This repository was archived by the owner on Dec 22, 2023. It is now read-only.

Commit 3afa60f

authored

Update economictimes_scraper.py

1 parent c43a74c commit 3afa60fCopy full SHA for 3afa60f

File tree

-7

lines changed

-7

lines changed

Lines changed: 6 additions & 7 deletions

Original file line number	Diff line number	Diff line change
`@@ -5,7 +5,7 @@`
`5`	`5`	`import datetime`
`6`	`6`	`import sys`
`7`	`7`
`8`		`-## Util`
	`8`	`+# Util`
`9`	`9`	`def datestr_to_date(datestr):`
`10`	`10`	`[year, month, day] = datestr.split('-')`
`11`	`11`	`return datetime.date(`
`@@ -14,8 +14,8 @@ def datestr_to_date(datestr):`
`14`	`14`	`day=int(day)`
`15`	`15`	`)`
`16`	`16`
`17`		`-## Reference dates`
`18`		`-reference_date = datetime.date(2001, 1, 1) ## 2001 Jan 1`
	`17`	`+# Reference dates`
	`18`	`+reference_date = datetime.date(2001, 1, 1) # 2001 Jan 1`
`19`	`19`	`reference_date_id = 36892`
`20`	`20`
`21`	`21`	`if len(sys.argv) < 3:`
`@@ -35,7 +35,7 @@ def datestr_to_date(datestr):`
`35`	`35`	`sys.exit(1)`
`36`	`36`
`37`	`37`
`38`		`-## Gets News article metadata from article url`
	`38`	`+# Gets News article metadata from article url`
`39`	`39`	`def fetchNewsArticle(url):`
`40`	`40`	`html = requests.get(url).content`
`41`	`41`	`root = defusedxml.HTML(html)`
`@@ -66,6 +66,5 @@ def fetchNewsArticle(url):`
`66`	`66`	`})`
`67`	`67`
`68`	`68`	`out_filename = 'ET_NewsData_{}_{}.json'.format(start_date, end_date)`
`69`		`-output_file = open(out_filename, 'w+')`
`70`		`-output_file.write(json.dumps(fetched_data, indent=2))`
`71`		`-output_file.close()`
	`69`	`+with open(out_filename, 'w+') as output_file:`
	`70`	`+ output_file.write(json.dumps(fetched_data, indent=2))`

Comments

(0)