diff --git a/com.oracle.truffle.r.nodes.builtin/src/com/oracle/truffle/r/nodes/builtin/base/GrepFunctions.java b/com.oracle.truffle.r.nodes.builtin/src/com/oracle/truffle/r/nodes/builtin/base/GrepFunctions.java
index f36dcf3633f70c228f60ac3fcd56f68dd2341ae4..ae5ec546788a7b82dc55dc25c92317349582fcfb 100644
--- a/com.oracle.truffle.r.nodes.builtin/src/com/oracle/truffle/r/nodes/builtin/base/GrepFunctions.java
+++ b/com.oracle.truffle.r.nodes.builtin/src/com/oracle/truffle/r/nodes/builtin/base/GrepFunctions.java
@@ -565,6 +565,16 @@ public class GrepFunctions {
@RBuiltin(name = "regexpr", kind = INTERNAL, parameterNames = {"pattern", "text", "ignore.case", "perl", "fixed", "useBytes"}, behavior = PURE)
public abstract static class Regexp extends CommonCodeAdapter {
+ protected static final class IndexAndSize {
+ protected int index;
+ protected int size;
+
+ public IndexAndSize(int index, int size) {
+ this.index = index;
+ this.size = size;
+ }
+ }
+
@Specialization
@TruffleBoundary
protected Object regexp(RAbstractStringVector patternArg, RAbstractStringVector vector, byte ignoreCaseL, byte perlL, byte fixedL, byte useBytesL) {
@@ -572,34 +582,45 @@ public class GrepFunctions {
boolean ignoreCase = RRuntime.fromLogical(ignoreCaseL);
String pattern = RegExp.checkPreDefinedClasses(patternArg.getDataAt(0));
int[] result = new int[vector.getLength()];
+ int[] matchLength = new int[vector.getLength()];
for (int i = 0; i < vector.getLength(); i++) {
- result[i] = findIndex(pattern, vector.getDataAt(i), ignoreCase, fixedL == RRuntime.LOGICAL_TRUE).get(0);
+ IndexAndSize res = findIndexAndSize(pattern, vector.getDataAt(i), ignoreCase, fixedL == RRuntime.LOGICAL_TRUE).get(0);
+ result[i] = res.index;
+ matchLength[i] = res.size;
}
- // TODO attribute as per spec
- return RDataFactory.createIntVector(result, RDataFactory.COMPLETE_VECTOR);
+ // TODO useBytes attribute as per spec
+ RIntVector ret = RDataFactory.createIntVector(result, RDataFactory.COMPLETE_VECTOR);
+ ret.setAttr("match.length", RDataFactory.createIntVector(matchLength, RDataFactory.COMPLETE_VECTOR));
+ return ret;
}
- protected static List<Integer> findIndex(String pattern, String text, boolean ignoreCase, boolean fixed) {
- List<Integer> list = new ArrayList<>();
+ protected static List<IndexAndSize> findIndexAndSize(String pattern, String text, boolean ignoreCase, boolean fixed) {
+ List<IndexAndSize> list = new ArrayList<>();
if (fixed) {
- int index;
- if (ignoreCase) {
- index = text.toLowerCase().indexOf(pattern.toLowerCase());
- } else {
- index = text.indexOf(pattern);
+ int index = 0;
+ while (true) {
+ if (ignoreCase) {
+ index = text.toLowerCase().indexOf(pattern.toLowerCase(), index);
+ } else {
+ index = text.indexOf(pattern, index);
+ }
+ if (index == -1) {
+ break;
+ }
+ list.add(new IndexAndSize(index + 1, pattern.length()));
+ index += pattern.length();
}
- list.add(index == -1 ? index : index + 1);
} else {
Matcher m = getPatternMatcher(pattern, text, ignoreCase);
while (m.find()) {
// R starts counting at index 1
- list.add(m.start() + 1);
- }
- if (list.size() > 0) {
- return list;
+ list.add(new IndexAndSize(m.start() + 1, m.end() - m.start()));
}
- list.add(-1);
}
+ if (list.size() > 0) {
+ return list;
+ }
+ list.add(new IndexAndSize(-1, -1));
return list;
}
@@ -622,17 +643,22 @@ public class GrepFunctions {
boolean fixed = RRuntime.fromLogical(fixedL);
Object[] result = new Object[vector.getLength()];
for (int i = 0; i < vector.getLength(); i++) {
- int[] data = toIntArray(findIndex(pattern, vector.getDataAt(i), ignoreCase, fixed));
- result[i] = RDataFactory.createIntVector(data, RDataFactory.COMPLETE_VECTOR);
- // TODO attributes as per spec
+ List<IndexAndSize> l = findIndexAndSize(pattern, vector.getDataAt(i), ignoreCase, fixed);
+ int[] indexes = toIndexOrSizeArray(l, true);
+ int[] sizes = toIndexOrSizeArray(l, false);
+ RIntVector res = RDataFactory.createIntVector(indexes, RDataFactory.COMPLETE_VECTOR);
+ res.setAttr("match.length", RDataFactory.createIntVector(sizes, RDataFactory.COMPLETE_VECTOR));
+ result[i] = res;
+ // TODO useBytes attributes as per spec
}
return RDataFactory.createList(result);
}
- private static int[] toIntArray(List<Integer> list) {
+ private static int[] toIndexOrSizeArray(List<IndexAndSize> list, boolean index) {
int[] arr = new int[list.size()];
for (int i = 0; i < list.size(); i++) {
- arr[i] = list.get(i);
+ IndexAndSize res = list.get(i);
+ arr[i] = index ? res.index : res.size;
}
return arr;
}
diff --git a/com.oracle.truffle.r.test/src/com/oracle/truffle/r/test/ExpectedTestOutput.test b/com.oracle.truffle.r.test/src/com/oracle/truffle/r/test/ExpectedTestOutput.test
index 01e3fcd48ae14532a5137cb0329f0b4184b4e46e..747b14fce59ab65bc0324e28d1ca11432e604e52 100644
--- a/com.oracle.truffle.r.test/src/com/oracle/truffle/r/test/ExpectedTestOutput.test
+++ b/com.oracle.truffle.r.test/src/com/oracle/truffle/r/test/ExpectedTestOutput.test
@@ -20692,6 +20692,32 @@ attr(,"useBytes")
[1] TRUE
+##com.oracle.truffle.r.test.builtins.TestBuiltin_gregexpr.testRegExpr
+#{ x<-gregexpr("foo", c("bar foo foo", "foo"), fixed=F); as.integer(c(x[[1]], x[[2]])) }
+[1] 5 9 1
+
+##com.oracle.truffle.r.test.builtins.TestBuiltin_gregexpr.testRegExpr
+#{ x<-gregexpr("foo", c("bar foo foo", "foo"), fixed=F); list(attr(x[[1]], "match.length"), attr(x[[2]], "match.length")) }
+[[1]]
+[1] 3 3
+
+[[2]]
+[1] 3
+
+
+##com.oracle.truffle.r.test.builtins.TestBuiltin_gregexpr.testRegExpr
+#{ x<-gregexpr("foo", c("bar foo foo", "foo"), fixed=T); as.integer(c(x[[1]], x[[2]])) }
+[1] 5 9 1
+
+##com.oracle.truffle.r.test.builtins.TestBuiltin_gregexpr.testRegExpr
+#{ x<-gregexpr("foo", c("bar foo foo", "foo"), fixed=T); list(attr(x[[1]], "match.length"), attr(x[[2]], "match.length")) }
+[[1]]
+[1] 3 3
+
+[[2]]
+[1] 3
+
+
##com.oracle.truffle.r.test.builtins.TestBuiltin_gregexpr.testgregexpr1
#argv <- list('', 'abc', FALSE, FALSE, FALSE, FALSE); .Internal(gregexpr(argv[[1]], argv[[2]], argv[[3]], argv[[4]], argv[[5]], argv[[6]]))
[[1]]
@@ -40138,6 +40164,14 @@ attr(,"match.length")
attr(,"useBytes")
[1] TRUE
+##com.oracle.truffle.r.test.builtins.TestBuiltin_regexpr.testRegExpr
+#{ as.integer(regexpr("foo", c("bar foo foo", "foo"), fixed=F)) }
+[1] 5 1
+
+##com.oracle.truffle.r.test.builtins.TestBuiltin_regexpr.testRegExpr
+#{ as.integer(regexpr("foo", c("bar foo foo", "foo"), fixed=T)) }
+[1] 5 1
+
##com.oracle.truffle.r.test.builtins.TestBuiltin_regexpr.testRegExpr
#{ regexpr("aaa", "bbbaaaccc", fixed=TRUE) }
[1] 4
@@ -40178,6 +40212,14 @@ attr(,"useBytes")
#{ x<-regexpr("aaa", c("bbbaaaccc", "hah"), fixed=TRUE); c(x[1], x[2]) }
[1] 4 -1
+##com.oracle.truffle.r.test.builtins.TestBuiltin_regexpr.testRegExpr
+#{ x<-regexpr("foo", c("bar foo foo", "foo")); attr(x, "match.length") }
+[1] 3 3
+
+##com.oracle.truffle.r.test.builtins.TestBuiltin_regexpr.testRegExpr
+#{ x<-regexpr("foo", c("bar foo foo", "foo"), fixed=T); attr(x, "match.length") }
+[1] 3 3
+
##com.oracle.truffle.r.test.builtins.TestBuiltin_regexpr.testregexpr1
#argv <- list('package:', 'exNSS4', FALSE, FALSE, TRUE, FALSE); .Internal(regexpr(argv[[1]], argv[[2]], argv[[3]], argv[[4]], argv[[5]], argv[[6]]))
[1] -1
diff --git a/com.oracle.truffle.r.test/src/com/oracle/truffle/r/test/builtins/TestBuiltin_gregexpr.java b/com.oracle.truffle.r.test/src/com/oracle/truffle/r/test/builtins/TestBuiltin_gregexpr.java
index 32e9867ee52e84d3ea7a31f20410ae095cd518eb..b4e870f0c06039804a2331ef84ca312cf2109bc2 100644
--- a/com.oracle.truffle.r.test/src/com/oracle/truffle/r/test/builtins/TestBuiltin_gregexpr.java
+++ b/com.oracle.truffle.r.test/src/com/oracle/truffle/r/test/builtins/TestBuiltin_gregexpr.java
@@ -103,5 +103,10 @@ public class TestBuiltin_gregexpr extends TestBase {
assertEval(Ignored.Unknown, "gregexpr(\"e\",c(\"arm\",\"foot\",\"lefroo\", \"bafoobar\"))");
// NOTE: this is without attributes
assertEval(Ignored.Unknown, "gregexpr(\"(a)[^a]\\\\1\", c(\"andrea apart\", \"amadeus\", NA))");
+
+ assertEval("{ x<-gregexpr(\"foo\", c(\"bar foo foo\", \"foo\"), fixed=T); as.integer(c(x[[1]], x[[2]])) }");
+ assertEval("{ x<-gregexpr(\"foo\", c(\"bar foo foo\", \"foo\"), fixed=F); as.integer(c(x[[1]], x[[2]])) }");
+ assertEval("{ x<-gregexpr(\"foo\", c(\"bar foo foo\", \"foo\"), fixed=T); list(attr(x[[1]], \"match.length\"), attr(x[[2]], \"match.length\")) }");
+ assertEval("{ x<-gregexpr(\"foo\", c(\"bar foo foo\", \"foo\"), fixed=F); list(attr(x[[1]], \"match.length\"), attr(x[[2]], \"match.length\")) }");
}
}
diff --git a/com.oracle.truffle.r.test/src/com/oracle/truffle/r/test/builtins/TestBuiltin_regexpr.java b/com.oracle.truffle.r.test/src/com/oracle/truffle/r/test/builtins/TestBuiltin_regexpr.java
index d0af20a2ac660aa0f515e4c91db64514caa41f54..262c013c9f7371a5e5391860b99479975dbe3450 100644
--- a/com.oracle.truffle.r.test/src/com/oracle/truffle/r/test/builtins/TestBuiltin_regexpr.java
+++ b/com.oracle.truffle.r.test/src/com/oracle/truffle/r/test/builtins/TestBuiltin_regexpr.java
@@ -99,5 +99,10 @@ public class TestBuiltin_regexpr extends TestBase {
assertEval("{ x<-regexpr(\"aaa\", c(\"bbbaaaccc\", \"hah\"), fixed=TRUE); c(x[1], x[2]) }");
assertEval("{ x <- \"methods.html\"; pos <- regexpr(\"\\\\.([[:alnum:]]+)$\", x); substring(x, pos + 1L) }");
+
+ assertEval("{ as.integer(regexpr(\"foo\", c(\"bar foo foo\", \"foo\"), fixed=T)) }");
+ assertEval("{ as.integer(regexpr(\"foo\", c(\"bar foo foo\", \"foo\"), fixed=F)) }");
+ assertEval("{ x<-regexpr(\"foo\", c(\"bar foo foo\", \"foo\"), fixed=T); attr(x, \"match.length\") }");
+ assertEval("{ x<-regexpr(\"foo\", c(\"bar foo foo\", \"foo\")); attr(x, \"match.length\") }");
}
}