Added some minor optimisations for a 6% perf improvement

Author: carlvoller

Cargo.lock

@@ -3,21 +3,21 @@ members = ["cli-excel-rs", "crates/*", "py-excel-rs"]
 resolver = "2"
 
 [workspace.package]
-version = "0.5.5"
+version = "0.6.0"
 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 = "0.5.5", 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 }
+excel-rs-xlsx = { version = "0.6.0", path = "crates/excel-rs-xlsx", default-features = false }
+excel-rs-csv = { version = "0.6.0", path = "crates/excel-rs-csv", default-features = false }
+excel-rs-postgres = { version = "0.6.0", path = "crates/excel-rs-postgres", default-features = false }
 
 [profile.release]
 opt-level = 3
 lto = "fat"
 debug = true
 overflow-checks = false
-debug-assertions = false
+debug-assertions = false

Cargo.toml

@@ -1,14 +1,7 @@
 import py_excel_rs
-from datetime import datetime, timezone, timedelta
-import pandas as pd
 
-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"])
-
-xlsx = py_excel_rs.df_to_xlsx(df, should_infer_types=False)
+f = open('organizations-1000000.csv', 'rb')
+xlsx = py_excel_rs.csv_to_xlsx(f.read())
 
 with open('report.xlsx', 'wb') as f:
     f.write(xlsx)

benchmarks/test-py-excel-rs.py

@@ -1,6 +1,6 @@
 [package]
 name = "cli-excel-rs"
-version = "0.3.0"
+version = "0.4.0"
 authors.workspace = true
 edition.workspace = true
 homepage.workspace = true

cli-excel-rs/Cargo.toml

@@ -32,7 +32,13 @@ fn main() {
 
             let output_buffer = vec![];
             let mut workbook = WorkBook::new(Cursor::new(output_buffer));
-            let mut worksheet = workbook.get_worksheet(String::from("Sheet 1"));
+            let worksheet = workbook.get_worksheet(String::from("Sheet 1"));
+
+            if let Err(e) = worksheet {
+                panic!("{e}");
+            }
+
+            let mut worksheet = worksheet.unwrap();
 
             let mut reader = bytes_to_csv(data.as_slice());
             let headers = get_headers(&mut reader);

cli-excel-rs/src/main.rs

@@ -14,7 +14,7 @@ pub use sql_impl::{ExcelBytes, ExcelBytesBorrowed};
 pub fn postgres_to_xlsx<'a>(mut iter: RowIter<'a>) -> Result<Vec<u8>> {
     let output_buffer = vec![];
     let mut workbook = WorkBook::new(Cursor::new(output_buffer));
-    let mut worksheet = workbook.get_worksheet(String::from("Sheet 1"));
+    let mut worksheet = workbook.get_worksheet(String::from("Sheet 1"))?;
 
     let headers = iter.next().ok().unwrap().unwrap();
     let len = headers.len();

crates/excel-rs-postgres/src/lib.rs

@@ -9,8 +9,8 @@ readme = "../../README.md"
 description = "xlsx processor for excel-rs"
 
 [dependencies]
-zip = { version = "2.2.0", default-features = false, features = [
+zip = { version = "4.6.1", default-features = false, features = [
     "deflate-flate2",
-    "deflate-zlib-ng",
+    "deflate-flate2-zlib-rs",
 ] }
 anyhow = "1.0.86"

crates/excel-rs-xlsx/Cargo.toml

@@ -27,7 +27,9 @@ impl<W: Write + Seek> XlsxFormatter<W> {
     // }
 
     pub fn finish(mut self, num_of_sheets: u16) -> Result<W> {
-        let options = SimpleFileOptions::default();
+        let options = SimpleFileOptions::default()
+            .compression_method(zip::CompressionMethod::Deflated)
+            .compression_level(Some(1));
         self.write_content_type(&options, num_of_sheets)?;
         self.write_rels(&options)?;
         self.write_doc_props(&options)?;