From 41339a76811af0c7d8b90be241e1b287b5313317 Mon Sep 17 00:00:00 2001
From: Lukas Stadler <lukas.stadler@oracle.com>
Date: Thu, 23 Aug 2018 16:27:57 +0200
Subject: [PATCH] treat CR and LF in readLine similar to GNU R
---
.../r/runtime/conn/ConnectionSupport.java | 6 +-
.../r/runtime/conn/TextConnections.java | 55 +++++++++++++++++-
.../truffle/r/test/ExpectedTestOutput.test | 56 +++++++++++++++++++
.../r/test/library/base/TestConnections.java | 4 ++
4 files changed, 116 insertions(+), 5 deletions(-)
diff --git a/com.oracle.truffle.r.runtime/src/com/oracle/truffle/r/runtime/conn/ConnectionSupport.java b/com.oracle.truffle.r.runtime/src/com/oracle/truffle/r/runtime/conn/ConnectionSupport.java
index 27aa8990f6..17d31db470 100644
--- a/com.oracle.truffle.r.runtime/src/com/oracle/truffle/r/runtime/conn/ConnectionSupport.java
+++ b/com.oracle.truffle.r.runtime/src/com/oracle/truffle/r/runtime/conn/ConnectionSupport.java
@@ -902,7 +902,7 @@ public class ConnectionSupport {
if (s == null) {
return null;
} else {
- String[] lines = s.split("\n", 2);
+ String[] lines = TextConnections.splitLines(s, 2);
if (lines.length == 2) {
// we hit end of the line
if (lines[1].length() != 0) {
@@ -923,7 +923,7 @@ public class ConnectionSupport {
break;
}
- lines = s.split("\n", 2);
+ lines = TextConnections.splitLines(s, 2);
if (lines.length == 2) {
// we hit end of the line
if (lines[1].length() != 0) {
@@ -947,7 +947,7 @@ public class ConnectionSupport {
*/
@TruffleBoundary
private String[] readLinesWithPushBack(int n, EnumSet<ReadLineWarning> warn, boolean skipNul) throws IOException {
- // NOTE: 'n' may be negative indicating to read as much lines as available
+ // NOTE: 'n' may be negative indicating to read as many lines as available
final List<String> res;
if (n >= 0) {
res = new ArrayList<>(n);
diff --git a/com.oracle.truffle.r.runtime/src/com/oracle/truffle/r/runtime/conn/TextConnections.java b/com.oracle.truffle.r.runtime/src/com/oracle/truffle/r/runtime/conn/TextConnections.java
index 7695a0d5be..418df719ce 100644
--- a/com.oracle.truffle.r.runtime/src/com/oracle/truffle/r/runtime/conn/TextConnections.java
+++ b/com.oracle.truffle.r.runtime/src/com/oracle/truffle/r/runtime/conn/TextConnections.java
@@ -41,6 +41,55 @@ import com.oracle.truffle.r.runtime.env.REnvironment;
import com.oracle.truffle.r.runtime.env.REnvironment.PutException;
public class TextConnections {
+
+ public static String[] splitLines(String value, int limit) {
+ assert limit != 0;
+ ArrayList<String> strings = null;
+
+ int lastPos = 0;
+ int pos = 0;
+ while (pos < value.length()) {
+ char c = value.charAt(pos);
+ if (c == '\n' || c == '\r') {
+ if (strings == null) {
+ strings = new ArrayList<>();
+ }
+ if (limit != -1 && strings.size() == limit - 1) {
+ strings.add(value.substring(lastPos));
+ return strings.toArray(new String[strings.size()]);
+ }
+ strings.add(value.substring(lastPos, pos));
+ // skip "\r\n" combination
+ if (c == '\r') {
+ if ((pos + 1) < value.length()) {
+ c = value.charAt(pos + 1);
+ if (c == '\r') {
+ // bug in GNU R: a second "\r" is immediately treated as a EOL
+ if (limit != -1 && strings.size() == limit - 1) {
+ strings.add(value.substring(pos + 2));
+ return strings.toArray(new String[strings.size()]);
+ }
+ strings.add("");
+ } else if (c == '\n') {
+ pos++;
+ }
+ } else {
+ // bug in GNU R: a final "\r" will be ignored
+ return strings.toArray(new String[strings.size()]);
+ }
+ }
+ lastPos = pos + 1;
+ }
+ pos++;
+ }
+ if (strings == null) {
+ return new String[]{value};
+ } else {
+ strings.add(value.substring(lastPos));
+ return strings.toArray(new String[strings.size()]);
+ }
+ }
+
public static class TextRConnection extends BaseRConnection {
protected String description;
private final RAbstractStringVector object;
@@ -97,11 +146,13 @@ public class TextConnections {
assert object != null;
StringBuilder sb = new StringBuilder();
for (int i = 0; i < object.getLength(); i++) {
+ if (i > 0) {
+ sb.append('\n');
+ }
sb.append(object.getDataAt(i));
// vector elements are implicitly terminated with a newline
- sb.append('\n');
}
- lines = sb.toString().split("\\n");
+ lines = splitLines(sb.toString(), -1);
}
@Override
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 bdb20bc4c5..32b76d9697 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
@@ -86035,6 +86035,62 @@ In writeChar(x, zz, nc, eos = NULL) :
#{ s <- "äöüß"; rc <- rawConnection(raw(0), "w"); writeChar(s, rc); rawConnectionValue(rc) }
[1] c3 a4 c3 b6 c3 bc c3 9f 00
+##com.oracle.truffle.r.test.library.base.TestConnections.testReadLines#
+#readLines(textConnection("foo"))
+[1] "foo"
+
+##com.oracle.truffle.r.test.library.base.TestConnections.testReadLines#
+#readLines(textConnection("foo\n"))
+[1] "foo" ""
+
+##com.oracle.truffle.r.test.library.base.TestConnections.testReadLines#
+#readLines(textConnection("foo\n\n\r"))
+[1] "foo" "" ""
+
+##com.oracle.truffle.r.test.library.base.TestConnections.testReadLines#
+#readLines(textConnection("foo\n\n\rfoo"))
+[1] "foo" "" "" "foo"
+
+##com.oracle.truffle.r.test.library.base.TestConnections.testReadLines#
+#readLines(textConnection("foo\n\r"))
+[1] "foo" ""
+
+##com.oracle.truffle.r.test.library.base.TestConnections.testReadLines#
+#readLines(textConnection("foo\n\rfoo"))
+[1] "foo" "" "foo"
+
+##com.oracle.truffle.r.test.library.base.TestConnections.testReadLines#
+#readLines(textConnection("foo\nfoo"))
+[1] "foo" "foo"
+
+##com.oracle.truffle.r.test.library.base.TestConnections.testReadLines#
+#readLines(textConnection("foo\r"))
+[1] "foo"
+
+##com.oracle.truffle.r.test.library.base.TestConnections.testReadLines#
+#readLines(textConnection("foo\r\n"))
+[1] "foo" ""
+
+##com.oracle.truffle.r.test.library.base.TestConnections.testReadLines#
+#readLines(textConnection("foo\r\nfoo"))
+[1] "foo" "foo"
+
+##com.oracle.truffle.r.test.library.base.TestConnections.testReadLines#
+#readLines(textConnection("foo\r\r\n"))
+[1] "foo" "" "" ""
+
+##com.oracle.truffle.r.test.library.base.TestConnections.testReadLines#
+#readLines(textConnection("foo\r\r\nfoo"))
+[1] "foo" "" "" "foo"
+
+##com.oracle.truffle.r.test.library.base.TestConnections.testReadLines#
+#readLines(textConnection("foo\rfoo"))
+[1] "foo" "foo"
+
+##com.oracle.truffle.r.test.library.base.TestConnections.testReadLines#
+#readLines(textConnection("foofoo"))
+[1] "foofoo"
+
##com.oracle.truffle.r.test.library.base.TestConnections.testReadLines#Output.MayIgnoreWarningContext#
#{ zz <- file('',"w+b", blocking=F); writeBin(as.raw(c(97,98,99,100,0,101)), zz, useBytes=T); seek(zz, 0); res <- readLines(zz, 2, warn=F, skipNul=F); close(zz); res }
[1] "abcd"
diff --git a/com.oracle.truffle.r.test/src/com/oracle/truffle/r/test/library/base/TestConnections.java b/com.oracle.truffle.r.test/src/com/oracle/truffle/r/test/library/base/TestConnections.java
index 55b38976ab..6601a88357 100644
--- a/com.oracle.truffle.r.test/src/com/oracle/truffle/r/test/library/base/TestConnections.java
+++ b/com.oracle.truffle.r.test/src/com/oracle/truffle/r/test/library/base/TestConnections.java
@@ -208,6 +208,10 @@ public class TestConnections extends TestRBase {
assertEval(Output.MayIgnoreWarningContext, TestBase.template(
"{ zz <- file('',\"w+b\", blocking=%0); writeBin(as.raw(%1), zz, useBytes=T); seek(zz, 0); res <- readLines(zz, 2, warn=%2, skipNul=%3); close(zz); res }",
LVAL, arr(lineWithNul, twoLinesOneNul, lineWithNulIncomp, twoLinesOneNulIncomp), LVAL, LVAL));
+
+ String[] endings = new String[]{"", "\\n", "\\r", "\\n\\r", "\\r\\n", "\\n\\n\\r", "\\r\\r\\n"};
+ String[] text = new String[]{"", "foo"};
+ assertEval(template("readLines(textConnection(\"foo%0%1\"))", endings, text));
}
@Test
--
GitLab