src/librustdoc/scrape_examples.rs

   1 //! This module analyzes crates to find call sites that can serve as examples in the documentation.
   2
   3 use crate::clean;
   4 use crate::config;
   5 use crate::formats;
   6 use crate::formats::renderer::FormatRenderer;
   7 use crate::html::render::Context;
   8
   9 use rustc_data_structures::fx::FxHashMap;
  10 use rustc_hir::{
  11     self as hir,
  12     intravisit::{self, Visitor},
  13 };
  14 use rustc_interface::interface;
  15 use rustc_macros::{Decodable, Encodable};
  16 use rustc_middle::hir::map::Map;
  17 use rustc_middle::hir::nested_filter;
  18 use rustc_middle::ty::{self, TyCtxt};
  19 use rustc_serialize::{
  20     opaque::{FileEncoder, MemDecoder},
  21     Decodable, Encodable,
  22 };
  23 use rustc_session::getopts;
  24 use rustc_span::{
  25     def_id::{CrateNum, DefPathHash, LOCAL_CRATE},
  26     edition::Edition,
  27     BytePos, FileName, SourceFile,
  28 };
  29
  30 use std::fs;
  31 use std::path::PathBuf;
  32
  33 #[derive(Debug, Clone)]
  34 pub(crate) struct ScrapeExamplesOptions {
  35     output_path: PathBuf,
  36     target_crates: Vec<String>,
  37     pub(crate) scrape_tests: bool,
  38 }
  39
  40 impl ScrapeExamplesOptions {
  41     pub(crate) fn new(
  42         matches: &getopts::Matches,
  43         diag: &rustc_errors::Handler,
  44     ) -> Result<Option<Self>, i32> {
  45         let output_path = matches.opt_str("scrape-examples-output-path");
  46         let target_crates = matches.opt_strs("scrape-examples-target-crate");
  47         let scrape_tests = matches.opt_present("scrape-tests");
  48         match (output_path, !target_crates.is_empty(), scrape_tests) {
  49             (Some(output_path), true, _) => Ok(Some(ScrapeExamplesOptions {
  50                 output_path: PathBuf::from(output_path),
  51                 target_crates,
  52                 scrape_tests,
  53             })),
  54             (Some(_), false, _) | (None, true, _) => {
  55                 diag.err("must use --scrape-examples-output-path and --scrape-examples-target-crate together");
  56                 Err(1)
  57             }
  58             (None, false, true) => {
  59                 diag.err("must use --scrape-examples-output-path and --scrape-examples-target-crate with --scrape-tests");
  60                 Err(1)
  61             }
  62             (None, false, false) => Ok(None),
  63         }
  64     }
  65 }
  66
  67 #[derive(Encodable, Decodable, Debug, Clone)]
  68 pub(crate) struct SyntaxRange {
  69     pub(crate) byte_span: (u32, u32),
  70     pub(crate) line_span: (usize, usize),
  71 }
  72
  73 impl SyntaxRange {
  74     fn new(span: rustc_span::Span, file: &SourceFile) -> Option<Self> {
  75         let get_pos = |bytepos: BytePos| file.original_relative_byte_pos(bytepos).0;
  76         let get_line = |bytepos: BytePos| file.lookup_line(bytepos);
  77
  78         Some(SyntaxRange {
  79             byte_span: (get_pos(span.lo()), get_pos(span.hi())),
  80             line_span: (get_line(span.lo())?, get_line(span.hi())?),
  81         })
  82     }
  83 }
  84
  85 #[derive(Encodable, Decodable, Debug, Clone)]
  86 pub(crate) struct CallLocation {
  87     pub(crate) call_expr: SyntaxRange,
  88     pub(crate) call_ident: SyntaxRange,
  89     pub(crate) enclosing_item: SyntaxRange,
  90 }
  91
  92 impl CallLocation {
  93     fn new(
  94         expr_span: rustc_span::Span,
  95         ident_span: rustc_span::Span,
  96         enclosing_item_span: rustc_span::Span,
  97         source_file: &SourceFile,
  98     ) -> Option<Self> {
  99         Some(CallLocation {
 100             call_expr: SyntaxRange::new(expr_span, source_file)?,
 101             call_ident: SyntaxRange::new(ident_span, source_file)?,
 102             enclosing_item: SyntaxRange::new(enclosing_item_span, source_file)?,
 103         })
 104     }
 105 }
 106
 107 #[derive(Encodable, Decodable, Debug, Clone)]
 108 pub(crate) struct CallData {
 109     pub(crate) locations: Vec<CallLocation>,
 110     pub(crate) url: String,
 111     pub(crate) display_name: String,
 112     pub(crate) edition: Edition,
 113 }
 114
 115 pub(crate) type FnCallLocations = FxHashMap<PathBuf, CallData>;
 116 pub(crate) type AllCallLocations = FxHashMap<DefPathHash, FnCallLocations>;
 117
 118 /// Visitor for traversing a crate and finding instances of function calls.
 119 struct FindCalls<'a, 'tcx> {
 120     tcx: TyCtxt<'tcx>,
 121     map: Map<'tcx>,
 122     cx: Context<'tcx>,
 123     target_crates: Vec<CrateNum>,
 124     calls: &'a mut AllCallLocations,
 125 }
 126
 127 impl<'a, 'tcx> Visitor<'tcx> for FindCalls<'a, 'tcx>
 128 where
 129     'tcx: 'a,
 130 {
 131     type NestedFilter = nested_filter::OnlyBodies;
 132
 133     fn nested_visit_map(&mut self) -> Self::Map {
 134         self.map
 135     }
 136
 137     fn visit_expr(&mut self, ex: &'tcx hir::Expr<'tcx>) {
 138         intravisit::walk_expr(self, ex);
 139
 140         let tcx = self.tcx;
 141
 142         // If we visit an item that contains an expression outside a function body,
 143         // then we need to exit before calling typeck (which will panic). See
 144         // test/run-make/rustdoc-scrape-examples-invalid-expr for an example.
 145         let hir = tcx.hir();
 146         if hir.maybe_body_owned_by(ex.hir_id.owner).is_none() {
 147             return;
 148         }
 149
 150         // Get type of function if expression is a function call
 151         let (ty, call_span, ident_span) = match ex.kind {
 152             hir::ExprKind::Call(f, _) => {
 153                 let types = tcx.typeck(ex.hir_id.owner);
 154
 155                 if let Some(ty) = types.node_type_opt(f.hir_id) {
 156                     (ty, ex.span, f.span)
 157                 } else {
 158                     trace!("node_type_opt({}) = None", f.hir_id);
 159                     return;
 160                 }
 161             }
 162             hir::ExprKind::MethodCall(path, _, call_span) => {
 163                 let types = tcx.typeck(ex.hir_id.owner);
 164                 let Some(def_id) = types.type_dependent_def_id(ex.hir_id) else {
 165                     trace!("type_dependent_def_id({}) = None", ex.hir_id);
 166                     return;
 167                 };
 168
 169                 let ident_span = path.ident.span;
 170                 (tcx.type_of(def_id), call_span, ident_span)
 171             }
 172             _ => {
 173                 return;
 174             }
 175         };
 176
 177         // If this span comes from a macro expansion, then the source code may not actually show
 178         // a use of the given item, so it would be a poor example. Hence, we skip all uses in macros.
 179         if call_span.from_expansion() {
 180             trace!("Rejecting expr from macro: {call_span:?}");
 181             return;
 182         }
 183
 184         // If the enclosing item has a span coming from a proc macro, then we also don't want to include
 185         // the example.
 186         let enclosing_item_span = tcx
 187             .hir()
 188             .span_with_body(tcx.hir().local_def_id_to_hir_id(tcx.hir().get_parent_item(ex.hir_id)));
 189         if enclosing_item_span.from_expansion() {
 190             trace!("Rejecting expr ({call_span:?}) from macro item: {enclosing_item_span:?}");
 191             return;
 192         }
 193
 194         // If the enclosing item doesn't actually enclose the call, this means we probably have a weird
 195         // macro issue even though the spans aren't tagged as being from an expansion.
 196         if !enclosing_item_span.contains(call_span) {
 197             warn!(
 198                 "Attempted to scrape call at [{call_span:?}] whose enclosing item [{enclosing_item_span:?}] doesn't contain the span of the call."
 199             );
 200             return;
 201         }
 202
 203         // Similarly for the call w/ the function ident.
 204         if !call_span.contains(ident_span) {
 205             warn!(
 206                 "Attempted to scrape call at [{call_span:?}] whose identifier [{ident_span:?}] was not contained in the span of the call."
 207             );
 208             return;
 209         }
 210
 211         // Save call site if the function resolves to a concrete definition
 212         if let ty::FnDef(def_id, _) = ty.kind() {
 213             if self.target_crates.iter().all(|krate| *krate != def_id.krate) {
 214                 trace!("Rejecting expr from crate not being documented: {call_span:?}");
 215                 return;
 216             }
 217
 218             let source_map = tcx.sess.source_map();
 219             let file = source_map.lookup_char_pos(call_span.lo()).file;
 220             let file_path = match file.name.clone() {
 221                 FileName::Real(real_filename) => real_filename.into_local_path(),
 222                 _ => None,
 223             };
 224
 225             if let Some(file_path) = file_path {
 226                 let abs_path = match fs::canonicalize(file_path.clone()) {
 227                     Ok(abs_path) => abs_path,
 228                     Err(_) => {
 229                         trace!("Could not canonicalize file path: {}", file_path.display());
 230                         return;
 231                     }
 232                 };
 233
 234                 let cx = &self.cx;
 235                 let clean_span = crate::clean::types::Span::new(call_span);
 236                 let url = match cx.href_from_span(clean_span, false) {
 237                     Some(url) => url,
 238                     None => {
 239                         trace!(
 240                             "Rejecting expr ({call_span:?}) whose clean span ({clean_span:?}) cannot be turned into a link"
 241                         );
 242                         return;
 243                     }
 244                 };
 245
 246                 let mk_call_data = || {
 247                     let display_name = file_path.display().to_string();
 248                     let edition = call_span.edition();
 249                     CallData { locations: Vec::new(), url, display_name, edition }
 250                 };
 251
 252                 let fn_key = tcx.def_path_hash(*def_id);
 253                 let fn_entries = self.calls.entry(fn_key).or_default();
 254
 255                 trace!("Including expr: {:?}", call_span);
 256                 let enclosing_item_span =
 257                     source_map.span_extend_to_prev_char(enclosing_item_span, '\n', false);
 258                 let location =
 259                     match CallLocation::new(call_span, ident_span, enclosing_item_span, &file) {
 260                         Some(location) => location,
 261                         None => {
 262                             trace!("Could not get serializable call location for {call_span:?}");
 263                             return;
 264                         }
 265                     };
 266                 fn_entries.entry(abs_path).or_insert_with(mk_call_data).locations.push(location);
 267             }
 268         }
 269     }
 270 }
 271
 272 pub(crate) fn run(
 273     krate: clean::Crate,
 274     mut renderopts: config::RenderOptions,
 275     cache: formats::cache::Cache,
 276     tcx: TyCtxt<'_>,
 277     options: ScrapeExamplesOptions,
 278 ) -> interface::Result<()> {
 279     let inner = move || -> Result<(), String> {
 280         // Generates source files for examples
 281         renderopts.no_emit_shared = true;
 282         let (cx, _) = Context::init(krate, renderopts, cache, tcx).map_err(|e| e.to_string())?;
 283
 284         // Collect CrateIds corresponding to provided target crates
 285         // If two different versions of the crate in the dependency tree, then examples will be collcted from both.
 286         let all_crates = tcx
 287             .crates(())
 288             .iter()
 289             .chain([&LOCAL_CRATE])
 290             .map(|crate_num| (crate_num, tcx.crate_name(*crate_num)))
 291             .collect::<Vec<_>>();
 292         let target_crates = options
 293             .target_crates
 294             .into_iter()
 295             .flat_map(|target| all_crates.iter().filter(move |(_, name)| name.as_str() == target))
 296             .map(|(crate_num, _)| **crate_num)
 297             .collect::<Vec<_>>();
 298
 299         debug!("All crates in TyCtxt: {all_crates:?}");
 300         debug!("Scrape examples target_crates: {target_crates:?}");
 301
 302         // Run call-finder on all items
 303         let mut calls = FxHashMap::default();
 304         let mut finder = FindCalls { calls: &mut calls, tcx, map: tcx.hir(), cx, target_crates };
 305         tcx.hir().visit_all_item_likes_in_crate(&mut finder);
 306
 307         // The visitor might have found a type error, which we need to
 308         // promote to a fatal error
 309         if tcx.sess.diagnostic().has_errors_or_lint_errors().is_some() {
 310             return Err(String::from("Compilation failed, aborting rustdoc"));
 311         }
 312
 313         // Sort call locations within a given file in document order
 314         for fn_calls in calls.values_mut() {
 315             for file_calls in fn_calls.values_mut() {
 316                 file_calls.locations.sort_by_key(|loc| loc.call_expr.byte_span.0);
 317             }
 318         }
 319
 320         // Save output to provided path
 321         let mut encoder = FileEncoder::new(options.output_path).map_err(|e| e.to_string())?;
 322         calls.encode(&mut encoder);
 323         encoder.finish().map_err(|e| e.to_string())?;
 324
 325         Ok(())
 326     };
 327
 328     if let Err(e) = inner() {
 329         tcx.sess.fatal(&e);
 330     }
 331
 332     Ok(())
 333 }
 334
 335 // Note: the Handler must be passed in explicitly because sess isn't available while parsing options
 336 pub(crate) fn load_call_locations(
 337     with_examples: Vec<String>,
 338     diag: &rustc_errors::Handler,
 339 ) -> Result<AllCallLocations, i32> {
 340     let inner = || {
 341         let mut all_calls: AllCallLocations = FxHashMap::default();
 342         for path in with_examples {
 343             let bytes = fs::read(&path).map_err(|e| format!("{} (for path {})", e, path))?;
 344             let mut decoder = MemDecoder::new(&bytes, 0);
 345             let calls = AllCallLocations::decode(&mut decoder);
 346
 347             for (function, fn_calls) in calls.into_iter() {
 348                 all_calls.entry(function).or_default().extend(fn_calls.into_iter());
 349             }
 350         }
 351
 352         Ok(all_calls)
 353     };
 354
 355     inner().map_err(|e: String| {
 356         diag.err(&format!("failed to load examples: {}", e));
 357         1
 358     })
 359 }