rust/clippy_lints/src/write.rs

476 lines
16 KiB
Rust
Raw Normal View History

use crate::utils::{snippet_with_applicability, span_lint, span_lint_and_sugg, span_lint_and_then};
use rustc::lint::{EarlyContext, EarlyLintPass, LintArray, LintPass};
2019-04-08 15:43:55 -05:00
use rustc::{declare_lint_pass, declare_tool_lint};
use rustc_errors::Applicability;
use std::borrow::Cow;
use syntax::ast::*;
use syntax::parse::{parser, token};
use syntax::tokenstream::TokenStream;
use syntax_pos::{BytePos, Span};
declare_clippy_lint! {
/// **What it does:** This lint warns when you use `println!("")` to
/// print a newline.
///
/// **Why is this bad?** You should use `println!()`, which is simpler.
///
/// **Known problems:** None.
///
/// **Example:**
/// ```rust
/// println!("");
/// ```
pub PRINTLN_EMPTY_STRING,
style,
"using `println!(\"\")` with an empty string"
}
declare_clippy_lint! {
/// **What it does:** This lint warns when you use `print!()` with a format
/// string that
/// ends in a newline.
///
/// **Why is this bad?** You should use `println!()` instead, which appends the
/// newline.
///
/// **Known problems:** None.
///
/// **Example:**
/// ```rust
/// # let name = "World";
/// print!("Hello {}!\n", name);
/// ```
/// use println!() instead
/// ```rust
/// # let name = "World";
/// println!("Hello {}!", name);
/// ```
pub PRINT_WITH_NEWLINE,
style,
"using `print!()` with a format string that ends in a single newline"
}
declare_clippy_lint! {
/// **What it does:** Checks for printing on *stdout*. The purpose of this lint
/// is to catch debugging remnants.
///
/// **Why is this bad?** People often print on *stdout* while debugging an
/// application and might forget to remove those prints afterward.
///
/// **Known problems:** Only catches `print!` and `println!` calls.
///
/// **Example:**
/// ```rust
/// println!("Hello world!");
/// ```
pub PRINT_STDOUT,
restriction,
"printing on stdout"
}
declare_clippy_lint! {
/// **What it does:** Checks for use of `Debug` formatting. The purpose of this
/// lint is to catch debugging remnants.
///
/// **Why is this bad?** The purpose of the `Debug` trait is to facilitate
/// debugging Rust code. It should not be used in in user-facing output.
///
/// **Example:**
/// ```rust
/// println!("{:?}", foo);
/// ```
pub USE_DEBUG,
restriction,
"use of `Debug`-based formatting"
}
declare_clippy_lint! {
/// **What it does:** This lint warns about the use of literals as `print!`/`println!` args.
///
/// **Why is this bad?** Using literals as `println!` args is inefficient
/// (c.f., https://github.com/matthiaskrgr/rust-str-bench) and unnecessary
/// (i.e., just put the literal in the format string)
///
/// **Known problems:** Will also warn with macro calls as arguments that expand to literals
/// -- e.g., `println!("{}", env!("FOO"))`.
///
/// **Example:**
/// ```rust
/// println!("{}", "foo");
/// ```
/// use the literal without formatting:
/// ```rust
/// println!("foo");
/// ```
pub PRINT_LITERAL,
style,
"printing a literal with a format string"
}
declare_clippy_lint! {
/// **What it does:** This lint warns when you use `writeln!(buf, "")` to
/// print a newline.
///
/// **Why is this bad?** You should use `writeln!(buf)`, which is simpler.
///
/// **Known problems:** None.
///
/// **Example:**
/// ```rust
/// # use std::fmt::Write;
/// # let mut buf = String::new();
2019-03-05 16:23:50 -06:00
/// writeln!(buf, "");
/// ```
pub WRITELN_EMPTY_STRING,
style,
2019-03-05 16:23:50 -06:00
"using `writeln!(buf, \"\")` with an empty string"
}
declare_clippy_lint! {
/// **What it does:** This lint warns when you use `write!()` with a format
/// string that
/// ends in a newline.
///
/// **Why is this bad?** You should use `writeln!()` instead, which appends the
/// newline.
///
/// **Known problems:** None.
///
/// **Example:**
/// ```rust
/// # use std::fmt::Write;
/// # let mut buf = String::new();
/// # let name = "World";
/// write!(buf, "Hello {}!\n", name);
/// ```
pub WRITE_WITH_NEWLINE,
style,
"using `write!()` with a format string that ends in a single newline"
}
declare_clippy_lint! {
/// **What it does:** This lint warns about the use of literals as `write!`/`writeln!` args.
///
/// **Why is this bad?** Using literals as `writeln!` args is inefficient
/// (c.f., https://github.com/matthiaskrgr/rust-str-bench) and unnecessary
/// (i.e., just put the literal in the format string)
///
/// **Known problems:** Will also warn with macro calls as arguments that expand to literals
/// -- e.g., `writeln!(buf, "{}", env!("FOO"))`.
///
/// **Example:**
/// ```rust
/// # use std::fmt::Write;
/// # let mut buf = String::new();
/// writeln!(buf, "{}", "foo");
/// ```
pub WRITE_LITERAL,
style,
"writing a literal with a format string"
}
2019-04-08 15:43:55 -05:00
declare_lint_pass!(Write => [
PRINT_WITH_NEWLINE,
PRINTLN_EMPTY_STRING,
PRINT_STDOUT,
USE_DEBUG,
PRINT_LITERAL,
WRITE_WITH_NEWLINE,
WRITELN_EMPTY_STRING,
WRITE_LITERAL
]);
2019-04-08 15:43:55 -05:00
impl EarlyLintPass for Write {
2018-07-23 06:01:12 -05:00
fn check_mac(&mut self, cx: &EarlyContext<'_>, mac: &Mac) {
2019-05-17 16:53:54 -05:00
if mac.node.path == sym!(println) {
span_lint(cx, PRINT_STDOUT, mac.span, "use of `println!`");
if let (Some(fmt_str), _) = check_tts(cx, &mac.node.tts, false) {
if fmt_str.contents.is_empty() {
span_lint_and_sugg(
cx,
PRINTLN_EMPTY_STRING,
mac.span,
"using `println!(\"\")`",
"replace it with",
"println!()".to_string(),
Applicability::MachineApplicable,
);
}
}
2019-05-17 16:53:54 -05:00
} else if mac.node.path == sym!(print) {
span_lint(cx, PRINT_STDOUT, mac.span, "use of `print!`");
if let (Some(fmt_str), _) = check_tts(cx, &mac.node.tts, false) {
if check_newlines(&fmt_str) {
span_lint_and_then(
2018-09-06 05:33:00 -05:00
cx,
PRINT_WITH_NEWLINE,
mac.span,
"using `print!()` with a format string that ends in a single newline",
|err| {
err.multipart_suggestion(
"use `println!` instead",
vec![
(mac.node.path.span, String::from("println")),
(fmt_str.newline_span(), String::new()),
],
Applicability::MachineApplicable,
);
},
2018-09-06 05:33:00 -05:00
);
}
}
2019-05-17 16:53:54 -05:00
} else if mac.node.path == sym!(write) {
if let (Some(fmt_str), _) = check_tts(cx, &mac.node.tts, true) {
if check_newlines(&fmt_str) {
span_lint_and_then(
2018-09-06 05:33:00 -05:00
cx,
WRITE_WITH_NEWLINE,
mac.span,
"using `write!()` with a format string that ends in a single newline",
|err| {
err.multipart_suggestion(
"use `writeln!()` instead",
vec![
(mac.node.path.span, String::from("writeln")),
(fmt_str.newline_span(), String::new()),
],
Applicability::MachineApplicable,
);
},
)
}
}
2019-05-17 16:53:54 -05:00
} else if mac.node.path == sym!(writeln) {
if let (Some(fmt_str), expr) = check_tts(cx, &mac.node.tts, true) {
if fmt_str.contents.is_empty() {
let mut applicability = Applicability::MachineApplicable;
let suggestion = expr.map_or_else(
move || {
applicability = Applicability::HasPlaceholders;
Cow::Borrowed("v")
},
move |expr| snippet_with_applicability(cx, expr.span, "v", &mut applicability),
);
span_lint_and_sugg(
cx,
WRITELN_EMPTY_STRING,
mac.span,
2018-08-20 08:50:15 -05:00
format!("using `writeln!({}, \"\")`", suggestion).as_str(),
"replace it with",
format!("writeln!({})", suggestion),
applicability,
);
}
}
}
}
}
/// The arguments of a `print[ln]!` or `write[ln]!` invocation.
struct FmtStr {
/// The contents of the format string (inside the quotes).
contents: String,
style: StrStyle,
/// The span of the format string, including quotes, the raw marker, and any raw hashes.
span: Span,
}
impl FmtStr {
/// Given a format string that ends in a newline and its span, calculates the span of the
/// newline.
fn newline_span(&self) -> Span {
let sp = self.span;
let newline_sp_hi = sp.hi()
- match self.style {
StrStyle::Cooked => BytePos(1),
StrStyle::Raw(hashes) => BytePos((1 + hashes).into()),
};
let newline_sp_len = if self.contents.ends_with('\n') {
BytePos(1)
} else if self.contents.ends_with(r"\n") {
BytePos(2)
} else {
panic!("expected format string to contain a newline");
};
sp.with_lo(newline_sp_hi - newline_sp_len).with_hi(newline_sp_hi)
}
}
/// Checks the arguments of `print[ln]!` and `write[ln]!` calls. It will return a tuple of two
/// `Option`s. The first `Option` of the tuple is the macro's format string. It includes
/// the contents of the string, whether it's a raw string, and the span of the literal in the
/// source. The second `Option` in the tuple is, in the `write[ln]!` case, the expression the
/// `format_str` should be written to.
///
/// Example:
///
/// Calling this function on
/// ```rust
/// # use std::fmt::Write;
/// # let mut buf = String::new();
/// # let something = "something";
/// writeln!(buf, "string to write: {}", something);
/// ```
/// will return
/// ```rust,ignore
/// (Some("string to write: {}"), Some(buf))
/// ```
fn check_tts<'a>(cx: &EarlyContext<'a>, tts: &TokenStream, is_write: bool) -> (Option<FmtStr>, Option<Expr>) {
use fmt_macros::*;
let tts = tts.clone();
let mut parser = parser::Parser::new(&cx.sess.parse_sess, tts, None, false, false, None);
let mut expr: Option<Expr> = None;
if is_write {
expr = match parser.parse_expr().map_err(|mut err| err.cancel()) {
Ok(p) => Some(p.into_inner()),
Err(_) => return (None, None),
};
// might be `writeln!(foo)`
if parser.expect(&token::Comma).map_err(|mut err| err.cancel()).is_err() {
return (None, expr);
}
}
let (fmtstr, fmtstyle) = match parser.parse_str().map_err(|mut err| err.cancel()) {
Ok((fmtstr, fmtstyle)) => (fmtstr.to_string(), fmtstyle),
Err(_) => return (None, expr),
};
let fmtspan = parser.prev_span;
let tmp = fmtstr.clone();
let mut args = vec![];
let mut fmt_parser = Parser::new(&tmp, None, Vec::new(), false);
while let Some(piece) = fmt_parser.next() {
if !fmt_parser.errors.is_empty() {
return (None, expr);
}
if let Piece::NextArgument(arg) = piece {
if arg.format.ty == "?" {
// FIXME: modify rustc's fmt string parser to give us the current span
span_lint(cx, USE_DEBUG, parser.prev_span, "use of `Debug`-based formatting");
}
args.push(arg);
}
}
2018-09-06 05:33:00 -05:00
let lint = if is_write { WRITE_LITERAL } else { PRINT_LITERAL };
let mut idx = 0;
loop {
2018-07-23 06:01:12 -05:00
const SIMPLE: FormatSpec<'_> = FormatSpec {
fill: None,
align: AlignUnknown,
flags: 0,
precision: CountImplied,
precision_span: None,
width: CountImplied,
width_span: None,
ty: "",
};
if !parser.eat(&token::Comma) {
return (
Some(FmtStr {
contents: fmtstr,
style: fmtstyle,
span: fmtspan,
}),
expr,
);
}
let token_expr = if let Ok(expr) = parser.parse_expr().map_err(|mut err| err.cancel()) {
expr
} else {
return (
Some(FmtStr {
contents: fmtstr,
style: fmtstyle,
span: fmtspan,
}),
None,
);
};
match &token_expr.node {
ExprKind::Lit(_) => {
let mut all_simple = true;
let mut seen = false;
for arg in &args {
match arg.position {
2018-11-27 14:14:15 -06:00
ArgumentImplicitlyIs(n) | ArgumentIs(n) => {
if n == idx {
all_simple &= arg.format == SIMPLE;
seen = true;
}
},
ArgumentNamed(_) => {},
}
}
if all_simple && seen {
span_lint(cx, lint, token_expr.span, "literal with an empty format string");
}
idx += 1;
},
ExprKind::Assign(lhs, rhs) => {
2018-07-25 16:31:17 -05:00
if let ExprKind::Lit(_) = rhs.node {
if let ExprKind::Path(_, p) = &lhs.node {
let mut all_simple = true;
let mut seen = false;
for arg in &args {
match arg.position {
2018-09-06 05:33:00 -05:00
ArgumentImplicitlyIs(_) | ArgumentIs(_) => {},
2018-11-27 14:14:15 -06:00
ArgumentNamed(name) => {
if *p == name {
2018-11-27 14:14:15 -06:00
seen = true;
all_simple &= arg.format == SIMPLE;
}
2018-07-25 16:31:17 -05:00
},
}
}
if all_simple && seen {
span_lint(cx, lint, rhs.span, "literal with an empty format string");
}
}
}
},
_ => idx += 1,
}
}
}
/// Checks if the format string constains a single newline that terminates it.
///
/// Literal and escaped newlines are both checked (only literal for raw strings).
fn check_newlines(fmt_str: &FmtStr) -> bool {
let s = &fmt_str.contents;
if s.ends_with('\n') {
return true;
} else if let StrStyle::Raw(_) = fmt_str.style {
return false;
}
if s.len() < 2 {
return false;
}
let bytes = s.as_bytes();
if bytes[bytes.len() - 2] != b'\\' || bytes[bytes.len() - 1] != b'n' {
return false;
}
let mut escaping = false;
for (index, &byte) in bytes.iter().enumerate() {
if escaping {
if byte == b'n' {
return index == bytes.len() - 1;
}
escaping = false;
} else if byte == b'\\' {
escaping = true;
}
}
false
}