. Done.

The prompt to give Claude Code: "Write Python that renames all PDF files in a folder. The current filenames are like 'Invoice_2024_HDFC_12345.pdf'. Rename them all to 'YYYY-MM_VendorName_InvoiceNumber.pdf' format, where the year and month come from today's date, the vendor name is everything after 'Invoice_' and before the first underscore after that, and the invoice number is the last number before '.pdf'. Show me a preview before making any changes."

Your Turn — Open Claude Code and paste this prompt exactly. Read the script it produces. Identify: (1) where the folder path is set, (2) where the new filename is constructed, (3) where it asks for your confirmation. These three things appear in almost every file automation script.

Problem: Organize a Downloads folder into subfolders by file type

The scenario: Your Downloads folder is chaos. PDFs, Excel files, images, Word documents all mixed together. One script to organize them all.

Your Turn — Give Claude Code this prompt: "Write Python that looks at every file in my Downloads folder and moves PDFs into a subfolder called 'PDFs', Excel files into 'Spreadsheets', and Word documents into 'Docs'. Create the folders if they don't exist. Show me what it plans to move before actually moving anything." Run the preview. Do not confirm until you've read every line.

Category 2: Excel and Data Automation

Problem: Combine 50 monthly Excel reports into one master sheet

The scenario: Each month, 12 branch offices send you an Excel file with their P&L data. You spend 2 hours every month copying and pasting them into a master sheet. This script does it in 10 seconds.

Your Turn — Give Claude Code this prompt: "Write Python using pandas that reads all Excel files in a folder, combines them into one master sheet with a new column showing which file each row came from, and saves the result as 'master_combined.xlsx'." This prompt is the template for every consolidation task you will ever do.

Problem: Find duplicate entries across multiple Excel files

The scenario: You have client data across 3 different Excel files and suspect there are duplicate entries (same PAN, same GST number, or same company name appearing in more than one file). Find them all.

Your Turn — Give Claude Code this prompt: "Write Python using pandas that reads three Excel files, finds any rows where the value in column 'PAN' appears in more than one file, and exports those duplicates to a new Excel file called 'duplicates.xlsx' with a column showing which files the duplicate appeared in."

Category 3: PDF Processing

Problem: Extract invoice totals from 200 PDF invoices

The scenario: You have 200 vendor invoices as PDFs. You need the invoice date, vendor name, and total amount from each one — exported to Excel for reconciliation. Doing this manually takes a full day.

Important note on PDF extraction: PDFs come in two types — text-based (where you can select and copy text) and image-based (scanned documents where text is a picture). pdfplumber works well on text-based PDFs. For scanned/image PDFs, you need OCR (Optical Character Recognition) — a more complex step. Always test on a sample of your actual PDFs first.

Problem: Merge multiple PDFs into one document

The scenario: You need to combine a client's 12 monthly bank statements (12 separate PDFs) into one PDF for a bank loan application. Doing it manually in Adobe Acrobat costs a license fee and takes time.

Your Turn — Give Claude Code this prompt: "Write Python using pypdf that takes all PDF files in a folder, merges them into one PDF in alphabetical filename order, and saves it as 'merged_statements.xlsx'. If there are fewer than 2 PDFs in the folder, print an error message." The filename-ordering detail prevents the months from being merged out of sequence.

Category 4: Email Automation

Problem: Send 100 personalized client emails from an Excel list

The scenario: Monthly statement dispatch. You have an Excel file with columns: Client Name, Email, Outstanding Amount, Due Date. You need to send each client a personalized email with their specific figures.

Gmail setup required: To send via Gmail, you need to enable "App Passwords" in your Google Account settings (under Security → 2-Step Verification → App Passwords). Do not use your main Google password in this script — create an App Password specifically for this.

Category 5: GST and Compliance

Problem: Validate GST numbers from a client list

The scenario: You receive a vendor list with 500 GST numbers from a client. Before processing GSTR-2A reconciliation, you need to validate that all the GST numbers are in the correct format (15-character alphanumeric with specific structure). Find the invalid ones immediately.

Your Turn — Give Claude Code this prompt: "Write Python that reads an Excel file with a column called 'GST Number', checks each value against the official GST format (2 digit state code + 10 digit PAN + 1 digit entity number + 1 letter Z + 1 check digit), and exports two Excel files: 'valid_gst.xlsx' and 'invalid_gst.xlsx' with the original row data preserved."

Problem: Validate PAN number formats in a list

The scenario: You receive a list of client PAN numbers and need to identify which ones are incorrectly formatted before submitting to the income tax portal.

Your Turn — Give Claude Code this prompt: "Write Python that reads an Excel file with a column called 'PAN', checks each value against the PAN format (5 uppercase letters + 4 digits + 1 uppercase letter), and creates two output files: 'valid_pan.xlsx' and 'invalid_pan.xlsx' with the original row data preserved." Test it with a few deliberately wrong values to confirm the validation catches them.

The Prompt Pattern for Getting Good Code

For every automation task, structure your Claude Code prompt like this:

What files are being read — exact folder path and filename pattern
What transformation to apply — specific logic, not vague goals
What the output should be — exact filename and format
What a preview looks like — ask to show changes before making them
What error handling you want — what to do if a file is missing or a field is blank

The more specific you are about column names, file paths, expected formats, and what "done" looks like — the less back-and-forth you need.

This prompt structure is not Python-specific. It works for any automation task you will ever give to Claude Code — whether it's Python, JavaScript, or any other language. The quality of the output is directly proportional to the specificity of the input.

Next: Your first automation — run it yourself, from scratch.