[common] Make quoting mean literal in array and row string cast rules

paimon-common/src/main/java/org/apache/paimon/casting/StringToArrayCastRule.java

@@ -26,11 +26,9 @@
 import org.apache.paimon.types.DataTypeFamily;
 import org.apache.paimon.types.DataTypeRoot;
 import org.apache.paimon.types.VarCharType;
-import org.apache.paimon.utils.StringUtils;
 
 import java.util.ArrayList;
 import java.util.List;
-import java.util.Stack;
 import java.util.regex.Matcher;
 import java.util.regex.Pattern;
 
@@ -114,52 +112,15 @@ private String extractArrayContent(String str) {
     private List<Object> parseArrayElements(
             String content, CastExecutor<BinaryString, Object> elementCastExecutor) {
         List<Object> elements = new ArrayList<>();
-        for (String token : splitArrayElements(content)) {
-            String trimmedToken = token.trim();
+        for (TokenSplitter.Token token : TokenSplitter.split(content)) {
+            String value = token.value();
+            // only an unquoted null is the null element; "null" is the four-character string
             Object element =
-                    "null".equals(trimmedToken)
+                    !token.literal() && "null".equals(value)
                             ? null
-                            : elementCastExecutor.cast(BinaryString.fromString(trimmedToken));
+                            : elementCastExecutor.cast(BinaryString.fromString(value));
             elements.add(element);
         }
         return elements;
     }
-
-    private List<String> splitArrayElements(String content) {
-        List<String> elements = new ArrayList<>();
-        StringBuilder current = new StringBuilder();
-        Stack<Character> bracketStack = new Stack<>();
-        boolean inQuotes = false;
-        boolean escaped = false;
-
-        for (char c : content.toCharArray()) {
-            if (escaped) {
-                escaped = false;
-            } else if (c == '\\') {
-                escaped = true;
-            } else if (c == '"') {
-                inQuotes = !inQuotes;
-            } else if (!inQuotes) {
-                if (StringUtils.isOpenBracket(c)) {
-                    bracketStack.push(c);
-                } else if (StringUtils.isCloseBracket(c) && !bracketStack.isEmpty()) {
-                    bracketStack.pop();
-                } else if (c == ',' && bracketStack.isEmpty()) {
-                    addCurrentElement(elements, current);
-                    continue;
-                }
-            }
-            current.append(c);
-        }
-
-        addCurrentElement(elements, current);
-        return elements;
-    }
-
-    private void addCurrentElement(List<String> elements, StringBuilder current) {
-        if (current.length() > 0) {
-            elements.add(current.toString());
-            current.setLength(0);
-        }
-    }
 }

paimon-common/src/main/java/org/apache/paimon/casting/StringToRowCastRule.java

@@ -26,11 +26,8 @@
 import org.apache.paimon.types.DataTypeRoot;
 import org.apache.paimon.types.RowType;
 import org.apache.paimon.types.VarCharType;
-import org.apache.paimon.utils.StringUtils;
 
-import java.util.ArrayList;
 import java.util.List;
-import java.util.Stack;
 import java.util.regex.Matcher;
 import java.util.regex.Pattern;
 
@@ -93,7 +90,7 @@ private InternalRow parseRow(
             if (content.isEmpty()) {
                 return createNullRow(fieldCount);
             }
-            List<String> fieldValues = splitRowFields(content);
+            List<TokenSplitter.Token> fieldValues = TokenSplitter.split(content);
             if (fieldValues.size() != fieldCount) {
                 throw new RuntimeException(
                         "Row field count mismatch. Expected: "
@@ -137,60 +134,22 @@ private GenericRow createNullRow(int fieldCount) {
     }
 
     private GenericRow createRowFromFields(
-            List<String> fieldValues,
+            List<TokenSplitter.Token> fieldValues,
             CastExecutor<BinaryString, Object>[] fieldCastExecutors,
             int fieldCount) {
         GenericRow row = new GenericRow(fieldCount);
         for (int i = 0; i < fieldCount; i++) {
-            String fieldValue = fieldValues.get(i).trim();
-            Object value = parseFieldValue(fieldValue, fieldCastExecutors[i]);
-            row.setField(i, value);
+            row.setField(i, parseFieldValue(fieldValues.get(i), fieldCastExecutors[i]));
         }
         return row;
     }
 
     private Object parseFieldValue(
-            String fieldValue, CastExecutor<BinaryString, Object> castExecutor) {
-        return "null".equals(fieldValue)
+            TokenSplitter.Token token, CastExecutor<BinaryString, Object> castExecutor) {
+        String value = token.value();
+        // only an unquoted null is the null field; "null" is the four-character string
+        return !token.literal() && "null".equals(value)
                 ? null
-                : castExecutor.cast(BinaryString.fromString(fieldValue));
-    }
-
-    private List<String> splitRowFields(String content) {
-        List<String> fields = new ArrayList<>();
-        StringBuilder current = new StringBuilder();
-        Stack<Character> bracketStack = new Stack<>();
-        boolean inQuotes = false;
-        boolean escaped = false;
-
-        for (char c : content.toCharArray()) {
-            if (escaped) {
-                escaped = false;
-            } else if (c == '\\') {
-                escaped = true;
-            } else if (c == '"') {
-                inQuotes = !inQuotes;
-            } else if (!inQuotes) {
-                if (StringUtils.isOpenBracket(c)) {
-                    bracketStack.push(c);
-                } else if (StringUtils.isCloseBracket(c) && !bracketStack.isEmpty()) {
-                    bracketStack.pop();
-                } else if (c == ',' && bracketStack.isEmpty()) {
-                    addCurrentField(fields, current);
-                    continue;
-                }
-            }
-            current.append(c);
-        }
-
-        addCurrentField(fields, current);
-        return fields;
-    }
-
-    private void addCurrentField(List<String> fields, StringBuilder current) {
-        if (current.length() > 0) {
-            fields.add(current.toString());
-            current.setLength(0);
-        }
+                : castExecutor.cast(BinaryString.fromString(value));
     }
 }

paimon-common/src/main/java/org/apache/paimon/casting/TokenSplitter.java

@@ -0,0 +1,141 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements.  See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership.  The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License.  You may obtain a copy of the License at
+ *
+ *     http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package org.apache.paimon.casting;
+
+import org.apache.paimon.utils.StringUtils;
+
+import java.util.ArrayList;
+import java.util.List;
+import java.util.Stack;
+
+/**
+ * Splits the comma-separated body of an array or row literal into its tokens, honouring quotes,
+ * escapes and nesting.
+ *
+ * <p>A separator only separates outside quotes and at bracket depth zero, so {@code "a,b"} and
+ * {@code [a, b]} each stay one token.
+ *
+ * <p>Quotes and backslashes are this level's syntax, so they are removed at depth zero and kept
+ * verbatim inside a nested literal, where they are the inner level's syntax and the rule for that
+ * element parses them again. Whether they appeared at depth zero is remembered: quoting or escaping
+ * is how the literal text {@code null} and the empty string are written, which are otherwise
+ * unrepresentable.
+ *
+ * <p>Whitespace around a token is dropped; whitespace inside quotes is kept.
+ */
+class TokenSplitter {
+
+    /** One token of a literal body, plus whether its value was written as a literal. */
+    static class Token {
+
+        private final String value;
+        private final boolean literal;
+
+        Token(String value, boolean literal) {
+            this.value = value;
+            this.literal = literal;
+        }
+
+        String value() {
+            return value;
+        }
+
+        /** Whether quotes or an escape made this a literal string rather than a bare word. */
+        boolean literal() {
+            return literal;
+        }
+    }
+
+    private TokenSplitter() {}
+
+    static List<Token> split(String content) {
+        List<Token> tokens = new ArrayList<>();
+        StringBuilder current = new StringBuilder();
+        Stack<Character> bracketStack = new Stack<>();
+        boolean inQuotes = false;
+        boolean escaped = false;
+        boolean literal = false;
+        // length of current up to the last character that was not unquoted whitespace
+        int end = 0;
+
+        for (char c : content.toCharArray()) {
+            boolean nested = !bracketStack.isEmpty();
+            if (escaped) {
+                // the escapee stands for itself and is never read as syntax
+                escaped = false;
+                current.append(c);
+                end = current.length();
+                continue;
+            }
+            if (c == '\\') {
+                escaped = true;
+                if (nested) {
+                    // the inner rule has to see the escape to protect its own separators
+                    current.append(c);
+                    end = current.length();
+                } else {
+                    literal = true;
+                }
+                continue;
+            }
+            if (c == '"') {
+                inQuotes = !inQuotes;
+                if (nested) {
+                    current.append(c);
+                    end = current.length();
+                } else {
+                    literal = true;
+                }
+                continue;
+            }
+            if (!inQuotes) {
+                if (StringUtils.isOpenBracket(c)) {
+                    bracketStack.push(c);
+                } else if (StringUtils.isCloseBracket(c) && !bracketStack.isEmpty()) {
+                    bracketStack.pop();
+                } else if (c == ',' && bracketStack.isEmpty()) {
+                    addToken(tokens, current, end, literal);
+                    current.setLength(0);
+                    end = 0;
+                    literal = false;
+                    continue;
+                } else if (Character.isWhitespace(c) && end == 0) {
+                    // leading whitespace outside quotes is not part of the token
+                    continue;
+                }
+            }
+            current.append(c);
+            if (inQuotes || !Character.isWhitespace(c)) {
+                end = current.length();
+            }
+        }
+
+        addToken(tokens, current, end, literal);
+        return tokens;
+    }
+
+    private static void addToken(
+            List<Token> tokens, StringBuilder current, int end, boolean literal) {
+        // whitespace is not part of a token, so one made only of whitespace was never written;
+        // quoting is how an empty value is written
+        if (end > 0 || literal) {
+            tokens.add(new Token(current.substring(0, end), literal));
+        }
+    }
+}