|
| 1 | +// Run clippy on a fixed set of crates and collect the warnings. |
| 2 | +// This helps observing the impact clippy changs have on a set of real-world code. |
| 3 | +// |
| 4 | +// When a new lint is introduced, we can search the results for new warnings and check for false |
| 5 | +// positives. |
| 6 | + |
| 7 | +#![cfg(feature = "lintcheck")] |
| 8 | +#![allow(clippy::filter_map)] |
| 9 | + |
| 10 | +use crate::clippy_project_root; |
| 11 | + |
| 12 | +use std::collections::HashMap; |
| 13 | +use std::process::Command; |
| 14 | +use std::{fmt, fs::write, path::PathBuf}; |
| 15 | + |
| 16 | +use clap::ArgMatches; |
| 17 | +use serde::{Deserialize, Serialize}; |
| 18 | +use serde_json::Value; |
| 19 | + |
| 20 | +// use this to store the crates when interacting with the crates.toml file |
| 21 | +#[derive(Debug, Serialize, Deserialize)] |
| 22 | +struct CrateList { |
| 23 | + crates: HashMap<String, Vec<String>>, |
| 24 | +} |
| 25 | + |
| 26 | +// crate data we stored in the toml, can have multiple versions per crate |
| 27 | +// A single TomlCrate is laster mapped to several CrateSources in that case |
| 28 | +struct TomlCrate { |
| 29 | + name: String, |
| 30 | + versions: Vec<String>, |
| 31 | +} |
| 32 | + |
| 33 | +// represents an archive we download from crates.io |
| 34 | +#[derive(Debug, Serialize, Deserialize, Eq, Hash, PartialEq)] |
| 35 | +struct CrateSource { |
| 36 | + name: String, |
| 37 | + version: String, |
| 38 | +} |
| 39 | + |
| 40 | +// represents the extracted sourcecode of a crate |
| 41 | +#[derive(Debug)] |
| 42 | +struct Crate { |
| 43 | + version: String, |
| 44 | + name: String, |
| 45 | + // path to the extracted sources that clippy can check |
| 46 | + path: PathBuf, |
| 47 | +} |
| 48 | + |
| 49 | +#[derive(Debug)] |
| 50 | +struct ClippyWarning { |
| 51 | + crate_name: String, |
| 52 | + crate_version: String, |
| 53 | + file: String, |
| 54 | + line: String, |
| 55 | + column: String, |
| 56 | + linttype: String, |
| 57 | + message: String, |
| 58 | +} |
| 59 | + |
| 60 | +impl std::fmt::Display for ClippyWarning { |
| 61 | + fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result { |
| 62 | + writeln!( |
| 63 | + f, |
| 64 | + r#"{}-{}/{}:{}:{} {} "{}""#, |
| 65 | + &self.crate_name, &self.crate_version, &self.file, &self.line, &self.column, &self.linttype, &self.message |
| 66 | + ) |
| 67 | + } |
| 68 | +} |
| 69 | + |
| 70 | +impl CrateSource { |
| 71 | + fn download_and_extract(&self) -> Crate { |
| 72 | + let extract_dir = PathBuf::from("target/lintcheck/crates"); |
| 73 | + let krate_download_dir = PathBuf::from("target/lintcheck/downloads"); |
| 74 | + |
| 75 | + // url to download the crate from crates.io |
| 76 | + let url = format!( |
| 77 | + "https://crates.io/api/v1/crates/{}/{}/download", |
| 78 | + self.name, self.version |
| 79 | + ); |
| 80 | + println!("Downloading and extracting {} {} from {}", self.name, self.version, url); |
| 81 | + let _ = std::fs::create_dir("target/lintcheck/"); |
| 82 | + let _ = std::fs::create_dir(&krate_download_dir); |
| 83 | + let _ = std::fs::create_dir(&extract_dir); |
| 84 | + |
| 85 | + let krate_file_path = krate_download_dir.join(format!("{}-{}.crate.tar.gz", &self.name, &self.version)); |
| 86 | + // don't download/extract if we already have done so |
| 87 | + if !krate_file_path.is_file() { |
| 88 | + // create a file path to download and write the crate data into |
| 89 | + let mut krate_dest = std::fs::File::create(&krate_file_path).unwrap(); |
| 90 | + let mut krate_req = ureq::get(&url).call().unwrap().into_reader(); |
| 91 | + // copy the crate into the file |
| 92 | + std::io::copy(&mut krate_req, &mut krate_dest).unwrap(); |
| 93 | + |
| 94 | + // unzip the tarball |
| 95 | + let ungz_tar = flate2::read::GzDecoder::new(std::fs::File::open(&krate_file_path).unwrap()); |
| 96 | + // extract the tar archive |
| 97 | + let mut archive = tar::Archive::new(ungz_tar); |
| 98 | + archive.unpack(&extract_dir).expect("Failed to extract!"); |
| 99 | + } |
| 100 | + // crate is extracted, return a new Krate object which contains the path to the extracted |
| 101 | + // sources that clippy can check |
| 102 | + Crate { |
| 103 | + version: self.version.clone(), |
| 104 | + name: self.name.clone(), |
| 105 | + path: extract_dir.join(format!("{}-{}/", self.name, self.version)), |
| 106 | + } |
| 107 | + } |
| 108 | +} |
| 109 | + |
| 110 | +impl Crate { |
| 111 | + fn run_clippy_lints(&self, cargo_clippy_path: &PathBuf) -> Vec<ClippyWarning> { |
| 112 | + println!("Linting {} {}...", &self.name, &self.version); |
| 113 | + let cargo_clippy_path = std::fs::canonicalize(cargo_clippy_path).unwrap(); |
| 114 | + |
| 115 | + let shared_target_dir = clippy_project_root().join("target/lintcheck/shared_target_dir/"); |
| 116 | + |
| 117 | + let all_output = std::process::Command::new(cargo_clippy_path) |
| 118 | + .env("CARGO_TARGET_DIR", shared_target_dir) |
| 119 | + // lint warnings will look like this: |
| 120 | + // src/cargo/ops/cargo_compile.rs:127:35: warning: usage of `FromIterator::from_iter` |
| 121 | + .args(&[ |
| 122 | + "--", |
| 123 | + "--message-format=json", |
| 124 | + "--", |
| 125 | + "--cap-lints=warn", |
| 126 | + "-Wclippy::pedantic", |
| 127 | + "-Wclippy::cargo", |
| 128 | + ]) |
| 129 | + .current_dir(&self.path) |
| 130 | + .output() |
| 131 | + .unwrap(); |
| 132 | + let stdout = String::from_utf8_lossy(&all_output.stdout); |
| 133 | + let output_lines = stdout.lines(); |
| 134 | + //dbg!(&output_lines); |
| 135 | + let warnings: Vec<ClippyWarning> = output_lines |
| 136 | + .into_iter() |
| 137 | + // get all clippy warnings |
| 138 | + .filter(|line| line.contains("clippy::")) |
| 139 | + .map(|json_msg| parse_json_message(json_msg, &self)) |
| 140 | + .collect(); |
| 141 | + warnings |
| 142 | + } |
| 143 | +} |
| 144 | + |
| 145 | +fn build_clippy() { |
| 146 | + Command::new("cargo") |
| 147 | + .arg("build") |
| 148 | + .output() |
| 149 | + .expect("Failed to build clippy!"); |
| 150 | +} |
| 151 | + |
| 152 | +// get a list of CrateSources we want to check from a "lintcheck_crates.toml" file. |
| 153 | +fn read_crates() -> Vec<CrateSource> { |
| 154 | + let toml_path = PathBuf::from("clippy_dev/lintcheck_crates.toml"); |
| 155 | + let toml_content: String = |
| 156 | + std::fs::read_to_string(&toml_path).unwrap_or_else(|_| panic!("Failed to read {}", toml_path.display())); |
| 157 | + let crate_list: CrateList = |
| 158 | + toml::from_str(&toml_content).unwrap_or_else(|e| panic!("Failed to parse {}: \n{}", toml_path.display(), e)); |
| 159 | + // parse the hashmap of the toml file into a list of crates |
| 160 | + let tomlcrates: Vec<TomlCrate> = crate_list |
| 161 | + .crates |
| 162 | + .into_iter() |
| 163 | + .map(|(name, versions)| TomlCrate { name, versions }) |
| 164 | + .collect(); |
| 165 | + |
| 166 | + // flatten TomlCrates into CrateSources (one TomlCrates may represent several versions of a crate => |
| 167 | + // multiple Cratesources) |
| 168 | + let mut crate_sources = Vec::new(); |
| 169 | + tomlcrates.into_iter().for_each(|tk| { |
| 170 | + tk.versions.iter().for_each(|ver| { |
| 171 | + crate_sources.push(CrateSource { |
| 172 | + name: tk.name.clone(), |
| 173 | + version: ver.to_string(), |
| 174 | + }); |
| 175 | + }) |
| 176 | + }); |
| 177 | + crate_sources |
| 178 | +} |
| 179 | + |
| 180 | +// extract interesting data from a json lint message |
| 181 | +fn parse_json_message(json_message: &str, krate: &Crate) -> ClippyWarning { |
| 182 | + let jmsg: Value = serde_json::from_str(&json_message).unwrap_or_else(|e| panic!("Failed to parse json:\n{:?}", e)); |
| 183 | + |
| 184 | + ClippyWarning { |
| 185 | + crate_name: krate.name.to_string(), |
| 186 | + crate_version: krate.version.to_string(), |
| 187 | + file: jmsg["message"]["spans"][0]["file_name"] |
| 188 | + .to_string() |
| 189 | + .trim_matches('"') |
| 190 | + .into(), |
| 191 | + line: jmsg["message"]["spans"][0]["line_start"] |
| 192 | + .to_string() |
| 193 | + .trim_matches('"') |
| 194 | + .into(), |
| 195 | + column: jmsg["message"]["spans"][0]["text"][0]["highlight_start"] |
| 196 | + .to_string() |
| 197 | + .trim_matches('"') |
| 198 | + .into(), |
| 199 | + linttype: jmsg["message"]["code"]["code"].to_string().trim_matches('"').into(), |
| 200 | + message: jmsg["message"]["message"].to_string().trim_matches('"').into(), |
| 201 | + } |
| 202 | +} |
| 203 | + |
| 204 | +// the main fn |
| 205 | +pub fn run(clap_config: &ArgMatches) { |
| 206 | + let cargo_clippy_path: PathBuf = PathBuf::from("target/debug/cargo-clippy"); |
| 207 | + |
| 208 | + println!("Compiling clippy..."); |
| 209 | + build_clippy(); |
| 210 | + println!("Done compiling"); |
| 211 | + |
| 212 | + // assert that clippy is found |
| 213 | + assert!( |
| 214 | + cargo_clippy_path.is_file(), |
| 215 | + "target/debug/cargo-clippy binary not found! {}", |
| 216 | + cargo_clippy_path.display() |
| 217 | + ); |
| 218 | + |
| 219 | + let clippy_ver = std::process::Command::new("target/debug/cargo-clippy") |
| 220 | + .arg("--version") |
| 221 | + .output() |
| 222 | + .map(|o| String::from_utf8_lossy(&o.stdout).into_owned()) |
| 223 | + .expect("could not get clippy version!"); |
| 224 | + |
| 225 | + // download and extract the crates, then run clippy on them and collect clippys warnings |
| 226 | + // flatten into one big list of warnings |
| 227 | + |
| 228 | + let crates = read_crates(); |
| 229 | + |
| 230 | + let clippy_warnings: Vec<ClippyWarning> = if let Some(only_one_crate) = clap_config.value_of("only") { |
| 231 | + // if we don't have the specified crated in the .toml, throw an error |
| 232 | + if !crates.iter().any(|krate| krate.name == only_one_crate) { |
| 233 | + eprintln!( |
| 234 | + "ERROR: could not find crate '{}' in clippy_dev/lintcheck_crates.toml", |
| 235 | + only_one_crate |
| 236 | + ); |
| 237 | + std::process::exit(1); |
| 238 | + } |
| 239 | + |
| 240 | + // only check a single crate that was passed via cmdline |
| 241 | + crates |
| 242 | + .into_iter() |
| 243 | + .map(|krate| krate.download_and_extract()) |
| 244 | + .filter(|krate| krate.name == only_one_crate) |
| 245 | + .map(|krate| krate.run_clippy_lints(&cargo_clippy_path)) |
| 246 | + .flatten() |
| 247 | + .collect() |
| 248 | + } else { |
| 249 | + // check all crates (default) |
| 250 | + crates |
| 251 | + .into_iter() |
| 252 | + .map(|krate| krate.download_and_extract()) |
| 253 | + .map(|krate| krate.run_clippy_lints(&cargo_clippy_path)) |
| 254 | + .flatten() |
| 255 | + .collect() |
| 256 | + }; |
| 257 | + |
| 258 | + // generate some stats: |
| 259 | + |
| 260 | + // count lint type occurrences |
| 261 | + let mut counter: HashMap<&String, usize> = HashMap::new(); |
| 262 | + clippy_warnings |
| 263 | + .iter() |
| 264 | + .for_each(|wrn| *counter.entry(&wrn.linttype).or_insert(0) += 1); |
| 265 | + |
| 266 | + // collect into a tupled list for sorting |
| 267 | + let mut stats: Vec<(&&String, &usize)> = counter.iter().map(|(lint, count)| (lint, count)).collect(); |
| 268 | + // sort by "000{count} {clippy::lintname}" |
| 269 | + // to not have a lint with 200 and 2 warnings take the same spot |
| 270 | + stats.sort_by_key(|(lint, count)| format!("{:0>4}, {}", count, lint)); |
| 271 | + |
| 272 | + let stats_formatted: String = stats |
| 273 | + .iter() |
| 274 | + .map(|(lint, count)| format!("{} {}\n", lint, count)) |
| 275 | + .collect::<String>(); |
| 276 | + |
| 277 | + let mut all_msgs: Vec<String> = clippy_warnings.iter().map(|warning| warning.to_string()).collect(); |
| 278 | + all_msgs.sort(); |
| 279 | + all_msgs.push("\n\n\n\nStats\n\n".into()); |
| 280 | + all_msgs.push(stats_formatted); |
| 281 | + |
| 282 | + // save the text into lintcheck-logs/logs.txt |
| 283 | + let mut text = clippy_ver; // clippy version number on top |
| 284 | + text.push_str(&format!("\n{}", all_msgs.join(""))); |
| 285 | + write("lintcheck-logs/logs.txt", text).unwrap(); |
| 286 | +} |
0 commit comments