feat: support PPDB embedded sources

CHANGELOG.md

@@ -5,6 +5,7 @@
 **Features**:
 
 - `PortablePdbDebugSession` now returns files referenced in the Portable PDB file. ([#729](https://github.com/getsentry/symbolic/pull/729))
+- `PortablePdbDebugSession` now returns source files embedded in the Portable PDB file. ([#734](https://github.com/getsentry/symbolic/pull/734))
 
 **Breaking changes**:
 

examples/dump_sources/src/main.rs

@@ -27,7 +27,7 @@ fn write_object_sources(path: &Path, output_path: &Path) -> Result<(), Box<dyn s
     for object in archive.objects() {
         match object {
             Ok(object) => {
-                let out = output_path.join(&format!("{}.zip", &object.debug_id()));
+                let out = output_path.join(format!("{}.zip", &object.debug_id()));
                 println!("  -> {}", out.display());
                 let writer = SourceBundleWriter::create(&out)?;
                 writer.write_object(&object, &path.file_name().unwrap().to_string_lossy())?;

symbolic-debuginfo/src/base.rs

@@ -513,7 +513,7 @@ impl<'data> FileInfo<'data> {
 }
 
 #[allow(clippy::ptr_arg)] // false positive https://github.com/rust-lang/rust-clippy/issues/9218
-fn from_utf8_cow_lossy<'data>(input: &Cow<'data, [u8]>) -> Cow<'data, str> {
+pub(crate) fn from_utf8_cow_lossy<'data>(input: &Cow<'data, [u8]>) -> Cow<'data, str> {
     // See https://github.com/rust-lang/rust/issues/32669
     match input {
         Cow::Borrowed(bytes) => String::from_utf8_lossy(bytes),

symbolic-debuginfo/src/ppdb.rs

@@ -1,9 +1,11 @@
 //! Support for Portable PDB Objects.
 use std::borrow::Cow;
+use std::collections::HashMap;
 use std::fmt;
 use std::iter;
 
 use symbolic_common::{Arch, CodeId, DebugId};
+use symbolic_ppdb::EmbeddedSource;
 use symbolic_ppdb::{FormatError, PortablePdb};
 
 use crate::base::*;
@@ -88,9 +90,7 @@ impl<'data: 'object, 'object> ObjectLike<'data, 'object> for PortablePdbObject<'
 
     /// Constructs a debugging session.
     fn debug_session(&self) -> Result<PortablePdbDebugSession<'data>, FormatError> {
-        Ok(PortablePdbDebugSession {
-            ppdb: self.ppdb.clone(),
-        })
+        PortablePdbDebugSession::new(&self.ppdb)
     }
 
     /// Determines whether this object contains stack unwinding information.
@@ -100,7 +100,10 @@ impl<'data: 'object, 'object> ObjectLike<'data, 'object> for PortablePdbObject<'
 
     /// Determines whether this object contains embedded source.
     fn has_sources(&self) -> bool {
-        false
+        match self.ppdb.get_embedded_sources() {
+            Ok(mut iter) => iter.any(|v| v.is_ok()),
+            Err(_) => false,
+        }
     }
 
     /// Determines whether this object is malformed and was only partially parsed.
@@ -138,9 +141,24 @@ impl fmt::Debug for PortablePdbObject<'_> {
 /// A debug session for a Portable PDB object.
 pub struct PortablePdbDebugSession<'data> {
     ppdb: PortablePdb<'data>,
+    sources: HashMap<String, EmbeddedSource<'data>>,
 }
 
 impl<'data> PortablePdbDebugSession<'data> {
+    fn new(ppdb: &'_ PortablePdb<'data>) -> Result<Self, FormatError> {
+        let mut sources: HashMap<String, EmbeddedSource<'data>> = HashMap::new();
+        for source in ppdb.get_embedded_sources()? {
+            match source {
+                Ok(source) => sources.insert(source.get_path().into(), source),
+                Err(e) => return Err(e),
+            };
+        }
+        Ok(PortablePdbDebugSession {
+            ppdb: ppdb.clone(),
+            sources,
+        })
+    }
+
     /// Returns an iterator over all functions in this debug file.
     pub fn functions(&self) -> PortablePdbFunctionIterator<'_> {
         iter::empty()
@@ -154,8 +172,13 @@ impl<'data> PortablePdbDebugSession<'data> {
     /// Looks up a file's source contents by its full canonicalized path.
     ///
     /// The given path must be canonicalized.
-    pub fn source_by_path(&self, _path: &str) -> Result<Option<Cow<'_, str>>, FormatError> {
-        Ok(None)
+    pub fn source_by_path(&self, path: &str) -> Result<Option<Cow<'_, str>>, FormatError> {
+        match self.sources.get(path) {
+            None => Ok(None),
+            Some(source) => source
+                .get_contents()
+                .map(|bytes| Some(from_utf8_cow_lossy(&bytes))),
+        }
     }
 }
 

symbolic-debuginfo/tests/test_objects.rs

@@ -670,6 +670,49 @@ fn test_ppdb_functions() -> Result<(), Error> {
     Ok(())
 }
 
+#[test]
+fn test_ppdb_has_sources() -> Result<(), Error> {
+    {
+        let view = ByteView::open(fixture("windows/portable.pdb"))?;
+        let object = Object::parse(&view)?;
+        assert_eq!(object.has_sources(), false);
+    }
+    {
+        let view = ByteView::open(fixture("windows/Sentry.Samples.Console.Basic.pdb"))?;
+        let object = Object::parse(&view)?;
+        assert_eq!(object.has_sources(), true);
+    }
+    Ok(())
+}
+
+#[test]
+fn test_ppdb_source_by_path() -> Result<(), Error> {
+    {
+        let view = ByteView::open(fixture("windows/portable.pdb"))?;
+        let object = Object::parse(&view)?;
+
+        let session = object.debug_session()?;
+        let source = session.source_by_path("foo/bar.cs").unwrap();
+        assert!(source.is_none());
+    }
+
+    {
+        let view = ByteView::open(fixture("windows/Sentry.Samples.Console.Basic.pdb"))?;
+        let object = Object::parse(&view)?;
+
+        let session = object.debug_session()?;
+        let source = session
+            .source_by_path(
+                "C:\\dev\\sentry-dotnet\\samples\\Sentry.Samples.Console.Basic\\Program.cs",
+            )
+            .unwrap();
+        let source_text = source.unwrap();
+        assert_eq!(source_text.len(), 204);
+    }
+
+    Ok(())
+}
+
 #[test]
 fn test_wasm_symbols() -> Result<(), Error> {
     let view = ByteView::open(fixture("wasm/simple.wasm"))?;

symbolic-ppdb/Cargo.toml

@@ -25,6 +25,7 @@ symbolic-common = { version = "10.2.1", path = "../symbolic-common" }
 watto = { version = "0.1.0", features = ["writer", "strings"] }
 thiserror = "1.0.31"
 uuid = "1.0.0"
+flate2 = { version ="1.0.13", default-features = false, features = [ "rust_backend" ] }
 
 [dev-dependencies]
 symbolic-testutils = { path = "../symbolic-testutils" }

symbolic-ppdb/src/cache/mod.rs

@@ -105,7 +105,7 @@ pub struct CacheError {
 }
 
 impl CacheError {
-    /// Creates a new SymCache error from a known kind of error as well as an
+    /// Creates a new CacheError from a known kind of error as well as an
     /// arbitrary error payload.
     pub(crate) fn new<E>(kind: CacheErrorKind, source: E) -> Self
     where