parse_bhel.py

import pdfplumber
import pandas as pd

def preprocess_rows(rows, expected_columns):
    aligned_rows = []
    buffer = []
    unalignable_rows = []  # Capture unaligned rows for inspection

    for row in rows:
        # Check if the row contains irrelevant metadata or headers
        if any(keyword in row[0] for keyword in ["GSTIN", "Currency", "Payment Terms", "General Terms", "Delivery Schedule"]):
            continue

        # If row matches expected length, add directly
        if len(row) == expected_columns:
            if buffer:
                aligned_rows.append(buffer)  # Add any buffered row first
                buffer = []  # Reset buffer
            aligned_rows.append(row)

        # If row contains part of an entry (such as "Material Number" or "HSN Code")
        elif "Material Number" in row[0] or "HSN Code" in row[0] or "IGST" in row[0]:
            if buffer:
                buffer[-1] += " " + row[0]  # Append to last column in buffer
            else:
                buffer = row  # Initialize buffer with this part-row
        else:
            # If unalignable, add to unalignable_rows for debugging
            unalignable_rows.append(row)

    # Log any remaining buffered content
    if buffer:
        aligned_rows.append(buffer)

    # Print unalignable rows for analysis
    for row in unalignable_rows:
        print(f"Unalignable row: {row}")

    return aligned_rows

def parse_bhel_pdf(pdf_path):
    columns = [
        "Purchase Order No", "Date", "Sl No", "Material Description", 
        "Unit", "Quantity", "Dely Qty", "Dely Date", "Unit Rate", "Value"
    ]
    expected_columns = len(columns)
    data = []

    with pdfplumber.open(pdf_path) as pdf:
        for page in pdf.pages:
            table = page.extract_table()
            if table:
                # Preprocess and align rows before DataFrame conversion
                rows = preprocess_rows(table[1:], expected_columns)
                for row in rows:
                    if len(row) == expected_columns:
                        data.append(row)
                    else:
                        print(f"Skipping unalignable row: {row}")

    # Convert aligned rows into a DataFrame
    df = pd.DataFrame(data, columns=columns)
    return df