Rollup merge of rust-lang#71048 - arlosi:normalize_ext_src, r=eddyb

Normalize source when loading external foreign source into SourceMap

The compiler normalizes source when reading files initially (removes BOMs, etc), but not when loading external sources.

This leads to the external source matching according to the `src_hash`, but differing internally because it was not normalized.

Fixes rust-lang#70874.

Author: Dylan-DPC

src/librustc_span/lib.rs

@@ -1192,8 +1192,10 @@ impl SourceFile {
                 kind: src_kind @ ExternalSourceKind::AbsentOk, ..
             } = &mut *external_src
             {
-                if let Some(src) = src {
+                if let Some(mut src) = src {
+                    // The src_hash needs to be computed on the pre-normalized src.
                     if self.src_hash.matches(&src) {
+                        normalize_src(&mut src, BytePos::from_usize(0));
                         *src_kind = ExternalSourceKind::Present(Lrc::new(src));
                         return true;
                     }

src/librustc_span/source_map/tests.rs

@@ -168,6 +168,62 @@ fn span_merging_fail() {
     assert!(sm.merge_spans(span1, span2).is_none());
 }
 
+/// Tests loading an external source file that requires normalization.
+#[test]
+fn t10() {
+    let sm = SourceMap::new(FilePathMapping::empty());
+    let unnormalized = "first line.\r\nsecond line";
+    let normalized = "first line.\nsecond line";
+
+    let src_file = sm.new_source_file(PathBuf::from("blork.rs").into(), unnormalized.to_string());
+
+    assert_eq!(src_file.src.as_ref().unwrap().as_ref(), normalized);
+    assert!(
+        src_file.src_hash.matches(unnormalized),
+        "src_hash should use the source before normalization"
+    );
+
+    let SourceFile {
+        name,
+        name_was_remapped,
+        src_hash,
+        start_pos,
+        end_pos,
+        lines,
+        multibyte_chars,
+        non_narrow_chars,
+        normalized_pos,
+        name_hash,
+        ..
+    } = (*src_file).clone();
+
+    let imported_src_file = sm.new_imported_source_file(
+        name,
+        name_was_remapped,
+        src_hash,
+        name_hash,
+        (end_pos - start_pos).to_usize(),
+        CrateNum::new(0),
+        lines,
+        multibyte_chars,
+        non_narrow_chars,
+        normalized_pos,
+        start_pos,
+        end_pos,
+    );
+
+    assert!(
+        imported_src_file.external_src.borrow().get_source().is_none(),
+        "imported source file should not have source yet"
+    );
+    imported_src_file.add_external_src(|| Some(unnormalized.to_string()));
+    assert_eq!(
+        imported_src_file.external_src.borrow().get_source().unwrap().as_ref(),
+        normalized,
+        "imported source file should be normalized"
+    );
+}
+
 /// Returns the span corresponding to the `n`th occurrence of `substring` in `source_text`.
 trait SourceMapExtension {
     fn span_substr(