From 4346955ef2c6963e2f9c30714102fd876c284351 Mon Sep 17 00:00:00 2001
From: Tomas Stupka <tomas.stupka@oracle.com>
Date: Thu, 25 May 2017 17:42:05 +0200
Subject: [PATCH] fixed quotes handling in scan
---
.../truffle/r/nodes/builtin/base/Scan.java | 11 +++++-----
.../r/test/builtins/TestBuiltin_scan.java | 22 +++++++++++++++++++
2 files changed, 28 insertions(+), 5 deletions(-)
diff --git a/com.oracle.truffle.r.nodes.builtin/src/com/oracle/truffle/r/nodes/builtin/base/Scan.java b/com.oracle.truffle.r.nodes.builtin/src/com/oracle/truffle/r/nodes/builtin/base/Scan.java
index db79d22096..5640789929 100644
--- a/com.oracle.truffle.r.nodes.builtin/src/com/oracle/truffle/r/nodes/builtin/base/Scan.java
+++ b/com.oracle.truffle.r.nodes.builtin/src/com/oracle/truffle/r/nodes/builtin/base/Scan.java
@@ -202,10 +202,10 @@ public abstract class Scan extends RBuiltinNode.Arg19 {
}
}
- private static int getFirstQuoteInd(String str, char sepChar) {
- int quoteInd = str.indexOf(sepChar);
+ private static int getFirstQuoteInd(String str, char quotechar, Character sepChar) {
+ int quoteInd = str.indexOf(quotechar);
if (quoteInd >= 0) {
- if (quoteInd == 0 || str.charAt(quoteInd - 1) == ' ' || str.charAt(quoteInd - 1) == '\t') {
+ if (quoteInd == 0 || (str.charAt(quoteInd - 1) == ' ' || str.charAt(quoteInd - 1) == '\t') || sepChar != null && str.charAt(quoteInd - 1) == sepChar) {
// it's a quote character if it starts the string or is preceded by a blank space
return quoteInd;
}
@@ -236,10 +236,11 @@ public abstract class Scan extends RBuiltinNode.Arg19 {
sepInd = str.indexOf(data.sepchar.charAt(0));
}
- int quoteInd = getFirstQuoteInd(str, data.quoteset.charAt(0));
+ Character sepChar = data.sepchar != null ? data.sepchar.charAt(0) : null;
+ int quoteInd = getFirstQuoteInd(str, data.quoteset.charAt(0), sepChar);
char quoteChar = data.quoteset.charAt(0);
for (int i = 1; i < data.quoteset.length(); i++) {
- int ind = getFirstQuoteInd(str, data.quoteset.charAt(i));
+ int ind = getFirstQuoteInd(str, data.quoteset.charAt(i), sepChar);
if (ind >= 0 && (quoteInd == -1 || (quoteInd >= 0 && ind < quoteInd))) {
// update quoteInd if either the new index is smaller or the previous one (for
// another separator) was not found
diff --git a/com.oracle.truffle.r.test/src/com/oracle/truffle/r/test/builtins/TestBuiltin_scan.java b/com.oracle.truffle.r.test/src/com/oracle/truffle/r/test/builtins/TestBuiltin_scan.java
index fd9a88eff6..3c3508a978 100644
--- a/com.oracle.truffle.r.test/src/com/oracle/truffle/r/test/builtins/TestBuiltin_scan.java
+++ b/com.oracle.truffle.r.test/src/com/oracle/truffle/r/test/builtins/TestBuiltin_scan.java
@@ -13,11 +13,23 @@ package com.oracle.truffle.r.test.builtins;
import org.junit.Test;
import com.oracle.truffle.r.test.TestBase;
+import java.io.File;
+import org.junit.After;
// Checkstyle: stop line length check
public class TestBuiltin_scan extends TestBase {
+ private static final String TEST_CVS_FILE = "__TestBuiltin_scan_testReadCsvTestFile.cvs";
+
+ @After
+ public void cleanup() {
+ File f = new File(TEST_CVS_FILE);
+ if (f.exists()) {
+ f.delete();
+ }
+ }
+
@Test
public void testScan() {
// from scan's documentation
@@ -45,12 +57,22 @@ public class TestBuiltin_scan extends TestBase {
assertEval("{ con<-textConnection(c(\"foo faz\", \"bar \\\"baz\\\"\")); scan(con, what=list(\"\", \"\")) }");
assertEval("{ con<-textConnection(c(\"foo, faz\", \"bar, baz\")); scan(con, what=list(\"\", \"\"), sep=\",\") }");
+ assertEval("con<-textConnection(c(\"foo,\\\"bar,bar\\\"\")); scan(con, what=list(\"\"), sep=',')");
+ assertEval("con<-textConnection(c(\"foo,'bar,bar'\")); scan(con, what=list(\"\"), sep=',')");
+
assertEval("{ con<-textConnection(c(\"bar'foo'\")); scan(con, what=list(\"\")) }");
assertEval("{ con<-textConnection(c(\"'foo'\")); scan(con, what=list(\"\")) }");
assertEval("{ con<-textConnection(c(\"bar 'foo'\")); scan(con, what=list(\"\")) }");
// sep should not be treated as a regex:
assertEval("con <- textConnection(\"A|B|C\\n1|2|3\\n4|5|6\"); read.csv(con, sep=\"|\")");
+
+ }
+
+ @Test
+ public void testReadCsv() {
+ String testData = "n1,n2\nv1,\"v5, v5\"\n";
+ assertEval("fileConn<-file('" + TEST_CVS_FILE + "'); writeLines(c('" + testData + "'), fileConn); m <- read.csv('" + TEST_CVS_FILE + "'); m");
}
@Test
--
GitLab