From e284bf83d07ed5063d1c330cf09ed93514b933d9 Mon Sep 17 00:00:00 2001
From: hcz <h.cz@qq.com>
Date: Wed, 1 Jul 2020 11:21:53 +0800
Subject: [PATCH] Optimize and fix

---
 .../Impl/TabSeparatedRawRowInputFormat.h      | 21 +++++--------------
 .../01324_insert_tsv_raw.reference            |  3 ++-
 .../0_stateless/01324_insert_tsv_raw.sql      |  7 ++++---
 3 files changed, 11 insertions(+), 20 deletions(-)

diff --git a/src/Processors/Formats/Impl/TabSeparatedRawRowInputFormat.h b/src/Processors/Formats/Impl/TabSeparatedRawRowInputFormat.h
index c5982714e34..fe113c60a0d 100644
--- a/src/Processors/Formats/Impl/TabSeparatedRawRowInputFormat.h
+++ b/src/Processors/Formats/Impl/TabSeparatedRawRowInputFormat.h
@@ -10,7 +10,7 @@ namespace DB
 {
 
 /** A stream to input data in tsv format, but without escaping individual values.
-  * It only supports one string column
+  * It only supports columns without '\n' or '\t'
   */
 class TabSeparatedRawRowInputFormat : public TabSeparatedRowInputFormat
 {
@@ -33,23 +33,12 @@ public:
 
     bool readField(IColumn & column, const DataTypePtr & type, bool) override
     {
-        // TODO: possible to optimize
-        std::string buf;
+        char * pos = find_first_symbols<'\n', '\t'>(in.position(), in.buffer().end());
+        ReadBufferFromMemory cell(in.position(), pos - in.position());
 
-        while (!in.eof())
-        {
-            char c = *in.position();
+        type->deserializeAsWholeText(column, cell, format_settings);
 
-            if (c == '\n' || c == '\t')
-                break;
-
-            in.ignore();
-            buf.push_back(c);
-        }
-
-        ReadBufferFromString line_in(buf);
-
-        type->deserializeAsWholeText(column, line_in, format_settings);
+        in.position() = pos;
 
         return true;
     }
diff --git a/tests/queries/0_stateless/01324_insert_tsv_raw.reference b/tests/queries/0_stateless/01324_insert_tsv_raw.reference
index f0f74866cd2..ce809415693 100644
--- a/tests/queries/0_stateless/01324_insert_tsv_raw.reference
+++ b/tests/queries/0_stateless/01324_insert_tsv_raw.reference
@@ -1 +1,2 @@
-"a 	1
+"a 	1	\	\\	"\""	"\\""
+["\"a ", "1", "\\", "\\\\", "\"\\\"\"", "\"\\\\\"\""]
diff --git a/tests/queries/0_stateless/01324_insert_tsv_raw.sql b/tests/queries/0_stateless/01324_insert_tsv_raw.sql
index 4bcc3a2268b..c3812730e5b 100644
--- a/tests/queries/0_stateless/01324_insert_tsv_raw.sql
+++ b/tests/queries/0_stateless/01324_insert_tsv_raw.sql
@@ -1,7 +1,8 @@
 drop table if exists tsv_raw;
-create table tsv_raw (a String, b Int64) engine = Memory;
-insert into tsv_raw format TSVRaw "a 	1
+create table tsv_raw (strval String, intval Int64, b1 String, b2 String, b3 String, b4 String) engine = Memory;
+insert into tsv_raw format TSVRaw "a 	1	\	\\	"\""	"\\""
 ;
 
-select * from tsv_raw;
+select * from tsv_raw format TSVRaw;
+select * from tsv_raw format JSONCompactEachRow;
 drop table tsv_raw;