From 3dd6d5dcc5eaefdb0855bb19f4982b787747722a Mon Sep 17 00:00:00 2001
From: Mahdi Dibaiee <mdibaiee@pm.me>
Date: Thu, 5 Oct 2023 17:17:55 +0100
Subject: [PATCH] parser: improve performance by first parsing a naive date

---
 Cargo.toml                                    |  1 -
 crates/parser/src/format/sanitize/datetime.rs | 36 +++++++++----------
 crates/parser/tests/sanitize_test.rs          |  2 --
 go/parser/config.go                           |  4 +--
 4 files changed, 20 insertions(+), 23 deletions(-)

diff --git a/Cargo.toml b/Cargo.toml
index 65f0b0e749..57b200f52e 100644
--- a/Cargo.toml
+++ b/Cargo.toml
@@ -40,7 +40,6 @@ byteorder = "1.4"
 caseless = "0.2"
 chardetng = "0.1"
 chrono = { version = "0.4", features = ["serde"] }
-chrono-tz = { version = "0.8" }
 clap = { version = "3.2", features = ["derive", "env"] }
 colored_json = "3"
 comfy-table = "6.1"
diff --git a/crates/parser/src/format/sanitize/datetime.rs b/crates/parser/src/format/sanitize/datetime.rs
index 8f616df962..b02bcd7be5 100644
--- a/crates/parser/src/format/sanitize/datetime.rs
+++ b/crates/parser/src/format/sanitize/datetime.rs
@@ -7,31 +7,33 @@ struct DatetimeSanitizer {
     default_offset: time::UtcOffset,
 }
 
-// Here we are trying to parse non-ambiguous, non-RFC3339 dates and formatting them as RFC3339
-// So we skip any valid RFC3339 in our processing and pass it as-is
+// Here we are trying to parse non-RFC3339 dates
 fn datetime_to_rfc3339(val: &mut Value, default_offset: time::UtcOffset) {
     match val {
         Value::String(s) => {
-            let offset_format = format_description!(
-                version = 2,
-                "[first
-                [[year]-[month]-[day] [hour]:[minute]:[second][optional [.[subsecond]]]Z]
-                [[year]-[month]-[day] [hour]:[minute]:[second][optional [.[subsecond]]]z]
-                [[year]-[month]-[day] [hour]:[minute]:[second][optional [.[subsecond]]][offset_hour]:[offset_minute]]
-                ]"
-            );
-
             let primitive_format = format_description!(
                 version = 2,
-                "[year]-[month]-[day][optional [T]][optional [ ]][hour]:[minute]:[second][optional [.[subsecond]]]"
+                "[year]-[month]-[day][optional [T]][optional [ ]][hour]:[minute]:[second][optional [.[subsecond]]][optional [Z]][optional [z]][optional [[offset_hour]:[offset_minute]]]"
             );
 
-            let parsed_with_tz = time::OffsetDateTime::parse(&s, offset_format);
-            let parsed_no_tz = time::PrimitiveDateTime::parse(&s, primitive_format);
+            let parsed_no_tz = time::PrimitiveDateTime::parse(&s, primitive_format).ok();
+
+            let parsed_with_tz = if parsed_no_tz.is_some() {
+                let offset_format = format_description!(
+                    version = 2,
+                    "[first
+                    [[year]-[month]-[day] [hour]:[minute]:[second][optional [.[subsecond]]]Z]
+                    [[year]-[month]-[day] [hour]:[minute]:[second][optional [.[subsecond]]]z]
+                    [[year]-[month]-[day] [hour]:[minute]:[second][optional [.[subsecond]]][offset_hour]:[offset_minute]]
+                    ]"
+                );
+
+                time::OffsetDateTime::parse(&s, offset_format).ok()
+            } else { None };
 
-            if let Ok(parsed) = parsed_with_tz {
+            if let Some(parsed) = parsed_with_tz {
                 *s = parsed.format(&time::format_description::well_known::Rfc3339).unwrap();
-            } else if let Ok(parsed) = parsed_no_tz {
+            } else if let Some(parsed) = parsed_no_tz {
                 *s = parsed.assume_offset(default_offset).format(&time::format_description::well_known::Rfc3339).unwrap();
             }
         }
@@ -76,9 +78,7 @@ pub enum DatetimeSanitizeError {
 }
 
 pub fn sanitize_datetime(config: &ParseConfig, output: Output) -> Result<Output, DatetimeSanitizeError> {
-    eprintln!("sanitize_datetime");
     let offset = time::UtcOffset::parse(&config.default_offset, format_description!("[offset_hour]:[offset_minute]")).map_err(DatetimeSanitizeError::OffsetParseError)?;
-    eprintln!("offset: {:?}", offset);
     let sanitizer = DatetimeSanitizer {
         from: output,
         default_offset: offset,
diff --git a/crates/parser/tests/sanitize_test.rs b/crates/parser/tests/sanitize_test.rs
index 8dacd082d9..bf6659ef20 100644
--- a/crates/parser/tests/sanitize_test.rs
+++ b/crates/parser/tests/sanitize_test.rs
@@ -1,7 +1,5 @@
 mod testutil;
 
-use chrono::DateTime;
-
 use parser::ParseConfig;
 use testutil::{input_for_file, run_test};
 
diff --git a/go/parser/config.go b/go/parser/config.go
index 5395eba9b4..e0f2ff53e5 100644
--- a/go/parser/config.go
+++ b/go/parser/config.go
@@ -18,7 +18,7 @@ type Config struct {
 	Compression     string                      `json:"compression,omitempty"`
 	ContentType     string                      `json:"contentType,omitempty"`
 	ContentEncoding string                      `json:"contentEncoding,omitempty"`
-	DefaultTimezone string                      `json:"defaultTimezone,omitempty"`
+	DefaultOffset   string                      `json:"defaultOffset,omitempty"`
 }
 
 func (c *Config) Copy() Config {
@@ -40,7 +40,7 @@ func (c *Config) Copy() Config {
 		Compression:     c.Compression,
 		ContentType:     c.ContentType,
 		ContentEncoding: c.ContentEncoding,
-		DefaultTimezone: c.DefaultTimezone,
+		DefaultOffset:   c.DefaultOffset,
 	}
 }