Mercurial Hosting > luan
annotate src/goodjava/html/Html.java @ 1764:527c53b91a50
lucene error handling
| author | Franklin Schmidt <fschmidt@gmail.com> |
|---|---|
| date | Mon, 22 May 2023 20:43:52 -0600 |
| parents | 31a82b0d0a87 |
| children | a045f30fa67d |
| rev | line source |
|---|---|
| 1712 | 1 package goodjava.html; |
| 2 | |
| 3 import java.util.List; | |
| 4 import java.util.ArrayList; | |
| 5 import java.util.Set; | |
| 6 import java.util.HashSet; | |
| 7 import java.util.Map; | |
| 8 import java.util.HashMap; | |
| 9 import java.util.Collections; | |
| 10 import java.util.regex.Pattern; | |
| 11 import java.util.regex.Matcher; | |
| 12 import goodjava.parser.Parser; | |
| 13 | |
| 14 | |
| 15 public final class Html { | |
| 16 | |
| 17 private static final Pattern entityPtn = Pattern.compile( | |
| 18 "&(#?[0-9a-zA-Z]+;)" | |
| 19 ); | |
| 20 | |
| 21 public static String encode(String s) { | |
| 22 //s = s.replace("&","&"); | |
| 23 s = entityPtn.matcher(s).replaceAll("&$1"); | |
| 24 s = s.replace("<","<"); | |
| 25 s = s.replace(">",">"); | |
| 26 s = s.replace("\"","""); | |
| 27 return s; | |
| 28 } | |
| 29 | |
| 30 private static final Pattern entityNumPtn = Pattern.compile( | |
| 31 "&#(\\d+);" | |
| 32 ); | |
| 33 | |
| 34 public static String decode(String s) { | |
| 35 Matcher m = entityNumPtn.matcher(s); | |
| 36 if( m.find() ) { | |
| 37 StringBuffer buf = new StringBuffer(); | |
| 38 do { | |
| 39 String entity = new String(new char[]{(char)Integer.parseInt(m.group(1))}); | |
| 40 m.appendReplacement(buf,entity); | |
| 41 } while( m.find() ); | |
| 42 m.appendTail(buf); | |
| 43 s = buf.toString(); | |
| 44 } | |
| 45 s = s.replace(" "," "); | |
| 46 s = s.replace(""","\""); | |
| 47 s = s.replace(">",">"); | |
| 48 s = s.replace("<","<"); | |
| 49 s = s.replace("&","&"); | |
| 50 return s; | |
| 51 } | |
| 52 | |
| 53 | |
| 54 public static final class Comment { | |
| 55 public final String text; | |
| 56 | |
| 57 private Comment(String text) { | |
| 58 this.text = text; | |
| 59 } | |
| 60 } | |
| 61 | |
| 62 public static final class CData { | |
| 63 public final String text; | |
| 64 | |
| 65 private CData(String text) { | |
| 66 this.text = text; | |
| 67 } | |
| 68 } | |
| 69 | |
| 70 public static final class Tag { | |
| 71 public final String name; | |
| 72 public final Map<String,Object> attributes; | |
| 73 public final boolean isEmpty; | |
| 74 public final String raw; | |
|
1714
31a82b0d0a87
bbcode and html work
Franklin Schmidt <fschmidt@gmail.com>
parents:
1712
diff
changeset
|
75 public final Map<String,String> style; |
| 1712 | 76 |
| 77 private Tag(String name,Map<String,Object> attributes,boolean isEmpty,String raw) { | |
|
1714
31a82b0d0a87
bbcode and html work
Franklin Schmidt <fschmidt@gmail.com>
parents:
1712
diff
changeset
|
78 this(name,attributes,isEmpty,raw,null); |
|
31a82b0d0a87
bbcode and html work
Franklin Schmidt <fschmidt@gmail.com>
parents:
1712
diff
changeset
|
79 } |
|
31a82b0d0a87
bbcode and html work
Franklin Schmidt <fschmidt@gmail.com>
parents:
1712
diff
changeset
|
80 |
|
31a82b0d0a87
bbcode and html work
Franklin Schmidt <fschmidt@gmail.com>
parents:
1712
diff
changeset
|
81 private Tag(String name,Map<String,Object> attributes,boolean isEmpty,String raw,Map<String,String> style) { |
| 1712 | 82 this.name = name; |
| 83 this.attributes = attributes; | |
| 84 this.isEmpty = isEmpty; | |
| 85 this.raw = raw; | |
|
1714
31a82b0d0a87
bbcode and html work
Franklin Schmidt <fschmidt@gmail.com>
parents:
1712
diff
changeset
|
86 this.style = style; |
| 1712 | 87 } |
| 88 } | |
| 89 | |
| 90 public static final class Container { | |
| 91 public final Tag tag; | |
| 92 public final String text; | |
| 93 | |
| 94 private Container(Tag tag,String text) { | |
| 95 this.tag = tag; | |
| 96 this.text = text; | |
| 97 } | |
| 98 } | |
| 99 | |
| 100 private static Set<String> defaultContainerTags = new HashSet<String>(); | |
| 101 static { | |
| 102 Collections.addAll( defaultContainerTags, "script", "style", "textarea" ); | |
| 103 } | |
| 104 | |
| 105 public static List parse(String text) { | |
| 106 return parse(text,defaultContainerTags); | |
| 107 } | |
| 108 | |
| 109 public static List parse(String text,Set<String> containerTags) { | |
| 110 return new Html(text,containerTags).parse(); | |
| 111 } | |
| 112 | |
| 113 private final Parser parser; | |
| 114 private final Set<String> containerTags; | |
| 115 | |
| 116 private Html(String text,Set<String> containerTags) { | |
| 117 this.parser = new Parser(text); | |
| 118 this.containerTags = containerTags; | |
| 119 } | |
| 120 | |
| 121 private List parse() { | |
| 122 List list = new ArrayList(); | |
| 123 StringBuilder sb = new StringBuilder(); | |
| 124 while( !parser.endOfInput() ) { | |
| 125 if( parser.test('<') ) { | |
| 126 Tag tag = parseTag(); | |
| 127 if( tag != null ) { | |
| 128 Object el = tag; | |
| 129 String tagName = tag.name; | |
| 130 if( containerTags.contains(tagName) ) { | |
| 131 Container container = parseContainer(tag); | |
| 132 if( container != null ) | |
| 133 el = container; | |
| 134 } | |
| 135 if( el != null | |
| 136 || (el = parseComment()) != null | |
| 137 || (el = parseCdata()) != null | |
| 138 ) { | |
| 139 add(list,sb); | |
| 140 list.add(el); | |
| 141 continue; | |
| 142 } | |
| 143 } | |
| 144 } | |
| 145 sb.append( parser.currentChar() ); | |
| 146 parser.anyChar(); | |
| 147 } | |
| 148 add(list,sb); | |
| 149 return list; | |
| 150 } | |
| 151 | |
| 152 private static void add(List list,StringBuilder sb) { | |
| 153 if( sb.length() > 0 ) { | |
| 154 list.add(decode(sb.toString())); | |
| 155 sb.setLength(0); | |
| 156 } | |
| 157 } | |
| 158 | |
| 159 private Comment parseComment() { | |
| 160 parser.begin(); | |
| 161 if( !parser.match("<!--") ) | |
| 162 return parser.failure(null); | |
| 163 int start = parser.currentIndex(); | |
| 164 while( !parser.test("-->") ) { | |
| 165 if( !parser.anyChar() ) | |
| 166 return parser.failure(null); | |
| 167 } | |
| 168 String text = parser.textFrom(start); | |
| 169 Comment comment = new Comment(text); | |
| 170 return parser.success(comment); | |
| 171 } | |
| 172 | |
| 173 private CData parseCdata() { | |
| 174 parser.begin(); | |
| 175 if( !parser.match("<![CDATA[") ) | |
| 176 return parser.failure(null); | |
| 177 int start = parser.currentIndex(); | |
| 178 while( !parser.test("]]>") ) { | |
| 179 if( !parser.anyChar() ) | |
| 180 return parser.failure(null); | |
| 181 } | |
| 182 String text = parser.textFrom(start); | |
| 183 CData cdata = new CData(text); | |
| 184 return parser.success(cdata); | |
| 185 } | |
| 186 | |
| 187 private Container parseContainer(Tag tag) { | |
| 188 String endTagName = '/' + tag.name; | |
| 189 int start = parser.begin(); | |
| 190 int end; | |
| 191 while(true) { | |
| 192 if( parser.test('<') ) { | |
| 193 end = parser.currentIndex(); | |
| 194 Tag tag2 = parseTag(); | |
| 195 if( tag2.name.equals(endTagName) ) | |
| 196 break; | |
| 197 } | |
| 198 if( !parser.anyChar() ) | |
| 199 return parser.failure(null); | |
| 200 } | |
| 201 String text = parser.text.substring(start,end); | |
| 202 Container container = new Container(tag,text); | |
| 203 return parser.success(container); | |
| 204 } | |
| 205 | |
| 206 private Tag parseTag() { | |
| 207 int tagStart = parser.begin(); | |
| 208 if( !parser.match('<') ) | |
| 209 return parser.failure(null); | |
| 210 int start = parser.currentIndex(); | |
| 211 parser.match('/'); | |
| 212 if( !matchNameChar() ) | |
| 213 return parser.failure(null); | |
| 214 while( matchNameChar() ); | |
| 215 String name = parser.textFrom(start).toLowerCase(); | |
| 216 Map<String,Object> attributes = new HashMap<String,Object>(); | |
| 217 String attrName; | |
|
1714
31a82b0d0a87
bbcode and html work
Franklin Schmidt <fschmidt@gmail.com>
parents:
1712
diff
changeset
|
218 Map<String,String> style = null; |
| 1712 | 219 while( (attrName = parseAttrName()) != null ) { |
| 220 String attrValue = parseAttrValue(); | |
| 221 attributes.put( attrName, attrValue!=null ? attrValue : true ); | |
|
1714
31a82b0d0a87
bbcode and html work
Franklin Schmidt <fschmidt@gmail.com>
parents:
1712
diff
changeset
|
222 if( attrName.equals("style") && attrValue!=null && style==null ) { |
|
31a82b0d0a87
bbcode and html work
Franklin Schmidt <fschmidt@gmail.com>
parents:
1712
diff
changeset
|
223 style = Css.style(attrValue); |
| 1712 | 224 } |
| 225 } | |
| 226 while( matchSpace() ); | |
| 227 boolean isEmpty = parser.match('/'); | |
| 228 if( !parser.match('>') ) | |
| 229 return parser.failure(null); | |
| 230 String raw = parser.textFrom(tagStart); | |
|
1714
31a82b0d0a87
bbcode and html work
Franklin Schmidt <fschmidt@gmail.com>
parents:
1712
diff
changeset
|
231 Tag tag = new Tag(name,attributes,isEmpty,raw,style); |
| 1712 | 232 return parser.success(tag); |
| 233 } | |
| 234 | |
| 235 private String parseAttrName() { | |
| 236 parser.begin(); | |
| 237 if( !matchSpace() ) | |
| 238 return parser.failure(null); | |
| 239 while( matchSpace() ); | |
| 240 int start = parser.currentIndex(); | |
| 241 if( !matchNameChar() ) | |
| 242 return parser.failure(null); | |
| 243 while( matchNameChar() ); | |
| 244 String name = parser.textFrom(start).toLowerCase(); | |
| 245 return parser.success(name); | |
| 246 } | |
| 247 | |
| 248 private String parseAttrValue() { | |
| 249 parser.begin(); | |
| 250 while( matchSpace() ); | |
| 251 if( !parser.match('=') ) | |
| 252 return parser.failure(null); | |
| 253 while( matchSpace() ); | |
| 254 if( parser.anyOf("\"'") ) { | |
| 255 char quote = parser.lastChar(); | |
| 256 int start = parser.currentIndex(); | |
| 257 while( !parser.test(quote) ) { | |
| 258 if( !parser.anyChar() ) | |
| 259 return parser.failure(null); | |
| 260 } | |
| 261 String value = parser.textFrom(start); | |
| 262 parser.match(quote); | |
| 263 value = decode(value); | |
| 264 return parser.success(value); | |
| 265 } | |
| 266 int start = parser.currentIndex(); | |
| 267 if( !matchValueChar() ) | |
| 268 return parser.failure(null); | |
| 269 while( matchValueChar() ); | |
| 270 String value = parser.textFrom(start); | |
| 271 value = decode(value); | |
| 272 return parser.success(value); | |
| 273 } | |
| 274 | |
| 275 private boolean matchNameChar() { | |
| 276 return parser.inCharRange('a','z') | |
| 277 || parser.inCharRange('A','Z') | |
| 278 || parser.inCharRange('0','9') | |
| 279 || parser.anyOf("_.-:") | |
| 280 ; | |
| 281 } | |
| 282 | |
| 283 private boolean matchValueChar() { | |
| 284 return parser.noneOf(" \t\r\n\"'>/="); | |
| 285 } | |
| 286 | |
| 287 private boolean matchSpace() { | |
| 288 return parser.anyOf(" \t\r\n"); | |
| 289 } | |
| 290 | |
| 291 } |