JAVA提取字符串中所有的URL链接，并加上a标签

 

工具类

Patterns.java

  1 package com.util;
  2 
  3 import java.util.regex.Matcher;
  4 import java.util.regex.Pattern;
  5 
  6 /**
  7  * Commonly used regular expression patterns.
  8  */
  9 public class Patterns {
 10     /**
 11      *  Regular expression to match all IANA top-level domains.
 12      *  List accurate as of 2011/07/18.  List taken from:
 13      *  http://data.iana.org/TLD/tlds-alpha-by-domain.txt
 14      *  This pattern is auto-generated by frameworks/ex/common/tools/make-iana-tld-pattern.py
 15      *
 16      *  @deprecated Due to the recent profileration of gTLDs, this API is
 17      *  expected to become out-of-date very quickly. Therefore it is now
 18      *  deprecated.
 19      */
 20     @Deprecated
 21     public static final String TOP_LEVEL_DOMAIN_STR =
 22             "((aero|arpa|asia|a[cdefgilmnoqrstuwxz])"
 23                     + "|(biz|b[abdefghijmnorstvwyz])"
 24                     + "|(cat|com|coop|c[acdfghiklmnoruvxyz])"
 25                     + "|d[ejkmoz]"
 26                     + "|(edu|e[cegrstu])"
 27                     + "|f[ijkmor]"
 28                     + "|(gov|g[abdefghilmnpqrstuwy])"
 29                     + "|h[kmnrtu]"
 30                     + "|(info|int|i[delmnoqrst])"
 31                     + "|(jobs|j[emop])"
 32                     + "|k[eghimnprwyz]"
 33                     + "|l[abcikrstuvy]"
 34                     + "|(mil|mobi|museum|m[acdeghklmnopqrstuvwxyz])"
 35                     + "|(name|net|n[acefgilopruz])"
 36                     + "|(org|om)"
 37                     + "|(pro|p[aefghklmnrstwy])"
 38                     + "|qa"
 39                     + "|r[eosuw]"
 40                     + "|s[abcdeghijklmnortuvyz]"
 41                     + "|(tel|travel|t[cdfghjklmnoprtvwz])"
 42                     + "|u[agksyz]"
 43                     + "|v[aceginu]"
 44                     + "|w[fs]"
 45                     + "|(\u03b4\u03bf\u03ba\u03b9\u03bc\u03ae|\u0438\u0441\u043f\u044b\u0442\u0430\u043d\u0438\u0435|\u0440\u0444|\u0441\u0440\u0431|\u05d8\u05e2\u05e1\u05d8|\u0622\u0632\u0645\u0627\u06cc\u0634\u06cc|\u0625\u062e\u062a\u0628\u0627\u0631|\u0627\u0644\u0627\u0631\u062f\u0646|\u0627\u0644\u062c\u0632\u0627\u0626\u0631|\u0627\u0644\u0633\u0639\u0648\u062f\u064a\u0629|\u0627\u0644\u0645\u063a\u0631\u0628|\u0627\u0645\u0627\u0631\u0627\u062a|\u0628\u06be\u0627\u0631\u062a|\u062a\u0648\u0646\u0633|\u0633\u0648\u0631\u064a\u0629|\u0641\u0644\u0633\u0637\u064a\u0646|\u0642\u0637\u0631|\u0645\u0635\u0631|\u092a\u0930\u0940\u0915\u094d\u0937\u093e|\u092d\u093e\u0930\u0924|\u09ad\u09be\u09b0\u09a4|\u0a2d\u0a3e\u0a30\u0a24|\u0aad\u0abe\u0ab0\u0aa4|\u0b87\u0ba8\u0bcd\u0ba4\u0bbf\u0baf\u0bbe|\u0b87\u0bb2\u0b99\u0bcd\u0b95\u0bc8|\u0b9a\u0bbf\u0b99\u0bcd\u0b95\u0baa\u0bcd\u0baa\u0bc2\u0bb0\u0bcd|\u0baa\u0bb0\u0bbf\u0b9f\u0bcd\u0b9a\u0bc8|\u0c2d\u0c3e\u0c30\u0c24\u0c4d|\u0dbd\u0d82\u0d9a\u0dcf|\u0e44\u0e17\u0e22|\u30c6\u30b9\u30c8|\u4e2d\u56fd|\u4e2d\u570b|\u53f0\u6e7e|\u53f0\u7063|\u65b0\u52a0\u5761|\u6d4b\u8bd5|\u6e2c\u8a66|\u9999\u6e2f|\ud14c\uc2a4\ud2b8|\ud55c\uad6d|xn\\-\\-0zwm56d|xn\\-\\-11b5bs3a9aj6g|xn\\-\\-3e0b707e|xn\\-\\-45brj9c|xn\\-\\-80akhbyknj4f|xn\\-\\-90a3ac|xn\\-\\-9t4b11yi5a|xn\\-\\-clchc0ea0b2g2a9gcd|xn\\-\\-deba0ad|xn\\-\\-fiqs8s|xn\\-\\-fiqz9s|xn\\-\\-fpcrj9c3d|xn\\-\\-fzc2c9e2c|xn\\-\\-g6w251d|xn\\-\\-gecrj9c|xn\\-\\-h2brj9c|xn\\-\\-hgbk6aj7f53bba|xn\\-\\-hlcj6aya9esc7a|xn\\-\\-j6w193g|xn\\-\\-jxalpdlp|xn\\-\\-kgbechtv|xn\\-\\-kprw13d|xn\\-\\-kpry57d|xn\\-\\-lgbbat1ad8j|xn\\-\\-mgbaam7a8h|xn\\-\\-mgbayh7gpa|xn\\-\\-mgbbh1a71e|xn\\-\\-mgbc0a9azcg|xn\\-\\-mgberp4a5d4ar|xn\\-\\-o3cw4h|xn\\-\\-ogbpf8fl|xn\\-\\-p1ai|xn\\-\\-pgbs0dh|xn\\-\\-s9brj9c|xn\\-\\-wgbh1c|xn\\-\\-wgbl6a|xn\\-\\-xkc2al3hye2a|xn\\-\\-xkc2dl3a5ee0h|xn\\-\\-yfro4i67o|xn\\-\\-ygbi2ammx|xn\\-\\-zckzah|xxx)"
 46                     + "|y[et]"
 47                     + "|z[amw])";
 48 
 49     /**
 50      *  Regular expression pattern to match all IANA top-level domains.
 51      *  @deprecated This API is deprecated. See {@link #TOP_LEVEL_DOMAIN_STR}.
 52      */
 53     @Deprecated
 54     public static final Pattern TOP_LEVEL_DOMAIN =
 55             Pattern.compile(TOP_LEVEL_DOMAIN_STR);
 56 
 57     /**
 58      *  Regular expression to match all IANA top-level domains for WEB_URL.
 59      *  List accurate as of 2011/07/18.  List taken from:
 60      *  http://data.iana.org/TLD/tlds-alpha-by-domain.txt
 61      *  This pattern is auto-generated by frameworks/ex/common/tools/make-iana-tld-pattern.py
 62      *
 63      *  @deprecated This API is deprecated. See {@link #TOP_LEVEL_DOMAIN_STR}.
 64      */
 65     @Deprecated
 66     public static final String TOP_LEVEL_DOMAIN_STR_FOR_WEB_URL =
 67             "(?:"
 68                     + "(?:aero|arpa|asia|a[cdefgilmnoqrstuwxz])"
 69                     + "|(?:biz|b[abdefghijmnorstvwyz])"
 70                     + "|(?:cat|com|coop|c[acdfghiklmnoruvxyz])"
 71                     + "|d[ejkmoz]"
 72                     + "|(?:edu|e[cegrstu])"
 73                     + "|f[ijkmor]"
 74                     + "|(?:gov|g[abdefghilmnpqrstuwy])"
 75                     + "|h[kmnrtu]"
 76                     + "|(?:info|int|i[delmnoqrst])"
 77                     + "|(?:jobs|j[emop])"
 78                     + "|k[eghimnprwyz]"
 79                     + "|l[abcikrstuvy]"
 80                     + "|(?:mil|mobi|museum|m[acdeghklmnopqrstuvwxyz])"
 81                     + "|(?:name|net|n[acefgilopruz])"
 82                     + "|(?:org|om)"
 83                     + "|(?:pro|p[aefghklmnrstwy])"
 84                     + "|qa"
 85                     + "|r[eosuw]"
 86                     + "|s[abcdeghijklmnortuvyz]"
 87                     + "|(?:tel|travel|t[cdfghjklmnoprtvwz])"
 88                     + "|u[agksyz]"
 89                     + "|v[aceginu]"
 90                     + "|w[fs]"
 91                     + "|(?:\u03b4\u03bf\u03ba\u03b9\u03bc\u03ae|\u0438\u0441\u043f\u044b\u0442\u0430\u043d\u0438\u0435|\u0440\u0444|\u0441\u0440\u0431|\u05d8\u05e2\u05e1\u05d8|\u0622\u0632\u0645\u0627\u06cc\u0634\u06cc|\u0625\u062e\u062a\u0628\u0627\u0631|\u0627\u0644\u0627\u0631\u062f\u0646|\u0627\u0644\u062c\u0632\u0627\u0626\u0631|\u0627\u0644\u0633\u0639\u0648\u062f\u064a\u0629|\u0627\u0644\u0645\u063a\u0631\u0628|\u0627\u0645\u0627\u0631\u0627\u062a|\u0628\u06be\u0627\u0631\u062a|\u062a\u0648\u0646\u0633|\u0633\u0648\u0631\u064a\u0629|\u0641\u0644\u0633\u0637\u064a\u0646|\u0642\u0637\u0631|\u0645\u0635\u0631|\u092a\u0930\u0940\u0915\u094d\u0937\u093e|\u092d\u093e\u0930\u0924|\u09ad\u09be\u09b0\u09a4|\u0a2d\u0a3e\u0a30\u0a24|\u0aad\u0abe\u0ab0\u0aa4|\u0b87\u0ba8\u0bcd\u0ba4\u0bbf\u0baf\u0bbe|\u0b87\u0bb2\u0b99\u0bcd\u0b95\u0bc8|\u0b9a\u0bbf\u0b99\u0bcd\u0b95\u0baa\u0bcd\u0baa\u0bc2\u0bb0\u0bcd|\u0baa\u0bb0\u0bbf\u0b9f\u0bcd\u0b9a\u0bc8|\u0c2d\u0c3e\u0c30\u0c24\u0c4d|\u0dbd\u0d82\u0d9a\u0dcf|\u0e44\u0e17\u0e22|\u30c6\u30b9\u30c8|\u4e2d\u56fd|\u4e2d\u570b|\u53f0\u6e7e|\u53f0\u7063|\u65b0\u52a0\u5761|\u6d4b\u8bd5|\u6e2c\u8a66|\u9999\u6e2f|\ud14c\uc2a4\ud2b8|\ud55c\uad6d|xn\\-\\-0zwm56d|xn\\-\\-11b5bs3a9aj6g|xn\\-\\-3e0b707e|xn\\-\\-45brj9c|xn\\-\\-80akhbyknj4f|xn\\-\\-90a3ac|xn\\-\\-9t4b11yi5a|xn\\-\\-clchc0ea0b2g2a9gcd|xn\\-\\-deba0ad|xn\\-\\-fiqs8s|xn\\-\\-fiqz9s|xn\\-\\-fpcrj9c3d|xn\\-\\-fzc2c9e2c|xn\\-\\-g6w251d|xn\\-\\-gecrj9c|xn\\-\\-h2brj9c|xn\\-\\-hgbk6aj7f53bba|xn\\-\\-hlcj6aya9esc7a|xn\\-\\-j6w193g|xn\\-\\-jxalpdlp|xn\\-\\-kgbechtv|xn\\-\\-kprw13d|xn\\-\\-kpry57d|xn\\-\\-lgbbat1ad8j|xn\\-\\-mgbaam7a8h|xn\\-\\-mgbayh7gpa|xn\\-\\-mgbbh1a71e|xn\\-\\-mgbc0a9azcg|xn\\-\\-mgberp4a5d4ar|xn\\-\\-o3cw4h|xn\\-\\-ogbpf8fl|xn\\-\\-p1ai|xn\\-\\-pgbs0dh|xn\\-\\-s9brj9c|xn\\-\\-wgbh1c|xn\\-\\-wgbl6a|xn\\-\\-xkc2al3hye2a|xn\\-\\-xkc2dl3a5ee0h|xn\\-\\-yfro4i67o|xn\\-\\-ygbi2ammx|xn\\-\\-zckzah|xxx)"
 92                     + "|y[et]"
 93                     + "|z[amw]))";
 94 
 95     /**
 96      *  Regular expression to match all IANA top-level domains.
 97      *
 98      *  List accurate as of 2015/11/24.  List taken from:
 99      *  http://data.iana.org/TLD/tlds-alpha-by-domain.txt
100      *  This pattern is auto-generated by frameworks/ex/common/tools/make-iana-tld-pattern.py
101      *
102      *  @hide
103      */
104     static final String IANA_TOP_LEVEL_DOMAINS =
105             "(?:"
106                     + "(?:aaa|aarp|abb|abbott|abogado|academy|accenture|accountant|accountants|aco|active"
107                     + "|actor|ads|adult|aeg|aero|afl|agency|aig|airforce|airtel|allfinanz|alsace|amica|amsterdam"
108                     + "|android|apartments|app|apple|aquarelle|aramco|archi|army|arpa|arte|asia|associates"
109                     + "|attorney|auction|audio|auto|autos|axa|azure|a[cdefgilmoqrstuwxz])"
110                     + "|(?:band|bank|bar|barcelona|barclaycard|barclays|bargains|bauhaus|bayern|bbc|bbva"
111                     + "|bcn|beats|beer|bentley|berlin|best|bet|bharti|bible|bid|bike|bing|bingo|bio|biz|black"
112                     + "|blackfriday|bloomberg|blue|bms|bmw|bnl|bnpparibas|boats|bom|bond|boo|boots|boutique"
113                     + "|bradesco|bridgestone|broadway|broker|brother|brussels|budapest|build|builders|business"
114                     + "|buzz|bzh|b[abdefghijmnorstvwyz])"
115                     + "|(?:cab|cafe|cal|camera|camp|cancerresearch|canon|capetown|capital|car|caravan|cards"
116                     + "|care|career|careers|cars|cartier|casa|cash|casino|cat|catering|cba|cbn|ceb|center|ceo"
117                     + "|cern|cfa|cfd|chanel|channel|chat|cheap|chloe|christmas|chrome|church|cipriani|cisco"
118                     + "|citic|city|cityeats|claims|cleaning|click|clinic|clothing|cloud|club|clubmed|coach"
119                     + "|codes|coffee|college|cologne|com|commbank|community|company|computer|comsec|condos"
120                     + "|construction|consulting|contractors|cooking|cool|coop|corsica|country|coupons|courses"
121                     + "|credit|creditcard|creditunion|cricket|crown|crs|cruises|csc|cuisinella|cymru|cyou|c[acdfghiklmnoruvwxyz])"
122                     + "|(?:dabur|dad|dance|date|dating|datsun|day|dclk|deals|degree|delivery|dell|delta"
123                     + "|democrat|dental|dentist|desi|design|dev|diamonds|diet|digital|direct|directory|discount"
124                     + "|dnp|docs|dog|doha|domains|doosan|download|drive|durban|dvag|d[ejkmoz])"
125                     + "|(?:earth|eat|edu|education|email|emerck|energy|engineer|engineering|enterprises"
126                     + "|epson|equipment|erni|esq|estate|eurovision|eus|events|everbank|exchange|expert|exposed"
127                     + "|express|e[cegrstu])"
128                     + "|(?:fage|fail|fairwinds|faith|family|fan|fans|farm|fashion|feedback|ferrero|film"
129                     + "|final|finance|financial|firmdale|fish|fishing|fit|fitness|flights|florist|flowers|flsmidth"
130                     + "|fly|foo|football|forex|forsale|forum|foundation|frl|frogans|fund|furniture|futbol|fyi"
131                     + "|f[ijkmor])"
132                     + "|(?:gal|gallery|game|garden|gbiz|gdn|gea|gent|genting|ggee|gift|gifts|gives|giving"
133                     + "|glass|gle|global|globo|gmail|gmo|gmx|gold|goldpoint|golf|goo|goog|google|gop|gov|grainger"
134                     + "|graphics|gratis|green|gripe|group|gucci|guge|guide|guitars|guru|g[abdefghilmnpqrstuwy])"
135                     + "|(?:hamburg|hangout|haus|healthcare|help|here|hermes|hiphop|hitachi|hiv|hockey|holdings"
136                     + "|holiday|homedepot|homes|honda|horse|host|hosting|hoteles|hotmail|house|how|hsbc|hyundai"
137                     + "|h[kmnrtu])"
138                     + "|(?:ibm|icbc|ice|icu|ifm|iinet|immo|immobilien|industries|infiniti|info|ing|ink|institute"
139                     + "|insure|int|international|investments|ipiranga|irish|ist|istanbul|itau|iwc|i[delmnoqrst])"
140                     + "|(?:jaguar|java|jcb|jetzt|jewelry|jlc|jll|jobs|joburg|jprs|juegos|j[emop])"
141                     + "|(?:kaufen|kddi|kia|kim|kinder|kitchen|kiwi|koeln|komatsu|krd|kred|kyoto|k[eghimnprwyz])"
142                     + "|(?:lacaixa|lancaster|land|landrover|lasalle|lat|latrobe|law|lawyer|lds|lease|leclerc"
143                     + "|legal|lexus|lgbt|liaison|lidl|life|lifestyle|lighting|limited|limo|linde|link|live"
144                     + "|lixil|loan|loans|lol|london|lotte|lotto|love|ltd|ltda|lupin|luxe|luxury|l[abcikrstuvy])"
145                     + "|(?:madrid|maif|maison|man|management|mango|market|marketing|markets|marriott|mba"
146                     + "|media|meet|melbourne|meme|memorial|men|menu|meo|miami|microsoft|mil|mini|mma|mobi|moda"
147                     + "|moe|moi|mom|monash|money|montblanc|mormon|mortgage|moscow|motorcycles|mov|movie|movistar"
148                     + "|mtn|mtpc|mtr|museum|mutuelle|m[acdeghklmnopqrstuvwxyz])"
149                     + "|(?:nadex|nagoya|name|navy|nec|net|netbank|network|neustar|new|news|nexus|ngo|nhk"
150                     + "|nico|ninja|nissan|nokia|nra|nrw|ntt|nyc|n[acefgilopruz])"
151                     + "|(?:obi|office|okinawa|omega|one|ong|onl|online|ooo|oracle|orange|org|organic|osaka"
152                     + "|otsuka|ovh|om)"
153                     + "|(?:page|panerai|paris|partners|parts|party|pet|pharmacy|philips|photo|photography"
154                     + "|photos|physio|piaget|pics|pictet|pictures|ping|pink|pizza|place|play|playstation|plumbing"
155                     + "|plus|pohl|poker|porn|post|praxi|press|pro|prod|productions|prof|properties|property"
156                     + "|protection|pub|p[aefghklmnrstwy])"
157                     + "|(?:qpon|quebec|qa)"
158                     + "|(?:racing|realtor|realty|recipes|red|redstone|rehab|reise|reisen|reit|ren|rent|rentals"
159                     + "|repair|report|republican|rest|restaurant|review|reviews|rich|ricoh|rio|rip|rocher|rocks"
160                     + "|rodeo|rsvp|ruhr|run|rwe|ryukyu|r[eosuw])"
161                     + "|(?:saarland|sakura|sale|samsung|sandvik|sandvikcoromant|sanofi|sap|sapo|sarl|saxo"
162                     + "|sbs|sca|scb|schmidt|scholarships|school|schule|schwarz|science|scor|scot|seat|security"
163                     + "|seek|sener|services|seven|sew|sex|sexy|shiksha|shoes|show|shriram|singles|site|ski"
164                     + "|sky|skype|sncf|soccer|social|software|sohu|solar|solutions|sony|soy|space|spiegel|spreadbetting"
165                     + "|srl|stada|starhub|statoil|stc|stcgroup|stockholm|studio|study|style|sucks|supplies"
166                     + "|supply|support|surf|surgery|suzuki|swatch|swiss|sydney|systems|s[abcdeghijklmnortuvxyz])"
167                     + "|(?:tab|taipei|tatamotors|tatar|tattoo|tax|taxi|team|tech|technology|tel|telefonica"
168                     + "|temasek|tennis|thd|theater|theatre|tickets|tienda|tips|tires|tirol|today|tokyo|tools"
169                     + "|top|toray|toshiba|tours|town|toyota|toys|trade|trading|training|travel|trust|tui|t[cdfghjklmnortvwz])"
170                     + "|(?:ubs|university|uno|uol|u[agksyz])"
171                     + "|(?:vacations|vana|vegas|ventures|versicherung|vet|viajes|video|villas|vin|virgin"
172                     + "|vision|vista|vistaprint|viva|vlaanderen|vodka|vote|voting|voto|voyage|v[aceginu])"
173                     + "|(?:wales|walter|wang|watch|webcam|website|wed|wedding|weir|whoswho|wien|wiki|williamhill"
174                     + "|win|windows|wine|wme|work|works|world|wtc|wtf|w[fs])"
175                     + "|(?:\u03b5\u03bb|\u0431\u0435\u043b|\u0434\u0435\u0442\u0438|\u043a\u043e\u043c|\u043c\u043a\u0434"
176                     + "|\u043c\u043e\u043d|\u043c\u043e\u0441\u043a\u0432\u0430|\u043e\u043d\u043b\u0430\u0439\u043d"
177                     + "|\u043e\u0440\u0433|\u0440\u0443\u0441|\u0440\u0444|\u0441\u0430\u0439\u0442|\u0441\u0440\u0431"
178                     + "|\u0443\u043a\u0440|\u049b\u0430\u0437|\u0570\u0561\u0575|\u05e7\u05d5\u05dd|\u0627\u0631\u0627\u0645\u0643\u0648"
179                     + "|\u0627\u0644\u0627\u0631\u062f\u0646|\u0627\u0644\u062c\u0632\u0627\u0626\u0631|\u0627\u0644\u0633\u0639\u0648\u062f\u064a\u0629"
180                     + "|\u0627\u0644\u0645\u063a\u0631\u0628|\u0627\u0645\u0627\u0631\u0627\u062a|\u0627\u06cc\u0631\u0627\u0646"
181                     + "|\u0628\u0627\u0632\u0627\u0631|\u0628\u06be\u0627\u0631\u062a|\u062a\u0648\u0646\u0633"
182                     + "|\u0633\u0648\u062f\u0627\u0646|\u0633\u0648\u0631\u064a\u0629|\u0634\u0628\u0643\u0629"
183                     + "|\u0639\u0631\u0627\u0642|\u0639\u0645\u0627\u0646|\u0641\u0644\u0633\u0637\u064a\u0646"
184                     + "|\u0642\u0637\u0631|\u0643\u0648\u0645|\u0645\u0635\u0631|\u0645\u0644\u064a\u0633\u064a\u0627"
185                     + "|\u0645\u0648\u0642\u0639|\u0915\u0949\u092e|\u0928\u0947\u091f|\u092d\u093e\u0930\u0924"
186                     + "|\u0938\u0902\u0917\u0920\u0928|\u09ad\u09be\u09b0\u09a4|\u0a2d\u0a3e\u0a30\u0a24|\u0aad\u0abe\u0ab0\u0aa4"
187                     + "|\u0b87\u0ba8\u0bcd\u0ba4\u0bbf\u0baf\u0bbe|\u0b87\u0bb2\u0b99\u0bcd\u0b95\u0bc8|\u0b9a\u0bbf\u0b99\u0bcd\u0b95\u0baa\u0bcd\u0baa\u0bc2\u0bb0\u0bcd"
188                     + "|\u0c2d\u0c3e\u0c30\u0c24\u0c4d|\u0dbd\u0d82\u0d9a\u0dcf|\u0e04\u0e2d\u0e21|\u0e44\u0e17\u0e22"
189                     + "|\u10d2\u10d4|\u307f\u3093\u306a|\u30b0\u30fc\u30b0\u30eb|\u30b3\u30e0|\u4e16\u754c"
190                     + "|\u4e2d\u4fe1|\u4e2d\u56fd|\u4e2d\u570b|\u4e2d\u6587\u7f51|\u4f01\u4e1a|\u4f5b\u5c71"
191                     + "|\u4fe1\u606f|\u5065\u5eb7|\u516b\u5366|\u516c\u53f8|\u516c\u76ca|\u53f0\u6e7e|\u53f0\u7063"
192                     + "|\u5546\u57ce|\u5546\u5e97|\u5546\u6807|\u5728\u7ebf|\u5927\u62ff|\u5a31\u4e50|\u5de5\u884c"
193                     + "|\u5e7f\u4e1c|\u6148\u5584|\u6211\u7231\u4f60|\u624b\u673a|\u653f\u52a1|\u653f\u5e9c"
194                     + "|\u65b0\u52a0\u5761|\u65b0\u95fb|\u65f6\u5c1a|\u673a\u6784|\u6de1\u9a6c\u9521|\u6e38\u620f"
195                     + "|\u70b9\u770b|\u79fb\u52a8|\u7ec4\u7ec7\u673a\u6784|\u7f51\u5740|\u7f51\u5e97|\u7f51\u7edc"
196                     + "|\u8c37\u6b4c|\u96c6\u56e2|\u98de\u5229\u6d66|\u9910\u5385|\u9999\u6e2f|\ub2f7\ub137"
197                     + "|\ub2f7\ucef4|\uc0bc\uc131|\ud55c\uad6d|xbox"
198                     + "|xerox|xin|xn\\-\\-11b4c3d|xn\\-\\-1qqw23a|xn\\-\\-30rr7y|xn\\-\\-3bst00m|xn\\-\\-3ds443g"
199                     + "|xn\\-\\-3e0b707e|xn\\-\\-3pxu8k|xn\\-\\-42c2d9a|xn\\-\\-45brj9c|xn\\-\\-45q11c|xn\\-\\-4gbrim"
200                     + "|xn\\-\\-55qw42g|xn\\-\\-55qx5d|xn\\-\\-6frz82g|xn\\-\\-6qq986b3xl|xn\\-\\-80adxhks"
201                     + "|xn\\-\\-80ao21a|xn\\-\\-80asehdb|xn\\-\\-80aswg|xn\\-\\-90a3ac|xn\\-\\-90ais|xn\\-\\-9dbq2a"
202                     + "|xn\\-\\-9et52u|xn\\-\\-b4w605ferd|xn\\-\\-c1avg|xn\\-\\-c2br7g|xn\\-\\-cg4bki|xn\\-\\-clchc0ea0b2g2a9gcd"
203                     + "|xn\\-\\-czr694b|xn\\-\\-czrs0t|xn\\-\\-czru2d|xn\\-\\-d1acj3b|xn\\-\\-d1alf|xn\\-\\-efvy88h"
204                     + "|xn\\-\\-estv75g|xn\\-\\-fhbei|xn\\-\\-fiq228c5hs|xn\\-\\-fiq64b|xn\\-\\-fiqs8s|xn\\-\\-fiqz9s"
205                     + "|xn\\-\\-fjq720a|xn\\-\\-flw351e|xn\\-\\-fpcrj9c3d|xn\\-\\-fzc2c9e2c|xn\\-\\-gecrj9c"
206                     + "|xn\\-\\-h2brj9c|xn\\-\\-hxt814e|xn\\-\\-i1b6b1a6a2e|xn\\-\\-imr513n|xn\\-\\-io0a7i"
207                     + "|xn\\-\\-j1aef|xn\\-\\-j1amh|xn\\-\\-j6w193g|xn\\-\\-kcrx77d1x4a|xn\\-\\-kprw13d|xn\\-\\-kpry57d"
208                     + "|xn\\-\\-kput3i|xn\\-\\-l1acc|xn\\-\\-lgbbat1ad8j|xn\\-\\-mgb9awbf|xn\\-\\-mgba3a3ejt"
209                     + "|xn\\-\\-mgba3a4f16a|xn\\-\\-mgbaam7a8h|xn\\-\\-mgbab2bd|xn\\-\\-mgbayh7gpa|xn\\-\\-mgbbh1a71e"
210                     + "|xn\\-\\-mgbc0a9azcg|xn\\-\\-mgberp4a5d4ar|xn\\-\\-mgbpl2fh|xn\\-\\-mgbtx2b|xn\\-\\-mgbx4cd0ab"
211                     + "|xn\\-\\-mk1bu44c|xn\\-\\-mxtq1m|xn\\-\\-ngbc5azd|xn\\-\\-node|xn\\-\\-nqv7f|xn\\-\\-nqv7fs00ema"
212                     + "|xn\\-\\-nyqy26a|xn\\-\\-o3cw4h|xn\\-\\-ogbpf8fl|xn\\-\\-p1acf|xn\\-\\-p1ai|xn\\-\\-pgbs0dh"
213                     + "|xn\\-\\-pssy2u|xn\\-\\-q9jyb4c|xn\\-\\-qcka1pmc|xn\\-\\-qxam|xn\\-\\-rhqv96g|xn\\-\\-s9brj9c"
214                     + "|xn\\-\\-ses554g|xn\\-\\-t60b56a|xn\\-\\-tckwe|xn\\-\\-unup4y|xn\\-\\-vermgensberater\\-ctb"
215                     + "|xn\\-\\-vermgensberatung\\-pwb|xn\\-\\-vhquv|xn\\-\\-vuq861b|xn\\-\\-wgbh1c|xn\\-\\-wgbl6a"
216                     + "|xn\\-\\-xhq521b|xn\\-\\-xkc2al3hye2a|xn\\-\\-xkc2dl3a5ee0h|xn\\-\\-y9a3aq|xn\\-\\-yfro4i67o"
217                     + "|xn\\-\\-ygbi2ammx|xn\\-\\-zfr164b|xperia|xxx|xyz)"
218                     + "|(?:yachts|yamaxun|yandex|yodobashi|yoga|yokohama|youtube|y[et])"
219                     + "|(?:zara|zip|zone|zuerich|z[amw]))";
220 
221     /**
222      * Kept for backward compatibility reasons.
223      *
224      * @deprecated Deprecated since it does not include all IRI characters defined in RFC 3987
225      */
226     @Deprecated
227     public static final String GOOD_IRI_CHAR =
228             "a-zA-Z0-9\u00A0-\uD7FF\uF900-\uFDCF\uFDF0-\uFFEF";
229 
230     public static final Pattern IP_ADDRESS
231             = Pattern.compile(
232             "((25[0-5]|2[0-4][0-9]|[0-1][0-9]{2}|[1-9][0-9]|[1-9])\\.(25[0-5]|2[0-4]"
233                     + "[0-9]|[0-1][0-9]{2}|[1-9][0-9]|[1-9]|0)\\.(25[0-5]|2[0-4][0-9]|[0-1]"
234                     + "[0-9]{2}|[1-9][0-9]|[1-9]|0)\\.(25[0-5]|2[0-4][0-9]|[0-1][0-9]{2}"
235                     + "|[1-9][0-9]|[0-9]))");
236 
237     /**
238      * Valid UCS characters defined in RFC 3987. Excludes space characters.
239      */
240     private static final String UCS_CHAR = "[" +
241             "\u00A0-\uD7FF" +
242             "\uF900-\uFDCF" +
243             "\uFDF0-\uFFEF" +
244             "\uD800\uDC00-\uD83F\uDFFD" +
245             "\uD840\uDC00-\uD87F\uDFFD" +
246             "\uD880\uDC00-\uD8BF\uDFFD" +
247             "\uD8C0\uDC00-\uD8FF\uDFFD" +
248             "\uD900\uDC00-\uD93F\uDFFD" +
249             "\uD940\uDC00-\uD97F\uDFFD" +
250             "\uD980\uDC00-\uD9BF\uDFFD" +
251             "\uD9C0\uDC00-\uD9FF\uDFFD" +
252             "\uDA00\uDC00-\uDA3F\uDFFD" +
253             "\uDA40\uDC00-\uDA7F\uDFFD" +
254             "\uDA80\uDC00-\uDABF\uDFFD" +
255             "\uDAC0\uDC00-\uDAFF\uDFFD" +
256             "\uDB00\uDC00-\uDB3F\uDFFD" +
257             "\uDB44\uDC00-\uDB7F\uDFFD" +
258             "&&[^\u00A0[\u2000-\u200A]\u2028\u2029\u202F\u3000]]";
259 
260     /**
261      * Valid characters for IRI label defined in RFC 3987.
262      */
263     private static final String LABEL_CHAR = "a-zA-Z0-9" + UCS_CHAR;
264 
265     /**
266      * Valid characters for IRI TLD defined in RFC 3987.
267      */
268     private static final String TLD_CHAR = "a-zA-Z" + UCS_CHAR;
269 
270     /**
271      * RFC 1035 Section 2.3.4 limits the labels to a maximum 63 octets.
272      */
273     private static final String IRI_LABEL =
274             "[" + LABEL_CHAR + "](?:[" + LABEL_CHAR + "\\-]{0,61}[" + LABEL_CHAR + "]){0,1}";
275 
276     /**
277      * RFC 3492 references RFC 1034 and limits Punycode algorithm output to 63 characters.
278      */
279     private static final String PUNYCODE_TLD = "xn\\-\\-[\\w\\-]{0,58}\\w";
280 
281     private static final String TLD = "(" + PUNYCODE_TLD + "|" + "[" + TLD_CHAR + "]{2,63}" +")";
282 
283     private static final String HOST_NAME = "(" + IRI_LABEL + "\\.)+" + TLD;
284 
285     public static final Pattern DOMAIN_NAME
286             = Pattern.compile("(" + HOST_NAME + "|" + IP_ADDRESS + ")");
287 
288     private static final String PROTOCOL = "(?i:http|https|rtsp):\\/\\/";
289 
290     /* A word boundary or end of input.  This is to stop foo.sure from matching as foo.su */
291     private static final String WORD_BOUNDARY = "(?:\\b|$|^)";
292 
293     private static final String USER_INFO = "(?:[a-zA-Z0-9\\$\\-\\_\\.\\+\\!\\*\\'\\(\\)"
294             + "\\,\\;\\?\\&\\=]|(?:\\%[a-fA-F0-9]{2})){1,64}(?:\\:(?:[a-zA-Z0-9\\$\\-\\_"
295             + "\\.\\+\\!\\*\\'\\(\\)\\,\\;\\?\\&\\=]|(?:\\%[a-fA-F0-9]{2})){1,25})?\\@";
296 
297     private static final String PORT_NUMBER = "\\:\\d{1,5}";
298 
299     private static final String PATH_AND_QUERY = "\\/(?:(?:[" + LABEL_CHAR
300             + "\\;\\/\\?\\:\\@\\&\\=\\#\\~"  // plus optional query params
301             + "\\-\\.\\+\\!\\*\\'\\(\\)\\,\\_])|(?:\\%[a-fA-F0-9]{2}))*";
302 
303     /**
304      *  Regular expression pattern to match most part of RFC 3987
305      *  Internationalized URLs, aka IRIs.
306      */
307     public static final Pattern WEB_URL = Pattern.compile("("
308             + "("
309             + "(?:" + PROTOCOL + "(?:" + USER_INFO + ")?" + ")?"
310             + "(?:" + DOMAIN_NAME + ")"
311             + "(?:" + PORT_NUMBER + ")?"
312             + ")"
313             + "(" + PATH_AND_QUERY + ")?"
314             + WORD_BOUNDARY
315             + ")");
316 
317     /**
318      * Regular expression that matches known TLDs and punycode TLDs
319      */
320     private static final String STRICT_TLD = "(?:" +
321             IANA_TOP_LEVEL_DOMAINS + "|" + PUNYCODE_TLD + ")";
322 
323     /**
324      * Regular expression that matches host names using {@link #STRICT_TLD}
325      */
326     private static final String STRICT_HOST_NAME = "(?:(?:" + IRI_LABEL + "\\.)+"
327             + STRICT_TLD + ")";
328 
329     /**
330      * Regular expression that matches domain names using either {@link #STRICT_HOST_NAME} or
331      * {@link #IP_ADDRESS}
332      */
333     private static final Pattern STRICT_DOMAIN_NAME
334             = Pattern.compile("(?:" + STRICT_HOST_NAME + "|" + IP_ADDRESS + ")");
335 
336     /**
337      * Regular expression that matches domain names without a TLD
338      */
339     private static final String RELAXED_DOMAIN_NAME =
340             "(?:" + "(?:" + IRI_LABEL + "(?:\\.(?=\\S))" +"?)+" + "|" + IP_ADDRESS + ")";
341 
342     /**
343      * Regular expression to match strings that do not start with a supported protocol. The TLDs
344      * are expected to be one of the known TLDs.
345      */
346     private static final String WEB_URL_WITHOUT_PROTOCOL = "("
347             + WORD_BOUNDARY
348             + "(?<!:\\/\\/)"
349             + "("
350             + "(?:" + STRICT_DOMAIN_NAME + ")"
351             + "(?:" + PORT_NUMBER + ")?"
352             + ")"
353             + "(?:" + PATH_AND_QUERY + ")?"
354             + WORD_BOUNDARY
355             + ")";
356 
357     /**
358      * Regular expression to match strings that start with a supported protocol. Rules for domain
359      * names and TLDs are more relaxed. TLDs are optional.
360      */
361     private static final String WEB_URL_WITH_PROTOCOL = "("
362             + WORD_BOUNDARY
363             + "(?:"
364             + "(?:" + PROTOCOL + "(?:" + USER_INFO + ")?" + ")"
365             + "(?:" + RELAXED_DOMAIN_NAME + ")?"
366             + "(?:" + PORT_NUMBER + ")?"
367             + ")"
368             + "(?:" + PATH_AND_QUERY + ")?"
369             + WORD_BOUNDARY
370             + ")";
371 
372     /**
373      * Regular expression pattern to match IRIs. If a string starts with http(s):// the expression
374      * tries to match the URL structure with a relaxed rule for TLDs. If the string does not start
375      * with http(s):// the TLDs are expected to be one of the known TLDs.
376      *
377      * @hide
378      */
379     public static final Pattern AUTOLINK_WEB_URL = Pattern.compile(
380             "(" + WEB_URL_WITH_PROTOCOL + "|" + WEB_URL_WITHOUT_PROTOCOL + ")");
381 
382     /**
383      * Regular expression for valid email characters. Does not include some of the valid characters
384      * defined in RFC5321: #&~!^`{}/=$*?|
385      */
386     private static final String EMAIL_CHAR = LABEL_CHAR + "\\+\\-_%'";
387 
388     /**
389      * Regular expression for local part of an email address. RFC5321 section 4.5.3.1.1 limits
390      * the local part to be at most 64 octets.
391      */
392     private static final String EMAIL_ADDRESS_LOCAL_PART =
393             "[" + EMAIL_CHAR + "]" + "(?:[" + EMAIL_CHAR + "\\.]{1,62}[" + EMAIL_CHAR + "])?";
394 
395     /**
396      * Regular expression for the domain part of an email address. RFC5321 section 4.5.3.1.2 limits
397      * the domain to be at most 255 octets.
398      */
399     private static final String EMAIL_ADDRESS_DOMAIN =
400             "(?=.{1,255}(?:\\s|$|^))" + HOST_NAME;
401 
402     /**
403      * Regular expression pattern to match email addresses. It excludes double quoted local parts
404      * and the special characters #&~!^`{}/=$*?| that are included in RFC5321.
405      * @hide
406      */
407     public static final Pattern AUTOLINK_EMAIL_ADDRESS = Pattern.compile("(" + WORD_BOUNDARY +
408             "(?:" + EMAIL_ADDRESS_LOCAL_PART + "@" + EMAIL_ADDRESS_DOMAIN + ")" +
409             WORD_BOUNDARY + ")"
410     );
411 
412     public static final Pattern EMAIL_ADDRESS
413             = Pattern.compile(
414             "[a-zA-Z0-9\\+\\.\\_\\%\\-\\+]{1,256}" +
415                     "\\@" +
416                     "[a-zA-Z0-9][a-zA-Z0-9\\-]{0,64}" +
417                     "(" +
418                     "\\." +
419                     "[a-zA-Z0-9][a-zA-Z0-9\\-]{0,25}" +
420                     ")+"
421     );
422 
423     /**
424      * This pattern is intended for searching for things that look like they
425      * might be phone numbers in arbitrary text, not for validating whether
426      * something is in fact a phone number.  It will miss many things that
427      * are legitimate phone numbers.
428      *
429      * <p> The pattern matches the following:
430      * <ul>
431      * <li>Optionally, a + sign followed immediately by one or more digits. Spaces, dots, or dashes
432      * may follow.
433      * <li>Optionally, sets of digits in parentheses, separated by spaces, dots, or dashes.
434      * <li>A string starting and ending with a digit, containing digits, spaces, dots, and/or dashes.
435      * </ul>
436      */
437     public static final Pattern PHONE
438             = Pattern.compile(                      // sdd = space, dot, or dash
439             "(\\+[0-9]+[\\- \\.]*)?"        // +<digits><sdd>*
440                     + "(\\([0-9]+\\)[\\- \\.]*)?"   // (<digits>)<sdd>*
441                     + "([0-9][0-9\\- \\.]+[0-9])"); // <digit><digit|sdd>+<digit>
442 
443     /**
444      *  Convenience method to take all of the non-null matching groups in a
445      *  regex Matcher and return them as a concatenated string.
446      *
447      *  @param matcher      The Matcher object from which grouped text will
448      *                      be extracted
449      *
450      *  @return             A String comprising all of the non-null matched
451      *                      groups concatenated together
452      */
453     public static final String concatGroups(Matcher matcher) {
454         StringBuilder b = new StringBuilder();
455         final int numGroups = matcher.groupCount();
456 
457         for (int i = 1; i <= numGroups; i++) {
458             String s = matcher.group(i);
459 
460             if (s != null) {
461                 b.append(s);
462             }
463         }
464 
465         return b.toString();
466     }
467 
468     /**
469      * Convenience method to return only the digits and plus signs
470      * in the matching string.
471      *
472      * @param matcher      The Matcher object from which digits and plus will
473      *                     be extracted
474      *
475      * @return             A String comprising all of the digits and plus in
476      *                     the match
477      */
478     public static final String digitsAndPlusOnly(Matcher matcher) {
479         StringBuilder buffer = new StringBuilder();
480         String matchingRegion = matcher.group();
481 
482         for (int i = 0, size = matchingRegion.length(); i < size; i++) {
483             char character = matchingRegion.charAt(i);
484 
485             if (character == '+' || Character.isDigit(character)) {
486                 buffer.append(character);
487             }
488         }
489         return buffer.toString();
490     }
491 
492     /**
493      * Do not create this static utility class.
494      */
495     private Patterns() {}
496 }

 



调用方法

 

       Matcher matcher = Patterns.WEB_URL.matcher(message);
        while (matcher.find()) {
            //循环输出所有匹配到的链接,并加上链接
            String link = matcher.group();
            String restr = "<a href='" + link + "' target='_blank'>" + link + "</a>";
            message = message.replaceAll(link, restr);
        }

　　

 

 

说明：

　　1、message为要进行提取的字符串

　　2、http/https可以获取，但是在链接结尾不能和其他文字连在一起，不然获取不准确

　　即："你真的是https://www.cnblogs.com/pxblog博客网"会获取到"https://www.cnblogs.com/pxblog博客网"

　　  　 "你真的是https://www.cnblogs.com/pxblog 博客网"会获取到"https://www.cnblogs.com/pxblog"

posted @ 2020-04-01 02:29  yvioo  阅读(4259)  评论(0)  编辑  收藏  举报