[iotdb] 01/01: Revert "fix import csv split by comma bug (#3253)"

[ch...@apache.org]

This is an automated email from the ASF dual-hosted git repository.

chaow pushed a commit to branch revert_not_ready_fix_csv_pr
in repository https://gitbox.apache.org/repos/asf/iotdb.git

commit f56ad572887f9fb6c52079166ac2ad69b985811a
Author: chaow <94...@qq.com>
AuthorDate: Tue Jun 8 15:31:49 2021 +0800

    Revert "fix import csv split by comma bug (#3253)"
    
    This reverts commit 5ea3c2e52ade6658621d798a131367bcdf438406.
---
 .../main/java/org/apache/iotdb/tool/ImportCsv.java | 41 +++++++++++++++-------
 .../org/apache/iotdb/tool/CsvLineSplitTest.java    | 11 ++----
 2 files changed, 31 insertions(+), 21 deletions(-)

diff --git a/cli/src/main/java/org/apache/iotdb/tool/ImportCsv.java b/cli/src/main/java/org/apache/iotdb/tool/ImportCsv.java
index a3179d3..6959360 100644
--- a/cli/src/main/java/org/apache/iotdb/tool/ImportCsv.java
+++ b/cli/src/main/java/org/apache/iotdb/tool/ImportCsv.java
@@ -424,21 +424,38 @@ public class ImportCsv extends AbstractCsvTool {
 
   public static String[] splitCsvLine(String path) {
     List<String> nodes = new ArrayList<>();
-    int start = 0;
-    boolean singleQuotes = false;
-    boolean doubleQuotes = false;
-    // split by comma if the comma is followed by an even number of quotes
-    for (int i = 0; i < path.length(); i++) {
-      if (path.charAt(i) == '\"') {
-        doubleQuotes = !doubleQuotes; // toggle state
+    startIndex = 0;
+    for (i = 0; i < path.length(); i++) {
+      if (path.charAt(i) == ',') {
+        nodes.add(path.substring(startIndex, i));
+        startIndex = i + 1;
+      } else if (path.charAt(i) == '"') {
+        nextNode(path, nodes, '"');
       } else if (path.charAt(i) == '\'') {
-        singleQuotes = !singleQuotes;
-      } else if (path.charAt(i) == ',' && (!singleQuotes && !doubleQuotes)) {
-        nodes.add(path.substring(start, i));
-        start = i + 1;
+        nextNode(path, nodes, '\'');
       }
     }
-    nodes.add(path.substring(start));
+    if (path.charAt(path.length() - 1) == ',') {
+      nodes.add("");
+    }
+    if (startIndex <= path.length() - 1) {
+      nodes.add(path.substring(startIndex));
+    }
     return nodes.toArray(new String[0]);
   }
+
+  public static void nextNode(String path, List<String> nodes, char enclose) {
+    int endIndex = path.indexOf(enclose, i + 1);
+    // if a double quotes with escape character
+    while (endIndex != -1 && path.charAt(endIndex - 1) == '\\') {
+      endIndex = path.indexOf(enclose, endIndex + 1);
+    }
+    if (endIndex != -1 && (endIndex == path.length() - 1 || path.charAt(endIndex + 1) == ',')) {
+      nodes.add(path.substring(startIndex + 1, endIndex));
+      i = endIndex + 1;
+      startIndex = endIndex + 2;
+    } else {
+      throw new IllegalArgumentException("Illegal csv line" + path);
+    }
+  }
 }
diff --git a/cli/src/test/java/org/apache/iotdb/tool/CsvLineSplitTest.java b/cli/src/test/java/org/apache/iotdb/tool/CsvLineSplitTest.java
index 2b54434..fd1c9ba 100644
--- a/cli/src/test/java/org/apache/iotdb/tool/CsvLineSplitTest.java
+++ b/cli/src/test/java/org/apache/iotdb/tool/CsvLineSplitTest.java
@@ -26,15 +26,8 @@ public class CsvLineSplitTest {
   @Test
   public void testSplit() {
     Assert.assertArrayEquals(
-        new String[] {"", "a", "b", "c", "\\\""}, ImportCsv.splitCsvLine(",a,b,c,\\\""));
+        new String[] {"", "a", "b", "c", "\\\""}, ImportCsv.splitCsvLine(",a,b,c,\"\\\"\""));
     Assert.assertArrayEquals(
-        new String[] {"", "a", "b", "\\'"}, ImportCsv.splitCsvLine(",a,b,\\'"));
-    Assert.assertArrayEquals(
-        new String[] {"", "a\",\"a", "\"a,,\"", "'"}, ImportCsv.splitCsvLine(",a\",\"a,\"a,,\",'"));
-    Assert.assertArrayEquals(
-        new String[] {"True", "a=\",\"a''"}, ImportCsv.splitCsvLine("True,a=\",\"a''"));
-    Assert.assertArrayEquals(
-        new String[] {"True", "\"a=,,,a=z//z'a\""},
-        ImportCsv.splitCsvLine("True,\"a=,,,a=z//z'a\""));
+        new String[] {"", "a", "b", "\\'"}, ImportCsv.splitCsvLine(",a,b,\"\\'\""));
   }
 }