View Javadoc
1   /*
2    * Licensed to the Apache Software Foundation (ASF) under one or more
3    * contributor license agreements.  See the NOTICE file distributed with
4    * this work for additional information regarding copyright ownership.
5    * The ASF licenses this file to You under the Apache License, Version 2.0
6    * (the "License"); you may not use this file except in compliance with
7    * the License.  You may obtain a copy of the License at
8    *
9    *      http://www.apache.org/licenses/LICENSE-2.0
10   *
11   *  Unless required by applicable law or agreed to in writing, software
12   *  distributed under the License is distributed on an "AS IS" BASIS,
13   *  WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
14   *  See the License for the specific language governing permissions and
15   *  limitations under the License.
16   *
17   */
18  package org.apache.bcel.util;
19  
20  import java.util.ArrayList;
21  import java.util.HashMap;
22  import java.util.Iterator;
23  import java.util.List;
24  import java.util.Locale;
25  import java.util.Map;
26  import java.util.regex.Matcher;
27  import java.util.regex.Pattern;
28  
29  import org.apache.bcel.Const;
30  import org.apache.bcel.generic.ClassGenException;
31  import org.apache.bcel.generic.InstructionHandle;
32  import org.apache.bcel.generic.InstructionList;
33  
34  /**
35   * InstructionFinder is a tool to search for given instructions patterns, i.e.,
36   * match sequences of instructions in an instruction list via regular
37   * expressions. This can be used, e.g., in order to implement a peep hole
38   * optimizer that looks for code patterns and replaces them with faster
39   * equivalents.
40   * 
41   * <p>
42   * This class internally uses the java.util.regex
43   * package to search for regular expressions.
44   * 
45   * A typical application would look like this:
46   * 
47   * <pre>
48   * 
49   *  
50   *   InstructionFinder f   = new InstructionFinder(il);
51   *   String            pat = &quot;IfInstruction ICONST_0 GOTO ICONST_1 NOP (IFEQ|IFNE)&quot;;
52   *   
53   *   for (Iterator i = f.search(pat, constraint); i.hasNext(); ) {
54   *   InstructionHandle[] match = (InstructionHandle[])i.next();
55   *   ...
56   *   il.delete(match[1], match[5]);
57   *   ...
58   *   }
59   *   
60   *  
61   * </pre>
62   * 
63   * @version $Id: InstructionFinder.html 992795 2016-07-14 11:53:52Z britter $
64   * @see org.apache.bcel.generic.Instruction
65   * @see InstructionList
66   */
67  public class InstructionFinder {
68  
69      private static final int OFFSET = 32767; // char + OFFSET is outside of LATIN-1
70      private static final int NO_OPCODES = 256; // Potential number, some are not used
71      private static final Map<String, String> map = new HashMap<>();
72      private final InstructionList il;
73      private String il_string; // instruction list as string
74      private InstructionHandle[] handles; // map instruction
75  
76  
77      // list to array
78      /**
79       * @param il
80       *          instruction list to search for given patterns
81       */
82      public InstructionFinder(final InstructionList il) {
83          this.il = il;
84          reread();
85      }
86  
87  
88      /**
89       * Reread the instruction list, e.g., after you've altered the list upon a
90       * match.
91       */
92      public final void reread() {
93          final int size = il.getLength();
94          final char[] buf = new char[size]; // Create a string with length equal to il length
95          handles = il.getInstructionHandles();
96          // Map opcodes to characters
97          for (int i = 0; i < size; i++) {
98              buf[i] = makeChar(handles[i].getInstruction().getOpcode());
99          }
100         il_string = new String(buf);
101     }
102 
103 
104     /**
105      * Map symbolic instruction names like "getfield" to a single character.
106      * 
107      * @param pattern
108      *          instruction pattern in lower case
109      * @return encoded string for a pattern such as "BranchInstruction".
110      */
111     private static String mapName( final String pattern ) {
112         final String result = map.get(pattern);
113         if (result != null) {
114             return result;
115         }
116         for (short i = 0; i < NO_OPCODES; i++) {
117             if (pattern.equals(Const.getOpcodeName(i))) {
118                 return "" + makeChar(i);
119             }
120         }
121         throw new RuntimeException("Instruction unknown: " + pattern);
122     }
123 
124 
125     /**
126      * Replace symbolic names of instructions with the appropiate character and
127      * remove all white space from string. Meta characters such as +, * are
128      * ignored.
129      * 
130      * @param pattern
131      *          The pattern to compile
132      * @return translated regular expression string
133      */
134     private static String compilePattern( final String pattern ) {
135         //Bug: BCEL-77 - Instructions are assumed to be english, to avoid odd Locale issues
136         final String lower = pattern.toLowerCase(Locale.ENGLISH);
137         final StringBuilder buf = new StringBuilder();
138         final int size = pattern.length();
139         for (int i = 0; i < size; i++) {
140             char ch = lower.charAt(i);
141             if (Character.isLetterOrDigit(ch)) {
142                 final StringBuilder name = new StringBuilder();
143                 while ((Character.isLetterOrDigit(ch) || ch == '_') && i < size) {
144                     name.append(ch);
145                     if (++i < size) {
146                         ch = lower.charAt(i);
147                     } else {
148                         break;
149                     }
150                 }
151                 i--;
152                 buf.append(mapName(name.toString()));
153             } else if (!Character.isWhitespace(ch)) {
154                 buf.append(ch);
155             }
156         }
157         return buf.toString();
158     }
159 
160 
161     /**
162      * @return the matched piece of code as an array of instruction (handles)
163      */
164     private InstructionHandle[] getMatch( final int matched_from, final int match_length ) {
165         final InstructionHandle[] match = new InstructionHandle[match_length];
166         System.arraycopy(handles, matched_from, match, 0, match_length);
167         return match;
168     }
169 
170 
171     /**
172      * Search for the given pattern in the instruction list. You can search for
173      * any valid opcode via its symbolic name, e.g. "istore". You can also use a
174      * super class or an interface name to match a whole set of instructions, e.g.
175      * "BranchInstruction" or "LoadInstruction". "istore" is also an alias for all
176      * "istore_x" instructions. Additional aliases are "if" for "ifxx", "if_icmp"
177      * for "if_icmpxx", "if_acmp" for "if_acmpxx".
178      * 
179      * Consecutive instruction names must be separated by white space which will
180      * be removed during the compilation of the pattern.
181      * 
182      * For the rest the usual pattern matching rules for regular expressions
183      * apply.
184      * <P>
185      * Example pattern:
186      * 
187      * <pre>
188      * search(&quot;BranchInstruction NOP ((IfInstruction|GOTO)+ ISTORE Instruction)*&quot;);
189      * </pre>
190      * 
191      * <p>
192      * If you alter the instruction list upon a match such that other matching
193      * areas are affected, you should call reread() to update the finder and call
194      * search() again, because the matches are cached.
195      * 
196      * @param pattern
197      *          the instruction pattern to search for, where case is ignored
198      * @param from
199      *          where to start the search in the instruction list
200      * @param constraint
201      *          optional CodeConstraint to check the found code pattern for
202      *          user-defined constraints
203      * @return iterator of matches where e.nextElement() returns an array of
204      *         instruction handles describing the matched area
205      */
206     public final Iterator<InstructionHandle[]> search( final String pattern, final InstructionHandle from, final CodeConstraint constraint ) {
207         final String search = compilePattern(pattern);
208         int start = -1;
209         for (int i = 0; i < handles.length; i++) {
210             if (handles[i] == from) {
211                 start = i; // Where to start search from (index)
212                 break;
213             }
214         }
215         if (start == -1) {
216             throw new ClassGenException("Instruction handle " + from
217                     + " not found in instruction list.");
218         }
219         final Pattern regex = Pattern.compile(search);
220         final List<InstructionHandle[]> matches = new ArrayList<>();
221         final Matcher matcher = regex.matcher(il_string);
222         while (start < il_string.length() && matcher.find(start)) {
223             final int startExpr = matcher.start();
224             final int endExpr = matcher.end();
225             final int lenExpr = endExpr - startExpr;
226             final InstructionHandle[] match = getMatch(startExpr, lenExpr);
227             if ((constraint == null) || constraint.checkCode(match)) {
228                 matches.add(match);
229             }
230             start = endExpr;
231         }
232         return matches.iterator();
233     }
234 
235 
236     /**
237      * Start search beginning from the start of the given instruction list.
238      * 
239      * @param pattern
240      *          the instruction pattern to search for, where case is ignored
241      * @return iterator of matches where e.nextElement() returns an array of
242      *         instruction handles describing the matched area
243      */
244     public final Iterator<InstructionHandle[]> search( final String pattern ) {
245         return search(pattern, il.getStart(), null);
246     }
247 
248 
249     /**
250      * Start search beginning from `from'.
251      * 
252      * @param pattern
253      *          the instruction pattern to search for, where case is ignored
254      * @param from
255      *          where to start the search in the instruction list
256      * @return iterator of matches where e.nextElement() returns an array of
257      *         instruction handles describing the matched area
258      */
259     public final Iterator<InstructionHandle[]> search( final String pattern, final InstructionHandle from ) {
260         return search(pattern, from, null);
261     }
262 
263 
264     /**
265      * Start search beginning from the start of the given instruction list. Check
266      * found matches with the constraint object.
267      * 
268      * @param pattern
269      *          the instruction pattern to search for, case is ignored
270      * @param constraint
271      *          constraints to be checked on matching code
272      * @return instruction handle or `null' if the match failed
273      */
274     public final Iterator<InstructionHandle[]> search( final String pattern, final CodeConstraint constraint ) {
275         return search(pattern, il.getStart(), constraint);
276     }
277 
278 
279     /**
280      * Convert opcode number to char.
281      */
282     private static char makeChar( final short opcode ) {
283         return (char) (opcode + OFFSET);
284     }
285 
286 
287     /**
288      * @return the inquired instruction list
289      */
290     public final InstructionList getInstructionList() {
291         return il;
292     }
293 
294     /**
295      * Code patterns found may be checked using an additional user-defined
296      * constraint object whether they really match the needed criterion. I.e.,
297      * check constraints that can not expressed with regular expressions.
298      * 
299      */
300     public interface CodeConstraint {
301 
302         /**
303          * @param match
304          *          array of instructions matching the requested pattern
305          * @return true if the matched area is really useful
306          */
307         boolean checkCode( InstructionHandle[] match );
308     }
309 
310     // Initialize pattern map
311     static {
312         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)");
313         map.put("invokeinstruction", "(invokevirtual|invokeinterface|invokestatic|invokespecial|invokedynamic)");
314         map.put("arrayinstruction", "(baload|aastore|saload|caload|fastore|lastore|iaload|castore|iastore|aaload|bastore|sastore|faload|laload|daload|dastore)");
315         map.put("gotoinstruction", "(goto|goto_w)");
316         map.put("conversioninstruction", "(d2l|l2d|i2s|d2i|l2i|i2b|l2f|d2f|f2i|i2d|i2l|f2d|i2c|f2l|i2f)");
317         map.put("localvariableinstruction","(fstore|iinc|lload|dstore|dload|iload|aload|astore|istore|fload|lstore)");
318         map.put("loadinstruction", "(fload|dload|lload|iload|aload)");
319         map.put("fieldinstruction", "(getfield|putstatic|getstatic|putfield)");
320         map.put("cpinstruction", "(ldc2_w|invokeinterface|invokedynamic|multianewarray|putstatic|instanceof|getstatic|checkcast|getfield|invokespecial|ldc_w|invokestatic|invokevirtual|putfield|ldc|new|anewarray)");
321         map.put("stackinstruction", "(dup2|swap|dup2_x2|pop|pop2|dup|dup2_x1|dup_x2|dup_x1)");
322         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)");
323         map.put("returninstruction", "(lreturn|ireturn|freturn|dreturn|areturn|return)");
324         map.put("storeinstruction", "(istore|fstore|dstore|astore|lstore)");
325         map.put("select", "(tableswitch|lookupswitch)");
326         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)");
327         map.put("jsrinstruction", "(jsr|jsr_w)");
328         map.put("variablelengthinstruction", "(tableswitch|jsr|goto|lookupswitch)");
329         map.put("unconditionalbranch", "(goto|jsr|jsr_w|athrow|goto_w)");
330         map.put("constantpushinstruction", "(dconst|bipush|sipush|fconst|iconst|lconst)");
331         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)");
332         map.put("popinstruction", "(fstore|dstore|pop|pop2|astore|putstatic|istore|lstore)");
333         map.put("allocationinstruction", "(multianewarray|new|anewarray|newarray)");
334         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)");
335         map.put("pushinstruction", "(dup|lload|dup2|bipush|fload|ldc2_w|sipush|lconst|fconst|dload|getstatic|ldc_w|aconst_null|dconst|iload|ldc|iconst|aload)");
336         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)");
337         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)");
338         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)");
339         map.put("loadclass", "(multianewarray|invokeinterface|invokedynamic|instanceof|invokespecial|putfield|checkcast|putstatic|invokevirtual|new|getstatic|invokestatic|getfield|anewarray)");
340         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)");
341         // Some aliases
342         map.put("if_icmp", "(if_icmpne|if_icmpeq|if_icmple|if_icmpge|if_icmplt|if_icmpgt)");
343         map.put("if_acmp", "(if_acmpeq|if_acmpne)");
344         map.put("if", "(ifeq|ifne|iflt|ifge|ifgt|ifle)");
345         // Precompile some aliases first
346         map.put("iconst", precompile(Const.ICONST_0, Const.ICONST_5, Const.ICONST_M1));
347         map.put("lconst", new String(new char[] { '(', makeChar(Const.LCONST_0), '|', makeChar(Const.LCONST_1), ')' }));
348         map.put("dconst", new String(new char[] { '(', makeChar(Const.DCONST_0), '|', makeChar(Const.DCONST_1), ')' }));
349         map.put("fconst", new String(new char[] { '(', makeChar(Const.FCONST_0), '|', makeChar(Const.FCONST_1), '|', makeChar(Const.FCONST_2), ')' }));
350         map.put("lload", precompile(Const.LLOAD_0, Const.LLOAD_3, Const.LLOAD));
351         map.put("iload", precompile(Const.ILOAD_0, Const.ILOAD_3, Const.ILOAD));
352         map.put("dload", precompile(Const.DLOAD_0, Const.DLOAD_3, Const.DLOAD));
353         map.put("fload", precompile(Const.FLOAD_0, Const.FLOAD_3, Const.FLOAD));
354         map.put("aload", precompile(Const.ALOAD_0, Const.ALOAD_3, Const.ALOAD));
355         map.put("lstore", precompile(Const.LSTORE_0, Const.LSTORE_3, Const.LSTORE));
356         map.put("istore", precompile(Const.ISTORE_0, Const.ISTORE_3, Const.ISTORE));
357         map.put("dstore", precompile(Const.DSTORE_0, Const.DSTORE_3, Const.DSTORE));
358         map.put("fstore", precompile(Const.FSTORE_0, Const.FSTORE_3, Const.FSTORE));
359         map.put("astore", precompile(Const.ASTORE_0, Const.ASTORE_3, Const.ASTORE));
360         // Compile strings
361         for (final Map.Entry<String, String> entry : map.entrySet()) {
362             final String key = entry.getKey();
363             final String value = entry.getValue();
364             final char ch = value.charAt(1); // Omit already precompiled patterns
365             if (ch < OFFSET) {
366                 map.put(key, compilePattern(value)); // precompile all patterns
367             }
368         }
369         // Add instruction alias to match anything
370         final StringBuilder buf = new StringBuilder("(");
371         for (short i = 0; i < NO_OPCODES; i++) {
372             if (Const.getNoOfOperands(i) != Const.UNDEFINED) { // Not an invalid opcode
373                 buf.append(makeChar(i));
374                 if (i < NO_OPCODES - 1) {
375                     buf.append('|');
376                 }
377             }
378         }
379         buf.append(')');
380         map.put("instruction", buf.toString());
381     }
382 
383 
384     private static String precompile( final short from, final short to, final short extra ) {
385         final StringBuilder buf = new StringBuilder("(");
386         for (short i = from; i <= to; i++) {
387             buf.append(makeChar(i));
388             buf.append('|');
389         }
390         buf.append(makeChar(extra));
391         buf.append(")");
392         return buf.toString();
393     }
394 
395 
396     /*
397      * Internal debugging routines.
398      */
399 //    private static final String pattern2string( String pattern ) {
400 //        return pattern2string(pattern, true);
401 //    }
402 
403 
404 //    private static final String pattern2string( String pattern, boolean make_string ) {
405 //        StringBuffer buf = new StringBuffer();
406 //        for (int i = 0; i < pattern.length(); i++) {
407 //            char ch = pattern.charAt(i);
408 //            if (ch >= OFFSET) {
409 //                if (make_string) {
410 //                    buf.append(Constants.getOpcodeName(ch - OFFSET));
411 //                } else {
412 //                    buf.append((ch - OFFSET));
413 //                }
414 //            } else {
415 //                buf.append(ch);
416 //            }
417 //        }
418 //        return buf.toString();
419 //    }
420 }