1。 从表格下载图片
def download_excel_images(excel_path, url_path, output_dir):
# 设置路径
os.makedirs(output_dir, exist_ok=True)
position_log = []
res = []
# 方法2:用openpyxl记录单元格位置(需额外安装 pip install openpyxl)
wb = load_workbook(excel_path)
sheet = wb.active
url_matrix = {}
for i, img in enumerate(sheet._images):
# 获取图片位置锚点
anchor = img.anchor
row_idx = anchor._from.row # 0-based
col_idx = anchor._from.col # 0-based
top_left_cell = f"{chr(65+anchor._from.col)}{anchor._from.row+1}"
img_name = f"{sheet.title}_img_{i}.png"
with open(os.path.join(output_dir, img_name), "wb") as f:
f.write(img.ref.getvalue())
position_log.append(f"{img_name} | 位置: {sheet.title}!{top_left_cell}")
img_data = Image.open(img.ref)
img_url = upload_img(img_data)
# 按行号分组,每行是一个dict
if row_idx not in url_matrix:
url_matrix[row_idx] = {}
url_matrix[row_idx][col_idx] = img_url
# pdb.set_trace()
for row in sorted(url_matrix.keys()):
row_urls = [url_matrix[row].get(col, "") for col in range(3)]
res.append(row_urls)
df = pd.DataFrame(res, columns=["图片1_url", "图片2_url", "图片3_url"])
df.to_excel(url_path, index=False)
2. 通过图片链接将图片插入表格
def insert_img2excel(excel_path, url_path, output_dir):
data = pd.read_excel(excel_path)
# 创建一个工作簿
wb = Workbook()
# 获取默认的活动工作表
ws = wb.active
ws.column_dimensions['C'].width = 100
i = 2
for index, row in data.iterrows():
try:
response = requests.get(row["merge_image_url2"])
img = response.content
image_path = os.path.join(output_dir, hashlib.md5(img).hexdigest()[:5] + ".png")
# with open(image_path, 'wb') as f:
# f.write(img)
# pdb.set_trace()
except Exception as e:
continue
img = Image.open(BytesIO(img))
img.save(image_path)
image_data = image.Image(image_path)
image_data.width, image_data.height = 100, 100
ws.add_image(image_data, "C" + str(i))
ws.row_dimensions[i].height = 100
image_path = image_path
i = i + 1
wb.save(url_path)