Function get_links

instapy/like_util.py:851–919

(browser, page, logger, media, element)

def get_links(browser, page, logger, media, element):
    links = []
    post_href = None

    try:
        # Get image links in scope from hashtag, location and other pages
        link_elems = element.find_elements(By.XPATH, '//a[starts-with(@href, "/p/")]')
        sleep(random.randint(2, 5))

        if link_elems:
            for link_elem in link_elems:
                try:
                    post_href = link_elem.get_attribute("href")
                    post_elem = element.find_elements(
                        By.XPATH,
                        "//a[@href='/p/" + post_href.split("/")[-2] + "/']/child::div",
                    )

                    if len(post_elem) == 1 and MEDIA_PHOTO in media:
                        logger.info("Found media type: {}".format(MEDIA_PHOTO))
                        links.append(post_href)

                    if len(post_elem) == 2:
                        logger.info(
                            "Found media type: {} - {} - {}".format(
                                MEDIA_CAROUSEL, MEDIA_VIDEO, MEDIA_IGTV
                            )
                        )
                        # If you see "Cannot detect post media type. Skip https://www.instagram.com/p/CFvUn0gpaMZ/"
                        # consider updating the @class,'CzVzU', new format types could be added
                        # Media types from constants.py must be updated here, otherwise the links
                        # cannot be categorized.
                        post_category = element.find_element(
                            By.XPATH,
                            "//a[@href='/p/"
                            + post_href.split("/")[-2]
                            + "/']/div[contains(@class,'CzVzU')]/child::*/*[name()='svg']",
                        ).get_attribute("aria-label")

                        logger.info("Post category: {}".format(post_category))

                        if post_category in media:
                            links.append(post_href)

                except WebDriverException:
                    # If "post_href" is None skip the logger to avoid confusion,
                    # the links that are not empty will be catched into the next
                    # loop. Other case, the "post_href" is not empty and needs
                    # to be displayed to the STDOUT for further review.
                    if post_href:
                        logger.info(
                            "Cannot detect post media type. Skip {}".format(post_href)
                        )
        else:
            logger.info("'{}' page does not contain a picture".format(page))

    except BaseException as e:
        logger.error("link_elems error \n\t{}".format(str(e).encode("utf-8")))