diff --git a/FourmiCrawler/sources/PubChem.py b/FourmiCrawler/sources/PubChem.py index 0768612..0bc8b8c 100644 --- a/FourmiCrawler/sources/PubChem.py +++ b/FourmiCrawler/sources/PubChem.py @@ -16,9 +16,9 @@ class PubChem(Source): """ #PubChem has its data on compound name, properties and their values on different html pages, so different URLs used - website = 'https://.*\\.ncbi\\.nlm\\.nih\\.gov/.*' - website_www = 'https://www.ncbi.nlm.nih.gov/*' - website_pubchem = 'https://pubchem.ncbi.nlm.nih.gov/.*' + website = 'http://.*\\.ncbi\\.nlm\\.nih\\.gov/.*' + website_www = 'http://www.ncbi.nlm.nih.gov/*' + website_pubchem = 'http://pubchem.ncbi.nlm.nih.gov/.*' search = 'pccompound?term=%s' data_url = 'toc/summary_toc.cgi?tocid=27&cid=%s'