python 提取 pdf 表单域中的值-k8凯发天生赢家

pip install spire.pdf

from spire.pdf.common import *
from spire.pdf import *
inputfile = "pdf表单.pdf"
outputfile = "获取表单数据.txt"
# 创建pdfdocument对象
pdf = pdfdocument()
# 加载pdf文件
pdf.loadfromfile(inputfile)
# 获取pdf表单
pdfform = pdf.form
formwidget = pdfformwidget(pdfform)
sb = []
# 遍历所有表单域
if formwidget.fieldswidget.count > 0:
    for i in range(formwidget.fieldswidget.count):
        field = formwidget.fieldswidget.get_item(i)
        # 获取文本框的名称和值
        if isinstance(field, pdftextboxfieldwidget):
            textboxfield = field if isinstance(field, pdftextboxfieldwidget) else none
            name = textboxfield.name
            value = textboxfield.text
            sb.append("textbox name: "   name   "\r")
            sb.append("textbox name "   value   "\r\n")
        # 获取列表框的名称 
        if isinstance(field, pdflistboxwidgetfieldwidget):
            listboxfield = field if isinstance(field, pdflistboxwidgetfieldwidget) else none
            name = listboxfield.name
            sb.append("listbox name: "   name   "\r")
            # 获取列表框中的项目
            sb.append("listbox items: \r")
            items = listboxfield.values
            for i in range(items.count):
                item = items.get_item(i)
                sb.append(item.value   "\r")
            # 获取列表框的选中项  
            selectedvalue = listboxfield.selectedvalue
            sb.append("listbox selected value: "   selectedvalue   "\r\n")
        
        # 获取组合框的名称
        if isinstance(field, pdfcomboboxwidgetfieldwidget):
            comboxfield = field if isinstance(field, pdfcomboboxwidgetfieldwidget) else none
            name = comboxfield.name
            sb.append("combobox name: "   name   "\r");
            # 获取组合框中的项目
            sb.append("combobox items: \r");
            items = comboxfield.values
            for i in range(items.count):
                item = items.get_item(i)
                sb.append(item.value   "\r")
            
            # 获取组合框的选中项
            selectedvalue = comboxfield.selectedvalue
            sb.append("combobox selected value: "   selectedvalue   "\r\n")
        
        # 获取单选按钮的名称和选中项
        if isinstance(field, pdfradiobuttonlistfieldwidget):
            radiobtnfield = field if isinstance(field, pdfradiobuttonlistfieldwidget) else none
            name = radiobtnfield.name
            selectedvalue = radiobtnfield.selectedvalue
            sb.append("radio button name: "   name   "\r");
            sb.append("radio button selected value: "   selectedvalue   "\r\n")
       
       # 获取复选框的名称和状态
        if isinstance(field, pdfcheckboxwidgetfieldwidget):
            checkboxfield = field if isinstance(field, pdfcheckboxwidgetfieldwidget) else none
            name = checkboxfield.name
            sb.append("checkbox name: "   name   "\r")
            
            state = checkboxfield.checked
            statevalue = "yes" if state else "no"
            sb.append("if the checkbox is checked: "   statevalue   "\r\n")
# 将获取的数据写入txt文本文件
f2=open(outputfile,'w', encoding='utf-8')
for item in sb:
        f2.write(item)
f2.close()
pdf.close()