|
| 1 | +/* |
| 2 | + * To change this license header, choose License Headers in Project Properties. |
| 3 | + * To change this template file, choose Tools | Templates |
| 4 | + * and open the template in the editor. |
| 5 | + */ |
| 6 | +package com.oracle.graal.python.builtins.objects.str; |
| 7 | + |
| 8 | +import com.oracle.graal.python.nodes.PGuards; |
| 9 | +import com.oracle.graal.python.runtime.PythonOptions; |
| 10 | +import com.oracle.truffle.api.CompilerAsserts; |
| 11 | +import com.oracle.truffle.api.CompilerDirectives.TruffleBoundary; |
| 12 | +import com.oracle.truffle.api.profiles.ConditionProfile; |
| 13 | + |
| 14 | +public class LazyString implements CharSequence { |
| 15 | + |
| 16 | + protected static int MinLazyStringLength = PythonOptions.getMinLazyStringLength(); |
| 17 | + protected static boolean UseLazyStrings = PythonOptions.useLazyString(); |
| 18 | + |
| 19 | + public static int length(CharSequence cs, ConditionProfile profile1, ConditionProfile profile2) { |
| 20 | + if (profile1.profile(cs instanceof String)) { |
| 21 | + return ((String) cs).length(); |
| 22 | + } else if (profile2.profile(cs instanceof LazyString)) { |
| 23 | + return ((LazyString) cs).length(); |
| 24 | + } |
| 25 | + return lengthIntl(cs); |
| 26 | + } |
| 27 | + |
| 28 | + @TruffleBoundary |
| 29 | + private static int lengthIntl(CharSequence cs) { |
| 30 | + return cs.length(); |
| 31 | + } |
| 32 | + |
| 33 | + @TruffleBoundary |
| 34 | + public static CharSequence create(CharSequence left, CharSequence right) { |
| 35 | + assert PGuards.isString(left); |
| 36 | + assert PGuards.isString(right); |
| 37 | + if (UseLazyStrings) { |
| 38 | + if (left.length() == 0) { |
| 39 | + return right; |
| 40 | + } else if (right.length() == 0) { |
| 41 | + return left; |
| 42 | + } |
| 43 | + int resultLength = left.length() + right.length(); |
| 44 | + if (resultLength < MinLazyStringLength) { |
| 45 | + return left.toString() + right.toString(); |
| 46 | + } |
| 47 | + return new LazyString(left, right, resultLength); |
| 48 | + } else { |
| 49 | + return left.toString() + right.toString(); |
| 50 | + } |
| 51 | + } |
| 52 | + |
| 53 | + /** |
| 54 | + * Only use when invariants are checked already, e.g. from specializing nodes. |
| 55 | + */ |
| 56 | + @TruffleBoundary |
| 57 | + public static CharSequence createChecked(CharSequence left, CharSequence right, int length) { |
| 58 | + assert assertChecked(left, right, length); |
| 59 | + return new LazyString(left, right, length); |
| 60 | + } |
| 61 | + |
| 62 | + private static boolean assertChecked(CharSequence left, CharSequence right, int length) { |
| 63 | + assert UseLazyStrings; |
| 64 | + assert (PGuards.isString(left) || left instanceof LazyString) && (PGuards.isString(right) || right instanceof LazyString); |
| 65 | + assert length == left.length() + right.length(); |
| 66 | + assert left.length() > 0 && right.length() > 0; |
| 67 | + assert length >= MinLazyStringLength; |
| 68 | + return true; |
| 69 | + } |
| 70 | + |
| 71 | + /** |
| 72 | + * Variant of {@link #createChecked} that tries to concatenate a very short string to an already |
| 73 | + * short root leaf up-front, e.g. when appending single characters. |
| 74 | + */ |
| 75 | + @TruffleBoundary |
| 76 | + public static CharSequence createCheckedShort(CharSequence left, CharSequence right, int length) { |
| 77 | + assertChecked(left, right, length); |
| 78 | + final int tinyLimit = 1; |
| 79 | + final int appendToLeafLimit = MinLazyStringLength / 2; |
| 80 | + if (left instanceof LazyString && right instanceof String && right.length() <= tinyLimit) { |
| 81 | + CharSequence ll = ((LazyString) left).left; |
| 82 | + CharSequence lr = ((LazyString) left).right; |
| 83 | + if (lr != null && lr instanceof String && lr.length() + right.length() <= appendToLeafLimit) { |
| 84 | + return new LazyString(ll, lr.toString() + right.toString(), length); |
| 85 | + } |
| 86 | + } else if (left instanceof String && left.length() <= tinyLimit && right instanceof LazyString) { |
| 87 | + CharSequence ll = ((LazyString) right).left; |
| 88 | + CharSequence lr = ((LazyString) right).right; |
| 89 | + if (lr != null && ll instanceof String && left.length() + ll.length() <= appendToLeafLimit) { |
| 90 | + return new LazyString(left.toString() + ll.toString(), lr, length); |
| 91 | + } |
| 92 | + } |
| 93 | + return new LazyString(left, right, length); |
| 94 | + } |
| 95 | + |
| 96 | + private CharSequence left; |
| 97 | + private CharSequence right; |
| 98 | + private final int length; |
| 99 | + |
| 100 | + private LazyString(CharSequence left, CharSequence right, int length) { |
| 101 | + assert left.length() > 0 && right.length() > 0 && length == left.length() + right.length(); |
| 102 | + this.left = left; |
| 103 | + this.right = right; |
| 104 | + this.length = length; |
| 105 | + } |
| 106 | + |
| 107 | + private LazyString(CharSequence left, CharSequence right) { |
| 108 | + this(left, right, left.length() + right.length()); |
| 109 | + } |
| 110 | + |
| 111 | + @Override |
| 112 | + public int length() { |
| 113 | + return length; |
| 114 | + } |
| 115 | + |
| 116 | + @Override |
| 117 | + public String toString() { |
| 118 | + if (!isFlat()) { |
| 119 | + flatten(); |
| 120 | + } |
| 121 | + return (String) left; |
| 122 | + } |
| 123 | + |
| 124 | + private boolean isFlat() { |
| 125 | + return right == null; |
| 126 | + } |
| 127 | + |
| 128 | + @TruffleBoundary |
| 129 | + private void flatten() { |
| 130 | + char[] dst = new char[length]; |
| 131 | + flatten(this, 0, length, dst, 0); |
| 132 | + left = new String(dst); |
| 133 | + right = null; |
| 134 | + } |
| 135 | + |
| 136 | + private static void flatten(CharSequence src, int srcBegin, int srcEnd, char[] dst, int dstBegin) { |
| 137 | + CompilerAsserts.neverPartOfCompilation(); |
| 138 | + CharSequence str = src; |
| 139 | + int from = srcBegin; |
| 140 | + int to = srcEnd; |
| 141 | + int dstFrom = dstBegin; |
| 142 | + for (;;) { |
| 143 | + assert 0 <= from && from <= to && to <= str.length(); |
| 144 | + if (str instanceof LazyString) { |
| 145 | + LazyString lazyString = (LazyString) str; |
| 146 | + CharSequence left = lazyString.left; |
| 147 | + CharSequence right = lazyString.right; |
| 148 | + int mid = left.length(); |
| 149 | + |
| 150 | + if (to - mid >= mid - from) { |
| 151 | + // right is longer, recurse left |
| 152 | + if (from < mid) { |
| 153 | + if (left instanceof String) { |
| 154 | + ((String) left).getChars(from, mid, dst, dstFrom); |
| 155 | + } else { |
| 156 | + flatten(left, from, mid, dst, dstFrom); |
| 157 | + } |
| 158 | + dstFrom += mid - from; |
| 159 | + from = 0; |
| 160 | + } else { |
| 161 | + from -= mid; |
| 162 | + } |
| 163 | + to -= mid; |
| 164 | + str = right; |
| 165 | + } else { |
| 166 | + // left is longer, recurse right |
| 167 | + if (to > mid) { |
| 168 | + if (right instanceof String) { |
| 169 | + ((String) right).getChars(0, to - mid, dst, dstFrom + mid - from); |
| 170 | + } else { |
| 171 | + flatten(right, 0, to - mid, dst, dstFrom + mid - from); |
| 172 | + } |
| 173 | + to = mid; |
| 174 | + } |
| 175 | + str = left; |
| 176 | + } |
| 177 | + } else if (str instanceof String) { |
| 178 | + ((String) str).getChars(from, to, dst, dstFrom); |
| 179 | + return; |
| 180 | + } |
| 181 | + } |
| 182 | + } |
| 183 | + |
| 184 | + @Override |
| 185 | + public char charAt(int index) { |
| 186 | + return toString().charAt(index); |
| 187 | + } |
| 188 | + |
| 189 | + @Override |
| 190 | + public CharSequence subSequence(int start, int end) { |
| 191 | + return toString().subSequence(start, end); |
| 192 | + } |
| 193 | + |
| 194 | + public boolean isEmpty() { |
| 195 | + return length == 0; |
| 196 | + } |
| 197 | + |
| 198 | + // accessed via Java Interop, JDK-8062624.js |
| 199 | + @TruffleBoundary |
| 200 | + public boolean startsWith(String prefix) { |
| 201 | + return toString().startsWith(prefix); |
| 202 | + } |
| 203 | + |
| 204 | + // accessed via Java Interop, JDK-8062624.js |
| 205 | + @TruffleBoundary |
| 206 | + public boolean endsWith(String prefix) { |
| 207 | + return toString().endsWith(prefix); |
| 208 | + } |
| 209 | + |
| 210 | + // accessed via Java Interop, JDK-8062624.js |
| 211 | + @TruffleBoundary |
| 212 | + public byte[] getBytes() { |
| 213 | + return toString().getBytes(); |
| 214 | + } |
| 215 | +} |
0 commit comments