User:Ohconfucius/script/Sources.js: Difference between revisions

Revision as of 18:21, 23 June 2025

Revision as of 18:32, 23 June 2025

Line 581:

regex(/((?:[\[=]\s*)(?:https?:|ftp:))([^\]\|\}]*)(\s*[\]|}])/gi, protect_function);

regex(/((?:Category|Image|File):)([^|\]]*)([|\]])/gi, protect_function); //protect file names from being changed

regex(/(\{\{(?:FAA-airport|harv\w*|sfn\w*|cite ?book|listen)\s?\|)([^\}]+)(\}\})/gi, protect_function);

regex(/(\|\s*(?:contribution|quote|title)\s*=)([^|}]+)([\|\}])/gi, protect_function);

//whitelisting -against changes made by generic regex in Ohc_sourcepub (e.g. Fast Company)

Line 647:

// removing references to other WP articles and 'external' WP links

regex(/<ref[^<>]*>[^<>]*\|[ ]*url ?= ?https?:\/\/(?:\w{2}\.wikipedia\.org\/wiki|(?:www\.|)(?:blogger|blogspot|facebook|myspacefindagrave|twitterinstagram|mailonline|findagravemyspace|twitter|x)\.com)\/[^<>]*<\/ref>/gi, '');

regex(/<ref>https?:\/\/(?:en\.wikipedia\.org\/wiki|(?:www\.|)(?:blogger|blogspot|facebook|myspacefindagrave|twitterinstagram|mailonline|findagravemyspace|twitter|x)\.com)\/[^\s\]<]*<\/ref>/gi, '');

regex(/<ref>\[https?:\/\/(?:en\.wikipedia\.org\/wiki|(?:www\.|)(?:blogger|blogspot|facebook|myspacefindagrave|twitterinstagram|mailonline|findagravemyspace|twitter|x)\.com)\/[^\s\]]*[ ]+[\w\d][^\]]*\]<\/ref>/gi, '');

regex(/\|[ ]*url[ ]*=[ ]*https?:\/\/(?:\w{2}\.wikipedia\.org\/wiki|(?:www\.|)(?:blogger|blogspot|facebook|myspacefindagrave|twitterinstagram|mailonline|findagravemyspace|twitter|x)\.com)[^\s\|\{\}<]*(?=[ ]*[|}])/gi, '');

Line 659:

function ohc_ref_names() {

// give names to citations in templates that are without names

regex(/<ref>((?:[^<>]*\|w,\s*url\s*=|[.&]+\"?\[\s]?)https?:\/\/(?:[^|</]+\.)*(\w+)\.(?:(?:com?|org|net|gov|ac)(?:\.[a-z]{2}|))\/(?:[^<>|{}\d]*\D|)(\d{6,12})[^<>]*)(?=<\/ref>)/gi, "<ref name=$22_$3>$1"); //urls with non-date serial numbers

regex(/<ref>((?:[^<>]*\|\s*url\s*=|[\[\s]?)https?:\/\/(?:[^|</]+\.)*([-\w]+)\.(?:(?:com?|org|net|gov|ac)(?:\.[a-z]{2}|))\/(?:[^<>|{}\d]*\D|)(\d{6,12})[^<>]*)(?=<\/ref>)/gi, "<ref name=$22_$3>$1"); //urls with non-date serial numbers

regex(/<ref>((?:[^<>]*\|\s*url\s*=|[\[\s]?)https?:\/\/(?:[^|</]+\.)*([-\w]+)(?:\.[a-z]{2})\/[^<>|{}\d]*(\d{6,12})[^<>]*)(?=<\/ref>)/gi, "<ref name=$2_$3>$1"); //urls with non-date serial numbers

regex(/<ref>((?:[^<>]*\|\s*url\s*=|[\[\s]?)https?:\/\/(?:[^|</]+\.)*(reuters)\.com\/[^<>|{}\d]*(id\w{11})[^<>]*)(?=<\/ref>)/gi, "<ref name=$2_$3>$1"); //reuters urls with non-date serial numbers

ohc_regex(/<ref>((?:[^<>]*\|\s*url\s*=|[\[\s]?)https?:\/\/(?:[^|</]+\.)*([-\w]+)(?:\.[a-z]{2,3})\/[^<>|{}\d]*@YYYY[-\/]@MM[-\/]@DD\/(\w+)[^<>]*)(?=<\/ref>)/gi, "<ref name=$2_@Year@ZM@ZD$2-ZD_$3>$1"); //urls with non-date serial numbers

ohc_regex(/<ref>((?:[^<>]*\|\s*url\s*=|[\[\s]?)https?:\/\/(?:[^|</]+\.)*([-\w]+)(?:\.[a-z]{2,3})\/[^<>|{}\d]*@YYYY[-\/]@MonthMon[-\/]@DD\/(\w+)[^<>]*)(?=<\/ref>)/gi, "<ref name=$2_@Year@ZM@ZD$2-ZD_$3>$1"); //urls with non-date serial numbers

ohc_regex(/<ref>([^<>]*\|\s*url\s*=https?:\/\/(?:[^|</]+\.)*(\w+)\.(?:(?:com?|org|net|gov|ac|)(?:\.[a-z]{2}|))\/[^<>]*\|\s*date\s*=\s*@Day @Month @YYYY\s*[|}][^<>]*)(?=<\/ref>)/gi, "<ref name=$2_@Year@ZM@ZD$2>$1");

ohc_regex(/<ref>([^<>]*\|\s*url\s*=https?:\/\/(?:[^|</]+\.)*(\w+)\.(?:(?:com?|org|net|gov|ac|)(?:\.[a-z]{2}|))\/[^<>]*\|\s*date\s*=\s*@Month @Day, @YYYY\s*[|}][^<>]*)(?=<\/ref>)/gi, "<ref name=$2_@Year@ZM@ZD$2>$1"); //this line may cause unwarranted substitution of "$5" by a comma (e.g. in 2019–20 Hong Kong protests)

ohc_regex(/<ref>([^<>]*\|\s*url\s*=https?:\/\/(?:[^|</]+\.)*(\w+)\.(?:(?:com?|org|net|gov|ac|)(?:\.[a-z]{2}|))\/[^<>]*\|\s*date\s*=\s*=\s*@YYYY-@MM-@DD\s*[|}][^<>]*)(?=<\/ref>)/gi, "<ref name=$2_@Year@MMZM@DD$2ZD>$1");

ohc_regex(/<ref>((?:[^<>]*\|\s*url\s*=|\[?)https?:\/\/(?:[^|<]+\.)*(\w+)\.(?:(?:com?|org|net|gov|ac|[a-z]{2})(?:\.[a-z]{2}|))\/[^<>|{}]+@YYYY[\/-]?@Month[\/-]?@Day[^<>]*)(?=<\/ref>)/gi, "<ref name=$2_@YYYYYear@MMZM@DD$2ZD>$1"); //urls with embedded dates

}