mass-scraper/Makefile

31 lines
1.1 KiB
Makefile
Raw Normal View History

2018-03-01 18:16:11 +01:00
SHELL := /bin/bash
PREPARE_ENVIRONMENT := $(shell ./prepare-environment.sh > /tmp/makeenv)
include /tmp/makeenv
JOBS := 40
2018-03-01 18:16:11 +01:00
.PHONY: all update data clean clean-data
2018-03-01 18:16:11 +01:00
all: data
data: parishwebsites/spider-commands.txt parishwebsites/domain-blacklist.txt
rm -f parishwebsites/*processed.txt
cd parishwebsites && parallel --jobs $(JOBS) < spider-commands.txt 2> crawler-log.txt
2018-03-01 18:16:11 +01:00
parishwebsites/spider-commands.txt: parishes-with-urls.tsv parishwebsites/domain-blacklist.txt
cut -f3 $< | tail -n +2 | grep http | parishwebsites/generate_spider_commands.sh | sort -u | parishwebsites/remove_blacklisted.py $(word 2,$^) > $@
2018-03-01 18:16:11 +01:00
parishes-with-urls.tsv: apikey.txt parishes-deon.tsv scraper/get_parishes_urls.py
scraper/get_parishes_urls.py -a $< -p $(word 2,$^) >> $@ 2> get-parishes-urls.log
2018-03-01 18:16:11 +01:00
parishes-deon.tsv: scraper/crawl_deon.py
scraper/crawl_deon.py > $@ 2> crawl-deon.log
update: environment.yml
conda env update -f $<
2018-03-01 18:16:11 +01:00
clean:
rm -rf parishes-deon.tsv parishes-with-urls.tsv spider-commands.txt
clean-data:
rm -rf parishwebsites/{data,processed.txt,crawler-log.txt}