Unify diagnostic generation for utf8 failure

Author: estebank

compiler/rustc_builtin_macros/src/source_util.rs

@@ -13,7 +13,7 @@ use rustc_expand::base::{
 use rustc_expand::module::DirOwnership;
 use rustc_lint_defs::BuiltinLintDiag;
 use rustc_parse::parser::{ForceCollect, Parser};
-use rustc_parse::{new_parser_from_file, unwrap_or_emit_fatal};
+use rustc_parse::{new_parser_from_file, unwrap_or_emit_fatal, utf8_error};
 use rustc_session::lint::builtin::INCOMPLETE_INCLUDE;
 use rustc_span::source_map::SourceMap;
 use rustc_span::{Pos, Span, Symbol};
@@ -211,25 +211,7 @@ pub(crate) fn expand_include_str(
             }
             Err(utf8err) => {
                 let mut err = cx.dcx().struct_span_err(sp, format!("`{path}` wasn't a utf-8 file"));
-                let path = PathBuf::from(path.as_str());
-                let start = utf8err.valid_up_to();
-                let note = format!("invalid utf-8 at byte `{start}`");
-                let msg = if let Some(len) = utf8err.error_len() {
-                    format!("`{:?}` is not valid utf-8", &bytes[start..start + len])
-                } else {
-                    note.clone()
-                };
-                let contents = String::from_utf8_lossy(&bytes[..]).to_string();
-                let source = cx.source_map().new_source_file(path.into(), contents);
-                let span = Span::with_root_ctxt(
-                    source.normalized_byte_pos(start as u32),
-                    source.normalized_byte_pos(start as u32),
-                );
-                if span.is_dummy() {
-                    err.note(note);
-                } else {
-                    err.span_note(span, msg);
-                }
+                utf8_error(cx.source_map(), path.as_str(), None, &mut err, utf8err, &bytes[..]);
                 DummyResult::any(sp, err.emit())
             }
         },

compiler/rustc_parse/src/lib.rs

@@ -15,15 +15,17 @@
 #![warn(unreachable_pub)]
 // tidy-alphabetical-end
 
-use std::path::Path;
+use std::path::{Path, PathBuf};
+use std::str::Utf8Error;
 
 use rustc_ast as ast;
 use rustc_ast::tokenstream::TokenStream;
 use rustc_ast::{AttrItem, Attribute, MetaItemInner, token};
 use rustc_ast_pretty::pprust;
 use rustc_data_structures::sync::Lrc;
-use rustc_errors::{Diag, FatalError, PResult};
+use rustc_errors::{Diag, EmissionGuarantee, FatalError, PResult};
 use rustc_session::parse::ParseSess;
+use rustc_span::source_map::SourceMap;
 use rustc_span::{FileName, SourceFile, Span};
 pub use unicode_normalization::UNICODE_VERSION as UNICODE_NORMALIZATION_VERSION;
 
@@ -74,36 +76,21 @@ pub fn new_parser_from_file<'a>(
     path: &Path,
     sp: Option<Span>,
 ) -> Result<Parser<'a>, Vec<Diag<'a>>> {
-    let source_file = psess.source_map().load_file(path).unwrap_or_else(|e| {
+    let sm = psess.source_map();
+    let source_file = sm.load_file(path).unwrap_or_else(|e| {
         let msg = format!("couldn't read `{}`: {}", path.display(), e);
         let mut err = psess.dcx().struct_fatal(msg);
         if let Ok(contents) = std::fs::read(path)
-            && let Err(utf8err) = String::from_utf8(contents)
+            && let Err(utf8err) = String::from_utf8(contents.clone())
         {
-            // The file exists, but it wasn't valid UTF-8.
-            let start = utf8err.utf8_error().valid_up_to();
-            let note = format!("invalid utf-8 at byte `{start}`");
-            let msg = if let Some(len) = utf8err.utf8_error().error_len() {
-                format!("`{:?}` is not valid utf-8", &utf8err.as_bytes()[start..start + len])
-            } else {
-                note.clone()
-            };
-            let contents = utf8err.into_utf8_lossy();
-            let source = psess.source_map().new_source_file(path.to_owned().into(), contents);
-            let span = Span::with_root_ctxt(
-                source.normalized_byte_pos(start as u32),
-                source.normalized_byte_pos(start as u32),
+            utf8_error(
+                sm,
+                &path.display().to_string(),
+                sp,
+                &mut err,
+                utf8err.utf8_error(),
+                &contents,
             );
-            if span.is_dummy() {
-                err.note(note);
-            } else {
-                if sp.is_some() {
-                    err.span_note(span, msg);
-                } else {
-                    err.span(span);
-                    err.span_label(span, msg);
-                }
-            }
         }
         if let Some(sp) = sp {
             err.span(sp);
@@ -113,6 +100,40 @@ pub fn new_parser_from_file<'a>(
     new_parser_from_source_file(psess, source_file)
 }
 
+pub fn utf8_error<E: EmissionGuarantee>(
+    sm: &SourceMap,
+    path: &str,
+    sp: Option<Span>,
+    err: &mut Diag<'_, E>,
+    utf8err: Utf8Error,
+    contents: &[u8],
+) {
+    // The file exists, but it wasn't valid UTF-8.
+    let start = utf8err.valid_up_to();
+    let note = format!("invalid utf-8 at byte `{start}`");
+    let msg = if let Some(len) = utf8err.error_len() {
+        format!("`{:?}` is not valid utf-8", &contents[start..start + len])
+    } else {
+        note.clone()
+    };
+    let contents = String::from_utf8_lossy(contents).to_string();
+    let source = sm.new_source_file(PathBuf::from(path).into(), contents);
+    let span = Span::with_root_ctxt(
+        source.normalized_byte_pos(start as u32),
+        source.normalized_byte_pos(start as u32),
+    );
+    if span.is_dummy() {
+        err.note(note);
+    } else {
+        if sp.is_some() {
+            err.span_note(span, msg);
+        } else {
+            err.span(span);
+            err.span_label(span, msg);
+        }
+    }
+}
+
 /// Given a session and a `source_file`, return a parser. Returns any buffered errors from lexing
 /// the initial token stream.
 fn new_parser_from_source_file(