1/* 2 * reserved comment block 3 * DO NOT REMOVE OR ALTER! 4 */ 5package com.sun.org.apache.bcel.internal.util; 6 7import java.util.ArrayList; 8import java.util.HashMap; 9import java.util.Iterator; 10import java.util.List; 11import java.util.Locale; 12import java.util.Map; 13import java.util.regex.Matcher; 14import java.util.regex.Pattern; 15 16import com.sun.org.apache.bcel.internal.Const; 17import com.sun.org.apache.bcel.internal.generic.ClassGenException; 18import com.sun.org.apache.bcel.internal.generic.InstructionHandle; 19import com.sun.org.apache.bcel.internal.generic.InstructionList; 20 21/** 22 * InstructionFinder is a tool to search for given instructions patterns, i.e., 23 * match sequences of instructions in an instruction list via regular 24 * expressions. This can be used, e.g., in order to implement a peep hole 25 * optimizer that looks for code patterns and replaces them with faster 26 * equivalents. 27 * 28 * <p> 29 * This class internally uses the java.util.regex 30 * package to search for regular expressions. 31 * 32 * A typical application would look like this: 33 * 34 * <pre> 35 * 36 * 37 * InstructionFinder f = new InstructionFinder(il); 38 * String pat = "IfInstruction ICONST_0 GOTO ICONST_1 NOP (IFEQ|IFNE)"; 39 * 40 * for (Iterator i = f.search(pat, constraint); i.hasNext(); ) { 41 * InstructionHandle[] match = (InstructionHandle[])i.next(); 42 * ... 43 * il.delete(match[1], match[5]); 44 * ... 45 * } 46 * 47 * 48 * </pre> 49 * 50 * @version $Id: InstructionFinder.java 1749603 2016-06-21 20:50:19Z ggregory $ 51 * @see com.sun.org.apache.bcel.internal.generic.Instruction 52 * @see InstructionList 53 */ 54public class InstructionFinder { 55 56 private static final int OFFSET = 32767; // char + OFFSET is outside of LATIN-1 57 private static final int NO_OPCODES = 256; // Potential number, some are not used 58 private static final Map<String, String> map = new HashMap<>(); 59 private final InstructionList il; 60 private String il_string; // instruction list as string 61 private InstructionHandle[] handles; // map instruction 62 63 64 // list to array 65 /** 66 * @param il 67 * instruction list to search for given patterns 68 */ 69 public InstructionFinder(final InstructionList il) { 70 this.il = il; 71 reread(); 72 } 73 74 75 /** 76 * Reread the instruction list, e.g., after you've altered the list upon a 77 * match. 78 */ 79 public final void reread() { 80 final int size = il.getLength(); 81 final char[] buf = new char[size]; // Create a string with length equal to il length 82 handles = il.getInstructionHandles(); 83 // Map opcodes to characters 84 for (int i = 0; i < size; i++) { 85 buf[i] = makeChar(handles[i].getInstruction().getOpcode()); 86 } 87 il_string = new String(buf); 88 } 89 90 91 /** 92 * Map symbolic instruction names like "getfield" to a single character. 93 * 94 * @param pattern 95 * instruction pattern in lower case 96 * @return encoded string for a pattern such as "BranchInstruction". 97 */ 98 private static String mapName( final String pattern ) { 99 final String result = map.get(pattern); 100 if (result != null) { 101 return result; 102 } 103 for (short i = 0; i < NO_OPCODES; i++) { 104 if (pattern.equals(Const.getOpcodeName(i))) { 105 return "" + makeChar(i); 106 } 107 } 108 throw new RuntimeException("Instruction unknown: " + pattern); 109 } 110 111 112 /** 113 * Replace symbolic names of instructions with the appropiate character and 114 * remove all white space from string. Meta characters such as +, * are 115 * ignored. 116 * 117 * @param pattern 118 * The pattern to compile 119 * @return translated regular expression string 120 */ 121 private static String compilePattern( final String pattern ) { 122 //Bug: BCEL-77 - Instructions are assumed to be english, to avoid odd Locale issues 123 final String lower = pattern.toLowerCase(Locale.ENGLISH); 124 final StringBuilder buf = new StringBuilder(); 125 final int size = pattern.length(); 126 for (int i = 0; i < size; i++) { 127 char ch = lower.charAt(i); 128 if (Character.isLetterOrDigit(ch)) { 129 final StringBuilder name = new StringBuilder(); 130 while ((Character.isLetterOrDigit(ch) || ch == '_') && i < size) { 131 name.append(ch); 132 if (++i < size) { 133 ch = lower.charAt(i); 134 } else { 135 break; 136 } 137 } 138 i--; 139 buf.append(mapName(name.toString())); 140 } else if (!Character.isWhitespace(ch)) { 141 buf.append(ch); 142 } 143 } 144 return buf.toString(); 145 } 146 147 148 /** 149 * @return the matched piece of code as an array of instruction (handles) 150 */ 151 private InstructionHandle[] getMatch( final int matched_from, final int match_length ) { 152 final InstructionHandle[] match = new InstructionHandle[match_length]; 153 System.arraycopy(handles, matched_from, match, 0, match_length); 154 return match; 155 } 156 157 158 /** 159 * Search for the given pattern in the instruction list. You can search for 160 * any valid opcode via its symbolic name, e.g. "istore". You can also use a 161 * super class or an interface name to match a whole set of instructions, e.g. 162 * "BranchInstruction" or "LoadInstruction". "istore" is also an alias for all 163 * "istore_x" instructions. Additional aliases are "if" for "ifxx", "if_icmp" 164 * for "if_icmpxx", "if_acmp" for "if_acmpxx". 165 * 166 * Consecutive instruction names must be separated by white space which will 167 * be removed during the compilation of the pattern. 168 * 169 * For the rest the usual pattern matching rules for regular expressions 170 * apply. 171 * <P> 172 * Example pattern: 173 * 174 * <pre> 175 * search("BranchInstruction NOP ((IfInstruction|GOTO)+ ISTORE Instruction)*"); 176 * </pre> 177 * 178 * <p> 179 * If you alter the instruction list upon a match such that other matching 180 * areas are affected, you should call reread() to update the finder and call 181 * search() again, because the matches are cached. 182 * 183 * @param pattern 184 * the instruction pattern to search for, where case is ignored 185 * @param from 186 * where to start the search in the instruction list 187 * @param constraint 188 * optional CodeConstraint to check the found code pattern for 189 * user-defined constraints 190 * @return iterator of matches where e.nextElement() returns an array of 191 * instruction handles describing the matched area 192 */ 193 public final Iterator<InstructionHandle[]> search( final String pattern, 194 final InstructionHandle from, final CodeConstraint constraint ) { 195 final String search = compilePattern(pattern); 196 int start = -1; 197 for (int i = 0; i < handles.length; i++) { 198 if (handles[i] == from) { 199 start = i; // Where to start search from (index) 200 break; 201 } 202 } 203 if (start == -1) { 204 throw new ClassGenException("Instruction handle " + from 205 + " not found in instruction list."); 206 } 207 final Pattern regex = Pattern.compile(search); 208 final List<InstructionHandle[]> matches = new ArrayList<>(); 209 final Matcher matcher = regex.matcher(il_string); 210 while (start < il_string.length() && matcher.find(start)) { 211 final int startExpr = matcher.start(); 212 final int endExpr = matcher.end(); 213 final int lenExpr = endExpr - startExpr; 214 final InstructionHandle[] match = getMatch(startExpr, lenExpr); 215 if ((constraint == null) || constraint.checkCode(match)) { 216 matches.add(match); 217 } 218 start = endExpr; 219 } 220 return matches.iterator(); 221 } 222 223 224 /** 225 * Start search beginning from the start of the given instruction list. 226 * 227 * @param pattern 228 * the instruction pattern to search for, where case is ignored 229 * @return iterator of matches where e.nextElement() returns an array of 230 * instruction handles describing the matched area 231 */ 232 public final Iterator<InstructionHandle[]> search( final String pattern ) { 233 return search(pattern, il.getStart(), null); 234 } 235 236 237 /** 238 * Start search beginning from `from'. 239 * 240 * @param pattern 241 * the instruction pattern to search for, where case is ignored 242 * @param from 243 * where to start the search in the instruction list 244 * @return iterator of matches where e.nextElement() returns an array of 245 * instruction handles describing the matched area 246 */ 247 public final Iterator<InstructionHandle[]> search( final String pattern, 248 final InstructionHandle from ) { 249 return search(pattern, from, null); 250 } 251 252 253 /** 254 * Start search beginning from the start of the given instruction list. Check 255 * found matches with the constraint object. 256 * 257 * @param pattern 258 * the instruction pattern to search for, case is ignored 259 * @param constraint 260 * constraints to be checked on matching code 261 * @return instruction handle or `null' if the match failed 262 */ 263 public final Iterator<InstructionHandle[]> search( final String pattern, 264 final CodeConstraint constraint ) { 265 return search(pattern, il.getStart(), constraint); 266 } 267 268 269 /** 270 * Convert opcode number to char. 271 */ 272 private static char makeChar( final short opcode ) { 273 return (char) (opcode + OFFSET); 274 } 275 276 277 /** 278 * @return the inquired instruction list 279 */ 280 public final InstructionList getInstructionList() { 281 return il; 282 } 283 284 /** 285 * Code patterns found may be checked using an additional user-defined 286 * constraint object whether they really match the needed criterion. I.e., 287 * check constraints that can not expressed with regular expressions. 288 * 289 */ 290 public interface CodeConstraint { 291 292 /** 293 * @param match 294 * array of instructions matching the requested pattern 295 * @return true if the matched area is really useful 296 */ 297 boolean checkCode( InstructionHandle[] match ); 298 } 299 300 // Initialize pattern map 301 static { 302 map.put("arithmeticinstruction","(irem|lrem|iand|ior|ineg|isub|lneg|fneg|fmul|ldiv|fadd|lxor|frem|idiv|land|ixor|ishr|fsub|lshl|fdiv|iadd|lor|dmul|lsub|ishl|imul|lmul|lushr|dneg|iushr|lshr|ddiv|drem|dadd|ladd|dsub)"); 303 map.put("invokeinstruction", "(invokevirtual|invokeinterface|invokestatic|invokespecial|invokedynamic)"); 304 map.put("arrayinstruction", "(baload|aastore|saload|caload|fastore|lastore|iaload|castore|iastore|aaload|bastore|sastore|faload|laload|daload|dastore)"); 305 map.put("gotoinstruction", "(goto|goto_w)"); 306 map.put("conversioninstruction", "(d2l|l2d|i2s|d2i|l2i|i2b|l2f|d2f|f2i|i2d|i2l|f2d|i2c|f2l|i2f)"); 307 map.put("localvariableinstruction","(fstore|iinc|lload|dstore|dload|iload|aload|astore|istore|fload|lstore)"); 308 map.put("loadinstruction", "(fload|dload|lload|iload|aload)"); 309 map.put("fieldinstruction", "(getfield|putstatic|getstatic|putfield)"); 310 map.put("cpinstruction", "(ldc2_w|invokeinterface|invokedynamic|multianewarray|putstatic|instanceof|getstatic|checkcast|getfield|invokespecial|ldc_w|invokestatic|invokevirtual|putfield|ldc|new|anewarray)"); 311 map.put("stackinstruction", "(dup2|swap|dup2_x2|pop|pop2|dup|dup2_x1|dup_x2|dup_x1)"); 312 map.put("branchinstruction", "(ifle|if_acmpne|if_icmpeq|if_acmpeq|ifnonnull|goto_w|iflt|ifnull|if_icmpne|tableswitch|if_icmple|ifeq|if_icmplt|jsr_w|if_icmpgt|ifgt|jsr|goto|ifne|ifge|lookupswitch|if_icmpge)"); 313 map.put("returninstruction", "(lreturn|ireturn|freturn|dreturn|areturn|return)"); 314 map.put("storeinstruction", "(istore|fstore|dstore|astore|lstore)"); 315 map.put("select", "(tableswitch|lookupswitch)"); 316 map.put("ifinstruction", "(ifeq|ifgt|if_icmpne|if_icmpeq|ifge|ifnull|ifne|if_icmple|if_icmpge|if_acmpeq|if_icmplt|if_acmpne|ifnonnull|iflt|if_icmpgt|ifle)"); 317 map.put("jsrinstruction", "(jsr|jsr_w)"); 318 map.put("variablelengthinstruction", "(tableswitch|jsr|goto|lookupswitch)"); 319 map.put("unconditionalbranch", "(goto|jsr|jsr_w|athrow|goto_w)"); 320 map.put("constantpushinstruction", "(dconst|bipush|sipush|fconst|iconst|lconst)"); 321 map.put("typedinstruction", "(imul|lsub|aload|fload|lor|new|aaload|fcmpg|iand|iaload|lrem|idiv|d2l|isub|dcmpg|dastore|ret|f2d|f2i|drem|iinc|i2c|checkcast|frem|lreturn|astore|lushr|daload|dneg|fastore|istore|lshl|ldiv|lstore|areturn|ishr|ldc_w|invokeinterface|invokedynamic|aastore|lxor|ishl|l2d|i2f|return|faload|sipush|iushr|caload|instanceof|invokespecial|putfield|fmul|ireturn|laload|d2f|lneg|ixor|i2l|fdiv|lastore|multianewarray|i2b|getstatic|i2d|putstatic|fcmpl|saload|ladd|irem|dload|jsr_w|dconst|dcmpl|fsub|freturn|ldc|aconst_null|castore|lmul|ldc2_w|dadd|iconst|f2l|ddiv|dstore|land|jsr|anewarray|dmul|bipush|dsub|sastore|d2i|i2s|lshr|iadd|l2i|lload|bastore|fstore|fneg|iload|fadd|baload|fconst|ior|ineg|dreturn|l2f|lconst|getfield|invokevirtual|invokestatic|iastore)"); 322 map.put("popinstruction", "(fstore|dstore|pop|pop2|astore|putstatic|istore|lstore)"); 323 map.put("allocationinstruction", "(multianewarray|new|anewarray|newarray)"); 324 map.put("indexedinstruction", "(lload|lstore|fload|ldc2_w|invokeinterface|invokedynamic|multianewarray|astore|dload|putstatic|instanceof|getstatic|checkcast|getfield|invokespecial|dstore|istore|iinc|ldc_w|ret|fstore|invokestatic|iload|putfield|invokevirtual|ldc|new|aload|anewarray)"); 325 map.put("pushinstruction", "(dup|lload|dup2|bipush|fload|ldc2_w|sipush|lconst|fconst|dload|getstatic|ldc_w|aconst_null|dconst|iload|ldc|iconst|aload)"); 326 map.put("stackproducer", "(imul|lsub|aload|fload|lor|new|aaload|fcmpg|iand|iaload|lrem|idiv|d2l|isub|dcmpg|dup|f2d|f2i|drem|i2c|checkcast|frem|lushr|daload|dneg|lshl|ldiv|ishr|ldc_w|invokeinterface|invokedynamic|lxor|ishl|l2d|i2f|faload|sipush|iushr|caload|instanceof|invokespecial|fmul|laload|d2f|lneg|ixor|i2l|fdiv|getstatic|i2b|swap|i2d|dup2|fcmpl|saload|ladd|irem|dload|jsr_w|dconst|dcmpl|fsub|ldc|arraylength|aconst_null|tableswitch|lmul|ldc2_w|iconst|dadd|f2l|ddiv|land|jsr|anewarray|dmul|bipush|dsub|d2i|newarray|i2s|lshr|iadd|lload|l2i|fneg|iload|fadd|baload|fconst|lookupswitch|ior|ineg|lconst|l2f|getfield|invokevirtual|invokestatic)"); 327 map.put("stackconsumer", "(imul|lsub|lor|iflt|fcmpg|if_icmpgt|iand|ifeq|if_icmplt|lrem|ifnonnull|idiv|d2l|isub|dcmpg|dastore|if_icmpeq|f2d|f2i|drem|i2c|checkcast|frem|lreturn|astore|lushr|pop2|monitorexit|dneg|fastore|istore|lshl|ldiv|lstore|areturn|if_icmpge|ishr|monitorenter|invokeinterface|invokedynamic|aastore|lxor|ishl|l2d|i2f|return|iushr|instanceof|invokespecial|fmul|ireturn|d2f|lneg|ixor|pop|i2l|ifnull|fdiv|lastore|i2b|if_acmpeq|ifge|swap|i2d|putstatic|fcmpl|ladd|irem|dcmpl|fsub|freturn|ifgt|castore|lmul|dadd|f2l|ddiv|dstore|land|if_icmpne|if_acmpne|dmul|dsub|sastore|ifle|d2i|i2s|lshr|iadd|l2i|bastore|fstore|fneg|fadd|ior|ineg|ifne|dreturn|l2f|if_icmple|getfield|invokevirtual|invokestatic|iastore)"); 328 map.put("exceptionthrower","(irem|lrem|laload|putstatic|baload|dastore|areturn|getstatic|ldiv|anewarray|iastore|castore|idiv|saload|lastore|fastore|putfield|lreturn|caload|getfield|return|aastore|freturn|newarray|instanceof|multianewarray|athrow|faload|iaload|aaload|dreturn|monitorenter|checkcast|bastore|arraylength|new|invokevirtual|sastore|ldc_w|ireturn|invokespecial|monitorexit|invokeinterface|invokedynamic|ldc|invokestatic|daload)"); 329 map.put("loadclass", "(multianewarray|invokeinterface|invokedynamic|instanceof|invokespecial|putfield|checkcast|putstatic|invokevirtual|new|getstatic|invokestatic|getfield|anewarray)"); 330 map.put("instructiontargeter", "(ifle|if_acmpne|if_icmpeq|if_acmpeq|ifnonnull|goto_w|iflt|ifnull|if_icmpne|tableswitch|if_icmple|ifeq|if_icmplt|jsr_w|if_icmpgt|ifgt|jsr|goto|ifne|ifge|lookupswitch|if_icmpge)"); 331 // Some aliases 332 map.put("if_icmp", "(if_icmpne|if_icmpeq|if_icmple|if_icmpge|if_icmplt|if_icmpgt)"); 333 map.put("if_acmp", "(if_acmpeq|if_acmpne)"); 334 map.put("if", "(ifeq|ifne|iflt|ifge|ifgt|ifle)"); 335 // Precompile some aliases first 336 map.put("iconst", precompile(Const.ICONST_0, Const.ICONST_5, Const.ICONST_M1)); 337 map.put("lconst", new String(new char[] { '(', makeChar(Const.LCONST_0), '|', makeChar(Const.LCONST_1), ')' })); 338 map.put("dconst", new String(new char[] { '(', makeChar(Const.DCONST_0), '|', makeChar(Const.DCONST_1), ')' })); 339 map.put("fconst", new String(new char[] { '(', makeChar(Const.FCONST_0), '|', makeChar(Const.FCONST_1), '|', makeChar(Const.FCONST_2), ')' })); 340 map.put("lload", precompile(Const.LLOAD_0, Const.LLOAD_3, Const.LLOAD)); 341 map.put("iload", precompile(Const.ILOAD_0, Const.ILOAD_3, Const.ILOAD)); 342 map.put("dload", precompile(Const.DLOAD_0, Const.DLOAD_3, Const.DLOAD)); 343 map.put("fload", precompile(Const.FLOAD_0, Const.FLOAD_3, Const.FLOAD)); 344 map.put("aload", precompile(Const.ALOAD_0, Const.ALOAD_3, Const.ALOAD)); 345 map.put("lstore", precompile(Const.LSTORE_0, Const.LSTORE_3, Const.LSTORE)); 346 map.put("istore", precompile(Const.ISTORE_0, Const.ISTORE_3, Const.ISTORE)); 347 map.put("dstore", precompile(Const.DSTORE_0, Const.DSTORE_3, Const.DSTORE)); 348 map.put("fstore", precompile(Const.FSTORE_0, Const.FSTORE_3, Const.FSTORE)); 349 map.put("astore", precompile(Const.ASTORE_0, Const.ASTORE_3, Const.ASTORE)); 350 // Compile strings 351 for (final Map.Entry<String, String> entry : map.entrySet()) { 352 final String key = entry.getKey(); 353 final String value = entry.getValue(); 354 final char ch = value.charAt(1); // Omit already precompiled patterns 355 if (ch < OFFSET) { 356 map.put(key, compilePattern(value)); // precompile all patterns 357 } 358 } 359 // Add instruction alias to match anything 360 final StringBuilder buf = new StringBuilder("("); 361 for (short i = 0; i < NO_OPCODES; i++) { 362 if (Const.getNoOfOperands(i) != Const.UNDEFINED) { // Not an invalid opcode 363 buf.append(makeChar(i)); 364 if (i < NO_OPCODES - 1) { 365 buf.append('|'); 366 } 367 } 368 } 369 buf.append(')'); 370 map.put("instruction", buf.toString()); 371 } 372 373 374 private static String precompile( final short from, final short to, final short extra ) { 375 final StringBuilder buf = new StringBuilder("("); 376 for (short i = from; i <= to; i++) { 377 buf.append(makeChar(i)); 378 buf.append('|'); 379 } 380 buf.append(makeChar(extra)); 381 buf.append(")"); 382 return buf.toString(); 383 } 384 385 386 /* 387 * Internal debugging routines. 388 */ 389// private static final String pattern2string( String pattern ) { 390// return pattern2string(pattern, true); 391// } 392 393 394// private static final String pattern2string( String pattern, boolean make_string ) { 395// StringBuffer buf = new StringBuffer(); 396// for (int i = 0; i < pattern.length(); i++) { 397// char ch = pattern.charAt(i); 398// if (ch >= OFFSET) { 399// if (make_string) { 400// buf.append(Constants.getOpcodeName(ch - OFFSET)); 401// } else { 402// buf.append((ch - OFFSET)); 403// } 404// } else { 405// buf.append(ch); 406// } 407// } 408// return buf.toString(); 409// } 410} 411