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