From b3e47584bb075fb4d8e2091bb7cc2d5ed0b33293 Mon Sep 17 00:00:00 2001
From: jakevin <30525741+jackwener@users.noreply.github.com>
Date: Thu, 27 Jan 2022 19:09:32 +0800
Subject: [PATCH] Enhance date parser (#3797)

* parse date

* enhance datetime

* add test

* fix review

* rename

Co-authored-by: Sophie <84560950+Sophie-Xie@users.noreply.github.com>
---
 src/common/time/parser/DatetimeReader.h       |  7 ++
 src/common/time/parser/datetime_parser.yy     | 56 ++++++-----
 src/common/time/parser/datetime_scanner.lex   |  4 +
 .../time/parser/test/DateTimeParserTest.cpp   | 95 +++++++++++++------
 4 files changed, 110 insertions(+), 52 deletions(-)

diff --git a/src/common/time/parser/DatetimeReader.h b/src/common/time/parser/DatetimeReader.h
index dfa8340f1f8..fda6f824301 100644
--- a/src/common/time/parser/DatetimeReader.h
+++ b/src/common/time/parser/DatetimeReader.h
@@ -33,16 +33,19 @@ class DatetimeReader {
   }
 
   StatusOr<DateTime> readDatetime(std::string input) {
+    input = kDatetimePrefix + input;
     return read(std::move(input));
   }
 
   StatusOr<Date> readDate(std::string input) {
+    input = kDatePrefix + input;
     auto result = read(std::move(input));
     NG_RETURN_IF_ERROR(result);
     return result.value().date();
   }
 
   StatusOr<Time> readTime(std::string input) {
+    input = kTimePrefix + input;
     auto result = read(std::move(input));
     NG_RETURN_IF_ERROR(result);
     return result.value().time();
@@ -60,6 +63,10 @@ class DatetimeReader {
   DatetimeParser parser_;
   std::string error_;
   DateTime *dt_{nullptr};
+
+  inline static const std::string kDatetimePrefix = "DATETIME__";
+  inline static const std::string kDatePrefix = "DATE__";
+  inline static const std::string kTimePrefix = "TIME__";
 };
 
 }  // namespace time
diff --git a/src/common/time/parser/datetime_parser.yy b/src/common/time/parser/datetime_parser.yy
index f6d04d957e9..db054fec422 100644
--- a/src/common/time/parser/datetime_parser.yy
+++ b/src/common/time/parser/datetime_parser.yy
@@ -59,7 +59,7 @@ enum class Type {
 %token KW_TIME_ID
 
 /* symbols */
-%token TIME_DELIMITER SPACE POSITIVE NEGATIVE
+%token TIME_DELIMITER SPACE POSITIVE NEGATIVE KW_DATETIME KW_DATE KW_TIME
 
 /* token type specification */
 %token <intVal> INTEGER
@@ -78,34 +78,26 @@ enum class Type {
 %%
 
 datetime
-  : date date_time_delimiter time opt_time_zone {
-    if (outputType != nebula::time::Type::kDateTime) {
-      delete $1;
-      delete $3;
-      throw DatetimeParser::syntax_error(@1, "Mismatched date time type.");
-    }
-    $$ = new DateTime(TimeConversion::dateTimeShift(DateTime(*$1, *$3), -$4));
+  : KW_DATETIME date date_time_delimiter time opt_time_zone {
+    $$ = new DateTime(TimeConversion::dateTimeShift(DateTime(*$2, *$4), -$5));
     *output = $$;
-    delete $1;
-    delete $3;
+    delete $2;
+    delete $4;
   }
-  | date {
-    if (outputType != nebula::time::Type::kDate) {
-      delete $1;
-      throw DatetimeParser::syntax_error(@1, "Mismatched date time type.");
-    }
-    $$ = new DateTime(*$1);
+  | KW_DATETIME date {
+    $$ = new DateTime(*$2);
     *output = $$;
-    delete $1;
+    delete $2;
   }
-  | time opt_time_zone {
-    if (outputType != nebula::time::Type::kTime) {
-      delete $1;
-      throw DatetimeParser::syntax_error(@1, "Mismatched date time type.");
-    }
-    $$ = new DateTime(TimeConversion::dateTimeShift(DateTime(1970, 1, 1, $1->hour, $1->minute, $1->sec, $1->microsec), -$2));
+  | KW_DATE date {
+    $$ = new DateTime(*$2);
     *output = $$;
-    delete $1;
+    delete $2;
+  }
+  | KW_TIME time opt_time_zone {
+    $$ = new DateTime(TimeConversion::dateTimeShift(DateTime(1970, 1, 1, $2->hour, $2->minute, $2->sec, $2->microsec), -$3));
+    *output = $$;
+    delete $2;
   }
   ;
 
@@ -131,6 +123,14 @@ date
       throw DatetimeParser::syntax_error(@1, result.toString());
     }
   }
+  | INTEGER {
+    $$ = new nebula::Date($1, 1, 1);
+    auto result = nebula::time::TimeUtils::validateDate(*$$);
+    if (!result.ok()) {
+      delete $$;
+      throw DatetimeParser::syntax_error(@1, result.toString());
+    }
+  }
   ;
 
 time
@@ -160,6 +160,14 @@ time
       throw DatetimeParser::syntax_error(@1, result.toString());
     }
   }
+  | INTEGER {
+    $$ = new nebula::Time($1, 0, 0, 0);
+    auto result = nebula::time::TimeUtils::validateTime(*$$);
+    if (!result.ok()) {
+      delete $$;
+      throw DatetimeParser::syntax_error(@1, result.toString());
+    }
+  }
   ;
 
 opt_time_zone
diff --git a/src/common/time/parser/datetime_scanner.lex b/src/common/time/parser/datetime_scanner.lex
index 6df2f8321d4..a1fec9576dc 100644
--- a/src/common/time/parser/datetime_scanner.lex
+++ b/src/common/time/parser/datetime_scanner.lex
@@ -31,6 +31,10 @@ R_BRACKET                   "]"
 "+"                     { return TokenType::POSITIVE; }
 "-"                     { return TokenType::NEGATIVE; }
 
+"DATETIME__"            { return TokenType::KW_DATETIME; }
+"DATE__"                { return TokenType::KW_DATE; }
+"TIME__"                { return TokenType::KW_TIME; }
+
 
 {DEC}+                  {
                           try {
diff --git a/src/common/time/parser/test/DateTimeParserTest.cpp b/src/common/time/parser/test/DateTimeParserTest.cpp
index 939441b48b4..a51d39a9a72 100644
--- a/src/common/time/parser/test/DateTimeParserTest.cpp
+++ b/src/common/time/parser/test/DateTimeParserTest.cpp
@@ -32,14 +32,25 @@ TEST(DatetimeReader, DateTime) {
     ASSERT_TRUE(result.ok()) << result.status();
     EXPECT_EQ(nebula::DateTime(2019, 1, 1, 22, 22, 3, 233300), result.value());
   }
-  // TODO
   // lack month
-  // {
-  // auto parser = time::DatetimeReader::makeDateTimeReader();
-  // auto result = parser.readDatetime("2019T22:22:3.2333");
-  // ASSERT_TRUE(result.ok()) << result.status();
-  // EXPECT_EQ(nebula::DateTime(2019, 1, 1, 22, 22, 3, 233300), result.value());
-  // }
+  {
+    auto parser = time::DatetimeReader::makeDateTimeReader();
+    auto result = parser.readDatetime("2019T22:22:3.2333");
+    ASSERT_TRUE(result.ok()) << result.status();
+    EXPECT_EQ(nebula::DateTime(2019, 1, 1, 22, 22, 3, 233300), result.value());
+  }
+  {
+    auto parser = time::DatetimeReader::makeDateTimeReader();
+    auto result = parser.readDatetime("2019T22:22");
+    ASSERT_TRUE(result.ok()) << result.status();
+    EXPECT_EQ(nebula::DateTime(2019, 1, 1, 22, 22, 0, 0), result.value());
+  }
+  {
+    auto parser = time::DatetimeReader::makeDateTimeReader();
+    auto result = parser.readDatetime("2019T22");
+    ASSERT_TRUE(result.ok()) << result.status();
+    EXPECT_EQ(nebula::DateTime(2019, 1, 1, 22, 0, 0, 0), result.value());
+  }
   // lack us
   {
     auto parser = time::DatetimeReader::makeDateTimeReader();
@@ -54,14 +65,32 @@ TEST(DatetimeReader, DateTime) {
     ASSERT_TRUE(result.ok()) << result.status();
     EXPECT_EQ(nebula::DateTime(2019, 1, 1, 22, 22, 0, 0), result.value());
   }
-  // TODO
   // lack minute
-  // {
-  // auto parser = time::DatetimeReader::makeDateTimeReader();
-  // auto result = parser.readDatetime("2019-1T22");
-  // ASSERT_TRUE(result.ok()) << result.status();
-  // EXPECT_EQ(nebula::DateTime(2019, 1, 1, 22, 0, 0, 0), result.value());
-  // }
+  {
+    auto parser = time::DatetimeReader::makeDateTimeReader();
+    auto result = parser.readDatetime("2019-1T22");
+    ASSERT_TRUE(result.ok()) << result.status();
+    EXPECT_EQ(nebula::DateTime(2019, 1, 1, 22, 0, 0, 0), result.value());
+  }
+  // datetime just include date
+  {
+    auto parser = time::DatetimeReader::makeDateTimeReader();
+    auto result = parser.readDatetime("2019-1-1");
+    ASSERT_TRUE(result.ok()) << result.status();
+    EXPECT_EQ(nebula::DateTime(2019, 1, 1, 0, 0, 0, 0), result.value());
+  }
+  {
+    auto parser = time::DatetimeReader::makeDateTimeReader();
+    auto result = parser.readDatetime("2019-1");
+    ASSERT_TRUE(result.ok()) << result.status();
+    EXPECT_EQ(nebula::DateTime(2019, 1, 1, 0, 0, 0, 0), result.value());
+  }
+  {
+    auto parser = time::DatetimeReader::makeDateTimeReader();
+    auto result = parser.readDatetime("2019");
+    ASSERT_TRUE(result.ok()) << result.status();
+    EXPECT_EQ(nebula::DateTime(2019, 1, 1, 0, 0, 0, 0), result.value());
+  }
 }
 
 TEST(DatetimeReader, DateTimeFailed) {
@@ -137,14 +166,20 @@ TEST(DatetimeReader, Date) {
     ASSERT_TRUE(result.ok()) << result.status();
     EXPECT_EQ(nebula::Date(2019, 1, 1), result.value());
   }
-  // TODO
+  // lack month and day
+  {
+    auto parser = time::DatetimeReader::makeDateReader();
+    auto result = parser.readDate("2019");
+    ASSERT_TRUE(result.ok()) << result.status();
+    EXPECT_EQ(nebula::Date(2019, 1, 1), result.value());
+  }
   // lack month
-  // {
-  // auto parser = time::DatetimeReader::makeDateReader();
-  // auto result = parser.readDate("2019");
-  // ASSERT_TRUE(result.ok()) << result.status();
-  // EXPECT_EQ(nebula::Date(2019, 1, 1), result.value());
-  // }
+  {
+    auto parser = time::DatetimeReader::makeDateReader();
+    auto result = parser.readDate("2019");
+    ASSERT_TRUE(result.ok()) << result.status();
+    EXPECT_EQ(nebula::Date(2019, 1, 1), result.value());
+  }
 }
 
 TEST(DatetimeReader, DateFailed) {
@@ -182,6 +217,11 @@ TEST(DatetimeReader, DateFailed) {
     auto result = parser.readDate("2019-01-");
     EXPECT_FALSE(result.ok()) << result.value();
   }
+  {
+    auto parser = time::DatetimeReader::makeDateReader();
+    auto result = parser.readDate("2019-");
+    EXPECT_FALSE(result.ok()) << result.value();
+  }
   // not exits prefix
   {
     auto parser = time::DatetimeReader::makeDateReader();
@@ -224,14 +264,13 @@ TEST(DatetimeReader, Time) {
     ASSERT_TRUE(result.ok()) << result.status();
     EXPECT_EQ(nebula::Time(22, 22, 0, 0), result.value());
   }
-  // TODO
   // lack minute
-  // {
-  // auto parser = time::DatetimeReader::makeTimeReader();
-  // auto result = parser.readTime("22");
-  // ASSERT_TRUE(result.ok()) << result.status();
-  // EXPECT_EQ(nebula::Time(22, 0, 0, 0), result.value());
-  // }
+  {
+    auto parser = time::DatetimeReader::makeTimeReader();
+    auto result = parser.readTime("22");
+    ASSERT_TRUE(result.ok()) << result.status();
+    EXPECT_EQ(nebula::Time(22, 0, 0, 0), result.value());
+  }
 }
 
 TEST(DatetimeReader, TimeFailed) {