rzmk/czv
1
0
Fork 0
mirror of https://github.com/rzmk/czv.git synced 2025-12-19 08:09:24 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

refactor: use options object for WASM function args, improve docs

Browse source 
- Add relevant links to all READMEs and source code
- Resolve clippy lints

czv:

- Add more docs for top-level items
- Add suggestion to use builder methods instead of functions
- Disable slice and slice tests until operation is complete

czv-wasm:

- Use tsify_next for allowing objects as parameters
- Add nodejs example and instructions
This commit is contained in:
rzmk 2024-06-20 16:28:53 -04:00
parent ce260e9491
commit e84c5bec8b
No known key found for this signature in database
20 changed files with 564 additions and 168 deletions
Download patch file Download diff file Expand all files Collapse all files

63
Cargo.lock generated
View file

@ -244,7 +244,10 @@ version = "0.0.0"
dependencies = [
"anyhow",
"csv",
"serde",
"serde-wasm-bindgen",
"thiserror",
"tsify-next",
"wasm-bindgen",
"wasm-bindgen-test",
]
@ -255,6 +258,19 @@ version = "1.12.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "3dca9240753cf90908d7e4aac30f630662b02aebaa1b58a3cadabdb23385b58b"
[[package]]
name = "gloo-utils"
version = "0.2.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "0b5555354113b18c547c1d3a98fbf7fb32a9ff4f6fa112ce823a21641a0ba3aa"
dependencies = [
"js-sys",
"serde",
"serde_json",
"wasm-bindgen",
"web-sys",
]
[[package]]
name = "half"
version = "2.4.1"
@ -628,6 +644,17 @@ dependencies = [
"serde_derive",
]
[[package]]
name = "serde-wasm-bindgen"
version = "0.6.5"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "8302e169f0eddcc139c70f139d19d6467353af16f9fce27e8c30158036a1e16b"
dependencies = [
"js-sys",
"serde",
"wasm-bindgen",
]
[[package]]
name = "serde_derive"
version = "1.0.203"
@ -639,6 +666,17 @@ dependencies = [
"syn",
]
[[package]]
name = "serde_derive_internals"
version = "0.29.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "18d26a20a969b9e3fdf2fc2d9f21eda6c40e2de84c9408bb5d3b05d499aae711"
dependencies = [
"proc-macro2",
"quote",
"syn",
]
[[package]]
name = "serde_json"
version = "1.0.117"
@ -709,6 +747,31 @@ dependencies = [
"serde_json",
]
[[package]]
name = "tsify-next"
version = "0.5.3"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "0407c98efe14c83fd025675246bf855f0ed750b6b57fa6ab839ade598ceb5374"
dependencies = [
"gloo-utils",
"serde",
"serde_json",
"tsify-next-macros",
"wasm-bindgen",
]
[[package]]
name = "tsify-next-macros"
version = "0.5.3"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "615b1b80acc6720d554777d55ed4ed35a5ceb0e0df7ff7cf9d3946ecbf74953e"
dependencies = [
"proc-macro2",
"quote",
"serde_derive_internals",
"syn",
]
[[package]]
name = "typed-builder"
version = "0.18.2"

23
README.md
View file

@ -2,8 +2,17 @@
czv is CSV content manipulation/analysis libraries with support for Rust, Python, and WebAssembly (JavaScript and TypeScript).
## Links
- czv GitHub repository: <https://github.com/rzmk/czv>
- Rust: [crates.io/crates/czv](https://crates.io/crates/czv) ([source code](https://github.com/rzmk/czv/tree/main/czv))
- WebAssembly (JavaScript/TypeScript): [npmjs.com/package/czv](https://www.npmjs.com/package/czv) ([source code](https://github.com/rzmk/czv/tree/main/czv-wasm))
- Python: [pypi.org/project/czv](https://pypi.org/project/czv/) ([source code](https://github.com/rzmk/czv/tree/main/czv-python))
## Installation and examples
In the following examples we'll get the total number of rows in the CSV data including the header row.
### Rust
```bash
@ -11,10 +20,7 @@ cargo install czv
```
```rust
use czv::{
count::RowCount,
Result
};
use czv::{RowCount, Result};
fn main() -> Result<()> {
let data = "\
@ -25,8 +31,9 @@ strawberry,1.50
";
let output = RowCount::new()
.file_data(data)
.include_header_row(true)
.execute()?;
println!("{output}"); // 3
println!("{output}"); // 4
Ok(())
}
```
@ -49,7 +56,11 @@ apple,2.50
banana,3.00
strawberry,1.50`;
const output = czv.rowCount(data);
const output = czv.rowCount({
file_data: data,
include_header_row: true,
});
console.log(output);
```

8
czv-python/README.md
View file

@ -2,8 +2,12 @@
Python library for [czv](https://github.com/rzmk/czv). czv is a library of CSV-related operations for data engineering and analysis tasks.
- For a Rust library see [czv](https://github.com/rzmk/czv/tree/main/czv).
- For a WebAssembly (JavaScript, TypeScript) library see [czv-wasm](https://github.com/rzmk/czv/tree/main/czv-wasm).
## Links
- czv GitHub repository: <https://github.com/rzmk/czv>
- Rust: [crates.io/crates/czv](https://crates.io/crates/czv) ([source code](https://github.com/rzmk/czv/tree/main/czv))
- WebAssembly (JavaScript/TypeScript): [npmjs.com/package/czv](https://www.npmjs.com/package/czv) ([source code](https://github.com/rzmk/czv/tree/main/czv-wasm))
- Python: [pypi.org/project/czv](https://pypi.org/project/czv/) ([source code](https://github.com/rzmk/czv/tree/main/czv-python))
## Installation and example

9
czv-python/czv.pyi
View file

@ -1,7 +1,7 @@
"""
# czv
Python library for [czv](https://github.com/rzmk/czv). CSV content manipulation and analysis.
Python library for [czv](https://github.com/rzmk/czv). CSV operations library for data engineering/analysis tasks.
## Install
@ -19,6 +19,13 @@ output = row_count(file_path="fruits.csv")
print(output)
```
## Links
- czv GitHub repository: <https://github.com/rzmk/czv>
- Rust: [crates.io/crates/czv](https://crates.io/crates/czv) ([source code](https://github.com/rzmk/czv/tree/main/czv))
- WebAssembly (JavaScript/TypeScript): [npmjs.com/package/czv](https://www.npmjs.com/package/czv) ([source code](https://github.com/rzmk/czv/tree/main/czv-wasm))
- Python: [pypi.org/project/czv](https://pypi.org/project/czv/) ([source code](https://github.com/rzmk/czv/tree/main/czv-python))
"""
from typing import Optional

3
czv-wasm/Cargo.toml
View file

@ -12,6 +12,9 @@ crate-type = ["cdylib", "rlib"]
[dependencies]
anyhow = "1.0.86"
csv = "1.3.0"
serde = { version = "1.0.203", features = ["derive"] }
serde-wasm-bindgen = "0.6.5"
thiserror = "1.0.61"
tsify-next = "0.5.3"
wasm-bindgen = { version = "0.2" }
wasm-bindgen-test = { version = "0.3.42" }

18
czv-wasm/README.md
View file

@ -2,8 +2,12 @@
WebAssembly (JavaScript and TypeScript) library for [czv](https://github.com/rzmk/czv). czv is a library of CSV-related operations for data engineering and analysis tasks.
- For a Rust library see [czv](https://github.com/rzmk/czv/tree/main/czv).
- For a Python library see [czv-python](https://github.com/rzmk/czv/tree/main/czv-python).
## Links
- czv GitHub repository: <https://github.com/rzmk/czv>
- Rust: [crates.io/crates/czv](https://crates.io/crates/czv) ([source code](https://github.com/rzmk/czv/tree/main/czv))
- WebAssembly (JavaScript/TypeScript): [npmjs.com/package/czv](https://www.npmjs.com/package/czv) ([source code](https://github.com/rzmk/czv/tree/main/czv-wasm))
- Python: [pypi.org/project/czv](https://pypi.org/project/czv/) ([source code](https://github.com/rzmk/czv/tree/main/czv-python))
## Installation and example
@ -23,7 +27,11 @@ apple,2.50
banana,3.00
strawberry,1.50`;
const output = czv.rowCount(data);
const output = czv.rowCount({
file_data: data,
include_header_row: true,
});
console.log(output);
```
@ -38,9 +46,11 @@ cargo install wasm-pack
### Build WASM for web
```bash
wasm-pack build --release --target web
wasm-pack build --release --target web --out-name czv
```
**Note**: Currently you must modify the `pkg/package.json` by replacing the name from `czv-wasm` to `czv`.
### Test WASM for browser
```bash

1
czv-wasm/examples/basic-demo/README.md Normal file
View file

@ -0,0 +1 @@
If you want to modify the TypeScript code then after you save your modifications you must run `tsc script.ts --module es2022 --target es2022` to generate a compatible `script.js` file.

25
czv-wasm/examples/basic-demo/index.html
View file

@ -40,30 +40,7 @@
</tr>
</tbody>
</table>
<script type="module">
const input = document.getElementById("upload")
const progress = document.getElementById("progress")
import init, * as czv from "../../pkg/czv.js";
// Must run `await init()` or `initSync()` first for web use
await init();
const fileReader = new FileReader()
fileReader.onloadstart = () => {
progress.style.display = "block";
}
fileReader.onloadend = () => {
document.getElementById("column-count").innerText = czv.columnCount(fileReader.result)
document.getElementById("row-count").innerText = czv.rowCount(fileReader.result)
progress.style.display = "none";
}
input.addEventListener("change", () => {
fileReader.readAsText(input.files[0])
})
</script>
<script src="script.js" type="module"></script>
</body>
</html>

27
czv-wasm/examples/basic-demo/script.js Normal file
View file

@ -0,0 +1,27 @@
const progress = document.getElementById("progress");
import init, * as czv from "../../pkg/czv.js";
// Must run `await init()` or `initSync()` first for web use
await init();
const fileReader = new FileReader();
fileReader.onloadstart = () => {
if (progress)
progress.style.display = "block";
};
fileReader.onloadend = () => {
const rowCountElement = document.getElementById("row-count");
const columnCountElement = document.getElementById("column-count");
if (rowCountElement)
rowCountElement.innerText = String(czv.rowCount({
file_data: fileReader.result,
}));
if (columnCountElement)
columnCountElement.innerText = String(czv.columnCount({ file_data: fileReader.result }));
if (progress)
progress.style.display = "none";
};
const input = document.getElementById("upload");
if (input)
input.addEventListener("change", () => {
// @ts-ignore
fileReader.readAsText(input.files[0]);
});

34
czv-wasm/examples/basic-demo/script.ts Normal file
View file

@ -0,0 +1,34 @@
const progress = document.getElementById("progress");
import init, * as czv from "../../pkg/czv.js";
// Must run `await init()` or `initSync()` first for web use
await init();
const fileReader = new FileReader();
fileReader.onloadstart = () => {
if (progress) progress.style.display = "block";
};
fileReader.onloadend = () => {
const rowCountElement = document.getElementById("row-count");
const columnCountElement = document.getElementById("column-count");
if (rowCountElement)
rowCountElement.innerText = String(
czv.rowCount({
file_data: fileReader.result as string,
})
);
if (columnCountElement)
columnCountElement.innerText = String(
czv.columnCount({ file_data: fileReader.result as string })
);
if (progress) progress.style.display = "none";
};
const input = document.getElementById("upload");
if (input)
input.addEventListener("change", () => {
// @ts-ignore
fileReader.readAsText(input.files[0]);
});

4
czv-wasm/examples/node-demo/README.md Normal file
View file

@ -0,0 +1,4 @@
To run this example:
1. Build the `nodejs` compatible bundle in the `czv-wasm` directory with `wasm-pack build -t nodejs --release --out-name czv-ts`.
2. Run `bunx tsx sample.ts` (or `npx`, `pnpx`, etc.).

12
czv-wasm/examples/node-demo/sample.ts Normal file
View file

@ -0,0 +1,12 @@
import czv from "../../pkg/czv-ts";
const options: czv.RowCountOptions = {
file_data: `fruit,price,
apple,2.50
banana,3.00
strawberry,1.50`,
};
const output = czv.rowCount(options);
console.log(output); // 4

52
czv-wasm/src/count.rs
View file

@ -1,32 +1,50 @@
use crate::Result;
use csv::ReaderBuilder;
use serde::{Deserialize, Serialize};
use tsify_next::Tsify;
use wasm_bindgen::prelude::*;
/// Options for `rowCount`.
#[derive(Tsify, Serialize, Deserialize)]
#[tsify(into_wasm_abi, from_wasm_abi)]
pub struct RowCountOptions {
/// CSV file data.
pub file_data: String,
#[tsify(optional)]
/// Specify whether to include the header row (first row) in the row count.
/// Default is false.
pub include_header_row: Option<bool>,
}
/// Returns a count of the total number of rows.
///
/// @param {string} `file_data` CSV file data.
/// @param {boolean | undefined} `include_header_row` Specify whether to include the header row (first row) in the row count. Default is false.
/// @returns {number}
#[wasm_bindgen(skip_jsdoc, js_name = rowCount)]
pub fn row_count(file_data: String, include_header_row: Option<bool>) -> Result<usize> {
pub fn row_count(options: RowCountOptions) -> Result<usize> {
let mut rdr = ReaderBuilder::new();
rdr.has_headers(!include_header_row.unwrap_or(false));
return Ok(rdr.from_reader(file_data.as_bytes()).records().count());
rdr.has_headers(!options.include_header_row.unwrap_or(false));
Ok(rdr
.from_reader(options.file_data.as_bytes())
.records()
.count())
}
/// Options for `columnCount`.
#[derive(Tsify, Serialize, Deserialize)]
#[tsify(into_wasm_abi, from_wasm_abi)]
pub struct ColumnCountOptions {
/// CSV file data.
pub file_data: String,
}
/// Returns a count of the total number of columns (fields).
///
/// ## Arguments
///
/// @param {string} `file_data` CSV file data.
#[wasm_bindgen(skip_jsdoc, js_name = columnCount)]
pub fn column_count(file_data: Option<String>) -> Result<usize> {
pub fn column_count(options: ColumnCountOptions) -> Result<usize> {
let rdr = ReaderBuilder::new();
if let Some(file_data) = file_data {
return Ok(rdr.from_reader(file_data.as_bytes()).headers()?.len());
} else {
bail!("Could not determine a file path or file data for column_count_builder.");
}
Ok(rdr
.from_reader(options.file_data.as_bytes())
.headers()?
.len())
}

54
czv-wasm/src/lib.rs
View file

@ -1,3 +1,45 @@
//! # czv
//!
//! WASM library for [czv](https://github.com/rzmk/czv). CSV operations library for data engineering/analysis tasks.
//!
//! # Example
//!
//! Let's say we want to print the total number of non-header rows in our data:
//!
//! ```js
//! import init, * as czv from "czv";
//! // Must run `await init()` or `initSync()` first for web use
//! await init();
//!
//! const data = `fruits,price
//! apple,2.50
//! banana,3.00
//! strawberry,1.50`;
//!
//! const output = czv.rowCount({
//! file_data: data,
//! include_header_row: true,
//! });
//!
//! console.log(output);
//! ```
//!
//! For a full website example see the example's source code here: https://www.npmjs.com/package/czv) <https://github.com/rzmk/czv/tree/main/czv-wasm/examples/basic-demo>
//!
//! # Links
//!
//! - czv GitHub repository: <https://github.com/rzmk/czv>
//! - Rust: [crates.io/crates/czv](https://crates.io/crates/czv) ([source code](https://github.com/rzmk/czv/tree/main/czv))
//! - WebAssembly (JavaScript/TypeScript): [npmjs.com/package/czv](https://www.npmjs.com/package/czv) ([source code](https://github.com/rzmk/czv/tree/main/czv-wasm))
//! - Python: [pypi.org/project/czv](https://pypi.org/project/czv/) ([source code](https://github.com/rzmk/czv/tree/main/czv-python))
#![allow(
// https://github.com/madonoharu/tsify/issues/42
non_snake_case,
// https://github.com/rustwasm/wasm-bindgen/issues/3945
clippy::empty_docs
)]
use wasm_bindgen::JsValue;
// Error-handling helpers
@ -11,9 +53,15 @@ impl From<csv::Error> for CzvError {
}
}
impl Into<JsValue> for CzvError {
fn into(self) -> JsValue {
JsValue::from_str(self.to_string().as_str())
impl From<serde_wasm_bindgen::Error> for CzvError {
fn from(value: serde_wasm_bindgen::Error) -> Self {
value.into()
}
}
impl From<CzvError> for JsValue {
fn from(val: CzvError) -> Self {
JsValue::from_str(val.to_string().as_str())
}
}

10
czv-wasm/tests/count.rs
View file

@ -1,4 +1,5 @@
use czv_wasm;
use czv_wasm::count::RowCountOptions;
use czv_wasm::Result;
use wasm_bindgen_test::*;
wasm_bindgen_test::wasm_bindgen_test_configure!(run_in_browser);
@ -6,14 +7,15 @@ wasm_bindgen_test::wasm_bindgen_test_configure!(run_in_browser);
#[test]
#[wasm_bindgen_test]
fn basic() -> Result<()> {
let result = czv_wasm::count::row_count(
"fruit,price
let options = RowCountOptions {
file_data: "fruit,price
apple,2.00
banana,1.50
strawberry,3.00"
.to_string(),
Some(false),
)?;
include_header_row: Some(false),
};
let result = czv_wasm::count::row_count(options)?;
assert_eq!(result, 3);
Ok(())
}

10
czv/README.md
View file

@ -2,8 +2,12 @@
Rust library for [czv](https://github.com/rzmk/czv). czv is a library of CSV-related operations for data engineering and analysis tasks.
- For a WebAssembly (JavaScript, TypeScript) library see [czv-wasm](https://github.com/rzmk/czv/tree/main/czv-wasm).
- For a Python library see [czv-python](https://github.com/rzmk/czv/tree/main/czv-python).
## Links
- czv GitHub repository: <https://github.com/rzmk/czv>
- Rust: [crates.io/crates/czv](https://crates.io/crates/czv) ([source code](https://github.com/rzmk/czv/tree/main/czv))
- WebAssembly (JavaScript/TypeScript): [npmjs.com/package/czv](https://www.npmjs.com/package/czv) ([source code](https://github.com/rzmk/czv/tree/main/czv-wasm))
- Python: [pypi.org/project/czv](https://pypi.org/project/czv/) ([source code](https://github.com/rzmk/czv/tree/main/czv-python))
## Usage
@ -16,7 +20,7 @@ cargo install czv
Let's say we want to print the total number of rows in a 4x3 CSV file `fruits.csv` including the header row:
```rust
use czv::{count::RowCount, Result};
use czv::{RowCount, Result};
fn main() -> Result<()> {
let data = "\

32
czv/src/count.rs
View file

@ -42,12 +42,16 @@ pub fn row_count(
#[derive(TypedBuilder)]
#[builder(doc, builder_method(name=new, doc="Returns a `RowCountBuilder` to customize row count options by running other methods before getting the row count with the `execute` method."), build_method(vis="", name=__build))]
/// Returns a count of the total number of rows.
///
/// The entry point for setting up a [`czv::count::RowCountBuilder`](crate::count::RowCountBuilder) by running [`RowCount::new()`](crate::count::RowCount::new).
///
/// For example:
/// # Example
///
/// Let's say we want to print the total number of non-header rows in our data:
///
/// ```rust
/// use czv::{count::RowCount, Result};
/// use czv::{RowCount, Result};
///
/// fn main() -> Result<()> {
/// let data = "\
@ -113,6 +117,26 @@ impl<
///
/// See [`czv::count::ColumnCount`](crate::count::ColumnCount) for a builder version (recommended) of this function.
///
/// # Example
///
/// Let's say we want to print the total number of columns in our data:
///
/// ```rust
/// use czv::{ColumnCount, Result};
///
/// fn main() -> Result<()> {
/// let data = "\
/// fruits,price
/// apple,2.50
/// banana,3.00
/// strawberry,1.50
/// ";
/// let output = ColumnCount::new().file_data(data).execute()?;
/// println!("{output}"); // 2
/// Ok(())
/// }
/// ```
///
/// ## Arguments
///
/// * `file_path` - CSV file path (alternative to `file_data`).
@ -122,7 +146,7 @@ pub fn column_count(file_path: Option<PathBuf>, file_data: Option<String>) -> Re
// file_path
if let Some(file_path) = file_path {
return Ok(rdr.from_path(file_path)?.headers()?.len());
Ok(rdr.from_path(file_path)?.headers()?.len())
}
// file_data
else if let Some(file_data) = file_data {
@ -136,6 +160,8 @@ pub fn column_count(file_path: Option<PathBuf>, file_data: Option<String>) -> Re
#[derive(TypedBuilder)]
#[builder(doc, builder_method(name=new, doc="Returns a `ColumnCountBuilder` to customize column count options by running other methods before getting the column count with the `execute` method."), build_method(vis="", name=__build))]
/// Returns a count of the total number of columns (fields).
///
/// The entry point for setting up a [`czv::count::ColumnCountBuilder`](crate::count::ColumnCountBuilder) by running [`ColumnCount::new()`](crate::count::ColumnCount::new).
///
/// For example:

45
czv/src/lib.rs
View file

@ -2,10 +2,12 @@
//!
//! CSV operations library for data engineering/analysis tasks.
//!
//! ## Example
//! # Example
//!
//! Let's say we want to print the total number of non-header rows in our data:
//!
//! ```rust
//! use czv::{count::RowCount, Result};
//! use czv::{RowCount, Result};
//!
//! fn main() -> Result<()> {
//! let data = "\
@ -20,21 +22,35 @@
//! }
//! ```
//!
//! ## Links
//! # Usage
//!
//! - Source code: <https://github.com/rzmk/czv>
//! - czv-wasm: <https://github.com/rzmk/czv>
//! - czv-python: <https://github.com/rzmk/czv>
//! It is recommended to use the builder structs rather than functions though both are provided.
//!
//! The builder structs are provided at the top-level for ease of use.
//!
//! For example use the [czv::RowCount](crate::RowCount) struct rather than the [czv::count::row_count](crate::count::row_count) function.
//!
//! # Links
//!
//! - czv GitHub repository: <https://github.com/rzmk/czv>
//! - Rust: [crates.io/crates/czv](https://crates.io/crates/czv) ([source code](https://github.com/rzmk/czv/tree/main/czv))
//! - WebAssembly (JavaScript/TypeScript): [npmjs.com/package/czv](https://www.npmjs.com/package/czv) ([source code](https://github.com/rzmk/czv/tree/main/czv-wasm))
//! - Python: [pypi.org/project/czv](https://pypi.org/project/czv/) ([source code](https://github.com/rzmk/czv/tree/main/czv-python))
/// Counting operations including row count and column count.
pub mod count;
#[doc(inline)]
pub use count::{ColumnCount, RowCount};
/// Extract a section of rows.
pub mod slice;
// pub mod slice;
// #[doc(inline)]
// pub use slice::Slice;
#[allow(dead_code)]
// Error-handling helpers
#[allow(dead_code)]
#[derive(thiserror::Error, Debug)]
#[error("{0}")]
/// Common catch-all error type based on [anyhow::Error].
pub struct CzvError(pub anyhow::Error);
impl From<anyhow::Error> for CzvError {
@ -49,10 +65,23 @@ impl From<csv::Error> for CzvError {
}
}
/// Common Result type based on [anyhow::Result] and [czv::CzvError](crate::CzvError).
pub type Result<T, E = CzvError> = anyhow::Result<T, E>;
#[macro_export]
#[allow(unused_macros)]
/// Function-like macro you may pass a `&str` to return a [czv::CzvError](crate::CzvError).
///
/// For example:
///
/// ```should_panic
/// use czv::{bail, CzvError, Result};
///
/// fn main() -> Result<()> {
/// bail!("This is an example of an error message using `bail!`.");
/// }
/// ```
#[allow(clippy::crate_in_macro_def)]
macro_rules! bail {
($err:expr $(,)?) => {
return Err(crate::CzvError(anyhow::anyhow!($err)))

139
czv/src/slice.rs
View file

@ -21,20 +21,20 @@ use typed_builder::TypedBuilder;
/// Let's print the first two non-header rows in the CSV file:
///
/// ```rust
/// use czv::{
/// Result,
/// slice::Slice
/// };
/// use czv::Result;
///
/// fn main() -> Result<()> {
/// let file_path = "tests/resources/fruits.csv";
///
/// let output: String = Slice::new()
/// .file_path(file_path)
/// .start(0)
/// .end(2) // exclusive
/// .include_header_row(false)
/// .execute()?;
/// let output: String = czv::slice::slice(
/// Some(file_path.into()), // file_path
/// None, // file_data
/// Some(0), // start
/// Some(2), // end
/// None, // length
/// None, // index
/// false // include_header_row
/// )?;
///
/// println!("{output}");
///
@ -106,14 +106,23 @@ pub fn slice(
}
None => None,
};
let end_line = match end {
Some(x) => {
let end_line = match (end, length) {
(Some(e), None) => {
if x >= records_count as i32 {
bail!("end value {x} cannot be greater than or equal to the number of records.")
} else if x >= 0 {
Some(x as usize)
bail!("end value {e} cannot be greater than or equal to the number of records.")
} else if e >= 0 {
Some(e as usize)
} else {
Some((records_count as i32 + x) as usize)
Some((records_count as i32 + e) as usize)
}
}
(None, Some(l)) => {
if l >= records_count as i32 {
bail!("end value {l} cannot be greater than or equal to the number of records.")
} else if l >= 0 {
Some(l as usize)
} else {
Some((records_count as i32 + l) as usize)
}
}
None => None,
@ -222,27 +231,108 @@ pub fn slice(
}
#[derive(TypedBuilder)]
#[builder(builder_method(name=new), build_method(vis="", name=__build))]
#[builder(doc, builder_method(name=new, doc="Returns a [`SliceBuilder`](crate::slice::SliceBuilder) to customize row count options by running other methods before getting the sliced rows with the [`execute`](crate::slice::SliceBuilder::execute) method."), build_method(vis="", name=__build))]
/// Returns a slice of rows from the CSV data.
///
/// ## Example
///
/// We have a file `fruits.csv` at the path `tests/resources/fruits.csv`:
///
/// ```csv
/// fruit,price
/// apple,2.50
/// banana,3.00
/// strawberry,1.50
/// ```
///
/// Let's print the first two non-header rows in the CSV file:
///
/// ```rust
/// use czv::{Result, Slice};
///
/// fn main() -> Result<()> {
/// let file_path = "tests/resources/fruits.csv";
///
/// let output: String = Slice::new()
/// .file_path(file_path)
/// .start(0)
/// .end(2) // exclusive
/// .include_header_row(false)
/// .execute()?;
///
/// println!("{output}");
///
/// Ok(())
/// }
/// ```
///
/// Printed to the terminal we get:
///
/// ```console
/// apple,2.50
/// banana,3.00
/// ```
///
/// ## Notes
///
/// - You may not use `file_path` and `file_data` together, only one may be specified.
/// - You may not use `end` and `length` together, only one may be specified.
/// - You may not use `index` with any of `start`, `end`, or `length`.
pub struct Slice {
#[builder(default, setter(strip_option, into))]
#[builder(
default,
setter(
doc = "CSV file path (alternative to `file_data`).",
strip_option,
into
)
)]
file_path: Option<PathBuf>,
#[builder(default, setter(strip_option, into))]
#[builder(
default,
setter(
doc = "CSV file data (alternative to `file_path`).",
strip_option,
into
)
)]
file_data: Option<String>,
#[builder(default, setter(strip_option))]
#[builder(
default,
setter(
doc = "The index of the record to slice from (0-indexed). If negative, starts from the last record.",
strip_option
)
)]
start: Option<i32>,
#[builder(default, setter(strip_option))]
#[builder(
default,
setter(doc = "The index of the record to slice to.", strip_option)
)]
end: Option<i32>,
#[builder(default, setter(strip_option))]
#[builder(
default,
setter(doc = "The length of the slice (alternative to `end`).", strip_option)
)]
length: Option<i32>,
#[builder(default, setter(strip_option))]
#[builder(
default,
setter(
doc = "Slice a single record. If negative, starts from the last record.",
strip_option
)
)]
index: Option<i32>,
#[builder(default = false)]
#[builder(
default = false,
setter(doc = "Specify whether to include the header row (first row) in the records.")
)]
include_header_row: bool,
}
@ -266,6 +356,7 @@ impl<
__include_header_row,
)>
{
/// Returns the sliced rows.
pub fn execute(self) -> Result<String> {
let builder = self.__build();
slice(

163
czv/tests/test_slice.rs
View file

@ -1,75 +1,100 @@
use czv;
use czv::Result;
// use czv;
// use czv::Result;
#[test]
fn test_slice_start_end() -> Result<()> {
let cases = vec![(
"tests/resources/fruits.csv",
1,
3,
"banana,3.00\nstrawberry,1.50".to_string(),
)];
for (file_name, start, end, expected) in cases {
let got = czv::slice::slice(
Some(file_name.into()),
None,
Some(start),
Some(end),
None,
None,
false,
)?;
assert_eq!(expected, got);
}
Ok(())
}
// #[test]
// fn test_slice() -> Result<()> {
// let got = czv::Slice::new()
// .file_path("tests/resources/fruits.csv")
// .execute()?;
// let expected = "fruit,price\napple,2.50\nbanana,3.00\nstrawberry,1.50".to_string();
// assert_eq!(expected, got);
// Ok(())
// }
#[test]
fn test_slice_start_end_data() -> Result<()> {
let cases = vec![(
"fruit,price\napple,2.50\nbanana,3.00\nstrawberry,1.50".to_string(),
1,
3,
"banana,3.00\nstrawberry,1.50".to_string(),
)];
for (file_data, start, end, expected) in cases {
let got = czv::slice::slice(
None,
Some(file_data),
Some(start),
Some(end),
None,
None,
false,
)?;
assert_eq!(expected, got);
}
Ok(())
}
// #[test]
// fn test_slice_start_end() -> Result<()> {
// let cases = vec![(
// "tests/resources/fruits.csv",
// 1,
// 3,
// "banana,3.00\nstrawberry,1.50".to_string(),
// )];
// for (file_name, start, end, expected) in cases {
// let got = czv::slice::slice(
// Some(file_name.into()),
// None,
// Some(start),
// Some(end),
// None,
// None,
// false,
// )?;
// assert_eq!(expected, got);
// }
// Ok(())
// }
#[test]
fn test_slice_start_0_end_3() -> Result<()> {
let expected = "apple,2.50\nbanana,3.00".to_string();
let got: String = czv::slice::Slice::new()
.file_path("tests/resources/fruits.csv")
.start(0)
.end(2) // exclusive
.include_header_row(false)
.execute()?;
// #[test]
// fn test_slice_start_end_data() -> Result<()> {
// let cases = vec![(
// "fruit,price\napple,2.50\nbanana,3.00\nstrawberry,1.50".to_string(),
// 1,
// 3,
// "banana,3.00\nstrawberry,1.50".to_string(),
// )];
// for (file_data, start, end, expected) in cases {
// let got = czv::slice::slice(
// None,
// Some(file_data),
// Some(start),
// Some(end),
// None,
// None,
// false,
// )?;
// assert_eq!(expected, got);
// }
// Ok(())
// }
assert_eq!(expected, got);
Ok(())
}
// #[test]
// fn test_slice_start_0_end_3() -> Result<()> {
// let expected = "apple,2.50\nbanana,3.00".to_string();
// let got: String = czv::slice::Slice::new()
// .file_path("tests/resources/fruits.csv")
// .start(0)
// .end(2) // exclusive
// .include_header_row(false)
// .execute()?;
#[test]
fn test_slice_index_2() -> Result<()> {
let expected = "strawberry,1.50".to_string();
let got: String = czv::slice::Slice::new()
.file_path("tests/resources/fruits.csv")
.index(2)
.include_header_row(false)
.execute()?;
// assert_eq!(expected, got);
// Ok(())
// }
assert_eq!(expected, got);
Ok(())
}
// #[test]
// fn test_slice_index_2() -> Result<()> {
// let expected = "strawberry,1.50".to_string();
// let got: String = czv::slice::Slice::new()
// .file_path("tests/resources/fruits.csv")
// .index(2)
// .include_header_row(false)
// .execute()?;
// assert_eq!(expected, got);
// Ok(())
// }
// #[test]
// fn test_slice_index_override() -> Result<()> {
// let got = czv::slice::Slice::new()
// .file_path("tests/resources/fruits.csv")
// .start(1)
// .end(3)
// .index(2)
// .include_header_row(false)
// .execute();
// // Error: CzvError(Cannot use index with start, end, or length.)
// assert!(got.is_err());
// Ok(())
// }