Fixed handling of timezone aware datetimes

Author: carlvoller

Cargo.lock

@@ -3,17 +3,17 @@ members = ["cli-excel-rs", "crates/*", "py-excel-rs"]
 resolver = "2"
 
 [workspace.package]
-version = "0.5.4"
+version = "0.5.5"
 authors = ["Carl Voller"]
 edition = "2021"
 homepage = "https://github.com/carlvoller/excel-rs"
 license = "MIT"
 repository = "https://github.com/carlvoller/excel-rs"
 
 [workspace.dependencies]
-excel-rs-xlsx = { version = "0.5.4", path = "crates/excel-rs-xlsx", default-features = false }
-excel-rs-csv = { version = "0.5.4", path = "crates/excel-rs-csv", default-features = false }
-excel-rs-postgres = { version = "0.5.4", path = "crates/excel-rs-postgres", default-features = false }
+excel-rs-xlsx = { version = "0.5.5", path = "crates/excel-rs-xlsx", default-features = false }
+excel-rs-csv = { version = "0.5.5", path = "crates/excel-rs-csv", default-features = false }
+excel-rs-postgres = { version = "0.5.5", path = "crates/excel-rs-postgres", default-features = false }
 
 [profile.release]
 opt-level = 3

Cargo.toml

@@ -1,15 +1,14 @@
 import py_excel_rs
-import datetime
+from datetime import datetime, timezone, timedelta
 import pandas as pd
 
-# f = open('organizations-1000000.csv', 'rb')
-# xlsx = py_excel_rs.csv_to_xlsx(f.read())
+naive_date = datetime.now()
+tz_date = datetime.now(timezone(timedelta(hours=8)))
 
+data = [[naive_date, tz_date, "hello", 10, 10.888]]
+df = pd.DataFrame(data, columns=["Date", "Timezone Date", "hi", "number1", "float2"])
 
-data = [[datetime.datetime.now(), "hello", 10, 10.888]]
-df = pd.DataFrame(data, columns=["Date", "hi", "number1", "float2"])
-
-xlsx = py_excel_rs.df_to_xlsx(df, should_infer_types=True)
+xlsx = py_excel_rs.df_to_xlsx(df, should_infer_types=False)
 
 with open('report.xlsx', 'wb') as f:
     f.write(xlsx)

benchmarks/test-py-excel-rs.py

@@ -4,7 +4,7 @@ build-backend = "maturin"
 
 [project]
 name = "py-excel-rs"
-version = "0.5.4"
+version = "0.5.5"
 description = "Some performant utility functions to convert common data structures to XLSX"
 dependencies = ["pandas", "numpy"]
 requires-python = ">=3.7"

py-excel-rs/pyproject.toml

@@ -3,13 +3,16 @@ mod utils;
 
 use std::io::Cursor;
 
-use chrono::NaiveDateTime;
+use chrono::{DateTime, NaiveDateTime, FixedOffset};
 use excel_rs_csv::{bytes_to_csv, get_headers, get_next_record};
 use excel_rs_xlsx::WorkBook;
 use numpy::PyReadonlyArray2;
 use postgres::PyPostgresClient;
-use utils::chrono_to_xlsx_date;
-use pyo3::{prelude::*, types::{PyBytes, PyList}};
+use pyo3::{
+    prelude::*,
+    types::{PyBytes, PyList},
+};
+use utils::{chrono_naive_to_xlsx_date, chrono_tz_to_xlsx_date};
 
 #[pymodule]
 fn _excel_rs<'py>(m: &Bound<'py, PyModule>) -> PyResult<()> {
@@ -68,7 +71,9 @@ fn _excel_rs<'py>(m: &Bound<'py, PyModule>) -> PyResult<()> {
                     }
                 } else {
                     if let Ok(inner_date) = x.extract::<NaiveDateTime>(py) {
-                        format!("{}", inner_date.format("%Y-%m-%d %r"))
+                        format!("{}", inner_date.format("%Y-%m-%d %H:%M:%S"))
+                    } else if let Ok(inner_date) = x.extract::<DateTime<FixedOffset>>(py) {
+                        format!("{}", inner_date.to_rfc3339())
                     } else {
                         String::from("")
                     }
@@ -117,9 +122,11 @@ fn _excel_rs<'py>(m: &Bound<'py, PyModule>) -> PyResult<()> {
                     }
                 } else {
                     if let Ok(inner_date) = x.extract::<NaiveDateTime>(py) {
-                        format!("{}", chrono_to_xlsx_date(inner_date))
+                        format!("{}", chrono_naive_to_xlsx_date(inner_date))
+                    } else if let Ok(inner_date) = x.extract::<DateTime<FixedOffset>>(py) {
+                        format!("{}", chrono_tz_to_xlsx_date(inner_date))
                     } else {
-                        String::from("")
+                        String::new()
                     }
                 }
             }

py-excel-rs/src/lib.rs

@@ -1,4 +1,4 @@
-use chrono::{NaiveDate, NaiveDateTime, NaiveTime};
+use chrono::{DateTime, FixedOffset, NaiveDate, NaiveDateTime, NaiveTime};
 
 const NANOSECONDS_IN_A_DAY: f64 = 1_000_000_000.0 * 60.0 * 60.0 * 24.0;
 
@@ -8,7 +8,7 @@ const NANOSECONDS_IN_A_DAY: f64 = 1_000_000_000.0 * 60.0 * 60.0 * 24.0;
 // const EXCEL_BUGGY_DATE_START: &str = "1900-03-01 00:00:00";
 // const DATE_TIME_FORMAT: &str = "%Y-%m-%d %H:%M:%S";
 
-pub fn chrono_to_xlsx_date(date: NaiveDateTime) -> f64 {
+pub fn chrono_naive_to_xlsx_date(date: NaiveDateTime) -> f64 {
     let start_date_only = NaiveDate::from_ymd_opt(1900, 1, 1).expect("Excel Epoch");
     let start_date = NaiveDateTime::new(
         start_date_only,
@@ -19,3 +19,19 @@ pub fn chrono_to_xlsx_date(date: NaiveDateTime) -> f64 {
     let true_delta: f64 = (delta as f64 + NANOSECONDS_IN_A_DAY + NANOSECONDS_IN_A_DAY) / NANOSECONDS_IN_A_DAY;
     return true_delta;
 }
+
+pub fn chrono_tz_to_xlsx_date(date: DateTime<FixedOffset>) -> f64 {
+    let offset = date.offset().to_owned();
+    let start_date_only = NaiveDate::from_ymd_opt(1900, 1, 1).expect("Excel Epoch");
+    let start_date = NaiveDateTime::new(
+        start_date_only,
+        NaiveTime::from_hms_opt(0, 0, 0).expect("Time Epoch"),
+    );
+
+    let start_date_utc = DateTime::<FixedOffset>::from_naive_utc_and_offset(start_date, offset);
+
+    let delta = (date - start_date_utc).num_nanoseconds().unwrap();
+
+    let true_delta: f64 = (delta as f64 + NANOSECONDS_IN_A_DAY + NANOSECONDS_IN_A_DAY) / NANOSECONDS_IN_A_DAY;
+    return true_delta;
+}