termux-app/termux-shared/src/main/java/com/termux/shared/termux/data/TermuxUrlUtils.java

105 lines
4.5 KiB
Java

package com.termux.shared.termux.data;
import java.util.LinkedHashSet;
import java.util.regex.Matcher;
import java.util.regex.Pattern;
public class TermuxUrlUtils {
public static Pattern URL_MATCH_REGEX;
public static Pattern getUrlMatchRegex() {
if (URL_MATCH_REGEX != null) return URL_MATCH_REGEX;
StringBuilder regex_sb = new StringBuilder();
regex_sb.append("("); // Begin first matching group.
regex_sb.append("(?:"); // Begin scheme group.
regex_sb.append("dav|"); // The DAV proto.
regex_sb.append("dict|"); // The DICT proto.
regex_sb.append("dns|"); // The DNS proto.
regex_sb.append("file|"); // File path.
regex_sb.append("finger|"); // The Finger proto.
regex_sb.append("ftp(?:s?)|"); // The FTP proto.
regex_sb.append("git|"); // The Git proto.
regex_sb.append("gemini|"); // The Gemini proto.
regex_sb.append("gopher|"); // The Gopher proto.
regex_sb.append("http(?:s?)|"); // The HTTP proto.
regex_sb.append("imap(?:s?)|"); // The IMAP proto.
regex_sb.append("irc(?:[6s]?)|"); // The IRC proto.
regex_sb.append("ip[fn]s|"); // The IPFS proto.
regex_sb.append("ldap(?:s?)|"); // The LDAP proto.
regex_sb.append("pop3(?:s?)|"); // The POP3 proto.
regex_sb.append("redis(?:s?)|"); // The Redis proto.
regex_sb.append("rsync|"); // The Rsync proto.
regex_sb.append("rtsp(?:[su]?)|"); // The RTSP proto.
regex_sb.append("sftp|"); // The SFTP proto.
regex_sb.append("smb(?:s?)|"); // The SAMBA proto.
regex_sb.append("smtp(?:s?)|"); // The SMTP proto.
regex_sb.append("svn(?:(?:\\+ssh)?)|"); // The Subversion proto.
regex_sb.append("tcp|"); // The TCP proto.
regex_sb.append("telnet|"); // The Telnet proto.
regex_sb.append("tftp|"); // The TFTP proto.
regex_sb.append("udp|"); // The UDP proto.
regex_sb.append("vnc|"); // The VNC proto.
regex_sb.append("ws(?:s?)"); // The Websocket proto.
regex_sb.append(")://"); // End scheme group.
regex_sb.append(")"); // End first matching group.
// Begin second matching group.
regex_sb.append("(");
// User name and/or password in format 'user:pass@'.
regex_sb.append("(?:\\S+(?::\\S*)?@)?");
// Begin host group.
regex_sb.append("(?:");
// IP address (from http://www.regular-expressions.info/examples.html).
regex_sb.append("(?:(?:25[0-5]|2[0-4][0-9]|[01]?[0-9][0-9]?)\\.){3}(?:25[0-5]|2[0-4][0-9]|[01]?[0-9][0-9]?)|");
// Host name or domain.
regex_sb.append("(?:(?:[a-z\\u00a1-\\uffff0-9]-*)*[a-z\\u00a1-\\uffff0-9]+)(?:(?:\\.(?:[a-z\\u00a1-\\uffff0-9]-*)*[a-z\\u00a1-\\uffff0-9]+)*(?:\\.(?:[a-z\\u00a1-\\uffff0-9]-*){1,}[a-z\\u00a1-\\uffff0-9]{1,}))?|");
// Just path. Used in case of 'file://' scheme.
regex_sb.append("/(?:(?:[a-z\\u00a1-\\uffff0-9]-*)*[a-z\\u00a1-\\uffff0-9]+)");
// End host group.
regex_sb.append(")");
// Port number.
regex_sb.append("(?::\\d{1,5})?");
// Resource path with optional query string.
regex_sb.append("(?:/[a-zA-Z0-9:@%\\-._~!$&()*+,;=?/]*)?");
// Fragment.
regex_sb.append("(?:#[a-zA-Z0-9:@%\\-._~!$&()*+,;=?/]*)?");
// End second matching group.
regex_sb.append(")");
URL_MATCH_REGEX = Pattern.compile(
regex_sb.toString(),
Pattern.CASE_INSENSITIVE | Pattern.MULTILINE | Pattern.DOTALL);
return URL_MATCH_REGEX;
}
public static LinkedHashSet<CharSequence> extractUrls(String text) {
LinkedHashSet<CharSequence> urlSet = new LinkedHashSet<>();
Matcher matcher = getUrlMatchRegex().matcher(text);
while (matcher.find()) {
int matchStart = matcher.start(1);
int matchEnd = matcher.end();
String url = text.substring(matchStart, matchEnd);
urlSet.add(url);
}
return urlSet;
}
}