feat: impl format tool.

kclvm/tools/Cargo.toml

@@ -8,10 +8,13 @@ edition = "2021"
 [dependencies]
 indexmap = "1.0"
 fancy-regex = "0.7.1"
+walkdir = "2"
+anyhow = "1.0"
 
 kclvm-ast = {path = "../ast", version = "0.1.0"}
 kclvm-error = {path = "../error", version = "0.1.0"}
 kclvm-parser = {path = "../parser", version = "0.1.0"}
+kclvm-config = {path = "../config", version = "0.1.0"}
 
 [dev-dependencies]
 pretty_assertions = "1.2.1"

kclvm/tools/src/format/mod.rs

@@ -0,0 +1,90 @@
+//! [kclvm_tools::format] module mainly contains some functions of language formatting,
+//! the main API function is `format`, which accepts a path to be formatted and
+//! formatted options.
+//!
+//! The basic principle is to call the [kclvm_parser::parse_file] function to parse the
+//! AST Module, and then use the AST printer [kclvm_tools::printer::print_ast_module]
+//! to print it as source code string.
+use anyhow::{anyhow, Result};
+use std::path::Path;
+
+use crate::{printer::print_ast_module, util::get_kcl_files};
+use kclvm_parser::parse_file;
+
+#[cfg(test)]
+mod tests;
+
+/// FormatOptions contains two options:
+/// - is_stdout: whether to output the formatted result to stdout.
+/// - recursively: whether to recursively traverse a folder and format all KCL files in it.
+#[derive(Debug, Default)]
+pub struct FormatOptions {
+    pub is_stdout: bool,
+    pub recursively: bool,
+}
+
+/// Formats kcl file or directory path contains kcl files and
+/// returns the changed file paths.
+///
+/// # Examples
+///
+/// ```no_run
+/// use kclvm_tools::format::{format, FormatOptions};
+///
+/// // Format a single file.
+/// format("path_to_a_single_file.k", &FormatOptions::default()).unwrap();
+/// // Format a folder contains kcl files
+/// format("path_to_a_folder", &FormatOptions::default()).unwrap();
+/// ```
+pub fn format<P: AsRef<Path>>(path: P, opts: &FormatOptions) -> Result<Vec<String>> {
+    let mut changed_paths: Vec<String> = vec![];
+    let path_ref = path.as_ref();
+    if path_ref.is_dir() {
+        for file in &get_kcl_files(path, opts.recursively)? {
+            if format_file(file, opts)? {
+                changed_paths.push(file.clone())
+            }
+        }
+    } else if path_ref.is_file() {
+        let file = path_ref.to_str().unwrap().to_string();
+        if format_file(&file, opts)? {
+            changed_paths.push(file)
+        }
+    }
+    if !opts.is_stdout {
+        let n = changed_paths.len();
+        println!(
+            "KCL format done and {} {} formatted:",
+            n,
+            if n <= 1 { "file was" } else { "files were" }
+        );
+        for p in &changed_paths {
+            println!("{}", p);
+        }
+    }
+    Ok(changed_paths)
+}
+
+/// Format a file and returns whether the file has been formatted and modified.
+fn format_file(file: &str, opts: &FormatOptions) -> Result<bool> {
+    let src = std::fs::read_to_string(file)?;
+    let (source, is_formatted) = format_source(&src)?;
+    if opts.is_stdout {
+        println!("{}", source);
+    } else {
+        std::fs::write(file, &source)?
+    }
+    Ok(is_formatted)
+}
+
+/// Format a code source and return the formatted source and
+/// whether the source is changed.
+fn format_source(src: &str) -> Result<(String, bool)> {
+    let module = match parse_file("", Some(src.to_string())) {
+        Ok(module) => module,
+        Err(err) => return Err(anyhow!("{}", err)),
+    };
+    let formatted_src = print_ast_module(&module);
+    let is_formatted = src != formatted_src;
+    Ok((formatted_src, is_formatted))
+}

kclvm/tools/src/format/test_data/format_data/assert.golden

@@ -0,0 +1,4 @@
+assert True if True, "message"
+# Comment
+assert False if data, "message"
+assert 1

kclvm/tools/src/format/test_data/format_data/assert.input

@@ -0,0 +1,3 @@
+assert True     if True, "message"
+assert  False   if   data   ,  "message"  # Comment
+assert   1

kclvm/tools/src/format/test_data/format_data/blankline.golden

@@ -0,0 +1,4 @@
+a = 1
+b = 2
+c = 3
+d = 4

kclvm/tools/src/format/test_data/format_data/blankline.input

@@ -0,0 +1,12 @@
+
+a=1
+
+
+b= 2
+
+
+c =3
+
+
+
+d  =  4

kclvm/tools/src/format/test_data/format_data/breakline.golden

@@ -0,0 +1,9 @@
+import math
+schema Base:
+    name: str
+
+schema Person(Base):
+    age: int
+
+person = Person {}
+

kclvm/tools/src/format/test_data/format_data/breakline.input

@@ -0,0 +1,6 @@
+import math
+schema Base:
+    name: str
+schema Person(Base):
+    age: int
+person = Person{}

kclvm/tools/src/format/test_data/format_data/check.golden

@@ -0,0 +1,13 @@
+schema Person:
+    firstName: str = "John"
+    lastName: str
+    times: int
+
+    check:
+        len(lastName) > 0 if times > 5
+
+JohnDoe = Person {
+    "lastName": "Doe"
+    "times": 10
+}
+

kclvm/tools/src/format/test_data/format_data/check.input

@@ -0,0 +1,12 @@
+
+schema Person:
+    firstName: str = "John"
+    lastName: str
+    times: int
+    check:
+        len(lastName)    > 0 if          times >     5
+
+JohnDoe = Person {
+    "lastName": "Doe"
+    "times":10
+}

kclvm/tools/src/format/test_data/format_data/codelayout.golden

@@ -0,0 +1,80 @@
+import math as alias_math
+schema Person(Base):
+    # inline comment
+    name: str
+    age: int
+
+    check:
+        age > 0 if age, "age must > 0"
+
+person = Person {
+    name: "Alice"
+    age: 18
+}
+
+if True:
+    a = 1
+elif True:
+    b = 2
+else:
+    c = 3
+d = 1 + 2
+e = (1 + 2)
+f = [1, 2, 3]
+g = {"key": "value"}
+# block comment
+print(1)
+dct = {"key": "value"}
+lst = [1, 2, 3]
+h = dct['key']
+i = lst[1]
+x = 1
+y = 2
+long_variable = 3
+i = i + 1
+submitted += 1
+x = x * 2 - 1
+hypot2 = x * x + y * y
+_c = (a + b) * (a - b)
+_b = 2
+_c = 3
+_d = 4
+_value = (1 + 2 * 3)
+_value = (1 + 2 * 3)
+_value = 1 + -2 * ~3
+_list = [1, 2, 3]
+_list = [*_list, [4, 5, 6]]
+_list = [*_list, [4, 5, 6]]
+_dict = {**{"k": "v"}, **{"k": "v"}}
+a = [1, 2, 3]
+b = [
+    1
+    2
+    3
+    4
+    5
+    6
+]
+_dict = {
+    "k1": "v1"
+    "k2": "v2"
+    "k3": "v3"
+    "k4": "v4"
+    "k5": "v5"
+}
+foo = 1
+if foo is not None:
+    _a = 1
+    _dict |= {}
+    hello = "world{}".format(1)[2:4:].lower()
+    range_int = [i for i in range(10) if i > 1]
+
+op = 1 + 2 - -3 + (3 - 1) // 3
+op += 1
+op -= 12 + 23
+print("  ", end='')
+log = math.log(12)
+aa = 1
+assert aa == 1 if aa, "message"
+aaaa = (1 + 2 / 2) if _a == 2 + +134.3 else ("a" * 3)
+bbbb = "{}".format(a)