Python Scrapy多级请求（三级）

Question

我是Python Scrapy的新手，我正在尝试从3层嵌套页面创建JSON文件。我有以下结构：

Page 1 (start): contains links of second page (called Mangas) Page 2: Contains nested Volumes and Chapters Page 3: Each Chapter contains multiple images

我的代码

import scrapy
import time
import items
import json

class GmangaSpider(scrapy.Spider):
    name = "gmanga"
    start_urls = [
        "http://gmanga.me/mangas"
    ]

    def parse(self, response):
        # mangas = []

        for manga in response.css('div.manga-item'):
            link = manga.css('a.manga-item-content').xpath('@href').extract_first()

            if link:
                page_link = "http://gmanga.me%s" % link

                mangas = items.Manga()
                mangas['cover'] =  manga.css('a.manga-item-content .manga-cover-container img').xpath('@src').extract_first()
                mangas['title'] = manga.css('a.manga-item-content .manga-cover-container img').xpath('@alt').extract_first()
                mangas['link'] = page_link
                mangas['volumes'] = []

                yield scrapy.Request(page_link, callback=self.parse_volumes, meta = {"mangas": mangas})

    def parse_volumes(self, response):
        mangas = response.meta['mangas']

        for manga in response.css('div.panel'):
            volume = items.Volume()
            volume['name'] = manga.css('div.panel-heading .panel-title a::text').extract_first()
            volume['chapters'] = []
            for tr in manga.css('div.panel-collapse .panel-body table tbody tr'):
                chapter = items.Chapter()
                chapter['name'] = tr.css('td:nth_child(1) div::text').extract_first()
                chapter_link = tr.css('td:nth_child(3) a::attr("href")').extract_first()
                chapter['link'] = chapter_link


                request = scrapy.Request("http://gmanga.me%s" % chapter_link, callback = self.parse_images, meta = {"chapter": chapter})
                yield request

                volume['chapters'].append(chapter)

            mangas['volumes'].append(volume)
        yield mangas


    def parse_images(self, response):
        chapter = response.meta['chapter']
        data = response.xpath("//script").re("alphanumSort\((.*])")
        if data:
            images = json.loads(data[0])
            chapter['images'] = images
        return chapter

我的Items.py

from scrapy import Item, Field
class Manga(Item):
    title = Field()
    cover = Field()
    link = Field()
    volumes = Field()

class Volume(Item):
    name = Field()
    chapters = Field()

class Chapter(Item):
    name = Field()
    images = Field()
    link = Field()

现在我在parse_volumes函数中有点困惑，在json文件中产生或返回以获取以下结构。

预期结果：

[{
"cover": "http://media.gmanga.me/uploads/manga/cover/151/medium_143061.jpg",
"link": "http://gmanga.me/mangas/gokko",
"volumes": [{
    "name": "xyz",
    "chapters": [{
        "link": "/mangas/gokko/4/3asq",
        "name": "4",
        "images": ["img1.jpg", "img2.jpg"]
    }, {
        "link": "/mangas/gokko/3/3asq",
        "name": "3",
        "images": ["img1.jpg", "img2.jpg"]
    }]
}],
"title": "Gokko"
}]

但我将图像节点作为单独的节点，它必须在卷的章节节点内：

[{
    "cover": "http://media.gmanga.me/uploads/manga/cover/10581/medium_I2.5HFzVh7e.png",
    "link": "http://gmanga.me/mangas/godess-creation-system",
    "volumes": [{
        "name": "\u0627\u0644\u0645\u062c\u0644\u062f ",
        "chapters": [{
            "link": "/mangas/godess-creation-system/1/ayou-cahn",
            "name": "1"
        }]
    }],
    "title": "Godess Creation System"
},
{
    "images": ["http://media.gmanga.me/uploads/releases/lolly-pop/047-20160111235059UXYGJACW/01.jpg?ak=p0skml", "http://media.gmanga.me/uploads/releases/lolly-pop/047-20160111235059UXYGJACW/02.jpg?ak=p0skml", "http://media.gmanga.me/uploads/releases/lolly-pop/047-20160111235059UXYGJACW/03.jpg?ak=p0skml", "http://media.gmanga.me/uploads/releases/lolly-pop/047-20160111235059UXYGJACW/04.jpg?ak=p0skml"],
    "link": "/mangas/reversal/1/Lolly-Pop",
    "name": "1"
}]

每个函数都是正确地单独提取数据，唯一的问题是JSON形成。它没有正确写入json文件。请带我到我错的地方。