From 3aab5a29214f3dcaeecd1ae01b7aab373f3489a5 Mon Sep 17 00:00:00 2001
From: wyb <wybb86@gmail.com>
Date: Thu, 16 Jan 2025 23:45:16 +0800
Subject: [PATCH] [BugFix] Fix ip string as double type in csv files()

Signed-off-by: wyb <wybb86@gmail.com>
---
 be/src/util/string_parser.hpp         |  8 ++++++++
 be/test/util/string_parser_test.cpp   |  2 ++
 test/sql/test_files/R/test_csv_ip     | 28 +++++++++++++++++++++++++++
 test/sql/test_files/T/test_csv_ip     | 12 ++++++++++++
 test/sql/test_files/csv_format/ip.csv |  1 +
 5 files changed, 51 insertions(+)
 create mode 100644 test/sql/test_files/R/test_csv_ip
 create mode 100644 test/sql/test_files/T/test_csv_ip
 create mode 100644 test/sql/test_files/csv_format/ip.csv

diff --git a/be/src/util/string_parser.hpp b/be/src/util/string_parser.hpp
index 951b690fa6eed..5a48edea7d12c 100644
--- a/be/src/util/string_parser.hpp
+++ b/be/src/util/string_parser.hpp
@@ -522,6 +522,14 @@ inline T StringParser::string_to_float_internal(const char* s, int len, ParseRes
     auto res = fast_float::from_chars(s + i, s + j + 1, val);
 
     if (LIKELY(res.ec == std::errc())) {
+        // 'res.ptr' is set to point right after the parsed number.
+        // if there are some chars left, treate it failure.
+        // for example,
+        // '10.11.12.13' is parsed as 10.11, res.ptr is '.12.13', and it is invalid.
+        if (res.ptr != s + j + 1) {
+            *result = PARSE_FAILURE;
+            return 0;
+        }
         if (UNLIKELY(val == std::numeric_limits<T>::infinity())) {
             *result = PARSE_OVERFLOW;
         } else {
diff --git a/be/test/util/string_parser_test.cpp b/be/test/util/string_parser_test.cpp
index 77d0550edb5a1..e7c52dbe1dc1f 100644
--- a/be/test/util/string_parser_test.cpp
+++ b/be/test/util/string_parser_test.cpp
@@ -540,6 +540,8 @@ TEST(StringToFloat, Basic) {
     test_all_float_variants("in finity", StringParser::PARSE_FAILURE);
     test_all_float_variants("na", StringParser::PARSE_FAILURE);
     test_all_float_variants("ThisIsANaN", StringParser::PARSE_FAILURE);
+    test_all_float_variants("10.1.2.3", StringParser::PARSE_FAILURE);
+    test_all_float_variants("10.1 max", StringParser::PARSE_FAILURE);
 }
 
 TEST(StringToFloat, InvalidLeadingTrailing) {
diff --git a/test/sql/test_files/R/test_csv_ip b/test/sql/test_files/R/test_csv_ip
new file mode 100644
index 0000000000000..8d0b9ac0c73e3
--- /dev/null
+++ b/test/sql/test_files/R/test_csv_ip
@@ -0,0 +1,28 @@
+-- name: test_csv_ip
+
+create database db_${uuid0};
+use db_${uuid0};
+
+shell: ossutil64 mkdir oss://${oss_bucket}/test_files/csv_format/${uuid0} >/dev/null || echo "exit 0" >/dev/null
+
+shell: ossutil64 cp --force ./sql/test_files/csv_format/ip.csv oss://${oss_bucket}/test_files/csv_format/${uuid0}/ | grep -Pv "(average|elapsed)"
+-- result:
+0
+
+Succeed: Total num: 1, size: 14. OK num: 1(upload 1 files).
+-- !result
+
+
+select * from files("path" = "oss://${oss_bucket}/test_files/csv_format/${uuid0}/*", "format" = "csv", "csv.column_separator" = "|", "csv.row_delimiter" = "\n");
+-- result:
+1	10.11.12.13
+-- !result
+
+desc files("path" = "oss://${oss_bucket}/test_files/csv_format/${uuid0}/*", "format" = "csv", "csv.column_separator" = "|", "csv.row_delimiter" = "\n");
+-- result:
+$1	bigint	YES
+$2	varchar(1048576)	YES
+-- !result
+
+
+shell: ossutil64 rm -rf oss://${oss_bucket}/test_files/csv_format/${uuid0}/ > /dev/null
diff --git a/test/sql/test_files/T/test_csv_ip b/test/sql/test_files/T/test_csv_ip
new file mode 100644
index 0000000000000..b62ffaa4e2d94
--- /dev/null
+++ b/test/sql/test_files/T/test_csv_ip
@@ -0,0 +1,12 @@
+-- name: test_csv_ip
+
+create database db_${uuid0};
+use db_${uuid0};
+
+shell: ossutil64 mkdir oss://${oss_bucket}/test_files/csv_format/${uuid0} >/dev/null || echo "exit 0" >/dev/null
+shell: ossutil64 cp --force ./sql/test_files/csv_format/ip.csv oss://${oss_bucket}/test_files/csv_format/${uuid0}/ | grep -Pv "(average|elapsed)"
+
+select * from files("path" = "oss://${oss_bucket}/test_files/csv_format/${uuid0}/*", "format" = "csv", "csv.column_separator" = "|", "csv.row_delimiter" = "\n");
+desc files("path" = "oss://${oss_bucket}/test_files/csv_format/${uuid0}/*", "format" = "csv", "csv.column_separator" = "|", "csv.row_delimiter" = "\n");
+
+shell: ossutil64 rm -rf oss://${oss_bucket}/test_files/csv_format/${uuid0}/ > /dev/null
diff --git a/test/sql/test_files/csv_format/ip.csv b/test/sql/test_files/csv_format/ip.csv
new file mode 100644
index 0000000000000..419eb79e83ba9
--- /dev/null
+++ b/test/sql/test_files/csv_format/ip.csv
@@ -0,0 +1 @@
+1|10.11.12.13