src/tools/rust-analyzer/crates/rust-analyzer/src/cli/analysis_stats.rs

   1 //! Fully type-check project and print various stats, like the number of type
   2 //! errors.
   3
   4 use std::{
   5     env,
   6     time::{SystemTime, UNIX_EPOCH},
   7 };
   8
   9 use hir::{
  10     db::{AstDatabase, DefDatabase, HirDatabase},
  11     AssocItem, Crate, Function, HasSource, HirDisplay, ModuleDef,
  12 };
  13 use hir_def::{
  14     body::{BodySourceMap, SyntheticSyntax},
  15     expr::ExprId,
  16     FunctionId,
  17 };
  18 use hir_ty::{TyExt, TypeWalk};
  19 use ide::{Analysis, AnalysisHost, LineCol, RootDatabase};
  20 use ide_db::base_db::{
  21     salsa::{self, debug::DebugQueryTable, ParallelDatabase},
  22     SourceDatabase, SourceDatabaseExt,
  23 };
  24 use itertools::Itertools;
  25 use oorandom::Rand32;
  26 use profile::{Bytes, StopWatch};
  27 use project_model::{CargoConfig, ProjectManifest, ProjectWorkspace};
  28 use rayon::prelude::*;
  29 use rustc_hash::FxHashSet;
  30 use stdx::format_to;
  31 use syntax::{AstNode, SyntaxNode};
  32 use vfs::{AbsPathBuf, Vfs, VfsPath};
  33
  34 use crate::cli::{
  35     flags::{self, OutputFormat},
  36     load_cargo::{load_workspace, LoadCargoConfig},
  37     print_memory_usage,
  38     progress_report::ProgressReport,
  39     report_metric, Result, Verbosity,
  40 };
  41
  42 /// Need to wrap Snapshot to provide `Clone` impl for `map_with`
  43 struct Snap<DB>(DB);
  44 impl<DB: ParallelDatabase> Clone for Snap<salsa::Snapshot<DB>> {
  45     fn clone(&self) -> Snap<salsa::Snapshot<DB>> {
  46         Snap(self.0.snapshot())
  47     }
  48 }
  49
  50 impl flags::AnalysisStats {
  51     pub fn run(self, verbosity: Verbosity) -> Result<()> {
  52         let mut rng = {
  53             let seed = SystemTime::now().duration_since(UNIX_EPOCH).unwrap().as_millis() as u64;
  54             Rand32::new(seed)
  55         };
  56
  57         let mut cargo_config = CargoConfig::default();
  58         cargo_config.no_sysroot = self.no_sysroot;
  59         let load_cargo_config = LoadCargoConfig {
  60             load_out_dirs_from_check: !self.disable_build_scripts,
  61             with_proc_macro: !self.disable_proc_macros,
  62             prefill_caches: false,
  63         };
  64         let no_progress = &|_| ();
  65
  66         let mut db_load_sw = self.stop_watch();
  67
  68         let path = AbsPathBuf::assert(env::current_dir()?.join(&self.path));
  69         let manifest = ProjectManifest::discover_single(&path)?;
  70
  71         let mut workspace = ProjectWorkspace::load(manifest, &cargo_config, no_progress)?;
  72         let metadata_time = db_load_sw.elapsed();
  73
  74         let build_scripts_time = if self.disable_build_scripts {
  75             None
  76         } else {
  77             let mut build_scripts_sw = self.stop_watch();
  78             let bs = workspace.run_build_scripts(&cargo_config, no_progress)?;
  79             workspace.set_build_scripts(bs);
  80             Some(build_scripts_sw.elapsed())
  81         };
  82
  83         let (host, vfs, _proc_macro) =
  84             load_workspace(workspace, &cargo_config, &load_cargo_config)?;
  85         let db = host.raw_database();
  86         eprint!("{:<20} {}", "Database loaded:", db_load_sw.elapsed());
  87         eprint!(" (metadata {}", metadata_time);
  88         if let Some(build_scripts_time) = build_scripts_time {
  89             eprint!("; build {}", build_scripts_time);
  90         }
  91         eprintln!(")");
  92
  93         let mut analysis_sw = self.stop_watch();
  94         let mut num_crates = 0;
  95         let mut visited_modules = FxHashSet::default();
  96         let mut visit_queue = Vec::new();
  97
  98         let mut krates = Crate::all(db);
  99         if self.randomize {
 100             shuffle(&mut rng, &mut krates);
 101         }
 102         for krate in krates {
 103             let module = krate.root_module(db);
 104             let file_id = module.definition_source(db).file_id;
 105             let file_id = file_id.original_file(db);
 106             let source_root = db.file_source_root(file_id);
 107             let source_root = db.source_root(source_root);
 108             if !source_root.is_library || self.with_deps {
 109                 num_crates += 1;
 110                 visit_queue.push(module);
 111             }
 112         }
 113
 114         if self.randomize {
 115             shuffle(&mut rng, &mut visit_queue);
 116         }
 117
 118         eprint!("  crates: {}", num_crates);
 119         let mut num_decls = 0;
 120         let mut funcs = Vec::new();
 121         while let Some(module) = visit_queue.pop() {
 122             if visited_modules.insert(module) {
 123                 visit_queue.extend(module.children(db));
 124
 125                 for decl in module.declarations(db) {
 126                     num_decls += 1;
 127                     if let ModuleDef::Function(f) = decl {
 128                         funcs.push(f);
 129                     }
 130                 }
 131
 132                 for impl_def in module.impl_defs(db) {
 133                     for item in impl_def.items(db) {
 134                         num_decls += 1;
 135                         if let AssocItem::Function(f) = item {
 136                             funcs.push(f);
 137                         }
 138                     }
 139                 }
 140             }
 141         }
 142         eprintln!(", mods: {}, decls: {}, fns: {}", visited_modules.len(), num_decls, funcs.len());
 143         eprintln!("{:<20} {}", "Item Collection:", analysis_sw.elapsed());
 144
 145         if self.randomize {
 146             shuffle(&mut rng, &mut funcs);
 147         }
 148
 149         if !self.skip_inference {
 150             self.run_inference(&host, db, &vfs, &funcs, verbosity);
 151         }
 152
 153         let total_span = analysis_sw.elapsed();
 154         eprintln!("{:<20} {}", "Total:", total_span);
 155         report_metric("total time", total_span.time.as_millis() as u64, "ms");
 156         if let Some(instructions) = total_span.instructions {
 157             report_metric("total instructions", instructions, "#instr");
 158         }
 159         if let Some(memory) = total_span.memory {
 160             report_metric("total memory", memory.allocated.megabytes() as u64, "MB");
 161         }
 162
 163         if env::var("RA_COUNT").is_ok() {
 164             eprintln!("{}", profile::countme::get_all());
 165         }
 166
 167         if self.source_stats {
 168             let mut total_file_size = Bytes::default();
 169             for e in ide_db::base_db::ParseQuery.in_db(db).entries::<Vec<_>>() {
 170                 total_file_size += syntax_len(db.parse(e.key).syntax_node())
 171             }
 172
 173             let mut total_macro_file_size = Bytes::default();
 174             for e in hir::db::ParseMacroExpansionQuery.in_db(db).entries::<Vec<_>>() {
 175                 if let Some((val, _)) = db.parse_macro_expansion(e.key).value {
 176                     total_macro_file_size += syntax_len(val.syntax_node())
 177                 }
 178             }
 179             eprintln!("source files: {}, macro files: {}", total_file_size, total_macro_file_size);
 180         }
 181
 182         if self.memory_usage && verbosity.is_verbose() {
 183             print_memory_usage(host, vfs);
 184         }
 185
 186         Ok(())
 187     }
 188
 189     fn run_inference(
 190         &self,
 191         host: &AnalysisHost,
 192         db: &RootDatabase,
 193         vfs: &Vfs,
 194         funcs: &[Function],
 195         verbosity: Verbosity,
 196     ) {
 197         let mut bar = match verbosity {
 198             Verbosity::Quiet | Verbosity::Spammy => ProgressReport::hidden(),
 199             _ if self.parallel || self.output.is_some() => ProgressReport::hidden(),
 200             _ => ProgressReport::new(funcs.len() as u64),
 201         };
 202
 203         if self.parallel {
 204             let mut inference_sw = self.stop_watch();
 205             let snap = Snap(db.snapshot());
 206             funcs
 207                 .par_iter()
 208                 .map_with(snap, |snap, &f| {
 209                     let f_id = FunctionId::from(f);
 210                     snap.0.body(f_id.into());
 211                     snap.0.infer(f_id.into());
 212                 })
 213                 .count();
 214             eprintln!("{:<20} {}", "Parallel Inference:", inference_sw.elapsed());
 215         }
 216
 217         let mut inference_sw = self.stop_watch();
 218         bar.tick();
 219         let mut num_exprs = 0;
 220         let mut num_exprs_unknown = 0;
 221         let mut num_exprs_partially_unknown = 0;
 222         let mut num_type_mismatches = 0;
 223         let analysis = host.analysis();
 224         for f in funcs.iter().copied() {
 225             let name = f.name(db);
 226             let full_name = f
 227                 .module(db)
 228                 .path_to_root(db)
 229                 .into_iter()
 230                 .rev()
 231                 .filter_map(|it| it.name(db))
 232                 .chain(Some(f.name(db)))
 233                 .join("::");
 234             if let Some(only_name) = self.only.as_deref() {
 235                 if name.to_string() != only_name && full_name != only_name {
 236                     continue;
 237                 }
 238             }
 239             let mut msg = format!("processing: {}", full_name);
 240             if verbosity.is_verbose() {
 241                 if let Some(src) = f.source(db) {
 242                     let original_file = src.file_id.original_file(db);
 243                     let path = vfs.file_path(original_file);
 244                     let syntax_range = src.value.syntax().text_range();
 245                     format_to!(msg, " ({} {:?})", path, syntax_range);
 246                 }
 247             }
 248             if verbosity.is_spammy() {
 249                 bar.println(msg.to_string());
 250             }
 251             bar.set_message(&msg);
 252             let f_id = FunctionId::from(f);
 253             let (body, sm) = db.body_with_source_map(f_id.into());
 254             let inference_result = db.infer(f_id.into());
 255             let (previous_exprs, previous_unknown, previous_partially_unknown) =
 256                 (num_exprs, num_exprs_unknown, num_exprs_partially_unknown);
 257             for (expr_id, _) in body.exprs.iter() {
 258                 let ty = &inference_result[expr_id];
 259                 num_exprs += 1;
 260                 let unknown_or_partial = if ty.is_unknown() {
 261                     num_exprs_unknown += 1;
 262                     if verbosity.is_spammy() {
 263                         if let Some((path, start, end)) =
 264                             expr_syntax_range(db, &analysis, vfs, &sm, expr_id)
 265                         {
 266                             bar.println(format!(
 267                                 "{} {}:{}-{}:{}: Unknown type",
 268                                 path,
 269                                 start.line + 1,
 270                                 start.col,
 271                                 end.line + 1,
 272                                 end.col,
 273                             ));
 274                         } else {
 275                             bar.println(format!("{}: Unknown type", name,));
 276                         }
 277                     }
 278                     true
 279                 } else {
 280                     let mut is_partially_unknown = false;
 281                     ty.walk(&mut |ty| {
 282                         if ty.is_unknown() {
 283                             is_partially_unknown = true;
 284                         }
 285                     });
 286                     if is_partially_unknown {
 287                         num_exprs_partially_unknown += 1;
 288                     }
 289                     is_partially_unknown
 290                 };
 291                 if self.only.is_some() && verbosity.is_spammy() {
 292                     // in super-verbose mode for just one function, we print every single expression
 293                     if let Some((_, start, end)) =
 294                         expr_syntax_range(db, &analysis, vfs, &sm, expr_id)
 295                     {
 296                         bar.println(format!(
 297                             "{}:{}-{}:{}: {}",
 298                             start.line + 1,
 299                             start.col,
 300                             end.line + 1,
 301                             end.col,
 302                             ty.display(db)
 303                         ));
 304                     } else {
 305                         bar.println(format!("unknown location: {}", ty.display(db)));
 306                     }
 307                 }
 308                 if unknown_or_partial && self.output == Some(OutputFormat::Csv) {
 309                     println!(
 310                         r#"{},type,"{}""#,
 311                         location_csv(db, &analysis, vfs, &sm, expr_id),
 312                         ty.display(db)
 313                     );
 314                 }
 315                 if let Some(mismatch) = inference_result.type_mismatch_for_expr(expr_id) {
 316                     num_type_mismatches += 1;
 317                     if verbosity.is_verbose() {
 318                         if let Some((path, start, end)) =
 319                             expr_syntax_range(db, &analysis, vfs, &sm, expr_id)
 320                         {
 321                             bar.println(format!(
 322                                 "{} {}:{}-{}:{}: Expected {}, got {}",
 323                                 path,
 324                                 start.line + 1,
 325                                 start.col,
 326                                 end.line + 1,
 327                                 end.col,
 328                                 mismatch.expected.display(db),
 329                                 mismatch.actual.display(db)
 330                             ));
 331                         } else {
 332                             bar.println(format!(
 333                                 "{}: Expected {}, got {}",
 334                                 name,
 335                                 mismatch.expected.display(db),
 336                                 mismatch.actual.display(db)
 337                             ));
 338                         }
 339                     }
 340                     if self.output == Some(OutputFormat::Csv) {
 341                         println!(
 342                             r#"{},mismatch,"{}","{}""#,
 343                             location_csv(db, &analysis, vfs, &sm, expr_id),
 344                             mismatch.expected.display(db),
 345                             mismatch.actual.display(db)
 346                         );
 347                     }
 348                 }
 349             }
 350             if verbosity.is_spammy() {
 351                 bar.println(format!(
 352                     "In {}: {} exprs, {} unknown, {} partial",
 353                     full_name,
 354                     num_exprs - previous_exprs,
 355                     num_exprs_unknown - previous_unknown,
 356                     num_exprs_partially_unknown - previous_partially_unknown
 357                 ));
 358             }
 359             bar.inc(1);
 360         }
 361
 362         bar.finish_and_clear();
 363         eprintln!(
 364             "  exprs: {}, ??ty: {} ({}%), ?ty: {} ({}%), !ty: {}",
 365             num_exprs,
 366             num_exprs_unknown,
 367             percentage(num_exprs_unknown, num_exprs),
 368             num_exprs_partially_unknown,
 369             percentage(num_exprs_partially_unknown, num_exprs),
 370             num_type_mismatches
 371         );
 372         report_metric("unknown type", num_exprs_unknown, "#");
 373         report_metric("type mismatches", num_type_mismatches, "#");
 374
 375         eprintln!("{:<20} {}", "Inference:", inference_sw.elapsed());
 376     }
 377
 378     fn stop_watch(&self) -> StopWatch {
 379         StopWatch::start().memory(self.memory_usage)
 380     }
 381 }
 382
 383 fn location_csv(
 384     db: &RootDatabase,
 385     analysis: &Analysis,
 386     vfs: &Vfs,
 387     sm: &BodySourceMap,
 388     expr_id: ExprId,
 389 ) -> String {
 390     let src = match sm.expr_syntax(expr_id) {
 391         Ok(s) => s,
 392         Err(SyntheticSyntax) => return "synthetic,,".to_string(),
 393     };
 394     let root = db.parse_or_expand(src.file_id).unwrap();
 395     let node = src.map(|e| e.to_node(&root).syntax().clone());
 396     let original_range = node.as_ref().original_file_range(db);
 397     let path = vfs.file_path(original_range.file_id);
 398     let line_index = analysis.file_line_index(original_range.file_id).unwrap();
 399     let text_range = original_range.range;
 400     let (start, end) =
 401         (line_index.line_col(text_range.start()), line_index.line_col(text_range.end()));
 402     format!("{},{}:{},{}:{}", path, start.line + 1, start.col, end.line + 1, end.col)
 403 }
 404
 405 fn expr_syntax_range(
 406     db: &RootDatabase,
 407     analysis: &Analysis,
 408     vfs: &Vfs,
 409     sm: &BodySourceMap,
 410     expr_id: ExprId,
 411 ) -> Option<(VfsPath, LineCol, LineCol)> {
 412     let src = sm.expr_syntax(expr_id);
 413     if let Ok(src) = src {
 414         let root = db.parse_or_expand(src.file_id).unwrap();
 415         let node = src.map(|e| e.to_node(&root).syntax().clone());
 416         let original_range = node.as_ref().original_file_range(db);
 417         let path = vfs.file_path(original_range.file_id);
 418         let line_index = analysis.file_line_index(original_range.file_id).unwrap();
 419         let text_range = original_range.range;
 420         let (start, end) =
 421             (line_index.line_col(text_range.start()), line_index.line_col(text_range.end()));
 422         Some((path, start, end))
 423     } else {
 424         None
 425     }
 426 }
 427
 428 fn shuffle<T>(rng: &mut Rand32, slice: &mut [T]) {
 429     for i in 0..slice.len() {
 430         randomize_first(rng, &mut slice[i..]);
 431     }
 432
 433     fn randomize_first<T>(rng: &mut Rand32, slice: &mut [T]) {
 434         assert!(!slice.is_empty());
 435         let idx = rng.rand_range(0..slice.len() as u32) as usize;
 436         slice.swap(0, idx);
 437     }
 438 }
 439
 440 fn percentage(n: u64, total: u64) -> u64 {
 441     (n * 100).checked_div(total).unwrap_or(100)
 442 }
 443
 444 fn syntax_len(node: SyntaxNode) -> usize {
 445     // Macro expanded code doesn't contain whitespace, so erase *all* whitespace
 446     // to make macro and non-macro code comparable.
 447     node.to_string().replace(|it: char| it.is_ascii_whitespace(), "").len()
 448 }