diff --git a/com.oracle.truffle.r.native/fficall/src/truffle_common/pcre_rffi.c b/com.oracle.truffle.r.native/fficall/src/truffle_common/pcre_rffi.c
index f70dca58caa5a535c9e99280f153e008bb04bde1..2f6e4e5b5ab9cc87180cc5056c8a66359eb3a880 100644
--- a/com.oracle.truffle.r.native/fficall/src/truffle_common/pcre_rffi.c
+++ b/com.oracle.truffle.r.native/fficall/src/truffle_common/pcre_rffi.c
@@ -74,7 +74,7 @@ int call_pcre_getcapturenames(void (*setcapturename)(int i, char *name), long co
if (res < 0) {
return res;
}
- // from GNU R's grep.c
+ // from GNU R's grep.c function do_regexpr()
for(int i = 0; i < nameCount; i++) {
char* entry = nameTable + nameEntrySize * i;
int captureNum = (entry[0] << 8) + entry[1] - 1;
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 6b3b9ef8cec8136a97daf6675ac75278faad1014..2799b9c753295d5c511ebfbe90c87763da88e9bb 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
@@ -765,6 +765,7 @@ public class GrepFunctions {
protected final int[] captureLength;
protected final String[] captureNames;
protected final boolean hasCapture;
+ protected final boolean hasCaptureResult;
public Info(int index, int size, int[] captureStart, int[] captureLength, String[] captureNames) {
this.index = index;
@@ -773,6 +774,7 @@ public class GrepFunctions {
this.captureLength = captureLength;
this.captureNames = captureNames;
this.hasCapture = captureStart != null && captureLength != null;
+ this.hasCaptureResult = captureNames != null && captureNames.length > 0;
}
}
@@ -799,7 +801,7 @@ public class GrepFunctions {
// TODO: useBytes normally depends on the value of the parameter and (if false) on
// whether the string is ASCII
boolean useBytes = true;
- boolean hasAnyCapture = false;
+ boolean hasCaptureResult = false;
int[] result = new int[vector.getLength()];
int[] matchLength = new int[vector.getLength()];
String[] captureNames = null;
@@ -810,30 +812,36 @@ public class GrepFunctions {
Arrays.fill(result, 1);
} else {
for (int i = 0; i < vector.getLength(); i++) {
- Info res = getInfo(common, pattern, vector.getDataAt(i), ignoreCase, perl, fixed).get(0);
+ Info res = getInfo(common, pattern, vector.getDataAt(i), ignoreCase, perl, fixed, true).get(0);
result[i] = res.index;
matchLength[i] = res.size;
if (res.hasCapture) {
- hasAnyCapture = true;
+ hasCaptureResult = true;
if (captureNames == null) {
// first time we see captures
captureNames = res.captureNames;
+ // length of res.captureNames gives the max amount of captures
captureStart = new int[captureNames.length * vector.getLength()];
captureLength = new int[captureNames.length * vector.getLength()];
- // previous matches had no capture - fill in result with -1-s
- for (int k = 0; k < i; k++) {
- setNoCaptureValues(captureStart, captureLength, captureNames.length, vector.getLength(), k);
- }
}
- assert captureNames.length == res.captureStart.length;
- assert captureNames.length == res.captureLength.length;
- for (int j = 0; j < captureNames.length; j++) {
+ assert captureNames.length == res.captureStart.length || captureNames.length - 1 == res.captureStart.length : captureNames.length + ", " + res.captureStart.length;
+ assert captureNames.length == res.captureLength.length || captureNames.length - 1 == res.captureLength.length : captureNames.length + ", " + res.captureLength.length;
+ for (int j = 0; j < res.captureStart.length; j++) {
+ // well, res.captureStart might be shorter then
+ // res.captureNames (but never more then by 1?),
+ // just ignore the remaining (zero) elements in captureStart
captureStart[j * vector.getLength() + i] = res.captureStart[j];
captureLength[j * vector.getLength() + i] = res.captureLength[j];
}
- } else if (hasAnyCapture) {
- // no capture for this part of the vector, but there are previous
- // captures
+ } else if (res.hasCaptureResult) {
+ // no capture for this part of the vector, but even then
+ // we want to return a "no capture" result
+ hasCaptureResult = true;
+ captureNames = res.captureNames;
+ if (captureStart == null) {
+ captureStart = new int[captureNames.length * vector.getLength()];
+ captureLength = new int[captureNames.length * vector.getLength()];
+ }
setNoCaptureValues(captureStart, captureLength, captureNames.length, vector.getLength(), i);
}
}
@@ -843,7 +851,7 @@ public class GrepFunctions {
if (useBytes) {
setUseBytesAttrNode.execute(ret, RRuntime.LOGICAL_TRUE);
}
- if (hasAnyCapture) {
+ if (hasCaptureResult) {
RStringVector captureNamesVec = RDataFactory.createStringVector(captureNames, RDataFactory.COMPLETE_VECTOR);
RIntVector captureStartVec = RDataFactory.createIntVector(captureStart, RDataFactory.COMPLETE_VECTOR, new int[]{vector.getLength(), captureNames.length});
setDimNamesAttrNode.execute(captureStartVec, RDataFactory.createList(new Object[]{RNull.instance, captureNamesVec.copy()}));
@@ -860,6 +868,10 @@ public class GrepFunctions {
}
protected List<Info> getInfo(CommonCodeNode common, String pattern, String text, boolean ignoreCase, boolean perl, boolean fixed) {
+ return getInfo(common, pattern, text, ignoreCase, perl, fixed, false);
+ }
+
+ protected List<Info> getInfo(CommonCodeNode common, String pattern, String text, boolean ignoreCase, boolean perl, boolean fixed, boolean onlyFirst) {
List<Info> list = new ArrayList<>();
if (fixed) {
int index = 0;
@@ -878,18 +890,25 @@ public class GrepFunctions {
} else if (perl) {
PCRERFFI.Result pcre = common.compilePerlPattern(pattern, ignoreCase);
int maxCaptureCount = getCaptureCountNode.execute(pcre.result, 0);
+ if (maxCaptureCount < 0) {
+ throw error(Message.PCRE_FULLINFO_RETURNED, maxCaptureCount);
+ }
+
+ String[] captureNames = getCaptureNamesNode.execute(pcre.result, 0, maxCaptureCount);
+ assert maxCaptureCount == captureNames.length;
+ for (int i = 0; i < captureNames.length; i++) {
+ if (captureNames[i] == null) {
+ captureNames[i] = "";
+ }
+ }
+
int[] ovector = new int[(maxCaptureCount + 1) * 3];
int offset = 0;
while (true) {
int captureCount = execNode.execute(pcre.result, 0, text, offset, 0, ovector);
if (captureCount >= 0) {
- String[] captureNames = getCaptureNamesNode.execute(pcre.result, 0, maxCaptureCount);
- for (int i = 0; i < captureNames.length; i++) {
- if (captureNames[i] == null) {
- captureNames[i] = "";
- }
- }
- assert captureCount - 1 == captureNames.length;
+ assert captureCount - 1 == captureNames.length || captureCount == captureNames.length : captureCount + ", " + captureNames.length;
+
int[] captureStart = null;
int[] captureLength = null;
if (captureCount > 1) {
@@ -904,11 +923,18 @@ public class GrepFunctions {
}
// R starts counting at index 1
list.add(new Info(ovector[0] + 1, ovector[1] - ovector[0], captureStart, captureLength, captureNames));
+ if (onlyFirst) {
+ break;
+ }
offset = ovector[1];
} else {
break;
}
}
+ if (list.isEmpty() && maxCaptureCount > 0) {
+ // at least a return array of emtpty string names, is necessary for output
+ list.add(new Info(-1, -1, null, null, captureNames));
+ }
} else {
Matcher m = getPatternMatcher(pattern, text, ignoreCase);
while (m.find()) {
diff --git a/com.oracle.truffle.r.runtime/src/com/oracle/truffle/r/runtime/RError.java b/com.oracle.truffle.r.runtime/src/com/oracle/truffle/r/runtime/RError.java
index 1253fd0cf72ca97ed5573ce9832a9b6545c36525..6238dd5937e4efcf9bef23976c01d12b20722e75 100644
--- a/com.oracle.truffle.r.runtime/src/com/oracle/truffle/r/runtime/RError.java
+++ b/com.oracle.truffle.r.runtime/src/com/oracle/truffle/r/runtime/RError.java
@@ -645,6 +645,7 @@ public final class RError extends RuntimeException implements TruffleException {
FIRST_ARGUMENT_MUST_BE_CHARACTER("the first argument must be of mode character"),
ALL_ATTRIBUTES_NAMES("all attributes must have names [%d does not]"),
INVALID_REGEXP("invalid regular expression '%s'"),
+ PCRE_FULLINFO_RETURNED("'pcre_fullinfo' returned '%s'"),
INVALID_REGEXP_REASON("invalid regular expression '%s': %s"),
COERCING_ARGUMENT("coercing argument of type '%s' to %s"),
MUST_BE_TRUE_FALSE_ENVIRONMENT("'%s' must be TRUE, FALSE or an environment"),
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 d1fe132529d3ad8f8f3dc5aa0036a2c8d7fcb8a2..663de5d632ac7d8f778b957f96ff41d8506483bd 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
@@ -50732,6 +50732,218 @@ attr(,"useBytes")
[1] TRUE
+##com.oracle.truffle.r.test.builtins.TestBuiltin_regexec.testregexec1#
+#regexpr("^((.*))$", 'A1', perl=T)
+[1] 1
+attr(,"match.length")
+[1] 2
+attr(,"useBytes")
+[1] TRUE
+attr(,"capture.start")
+
+[1,] 1 1
+attr(,"capture.length")
+
+[1,] 2 2
+attr(,"capture.names")
+[1] "" ""
+
+##com.oracle.truffle.r.test.builtins.TestBuiltin_regexec.testregexec1#
+#regexpr("^(([A-Z)|([a-z]))$", 'Aa', perl=T)
+[1] -1
+attr(,"match.length")
+[1] -1
+attr(,"useBytes")
+[1] TRUE
+attr(,"capture.start")
+
+[1,] -1 -1
+attr(,"capture.length")
+
+[1,] -1 -1
+attr(,"capture.names")
+[1] "" ""
+
+##com.oracle.truffle.r.test.builtins.TestBuiltin_regexec.testregexec1#
+#regexpr("^(([A-Z)|([a-z]))$", c('A', 'Aa'), perl=T)
+[1] 1 -1
+attr(,"match.length")
+[1] 1 -1
+attr(,"useBytes")
+[1] TRUE
+attr(,"capture.start")
+
+[1,] 1 1
+[2,] -1 -1
+attr(,"capture.length")
+
+[1,] 1 1
+[2,] -1 -1
+attr(,"capture.names")
+[1] "" ""
+
+##com.oracle.truffle.r.test.builtins.TestBuiltin_regexec.testregexec1#
+#regexpr("^(([A-Z)|([a-z]))$", c('Aa', 'A'), perl=T)
+[1] -1 1
+attr(,"match.length")
+[1] -1 1
+attr(,"useBytes")
+[1] TRUE
+attr(,"capture.start")
+
+[1,] -1 -1
+[2,] 1 1
+attr(,"capture.length")
+
+[1,] -1 -1
+[2,] 1 1
+attr(,"capture.names")
+[1] "" ""
+
+##com.oracle.truffle.r.test.builtins.TestBuiltin_regexec.testregexec1#
+#regexpr("^(.*)$", 'A1', perl=T)
+[1] 1
+attr(,"match.length")
+[1] 2
+attr(,"useBytes")
+[1] TRUE
+attr(,"capture.start")
+
+[1,] 1
+attr(,"capture.length")
+
+[1,] 2
+attr(,"capture.names")
+[1] ""
+
+##com.oracle.truffle.r.test.builtins.TestBuiltin_regexec.testregexec1#
+#regexpr("^(?:(?:^\\[([^\\]]+)\\])?(?:'?([^']+)'?!)?([a-zA-Z0-9:\\-$\\[\\]]+)|(.*))$", 'A1 A1', perl=T)
+[1] 1
+attr(,"match.length")
+[1] 5
+attr(,"useBytes")
+[1] TRUE
+attr(,"capture.start")
+
+[1,] 0 0 0 1
+attr(,"capture.length")
+
+[1,] 0 0 0 5
+attr(,"capture.names")
+[1] "" "" "" ""
+
+##com.oracle.truffle.r.test.builtins.TestBuiltin_regexec.testregexec1#
+#regexpr("^(?:(?:^\\[([^\\]]+)\\])?(?:'?([^']+)'?!)?([a-zA-Z0-9:\\-$\\[\\]]+)|(.*))$", 'A1', perl=T)
+[1] 1
+attr(,"match.length")
+[1] 2
+attr(,"useBytes")
+[1] TRUE
+attr(,"capture.start")
+
+[1,] 0 0 1 0
+attr(,"capture.length")
+
+[1,] 0 0 2 0
+attr(,"capture.names")
+[1] "" "" "" ""
+
+##com.oracle.truffle.r.test.builtins.TestBuiltin_regexec.testregexec1#
+#regexpr("^(?:(?:^\\[([^\\]]+)\\])?(?:'?([^']+)'?!)?([a-zA-Z0-9:\\-$\\[\\]]+)|(.*))$", 'A1A1', perl=T)
+[1] 1
+attr(,"match.length")
+[1] 4
+attr(,"useBytes")
+[1] TRUE
+attr(,"capture.start")
+
+[1,] 0 0 1 0
+attr(,"capture.length")
+
+[1,] 0 0 4 0
+attr(,"capture.names")
+[1] "" "" "" ""
+
+##com.oracle.truffle.r.test.builtins.TestBuiltin_regexec.testregexec1#
+#regexpr("^(?:(?:^\\[([^\\]]+)\\])?(?<n2>:'?([^']+)'?!)?([a-zA-Z0-9:\\-$\\[\\]]+)|(.*))$", 'A1', perl=T)
+[1] 1
+attr(,"match.length")
+[1] 2
+attr(,"useBytes")
+[1] TRUE
+attr(,"capture.start")
+ n2
+[1,] 0 0 0 1 0
+attr(,"capture.length")
+ n2
+[1,] 0 0 0 2 0
+attr(,"capture.names")
+[1] "" "n2" "" "" ""
+
+##com.oracle.truffle.r.test.builtins.TestBuiltin_regexec.testregexec1#
+#regexpr("^(?<n1>:(?:^\\[([^\\]]+)\\])?(?:'?([^']+)'?!)?([a-zA-Z0-9:\\-$\\[\\]]+)|(.*))$", 'A1', perl=T)
+[1] 1
+attr(,"match.length")
+[1] 2
+attr(,"useBytes")
+[1] TRUE
+attr(,"capture.start")
+ n1
+[1,] 1 0 0 0 1
+attr(,"capture.length")
+ n1
+[1,] 2 0 0 0 2
+attr(,"capture.names")
+[1] "n1" "" "" "" ""
+
+##com.oracle.truffle.r.test.builtins.TestBuiltin_regexec.testregexec1#
+#regexpr("^(?<n1>:(?:^\\[([^\\]]+)\\])?(?<n2>:'?([^']+)'?!)?([a-zA-Z0-9:\\-$\\[\\]]+)|(.*))$", 'A1', perl=T)
+[1] 1
+attr(,"match.length")
+[1] 2
+attr(,"useBytes")
+[1] TRUE
+attr(,"capture.start")
+ n1 n2
+[1,] 1 0 0 0 0 1
+attr(,"capture.length")
+ n1 n2
+[1,] 2 0 0 0 0 2
+attr(,"capture.names")
+[1] "n1" "" "n2" "" "" ""
+
+##com.oracle.truffle.r.test.builtins.TestBuiltin_regexec.testregexec1#
+#regexpr("^(?<n>(.*))$", 'A1', perl=T)
+[1] 1
+attr(,"match.length")
+[1] 2
+attr(,"useBytes")
+[1] TRUE
+attr(,"capture.start")
+ n
+[1,] 1 1
+attr(,"capture.length")
+ n
+[1,] 2 2
+attr(,"capture.names")
+[1] "n" ""
+
+##com.oracle.truffle.r.test.builtins.TestBuiltin_regexec.testregexec1#
+#regexpr("^(?<n>.*)$", 'A1', perl=T)
+[1] 1
+attr(,"match.length")
+[1] 2
+attr(,"useBytes")
+[1] TRUE
+attr(,"capture.start")
+ n
+[1,] 1
+attr(,"capture.length")
+ n
+[1,] 2
+attr(,"capture.names")
+[1] "n"
+
##com.oracle.truffle.r.test.builtins.TestBuiltin_regexpr.testRegExpr#Ignored.ImplementationError#
#regexpr("(a)[^a]\\1", c("andrea apart", "amadeus", NA))
[1] 6 1 NA
diff --git a/com.oracle.truffle.r.test/src/com/oracle/truffle/r/test/builtins/TestBuiltin_regexec.java b/com.oracle.truffle.r.test/src/com/oracle/truffle/r/test/builtins/TestBuiltin_regexec.java
index a65117eb42687fd2a04b1fe23f7aa9c2c2f0f5e5..7867572091284915835d72acea2eaa1150f8b545 100644
--- a/com.oracle.truffle.r.test/src/com/oracle/truffle/r/test/builtins/TestBuiltin_regexec.java
+++ b/com.oracle.truffle.r.test/src/com/oracle/truffle/r/test/builtins/TestBuiltin_regexec.java
@@ -25,5 +25,20 @@ public class TestBuiltin_regexec extends TestBase {
// docs do not mention that so ReferenceError for now.
assertEval(Ignored.ReferenceError,
"argv <- list('^(([^:]+)://)?([^:/]+)(:([0-9]+))?(/.*)', 'http://stat.umn.edu:80/xyz', FALSE, FALSE, FALSE); .Internal(regexec(argv[[1]], argv[[2]], argv[[3]], argv[[4]], argv[[5]]))");
+
+ assertEval("regexpr(\"^(?:(?:^\\\\[([^\\\\]]+)\\\\])?(?:'?([^']+)'?!)?([a-zA-Z0-9:\\\\-$\\\\[\\\\]]+)|(.*))$\", 'A1', perl=T)");
+ assertEval("regexpr(\"^(?:(?:^\\\\[([^\\\\]]+)\\\\])?(?:'?([^']+)'?!)?([a-zA-Z0-9:\\\\-$\\\\[\\\\]]+)|(.*))$\", 'A1A1', perl=T)");
+ assertEval("regexpr(\"^(?:(?:^\\\\[([^\\\\]]+)\\\\])?(?:'?([^']+)'?!)?([a-zA-Z0-9:\\\\-$\\\\[\\\\]]+)|(.*))$\", 'A1 A1', perl=T)");
+ assertEval("regexpr(\"^(?<n1>:(?:^\\\\[([^\\\\]]+)\\\\])?(?:'?([^']+)'?!)?([a-zA-Z0-9:\\\\-$\\\\[\\\\]]+)|(.*))$\", 'A1', perl=T)");
+ assertEval("regexpr(\"^(?<n1>:(?:^\\\\[([^\\\\]]+)\\\\])?(?<n2>:'?([^']+)'?!)?([a-zA-Z0-9:\\\\-$\\\\[\\\\]]+)|(.*))$\", 'A1', perl=T)");
+ assertEval("regexpr(\"^(?:(?:^\\\\[([^\\\\]]+)\\\\])?(?<n2>:'?([^']+)'?!)?([a-zA-Z0-9:\\\\-$\\\\[\\\\]]+)|(.*))$\", 'A1', perl=T)");
+ assertEval("regexpr(\"^((.*))$\", 'A1', perl=T)");
+ assertEval("regexpr(\"^(?<n>(.*))$\", 'A1', perl=T)");
+ assertEval("regexpr(\"^(.*)$\", 'A1', perl=T)");
+ assertEval("regexpr(\"^(?<n>.*)$\", 'A1', perl=T)");
+
+ assertEval("regexpr(\"^(([A-Z)|([a-z]))$\", 'Aa', perl=T)");
+ assertEval("regexpr(\"^(([A-Z)|([a-z]))$\", c('A', 'Aa'), perl=T)");
+ assertEval("regexpr(\"^(([A-Z)|([a-z]))$\", c('Aa', 'A'), perl=T)");
}
}