feat: performance improvement + bug fixes to conform with the openxml… (#41)

* feat: performance improvement + bug fixes to conform with the openxml sdk better

 Release-As: 0.7.0

Author: omarirfa

Cargo.toml

@@ -11,14 +11,23 @@ readme = "README.md"
 name = "jetxl"
 crate-type = ["cdylib"]
 
+[features]
+# Opt-in: use libdeflate (via the `libdeflater` crate) instead of miniz_oxide
+# for DEFLATE compression in the ZIP writer. libdeflate is typically the fastest
+# DEFLATE encoder available and still produces standard DEFLATE streams, so the
+# output remains a spec-valid .xlsx. It requires a C compiler (`cc`) at build
+# time but, unlike zlib-ng, does not require cmake. The compression ratio can
+# differ slightly from miniz_oxide at a given level, so benchmark both on your
+# data. Default build uses the pure-Rust miniz_oxide backend (no C toolchain).
+libdeflate = ["dep:libdeflater"]
+
 [dependencies]
 pyo3 = { version = "0.28", features = ["chrono", "chrono-tz","extension-module"] }
 pyo3-arrow = "0.17"
 arrow = "58"
 arrow-array = "58"
 arrow-schema = "58"
 arrow-data = "58"
-mtzip = { version = "4.0.3", features = ["rayon"] }
 chrono = "0.4"
 zmij = "1.0"
 itoa = "1.0"
@@ -28,12 +37,15 @@ base64 = "0.22"
 rand = "0.9"
 rayon = "1.10"
 rustc-hash = "2.0"
+mimalloc = { version = "0.1", default-features = false }
+miniz_oxide = "0.8"
+crc32fast = "1.4"
+libdeflater = { version = "1.24", optional = true }
 image = { version = "0.25", default-features = false, features = ["png", "jpeg", "gif", "bmp"] }
 
 [profile.release]
 opt-level = 3
 lto = "fat"
 codegen-units = 1
 panic = "abort"
-strip = true
-
+strip = true

README.md

@@ -27,7 +27,7 @@ Jetxl is a high-performance library for creating Excel files from Python with na
 
 At the time of the test, the following library versions were used:
 - Jetxl: 0.2.5
-- Polars: 1.40.1
+- Polars: 1.42.1
 - Pandas: 3.0.3
 - Pyexcelerate: 0.13.0
 - Rustpy-xlsxwriter: 0.4.4
@@ -38,7 +38,7 @@ At the time of the test, the following library versions were used:
 ### Library comparison summary
 | Library | 1M Rows | Speedup | Throughput | Memory |
 |---------|---------|---------|------------|--------|
-| **jetxl (arrow)** | **1.85s** | **1.0x** | **523K rows/s** | **~0 MB** |
+| **jetxl (arrow)** | **0.68s** | **1.0x** | **1468K rows/s** | **~0 MB** |
 | jetxl (dict) | 3.02s | 1.6x slower | 322K rows/s | ~0 MB |
 | rustpy_xlsxwriter | 7.51s | 4.0x slower | 125K rows/s | 267.5 MB |
 | xlsxwriter | 29.16s | **15x slower** | 31K rows/s | 947.6 MB |
@@ -2685,7 +2685,8 @@ import jetxl as jet
 
 df = pl.DataFrame({
     "Name": ["Alice", "Bob"],
-    "Sales": [1000, 1500]
+    "Sales": [1000, 1500],
+    "Region": ["North", "South"]
 })
 
 # Add title rows, metadata, spacing before DataFrame
@@ -3049,4 +3050,4 @@ except ValueError as e:
 ---
 
 
-Made with ❤️ and 🦀 by the Jetxl team
+Made with ❤️ and 🦀 by the Jetxl team