vis_json_cell.py

import json 
import cv2
import os

def get_visible_cell_coords(table_properties, table_x, table_y):
    rows = table_properties.get("rows", 0)
    columns = table_properties.get("columns", 0)
    column_widths = table_properties.get("columnWidths", {})
    row_heights = table_properties.get("rowHeights", {})
    merged_cells = table_properties.get("mergedCells", {})
    hidden_cells = table_properties.get("hiddenCells", {})

    DEFAULT_WIDTH = 100
    DEFAULT_HEIGHT = 30

    def get_col_width(col):
        return column_widths.get(str(col), DEFAULT_WIDTH)

    def get_row_height(row):
        return row_heights.get(str(row), DEFAULT_HEIGHT)

    # Convert merge cell coordinates to set of all spanned cells (excluding top-left)
    merged_spanned_cells = set()
    for key, merge_info in merged_cells.items():
        base_row, base_col = map(int, key.split('-'))
        rowspan = merge_info.get('rowspan', 1)
        colspan = merge_info.get('colspan', 1)
        for r in range(base_row, base_row + rowspan):
            for c in range(base_col, base_col + colspan):
                if (r, c) != (base_row, base_col):
                    merged_spanned_cells.add((r, c))

    result = {}

    for row in range(rows):
        for col in range(columns):
            coord_key = f"{row}-{col}"
            if hidden_cells.get(coord_key):
                continue  # Skip hidden cells
            if (row, col) in merged_spanned_cells:
                continue  # Skip cells covered by merged cells

            # Calculate x by summing widths of all previous columns
            x = sum(get_col_width(c) for c in range(col))
            y = sum(get_row_height(r) for r in range(row))

            # Check if it's a merged cell origin
            if coord_key in merged_cells:
                colspan = merged_cells[coord_key].get("colspan", 1)
                rowspan = merged_cells[coord_key].get("rowspan", 1)
            else:
                colspan = 1
                rowspan = 1

            width = sum(get_col_width(c) for c in range(col, col + colspan))
            height = sum(get_row_height(r) for r in range(row, row + rowspan))

            result[(row, col)] = {
                "x":  int(x + table_x),
                "y":  int(y + table_y),
                "width":  int(width),
                "height":  int(height)
            }

    return result

folder_path = "/Users/tuvn18/Desktop/tuvn18/dev/KIAI/dev/trace/train_table_1209"
save_folder = "cell_vis"
os.makedirs(save_folder, exist_ok=True)
for name in os.listdir(folder_path):
    if name.endswith("json"):
        json_file = os.path.join(folder_path, name)
        with open(json_file, "r") as f:
            data = json.load(f)
            
        img_name = name.split('.')[0] + ".png"
        print(img_name)
        image = cv2.imread(os.path.join(folder_path,img_name))
        items = data.get('items')
        for index in range(len(items)):

            table = items[index]
            if table.get('type') != 'table':
                continue
            table_x = table.get('x')
            table_y = table.get('y')

            table_prob = table.get('properties')
            boxes = get_visible_cell_coords(table_prob, table_x, table_y)

            for key, box in boxes.items():
                row, col = key
                x, y, w, h = int(box["x"]), int(box["y"]), int(box["width"]), int(box["height"])

                top_left = (x, y)
                bottom_right = (x + w, y + h)

                cv2.rectangle(image, top_left, bottom_right, (0, 255, 0), 2)
                # cv2.putText(image, f"{row},{col}", (x + 5, y + 20),
                #             cv2.FONT_HERSHEY_SIMPLEX, 0.5, (255, 0, 0), 1)

            # Save or show
            cv2.imwrite(os.path.join(save_folder, img_name), image)