How to separate pipeline processing code into files by spider in Scrapy

Introduction

If you increase the scraping target (spider file) in Scrapy, the pipeline processing code will increase in the same file, which will make the visibility worse and maintainability problems. I came. Finally, I was able to separate the pipeline implementation file for each spider, so I will introduce the method.

Provisional target method

A Scrapy project has a configuration file called setting.py. Even if there is a setting item of ʻITEM_PIPELINES in setting.py` and there are multiple spiders, Initially I thought I had no choice but to aggregate the pipeline processing into classes in the single implementation file specified here.

setting.py

ITEM_PIPELINES = {
    'example_project.pipelines.DBPipeline': 100,
}

--Pattern that processes and branches with the name of siper

I was routing with the spider name as the key, but it is clear that the visibility of the code gets worse as the spider increases.

pipelines.py

class DBPipeline(object):
    def process_item(self, item, spider):
        if spider.name in ['example_spider']:
            # example_Spider pipeline processing
        
        if spider.name in ['example_spider2']:
            # example_Pipeline processing of spider2

Conclusion

If you set the ʻITEM_PIPELINESitem incustom_settings` for each spider as shown below, The implementation file for pipeline processing can be individualized. [^ 1]

example_spider.py

class ExampleSpider(scrapy.Spider):
    custom_settings = {
        'ITEM_PIPELINES': {
            'example_project.example_pipelines.ValidationPipeline': 100,
            'example_project.example_pipelines.DBPipeline': 200,
        }
    }

example_spider2.py

class ExampleSpider２(scrapy.Spider):
    custom_settings = {
        'ITEM_PIPELINES': {
            'example_project.example_pipelines2.DBPipeline': 100,
        }
    }

Individually routed to the following pipeline processing as set in custom_settings.

example_pipelines.py

class ValidationPipeline(object):
    def process_item(self, item: scrapy.Item, spider: scrapy.Spider):
        # example_spider.Processed when running py

class DBPipeline(object):
    def process_item(self, item: scrapy.Item, spider: scrapy.Spider):
        # example_spider.Processed when running py

example_pipelines2.py

class DBPipeline(object):
    def process_item(self, item: scrapy.Item, spider: scrapy.Spider):
        # example_spider2.Processed when running py

With the above, even if the number of scraping targets (spiders) increases, the visibility of the pipeline processing code can be maintained in good condition.

[^ 1]: Similarly, it seems that you can customize other items such as SPIDER_MIDDLEWARES.