001/*
002 * Licensed to the Apache Software Foundation (ASF) under one or more
003 * contributor license agreements.  See the NOTICE file distributed with
004 * this work for additional information regarding copyright ownership.
005 * The ASF licenses this file to You under the Apache License, Version 2.0
006 * (the "License"); you may not use this file except in compliance with
007 * the License.  You may obtain a copy of the License at
008 *
009 *      http://www.apache.org/licenses/LICENSE-2.0
010 *
011 *  Unless required by applicable law or agreed to in writing, software
012 *  distributed under the License is distributed on an "AS IS" BASIS,
013 *  WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
014 *  See the License for the specific language governing permissions and
015 *  limitations under the License.
016 *
017 */
018package org.apache.bcel.util;
019
020import java.util.ArrayList;
021import java.util.HashMap;
022import java.util.Iterator;
023import java.util.List;
024import java.util.Locale;
025import java.util.Map;
026import java.util.regex.Matcher;
027import java.util.regex.Pattern;
028
029import org.apache.bcel.Const;
030import org.apache.bcel.generic.ClassGenException;
031import org.apache.bcel.generic.InstructionHandle;
032import org.apache.bcel.generic.InstructionList;
033
034/**
035 * InstructionFinder is a tool to search for given instructions patterns, i.e.,
036 * match sequences of instructions in an instruction list via regular
037 * expressions. This can be used, e.g., in order to implement a peep hole
038 * optimizer that looks for code patterns and replaces them with faster
039 * equivalents.
040 *
041 * <p>
042 * This class internally uses the java.util.regex
043 * package to search for regular expressions.
044 *
045 * A typical application would look like this:
046 *
047 * <pre>
048 *
049 *
050 *   InstructionFinder f   = new InstructionFinder(il);
051 *   String            pat = &quot;IfInstruction ICONST_0 GOTO ICONST_1 NOP (IFEQ|IFNE)&quot;;
052 *
053 *   for (Iterator i = f.search(pat, constraint); i.hasNext(); ) {
054 *   InstructionHandle[] match = (InstructionHandle[])i.next();
055 *   ...
056 *   il.delete(match[1], match[5]);
057 *   ...
058 *   }
059 *
060 *
061 * </pre>
062 *
063 * @see org.apache.bcel.generic.Instruction
064 * @see InstructionList
065 */
066public class InstructionFinder {
067
068    private static final int OFFSET = 32767; // char + OFFSET is outside of LATIN-1
069    private static final int NO_OPCODES = 256; // Potential number, some are not used
070    private static final Map<String, String> map = new HashMap<>();
071    private final InstructionList il;
072    private String ilString; // instruction list as string
073    private InstructionHandle[] handles; // map instruction
074
075
076    // list to array
077    /**
078     * @param il
079     *          instruction list to search for given patterns
080     */
081    public InstructionFinder(final InstructionList il) {
082        this.il = il;
083        reread();
084    }
085
086
087    /**
088     * Reread the instruction list, e.g., after you've altered the list upon a
089     * match.
090     */
091    public final void reread() {
092        final int size = il.getLength();
093        final char[] buf = new char[size]; // Create a string with length equal to il length
094        handles = il.getInstructionHandles();
095        // Map opcodes to characters
096        for (int i = 0; i < size; i++) {
097            buf[i] = makeChar(handles[i].getInstruction().getOpcode());
098        }
099        ilString = new String(buf);
100    }
101
102
103    /**
104     * Map symbolic instruction names like "getfield" to a single character.
105     *
106     * @param pattern
107     *          instruction pattern in lower case
108     * @return encoded string for a pattern such as "BranchInstruction".
109     */
110    private static String mapName( final String pattern ) {
111        final String result = map.get(pattern);
112        if (result != null) {
113            return result;
114        }
115        for (short i = 0; i < NO_OPCODES; i++) {
116            if (pattern.equals(Const.getOpcodeName(i))) {
117                return "" + makeChar(i);
118            }
119        }
120        throw new IllegalArgumentException("Instruction unknown: " + pattern);
121    }
122
123
124    /**
125     * Replace symbolic names of instructions with the appropiate character and
126     * remove all white space from string. Meta characters such as +, * are
127     * ignored.
128     *
129     * @param pattern
130     *          The pattern to compile
131     * @return translated regular expression string
132     */
133    private static String compilePattern( final String pattern ) {
134        //Bug: BCEL-77 - Instructions are assumed to be english, to avoid odd Locale issues
135        final String lower = pattern.toLowerCase(Locale.ENGLISH);
136        final StringBuilder buf = new StringBuilder();
137        final int size = pattern.length();
138        for (int i = 0; i < size; i++) {
139            char ch = lower.charAt(i);
140            if (Character.isLetterOrDigit(ch)) {
141                final StringBuilder name = new StringBuilder();
142                while ((Character.isLetterOrDigit(ch) || ch == '_') && i < size) {
143                    name.append(ch);
144                    if (++i < size) {
145                        ch = lower.charAt(i);
146                    } else {
147                        break;
148                    }
149                }
150                i--;
151                buf.append(mapName(name.toString()));
152            } else if (!Character.isWhitespace(ch)) {
153                buf.append(ch);
154            }
155        }
156        return buf.toString();
157    }
158
159
160    /**
161     * @return the matched piece of code as an array of instruction (handles)
162     */
163    private InstructionHandle[] getMatch( final int matched_from, final int match_length ) {
164        final InstructionHandle[] match = new InstructionHandle[match_length];
165        System.arraycopy(handles, matched_from, match, 0, match_length);
166        return match;
167    }
168
169
170    /**
171     * Search for the given pattern in the instruction list. You can search for
172     * any valid opcode via its symbolic name, e.g. "istore". You can also use a
173     * super class or an interface name to match a whole set of instructions, e.g.
174     * "BranchInstruction" or "LoadInstruction". "istore" is also an alias for all
175     * "istore_x" instructions. Additional aliases are "if" for "ifxx", "if_icmp"
176     * for "if_icmpxx", "if_acmp" for "if_acmpxx".
177     *
178     * Consecutive instruction names must be separated by white space which will
179     * be removed during the compilation of the pattern.
180     *
181     * For the rest the usual pattern matching rules for regular expressions
182     * apply.
183     * <P>
184     * Example pattern:
185     *
186     * <pre>
187     * search(&quot;BranchInstruction NOP ((IfInstruction|GOTO)+ ISTORE Instruction)*&quot;);
188     * </pre>
189     *
190     * <p>
191     * If you alter the instruction list upon a match such that other matching
192     * areas are affected, you should call reread() to update the finder and call
193     * search() again, because the matches are cached.
194     *
195     * @param pattern
196     *          the instruction pattern to search for, where case is ignored
197     * @param from
198     *          where to start the search in the instruction list
199     * @param constraint
200     *          optional CodeConstraint to check the found code pattern for
201     *          user-defined constraints
202     * @return iterator of matches where e.nextElement() returns an array of
203     *         instruction handles describing the matched area
204     */
205    public final Iterator<InstructionHandle[]> search( final String pattern, final InstructionHandle from, final CodeConstraint constraint ) {
206        final String search = compilePattern(pattern);
207        int start = -1;
208        for (int i = 0; i < handles.length; i++) {
209            if (handles[i] == from) {
210                start = i; // Where to start search from (index)
211                break;
212            }
213        }
214        if (start == -1) {
215            throw new ClassGenException("Instruction handle " + from
216                    + " not found in instruction list.");
217        }
218        final Pattern regex = Pattern.compile(search);
219        final List<InstructionHandle[]> matches = new ArrayList<>();
220        final Matcher matcher = regex.matcher(ilString);
221        while (start < ilString.length() && matcher.find(start)) {
222            final int startExpr = matcher.start();
223            final int endExpr = matcher.end();
224            final int lenExpr = endExpr - startExpr;
225            final InstructionHandle[] match = getMatch(startExpr, lenExpr);
226            if ((constraint == null) || constraint.checkCode(match)) {
227                matches.add(match);
228            }
229            start = endExpr;
230        }
231        return matches.iterator();
232    }
233
234
235    /**
236     * Start search beginning from the start of the given instruction list.
237     *
238     * @param pattern
239     *          the instruction pattern to search for, where case is ignored
240     * @return iterator of matches where e.nextElement() returns an array of
241     *         instruction handles describing the matched area
242     */
243    public final Iterator<InstructionHandle[]> search( final String pattern ) {
244        return search(pattern, il.getStart(), null);
245    }
246
247
248    /**
249     * Start search beginning from `from'.
250     *
251     * @param pattern
252     *          the instruction pattern to search for, where case is ignored
253     * @param from
254     *          where to start the search in the instruction list
255     * @return iterator of matches where e.nextElement() returns an array of
256     *         instruction handles describing the matched area
257     */
258    public final Iterator<InstructionHandle[]> search( final String pattern, final InstructionHandle from ) {
259        return search(pattern, from, null);
260    }
261
262
263    /**
264     * Start search beginning from the start of the given instruction list. Check
265     * found matches with the constraint object.
266     *
267     * @param pattern
268     *          the instruction pattern to search for, case is ignored
269     * @param constraint
270     *          constraints to be checked on matching code
271     * @return instruction handle or `null' if the match failed
272     */
273    public final Iterator<InstructionHandle[]> search( final String pattern, final CodeConstraint constraint ) {
274        return search(pattern, il.getStart(), constraint);
275    }
276
277
278    /**
279     * Convert opcode number to char.
280     */
281    private static char makeChar( final short opcode ) {
282        return (char) (opcode + OFFSET);
283    }
284
285
286    /**
287     * @return the inquired instruction list
288     */
289    public final InstructionList getInstructionList() {
290        return il;
291    }
292
293    /**
294     * Code patterns found may be checked using an additional user-defined
295     * constraint object whether they really match the needed criterion. I.e.,
296     * check constraints that can not expressed with regular expressions.
297     *
298     */
299    public interface CodeConstraint {
300
301        /**
302         * @param match
303         *          array of instructions matching the requested pattern
304         * @return true if the matched area is really useful
305         */
306        boolean checkCode( InstructionHandle[] match );
307    }
308
309    // Initialize pattern map
310    static {
311        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)");
312        map.put("invokeinstruction", "(invokevirtual|invokeinterface|invokestatic|invokespecial|invokedynamic)");
313        map.put("arrayinstruction", "(baload|aastore|saload|caload|fastore|lastore|iaload|castore|iastore|aaload|bastore|sastore|faload|laload|daload|dastore)");
314        map.put("gotoinstruction", "(goto|goto_w)");
315        map.put("conversioninstruction", "(d2l|l2d|i2s|d2i|l2i|i2b|l2f|d2f|f2i|i2d|i2l|f2d|i2c|f2l|i2f)");
316        map.put("localvariableinstruction","(fstore|iinc|lload|dstore|dload|iload|aload|astore|istore|fload|lstore)");
317        map.put("loadinstruction", "(fload|dload|lload|iload|aload)");
318        map.put("fieldinstruction", "(getfield|putstatic|getstatic|putfield)");
319        map.put("cpinstruction", "(ldc2_w|invokeinterface|invokedynamic|multianewarray|putstatic|instanceof|getstatic|checkcast|getfield|invokespecial|ldc_w|invokestatic|invokevirtual|putfield|ldc|new|anewarray)");
320        map.put("stackinstruction", "(dup2|swap|dup2_x2|pop|pop2|dup|dup2_x1|dup_x2|dup_x1)");
321        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)");
322        map.put("returninstruction", "(lreturn|ireturn|freturn|dreturn|areturn|return)");
323        map.put("storeinstruction", "(istore|fstore|dstore|astore|lstore)");
324        map.put("select", "(tableswitch|lookupswitch)");
325        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)");
326        map.put("jsrinstruction", "(jsr|jsr_w)");
327        map.put("variablelengthinstruction", "(tableswitch|jsr|goto|lookupswitch)");
328        map.put("unconditionalbranch", "(goto|jsr|jsr_w|athrow|goto_w)");
329        map.put("constantpushinstruction", "(dconst|bipush|sipush|fconst|iconst|lconst)");
330        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)");
331        map.put("popinstruction", "(fstore|dstore|pop|pop2|astore|putstatic|istore|lstore)");
332        map.put("allocationinstruction", "(multianewarray|new|anewarray|newarray)");
333        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)");
334        map.put("pushinstruction", "(dup|lload|dup2|bipush|fload|ldc2_w|sipush|lconst|fconst|dload|getstatic|ldc_w|aconst_null|dconst|iload|ldc|iconst|aload)");
335        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)");
336        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)");
337        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)");
338        map.put("loadclass", "(multianewarray|invokeinterface|invokedynamic|instanceof|invokespecial|putfield|checkcast|putstatic|invokevirtual|new|getstatic|invokestatic|getfield|anewarray)");
339        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)");
340        // Some aliases
341        map.put("if_icmp", "(if_icmpne|if_icmpeq|if_icmple|if_icmpge|if_icmplt|if_icmpgt)");
342        map.put("if_acmp", "(if_acmpeq|if_acmpne)");
343        map.put("if", "(ifeq|ifne|iflt|ifge|ifgt|ifle)");
344        // Precompile some aliases first
345        map.put("iconst", precompile(Const.ICONST_0, Const.ICONST_5, Const.ICONST_M1));
346        map.put("lconst", new String(new char[] { '(', makeChar(Const.LCONST_0), '|', makeChar(Const.LCONST_1), ')' }));
347        map.put("dconst", new String(new char[] { '(', makeChar(Const.DCONST_0), '|', makeChar(Const.DCONST_1), ')' }));
348        map.put("fconst", new String(new char[] { '(', makeChar(Const.FCONST_0), '|', makeChar(Const.FCONST_1), '|', makeChar(Const.FCONST_2), ')' }));
349        map.put("lload", precompile(Const.LLOAD_0, Const.LLOAD_3, Const.LLOAD));
350        map.put("iload", precompile(Const.ILOAD_0, Const.ILOAD_3, Const.ILOAD));
351        map.put("dload", precompile(Const.DLOAD_0, Const.DLOAD_3, Const.DLOAD));
352        map.put("fload", precompile(Const.FLOAD_0, Const.FLOAD_3, Const.FLOAD));
353        map.put("aload", precompile(Const.ALOAD_0, Const.ALOAD_3, Const.ALOAD));
354        map.put("lstore", precompile(Const.LSTORE_0, Const.LSTORE_3, Const.LSTORE));
355        map.put("istore", precompile(Const.ISTORE_0, Const.ISTORE_3, Const.ISTORE));
356        map.put("dstore", precompile(Const.DSTORE_0, Const.DSTORE_3, Const.DSTORE));
357        map.put("fstore", precompile(Const.FSTORE_0, Const.FSTORE_3, Const.FSTORE));
358        map.put("astore", precompile(Const.ASTORE_0, Const.ASTORE_3, Const.ASTORE));
359        // Compile strings
360        for (final Map.Entry<String, String> entry : map.entrySet()) {
361            final String key = entry.getKey();
362            final String value = entry.getValue();
363            final char ch = value.charAt(1); // Omit already precompiled patterns
364            if (ch < OFFSET) {
365                map.put(key, compilePattern(value)); // precompile all patterns
366            }
367        }
368        // Add instruction alias to match anything
369        final StringBuilder buf = new StringBuilder("(");
370        for (short i = 0; i < NO_OPCODES; i++) {
371            if (Const.getNoOfOperands(i) != Const.UNDEFINED) { // Not an invalid opcode
372                buf.append(makeChar(i));
373                if (i < NO_OPCODES - 1) {
374                    buf.append('|');
375                }
376            }
377        }
378        buf.append(')');
379        map.put("instruction", buf.toString());
380    }
381
382
383    private static String precompile( final short from, final short to, final short extra ) {
384        final StringBuilder buf = new StringBuilder("(");
385        for (short i = from; i <= to; i++) {
386            buf.append(makeChar(i));
387            buf.append('|');
388        }
389        buf.append(makeChar(extra));
390        buf.append(")");
391        return buf.toString();
392    }
393
394
395    /*
396     * Internal debugging routines.
397     */
398//    private static final String pattern2string( String pattern ) {
399//        return pattern2string(pattern, true);
400//    }
401
402
403//    private static final String pattern2string( String pattern, boolean make_string ) {
404//        StringBuffer buf = new StringBuffer();
405//        for (int i = 0; i < pattern.length(); i++) {
406//            char ch = pattern.charAt(i);
407//            if (ch >= OFFSET) {
408//                if (make_string) {
409//                    buf.append(Constants.getOpcodeName(ch - OFFSET));
410//                } else {
411//                    buf.append((ch - OFFSET));
412//                }
413//            } else {
414//                buf.append(ch);
415//            }
416//        }
417//        return buf.toString();
418//    }
419}