From 0faa5175b6f0de8835ed514615ac64135f406b29 Mon Sep 17 00:00:00 2001 From: "Benjamin J. Culkin" Date: Tue, 29 May 2018 15:35:02 -0300 Subject: Move files out of folder --- data/sample-grammars/web.gram | 227 ++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 227 insertions(+) create mode 100644 data/sample-grammars/web.gram (limited to 'data/sample-grammars/web.gram') diff --git a/data/sample-grammars/web.gram b/data/sample-grammars/web.gram new file mode 100644 index 0000000..357fc36 --- /dev/null +++ b/data/sample-grammars/web.gram @@ -0,0 +1,227 @@ +[generic-email] [email] + [fake-email] + +[email] [actual+email] + +[fake-email] [actual+fake-email] + +[username] [actual+username] + +[url] [actual+url] + +pragma initial-rule [generic-email] + +pragma export-rule [email] +pragma export-rule [fake-email] +pragma export-rule [username] +pragma export-rule [domain-name] +pragma export-rule [url] +pragma export-rule [big-tld] + +[actual+email] [email-initial] @ [email-trailing] + +[email-initial] [username] + [acronym] + [email-alias] + [full-name] + [username] + [acronym] + [email-alias] + [full-name] + [subdomain-name] + [email-misc] + [email-initial] + [email-initial] + +[email-trailing] [subdomain-name] . [domain] + mail. [domain] + [domain] + [subdomain-name] - [town-name] . [tld] + [last-name] . [tld] + [first-name] . [tld] + [last-name] - [first-name] . [tld] + [first-name] - [last-name] . [tld] + +[actual+fake-email] [email-initial] @ [fake-trailing] + +[fake-trailing] [subdomain-name] . [fake-domain] + [fake-domain] + +[email-alias] abuse + admin + administrator + billing + blog + ceo + cfo + contact + coo + cto + customer + customerservice + dnsadmin + dnsmaster + email + ftp + help + helpdesk + hostmaster + info + ipadmin + mail + mailer-daemon + marketing + marketingscum + news + nobody + noc + no-reply + postman + postmaster + root + sales + security + ssladmin + support + usenet + webjockey + webmaster + www + +[actual+username] [full-name] + [first-name] . [last-name] + [full-name] [1..999] + +[actual+url] http://www. [domain] / + http://www. [domain] / + http:// [subdomain-name] . [domain] / + http://www. [last-name] . [tld] / + http:// [username] . [domain] / + [url-with-path] + [url-with-path] + +[url-with-path] [url-protocol] :// www.geocities.com/ [neighborhood] / [digit-nonzero] [digit] [digit] [digit] / + [url-protocol] :// [url-begin] . [acronym] .net/ [url-username] / + [url-protocol] :// [url-begin] . [town-name] .net/ [url-username] + [url-protocol] :// [subdomain-name] . [acronym] .net/ [url-username] + [url-protocol] :// [subdomain-name] . [town-name] .net/ [url-username] + [url-protocol] :// [subdomain-name] . [last-name] . [big-tld] / [url-username] + [url-protocol] :// [username] . [acronym] .net/ + [url-protocol] :// [username] . [url-state] [us-postalstate] .edu/ + [freshmeat-url] + +[url-protocol] http + http + https + https + ftp + ssh + rsync + gopher + git + +[neighborhood] siliconvalley + heartland + timessquare + +[url-begin] www + www2 + users + members + +[url-username] [username] + ~ [username] + +[url-state] s + n + e + w + se + nw + ne + sw + univ + u + +[domain] [domain-name] . [tld] + [domain-name] . [big-tld] + +[fake-domain] [domain-name] [fake-ending] + +[fake-ending] .con + .con + .not + .null + .usa + .com.uk + .con + .dot + .fake + .faux + .spam + .tld + +[domain-name] [domain+name] + +[domain+name] [acronym] + [random-name] + [last-name] - [last-name] + [first-name] - [last-name] + [letter] - [last-name] + [consonant] [vowel] [consonant] + [consonant-group] [vowel-group] [consonant] + [vowel-group] [consonant-group] + [letter] [vowel] [letter] + [town-name] + [last-name] + [syllable] [syllable] + +[tld] com + com + com + com + com + net + net + org + org + edu + de + ca + com.au + co.uk + fr + nu + to + se + no + +[big-tld] com + com + net + org + +[subdomain-name] [planet] + [planet] + mail + www2 + www [digit] + alpha + beta + gamma + omega + [random-name] + [first-name] + [last-name] + [acronym] + [email-misc] + [email-initial] + [subdomain-name] - [subdomain-name] + +[email-misc] + [us-stateadj] + [town-name] + [complete-many-things] + [major-arcana] + [trademark+drug] + [trademark] + [city] -- cgit v1.2.3