在这个网站上，验证码可以被解析，而不需要求助于第三方服务。当你点击“验证码音频”按钮时，一个GET请求被发送到端点https://ipindiaservices.gov.in/PublicSearch/Captcha/CaptchaAudio。响应是一个字典{"CaptchaImageText":"hnnxd"}，你可以通过"Chrome Devtools Protocol"使用Network.getResponseBody方法从Selenium获取它，或者你可以使用requests库。要将数据保存为csv格式，例如，您可以使用标准库中包含的csv模块。
以下是一个可能的解决方案：

import re
import csv
import json
from time import sleep
from selenium import webdriver
from typing import Generator, List, Tuple
from selenium.webdriver.common.by import By
from selenium.webdriver import DesiredCapabilities
from selenium.webdriver.chrome.service import Service
from selenium.webdriver.support.ui import WebDriverWait
from selenium.webdriver.chrome.webdriver import WebDriver
from selenium.webdriver.support import expected_conditions as EC

def get_captcha_text(log: dict, driver: WebDriver) -> dict:
    log = json.loads(log["message"])["message"]
    if all([
        "Network.responseReceived" in log["method"], 
        "params" in log.keys(), 
        'CaptchaAudio' in str(log["params"].values())
        ]):
        return driver.execute_cdp_cmd('Network.getResponseBody', {'requestId': log["params"]["requestId"]})

def save_to_csv(table_data: Generator[None, None, List[Tuple[str, str, str, str]]]) -> None:
    for app_num, title, app_date, status in zip(*list(zip(*table_data))):
        data = {
            'Application Number': app_num,
            'Title': title,
            'Application Date': app_date,
            'Status': status
        }
        with open(file='ipindiaservices.csv', mode='a', encoding="utf-8") as f:
            writer = csv.writer(f, lineterminator='\n')
            writer.writerow([data['Application Number'], data['Title'], data['Application Date'], data['Status']])

def start_from_page(page_number: int, driver: WebDriver) -> None:
    driver.execute_script(
    f"""
    document.querySelector('button.next').value = {page_number}; 
    document.querySelector('button.next').click();
    """
    )

def titles_validation(driver: WebDriver) -> None:
    """replace empty title name with '_'"""
    driver.execute_script(
        """
        let titles = document.querySelectorAll('#tableData>tbody>tr>td.title')
        Array.from(titles).forEach((e) => {
            if (!e.textContent.trim()) {
                e.textContent = '_';
            }
        });
        """
    )

options = webdriver.ChromeOptions()
options.add_experimental_option("excludeSwitches", ["enable-automation", "enable-logging"])
capabilities = DesiredCapabilities.CHROME
capabilities["goog:loggingPrefs"] = {"performance": "ALL"}
service = Service(executable_path="path/to/your/chromedriver.exe")
driver = webdriver.Chrome(service=service, options=options, desired_capabilities=capabilities)
wait = WebDriverWait(driver, 10)

# regular expression to search for data in a table
pattern = r'^([0-9A-Z\/\-,]+) (.+)? ([0-9\/]+) (\w+)'

driver.get('https://ipindiaservices.gov.in/PublicSearch/')

# sometimes an alert with an error message("") may appear, so a small pause is used
sleep(1)
driver.find_element(By.CSS_SELECTOR, 'img[title="Captcha  Audio"]').click()
driver.find_element(By.ID, 'TextField6').send_keys('ltd')
# short pause is needed here to write the log, otherwise we will get an empty list
sleep(1)
logs = driver.get_log('performance')
# get request data that is generated when click on the button listen to the text of the captcha
responses = [get_captcha_text(log, driver) for log in logs if get_captcha_text(log, driver)]
# get captcha text
captcha_text = json.loads(responses[0]['body'])['CaptchaImageText']
# enter the captcha text and click "Search" button
driver.find_element(By.ID, 'CaptchaText').send_keys(captcha_text)
driver.find_element(By.CSS_SELECTOR, 'input[name="submit"]').click()

# the page where the search starts
start_from_page(1, driver)

while True:
    # get current page number
    current_page = wait.until(EC.presence_of_element_located((By.CSS_SELECTOR, 'span.Selected'))).text
    # print the current page number to the console
    print(f"Current page: {current_page}")
    # get all fields of the table
    table_web_elements = wait.until(EC.presence_of_all_elements_located((By.CSS_SELECTOR, '#tableData>tbody>tr')))
    # check title name
    titles_validation(driver)
    # get all table data on current page
    table_data = (re.findall(pattern, data.text)[0] for data in table_web_elements)
    # save table data to csv
    save_to_csv(table_data)
    # if the current page is equal to the specified one, then stop the search and close the driver
    if current_page == '3768':
        break
    # click next page
    driver.find_element(By.CSS_SELECTOR, 'button.next').click()

driver.quit()

此解决方案的性能约为每100页280秒。
收集所有数据大约需要2.5-3小时。
因此，添加了在特定页面（默认情况下，这是最后一页）上停止数据收集的功能：

if current_page == '3768':
     break

并从指定页面开始采集数据（默认情况下，这是第一页）：

start_from_page(1, driver)

输出为ipindiaservices.csv

202247057786,NON-AQUEOUS ELECTROLYTE SECONDARY BATTERY,10/10/2022,Published
202247057932,"COMMUNICATION METHOD, APPARATUS AND SYSTEM",10/10/2022,Published
202247057855,POLYOLEFIN RESIN FILM,10/10/2022,Published
202247057853,CEMENT COMPOSITION AND CURED PRODUCT THEREOF,10/10/2022,Published
...

要在Google Collab中使用此解决方案，请按照下列步骤操作：
1.安装Selenium和ChromeDriver

!pip install selenium
!apt-get update
!apt install chromium-chromedriver

1.进行必要的导入

import re
import csv
import json
from time import sleep
from typing import Generator, List, Tuple
from selenium import webdriver
from selenium.webdriver.common.by import By
from selenium.webdriver import DesiredCapabilities
from selenium.webdriver.support.ui import WebDriverWait
from selenium.webdriver.chrome.webdriver import WebDriver
from selenium.webdriver.support import expected_conditions as EC

1.设定选项

options = webdriver.ChromeOptions()
options.add_argument('--headless')
options.add_argument('--no-sandbox')
options.add_argument('--disable-dev-shm-usage')
capabilities = DesiredCapabilities.CHROME
capabilities["goog:loggingPrefs"] = {"performance": "ALL"}
driver = webdriver.Chrome('chromedriver', chrome_options=options, desired_capabilities=capabilities)

其他的一切都保持不变。不要把所有的代码放在一个单元格里。

更新：这是一种可能的解决方案，可从每个应用程序的表中收集所有信息：

import csv
import json
from time import sleep, time
from selenium import webdriver
from selenium.webdriver.common.by import By
from selenium.webdriver import DesiredCapabilities
from selenium.webdriver.chrome.service import Service
from selenium.webdriver.support.ui import WebDriverWait
from selenium.webdriver.chrome.webdriver import WebDriver
from selenium.webdriver.support import expected_conditions as EC
from selenium.common.exceptions import TimeoutException

def save_to_csv(data: list) -> None:
    with open(file='ipindiaservices.csv', mode='a', encoding="utf-8") as f:
        writer = csv.writer(f, lineterminator='\n')
        writer.writerow([*data])

def start_from_page(page_number: int, driver: WebDriver) -> None:
    driver.execute_script(
    f"""
    document.querySelector('button.next').value = {page_number}; 
    document.querySelector('button.next').click();
    """
    )

def titles_validation(driver: WebDriver) -> None:
    """replace empty title name with '_'"""
    driver.execute_script(
        """
        let titles = document.querySelectorAll('input+.tab-pane tr:not(:first-child)>td:last-child')
        Array.from(titles).forEach((e) => {
            if (!e.textContent.trim()) {
                e.textContent = '_';
            }
        });
        """
    )

def get_network_data(log: dict, driver: WebDriver) -> dict:
    log = json.loads(log["message"])["message"]
    if all([
        "Network.responseReceived" in log["method"], 
        "params" in log.keys(), 
        'CaptchaAudio' in str(log["params"].values())
        ]):
        return driver.execute_cdp_cmd('Network.getResponseBody', {'requestId': log["params"]["requestId"]})

def get_captcha_text(driver: WebDriver, timeout: float) -> str:
    """Return captcha text

    Arguments:
        - driver: WebDriver
        - timeout: pause before receiving data from the web driver log
    """
    driver.execute_script(
        """
        // document.querySelector('img[title="Captcha"]').click()
        document.querySelector('img[title="Captcha  Audio"]').click()
        """
        )
    sleep(timeout)
    logs = driver.get_log('performance')
    if responses := [get_network_data(log, driver) for log in logs if get_network_data(log, driver)]:
        return json.loads(responses[0]['body'])['CaptchaImageText']
    else:
        get_captcha_text(driver, timeout)

def submit_captcha(captcha_text: str, btn_name: str) -> None:
    """Submit captcha

    Arguments:
        - btn_name: captcha send button name["submit" or "search"]
    """
    if btn_name == 'search':
        captcha_locator = (By.CSS_SELECTOR, 'input[name="submit"]')
    elif btn_name == 'submit':
        captcha_locator = (By.ID, 'btnSubmit')
    wait.until(EC.visibility_of_element_located((By.ID, 'CaptchaText'))).send_keys(captcha_text)
    wait.until(EC.visibility_of_element_located(captcha_locator)).click()

options = webdriver.ChromeOptions()
options.add_argument('--headless')
options.add_experimental_option("excludeSwitches", ["enable-automation", "enable-logging"])
capabilities = DesiredCapabilities.CHROME
capabilities["goog:loggingPrefs"] = {"performance": "ALL"}
service = Service(executable_path="path/to/your/chromedriver.exe")
driver = webdriver.Chrome(service=service, options=options, desired_capabilities=capabilities)
wait = WebDriverWait(driver, 15)

table_values_locator = (By.CSS_SELECTOR, 'input+.tab-pane tr:not(:first-child)>td:last-child')
applicant_name_locator = (By.ID, 'TextField6')
page_number_locator = (By.CSS_SELECTOR, 'span.Selected')
app_satus_locator = (By.CSS_SELECTOR, 'button.btn')
next_btn_locator = (By.CSS_SELECTOR, 'button.next')

driver.get('https://ipindiaservices.gov.in/PublicSearch/')

# sometimes an alert with an error message("") may appear, so a small pause is used
sleep(1)
wait.until(EC.visibility_of_element_located(applicant_name_locator)).send_keys('ltd')
# on the start page and the page with the table, the names of the buttons are different
captcha_text = get_captcha_text(driver, 1)
submit_captcha(captcha_text, "search")
# the page where the search starts
start_from_page(1, driver)

while True:
    start = time()
    # get current page number
    current_page = wait.until(EC.visibility_of_element_located(page_number_locator)).text
    print(f"Current page: {current_page}")
    # get all application status WebElements
    app_status_elements = wait.until(EC.visibility_of_all_elements_located(app_satus_locator))

    for element in range(len(app_status_elements)):
        print(f"App number: {element}")
        # update application status WebElements
        app_status_elements = wait.until(EC.visibility_of_all_elements_located(app_satus_locator))
        # click on application status
        wait.until(EC.visibility_of(app_status_elements[element])).click()
        # wait 2 seconds for the captcha to change
        sleep(2)
        # get text and submit captcha
        captcha_text = get_captcha_text(driver, 1)
        submit_captcha(captcha_text, "submit")
        try:
            # get all table data values(without titles) WebElements
            table_data_values = wait.until(EC.visibility_of_all_elements_located(table_values_locator))
            # if there are empty rows in the table replace them with "_"
            titles_validation(driver)
            # save data to csv
            save_to_csv([val.text.replace('\n', ' ') for val in table_data_values])
        except TimeoutException:
            print("Application Number does not exist")
        finally:
            driver.back()
    # print the current page number to the console
    print(f"Time per page: {round(time()-start, 3)}")
    # if the current page is equal to the specified one, then stop the search and close the driver
    if current_page == '3776':
        break
    # click next page
    wait.until(EC.visibility_of_element_located(next_btn_locator)).click()

driver.quit()

输出为ipindiaservices.csv

202247059447,PCT NATIONAL PHASE APPLICATION,18/10/2022,"PANASONIC INTELLECTUAL PROPERTY MANAGEMENT CO., LTD.",SOLID ELECTROLYTE MATERIAL AND BATTERY USING SAME,ELECTRICAL,patent@depenning.com,_,_,PCT/JP2021/004427,05/02/2021,31/03/2020,--,21/10/2022
202247059470,PCT NATIONAL PHASE APPLICATION,18/10/2022,"SHENZHEN SKYWORTH-RGB ELECTRONIC CO., LTD.","ATTACHMENT FORCE ADJUSTMENT METHOD AND APPARATUS, DEVICE, AND STORAGE MEDIUM",COMPUTER SCIENCE,ipo@knspartners.com,_,_,PCT/CN2020/125555,30/10/2020,02/04/2020,18/10/2022,21/10/2022
202247058733,PCT NATIONAL PHASE APPLICATION,14/10/2022,"SUMITOMO ELECTRIC OPTIFRONTIER CO., LTD.","FUSION SPLICING SYSTEM, FUSION SPLICING DEVICE, AND DETERIORATION DETERMINATION METHOD",COMPUTER SCIENCE,patent@depenning.com,_,_,PCT/JP2021/017016,28/04/2021,30/04/2020,--,21/10/2022

此解决方案的性能约为每页230sec。
有时，应用程序状态页上可能没有数据（例如，通过编号“00054/CAL/1998”，我们会得到“Application Number does not exist”（应用程序编号不存在））。因此，脚本会忽略此应用程序。
2 sec timeout before receiving the captcha text is due to the fact that after clicking on "Application Status" one captcha is shown and after ~1sec it changes to another one which we must enter

修复：由于点击Application status后出现的验证码已从站点中删除，其脚本中的解决方案也被删除。此外，此资源的开发人员为Application Number和Application status创建了相同的类，因此我们需要更改Application status的css选择器。Google Colab的解决方案：

import csv
import json
from time import sleep, time
from selenium import webdriver
from selenium.webdriver.common.by import By
from selenium.webdriver import DesiredCapabilities
from selenium.webdriver.chrome.service import Service
from selenium.webdriver.support.ui import WebDriverWait
from selenium.webdriver.chrome.webdriver import WebDriver
from selenium.webdriver.support import expected_conditions as EC
from selenium.common.exceptions import TimeoutException

def save_to_csv(data: list) -> None:
    with open(file='ipindiaservices.csv', mode='a', encoding="utf-8") as f:
        writer = csv.writer(f, lineterminator='\n')
        writer.writerow([*data])

def start_from_page(page_number: int, driver: WebDriver) -> None:
    driver.execute_script(
    f"""
    document.querySelector('button.next').value = {page_number}; 
    document.querySelector('button.next').click();
    """
    )

def titles_validation(driver: WebDriver) -> None:
    """replace empty title name with '_'"""
    driver.execute_script(
        """
        let titles = document.querySelectorAll('input+.tab-pane tr:not(:first-child)>td:last-child')
        Array.from(titles).forEach((e) => {
            if (!e.textContent.trim()) {
                e.textContent = '_';
            }
        });
        """
    )

def get_network_data(log: dict, driver: WebDriver) -> dict:
    log = json.loads(log["message"])["message"]
    if all([
        "Network.responseReceived" in log["method"], 
        "params" in log.keys(), 
        'CaptchaAudio' in str(log["params"].values())
        ]):
        return driver.execute_cdp_cmd('Network.getResponseBody', {'requestId': log["params"]["requestId"]})

def get_captcha_text(driver: WebDriver, timeout: float) -> str:
    """Return captcha text

    Arguments:
        - driver: WebDriver
        - timeout: pause before receiving data from the web driver log
    """
    driver.execute_script(
        """
        // document.querySelector('img[title="Captcha"]').click()
        document.querySelector('img[title="Captcha  Audio"]').click()
        """
        )
    sleep(timeout)
    logs = driver.get_log('performance')
    responses = [get_network_data(log, driver) for log in logs if get_network_data(log, driver)]
    if responses:
        return json.loads(responses[0]['body'])['CaptchaImageText']
    else:
        get_captcha_text(driver, timeout)

def submit_captcha(captcha_text: str, btn_name: str) -> None:
    """Submit captcha

    Arguments:
        - btn_name: captcha send button name["submit" or "search"]
    """
    if btn_name == 'search':
        captcha_locator = (By.CSS_SELECTOR, 'input[name="submit"]')
    elif btn_name == 'submit':
        captcha_locator = (By.ID, 'btnSubmit')
    wait.until(EC.visibility_of_element_located((By.ID, 'CaptchaText'))).send_keys(captcha_text)
    wait.until(EC.visibility_of_element_located(captcha_locator)).click()

options = webdriver.ChromeOptions()
options.add_argument('--headless')
options.add_argument('--no-sandbox')
options.add_argument('--disable-dev-shm-usage')
options.add_experimental_option("excludeSwitches", ["enable-automation", "enable-logging"])
capabilities = DesiredCapabilities.CHROME
capabilities["goog:loggingPrefs"] = {"performance": "ALL"}
driver = webdriver.Chrome(executable_path='chromedriver', options=options, desired_capabilities=capabilities)
wait = WebDriverWait(driver, 10)

table_values_locator = (By.CSS_SELECTOR, 'input+.tab-pane tr:not(:first-child)>td:last-child')
applicant_name_locator = (By.ID, 'TextField6')
page_number_locator = (By.CSS_SELECTOR, 'span.Selected')
app_satus_locator = (By.CSS_SELECTOR, 'button[name="ApplicationSatus"]')
next_btn_locator = (By.CSS_SELECTOR, 'button.next')

driver.get('https://ipindiaservices.gov.in/PublicSearch/')

# sometimes an alert with an error message("") may appear, so a small pause is used
sleep(1)
wait.until(EC.visibility_of_element_located(applicant_name_locator)).send_keys('ltd')
# on the start page and the page with the table, the names of the buttons are different
captcha_text = get_captcha_text(driver, 1)
submit_captcha(captcha_text, "search")
# the page where the search starts
start_from_page(1, driver)

while True:
    start = time()
    # get current page number
    current_page = wait.until(EC.visibility_of_element_located(page_number_locator)).text
    print(f"Current page: {current_page}")
    # get all application status WebElements
    app_status_elements = wait.until(EC.visibility_of_all_elements_located(app_satus_locator))

    for element in range(len(app_status_elements)):
        print(f"App number: {element}")
        # update application status WebElements
        app_status_elements = wait.until(EC.visibility_of_all_elements_located(app_satus_locator))
        # click on application status
        wait.until(EC.visibility_of(app_status_elements[element])).click()
        try:
            # switch to new tab
            driver.switch_to.window(driver.window_handles[-1])
            # get all table data values(without titles) WebElements
            table_data_values = wait.until(EC.visibility_of_all_elements_located(table_values_locator))
            # if there are empty rows in the table replace them with "_"
            titles_validation(driver)
            # save data to csv
            save_to_csv([val.text.replace('\n', ' ') for val in table_data_values])
        except TimeoutException:
            print("Application Number does not exist")
        finally:
            # close new tab
            driver.close()
            # switch to main tab
            driver.switch_to.window(driver.window_handles[0])
    # print the current page number to the console
    print(f"Time per page: {round(time()-start, 3)}")
    # if the current page is equal to the specified one, then stop the search and close the driver
    if current_page == '100':
        break
    # click next page
    wait.until(EC.visibility_of_element_located(next_btn_locator)).click()

driver.quit()

这些修正已经在Google Colab（python 3.7.15）上测试过了。这个解决方案的性能大约是每页55sec。
保留以前版本的脚本，以便进行直观比较。

你可以使用任何capcta解决网站。在这些网站上，用户通常自己修复它，所以他们可能很慢，但它确实起到了作用。
Sample website (I didn't receive any ads)
您可以使用selenium来提取信息。在站点上提取和放置带有“id”标签的元素就足够了。Library for reading/writing excel in python