2

Just a quick question about json export formatting in Scrapy. My exported file looks like this. 

{"pages": {"title": "x", "text": "x", "tags": "x", "url": "x"}}
{"pages": {"title": "x", "text": "x", "tags": "x", "url": "x"}}
{"pages": {"title": "x", "text": "x", "tags": "x", "url": "x"}}

But I would like it to be in this exact format. Somehow I need to get all the other information under "pages".

{"pages": [
     {"title": "x", "text": "x", "tags": "x", "url": "x"},
     {"title": "x", "text": "x", "tags": "x", "url": "x"},
     {"title": "x", "text": "x", "tags": "x", "url": "x"}
]}

I'm not very experienced in scrapy or python, but I have gotten everything else done in my spider except the export format. This is my pipelines.py, which I just got working.

from scrapy.exporters import JsonItemExporter
import json

class RautahakuPipeline(object):

    def open_spider(self, spider):
        self.file = open('items.json', 'w')

    def close_spider(self, spider):
        self.file.close()

    def process_item(self, item, spider):
        line = json.dumps(dict(item)) + "\n"
        self.file.write(line)
        return item

These are the items in my spider.py I need to extract

        items = []
        for title, text, tags, url in zip(product_title, product_text, product_tags, product_url):
            item = TechbbsItem()
            item['pages'] = {}
            item['pages']['title'] = title
            item['pages']['text'] = text
            item['pages']['tags'] = tags
            item['pages']['url'] = url
            items.append(item)
        return items

Any help is greatly appreciated, as this is the last obstacle in my project.

EDIT

items = {'pages':[{'title':title,'text':text,'tags':tags,'url':url} for title, text, tags, url in zip(product_title, product_text, product_tags, product_url)]}

This extracts the .json in this format

{"pages": [{"title": "x", "text": "x", "tags": "x", "url": "x"}]} {"pages": [{"title": "x", "text": "x", "tags": "x", "url": "x"}]} {"pages": [{"title": "x", "text": "x", "tags": "x", "url": "x"}]}

This is getting better but I would still need only one "pages" on the start of the file and everything else inside an array under it.

EDIT 2

I think my spider.py is the reason why "pages" gets added to every line in the .json file and I should have originally posted the whole code of it. Here it is.

# -*- coding: utf-8 -*-
import scrapy
from urllib.parse import urljoin

class TechbbsItem(scrapy.Item):
    pages = scrapy.Field()
    title = scrapy.Field()
    text= scrapy.Field()
    tags= scrapy.Field()
    url = scrapy.Field()

class TechbbsSpider(scrapy.Spider):
    name = 'techbbs'
    allowed_domains = ['bbs.io-tech.fi']
    start_urls = ['https://bbs.io-tech.fi/forums/prosessorit-emolevyt-ja-muistit.73/?prefix_id=1' #This is a list page full of used pc-part listings
             ]
    def parse(self, response): #This visits product links in the product list page
        links = response.css('a.PreviewTooltip::attr(href)').extract()
        for l in links:
            url = response.urljoin(l)
            yield scrapy.Request(url, callback=self.parse_product)
        next_page_url = response.xpath('//a[contains(.,"Seuraava ")]/@href').extract_first()
        if next_page_url:
           next_page_url =  response.urljoin(next_page_url) 
           yield scrapy.Request(url=next_page_url, callback=self.parse)

    def parse_product(self, response): #This extracts data from inside the links
        product_title = response.xpath('normalize-space(//h1/span/following-sibling::text())').extract()
        product_text = response.xpath('//b[contains(.,"Hinta:")]/following-sibling::text()[1]').re('([0-9]+)')
        tags = "tags" #This is just a placeholder
        product_tags = tags
        product_url = response.xpath('//html/head/link[7]/@href').extract()

        items = []
        for title, text, tags, url in zip(product_title, product_text, product_tags, product_url):
            item = TechbbsItem()
            item['pages'] = {}
            item['pages']['title'] = title
            item['pages']['text'] = text
            item['pages']['tags'] = tags
            item['pages']['url'] = url
            items.append(item)
        return items

So my spider starts crawling from a page full of product listings. It visits every one of the 50 product links and scrapes 4 items, title, text, tags and url. After scraping every link in one page, it goes to next one and so on. I suspect the loops in the code prevent your suggestions from working for me.

I would like to get the .json export to the exact form mentioned in the original question. Se there would be {"pages": [ on the beginning of the file, then all the indented item lines {"title": "x", "text": "x", "tags": "x", "url": "x"}, and in the end ]}

Vadim Kotov
  • 8,084
  • 8
  • 48
  • 62
Janne Salmi
  • 33
  • 4

3 Answers3

6

In terms of memory usage, it's not a good practice, but an option is to keep an object and write it at the end of the process:

class RautahakuPipeline(object):

    def open_spider(self, spider):
        self.items = { "pages":[] }
        self.file = null # open('items.json', 'w')

    def close_spider(self, spider):
        self.file = open('items.json', 'w')
        self.file.write(json.dumps(self.items))
        self.file.close()

    def process_item(self, item, spider):            
        self.items["pages"].append(dict(item))
        return item

Then, if memory is an issue (must be treat with attention anyway), try writing the json file as follows: 

class RautahakuPipeline(object):

    def open_spider(self, spider):
        self.file = open('items.json', 'w')
        header='{"pages": ['
        self.file.write(header)

    def close_spider(self, spider):
        footer=']}'
        self.file.write(footer)
        self.file.close()

    def process_item(self, item, spider):
        line = json.dumps(dict(item)) + "\n"
        self.file.write(line)
        return item

I hope it helps.

Evhz
  • 8,852
  • 9
  • 51
  • 69
  • Thank you! Your first suggestion worked perfectly! Only thing I want to nitpick about is that every item is on the same line. Where should I add \n or is it done with some other way? – Janne Salmi Apr 05 '18 at 15:33
  • As the `json.dumps` is written at a time, you could not write a `\n`. In that case, try with the option two. – Evhz Apr 05 '18 at 16:04
1
items = {}
#item = TechbbsItem()  # not sure what this is doing?
items['pages'] = []
for title, text, tags, url in zip(product_title, product_text, product_tags, product_url):
    temp_dict = {}
    temp_dict['title'] = title
    temp_dict['text'] = text
    temp_dict['tags'] = tags
    temp_dict['url'] = url
    items["pages"].append(temp_dict)
return items
JacobIRR
  • 8,545
  • 8
  • 39
  • 68
1

Using list comprehension. I don´t know how looks your data , but using a toy example:

product_title = range(1,10)
product_text = range(10,20)
product_tags = range(20,30)
product_url = range(30,40)

item = {'pages':[{'title':title,'text':text,'tags':tags,'url':url} 
  for title, text, tags, url in zip(product_title, product_text, product_tags, product_url)]}

I get this result:

{'pages': [{'tags': 20, 'text': 10, 'title': 1, 'url': 30},
{'tags': 21, 'text': 11, 'title': 2, 'url': 31},
{'tags': 22, 'text': 12, 'title': 3, 'url': 32},
{'tags': 23, 'text': 13, 'title': 4, 'url': 33},
{'tags': 24, 'text': 14, 'title': 5, 'url': 34},
{'tags': 25, 'text': 15, 'title': 6, 'url': 35},
{'tags': 26, 'text': 16, 'title': 7, 'url': 36},
{'tags': 27, 'text': 17, 'title': 8, 'url': 37},
{'tags': 28, 'text': 18, 'title': 9, 'url': 38}]}
jjsantoso
  • 1,586
  • 1
  • 12
  • 17
  • I shared the whole code of my spider, because it's structure was probably the reason why your solution didn't work for me. Could you take a look at my second edit. – Janne Salmi Apr 04 '18 at 16:32