feat: Added debug IDs to source bundle javascript files (#762)

Author: mitsuhiko

CHANGELOG.md

@@ -2,6 +2,10 @@
 
 ## Unreleased
 
+**Features**:
+
+- Added debug IDs to source bundle JavaScript files and source maps. ([#762](https://github.com/getsentry/symbolic/pull/762))
+
 **Breaking changes**:
 
 - Change `DebugSession::source_by_path()` to return `SourceCode` enum with either file content or a URL to fetch it from. ([#758](https://github.com/getsentry/symbolic/pull/758))

symbolic-debuginfo/Cargo.toml

@@ -70,13 +70,15 @@ sourcebundle = [
     "regex",
     "serde_json",
     "zip",
+    "debugid/serde"
 ]
 # WASM processing
 wasm = ["bitvec", "dwarf", "wasmparser"]
 
 [dependencies]
 bitvec = { version = "1.0.0", optional = true, features = ["alloc"] }
 dmsort = "1.0.1"
+debugid = { version = "0.8.0" }
 elementtree = { version = "1.2.2", optional = true }
 elsa = { version = "1.4.0", optional = true }
 fallible-iterator = "0.2.0"

symbolic-debuginfo/src/base.rs

@@ -6,6 +6,8 @@ use std::str::FromStr;
 
 use symbolic_common::{clean_path, join_path, Arch, CodeId, DebugId, Name};
 
+use crate::sourcebundle::SourceFileDescriptor;
+
 pub(crate) trait Parse<'data>: Sized {
     type Error;
 
@@ -670,17 +672,6 @@ impl fmt::Debug for Function<'_> {
 /// A dynamically dispatched iterator over items with the given lifetime.
 pub type DynIterator<'a, T> = Box<dyn Iterator<Item = T> + 'a>;
 
-/// Represents a source file referenced by a debug information object file.
-#[non_exhaustive]
-#[derive(Debug, Clone)]
-pub enum SourceCode<'a> {
-    /// Verbatim source code/file contents.
-    Content(Cow<'a, str>),
-
-    /// Url (usually https) where the content can be fetched from.
-    Url(Cow<'a, str>),
-}
-
 /// A stateful session for interfacing with debug information.
 ///
 /// Debug sessions can be obtained via [`ObjectLike::debug_session`]. Since computing a session may
@@ -728,8 +719,10 @@ pub trait DebugSession<'session> {
     fn files(&'session self) -> Self::FileIterator;
 
     /// Looks up a file's source by its full canonicalized path.
-    /// Returns either source contents, if it was embedded, or a source link.
-    fn source_by_path(&self, path: &str) -> Result<Option<SourceCode<'_>>, Self::Error>;
+    ///
+    /// Returns a descriptor that has all the information available of the source.  It can
+    /// either contain the source contents directly, if it was embedded, or a source link.
+    fn source_by_path(&self, path: &str) -> Result<Option<SourceFileDescriptor<'_>>, Self::Error>;
 }
 
 /// An object containing debug information.

symbolic-debuginfo/src/breakpad.rs

@@ -13,6 +13,7 @@ use symbolic_common::{Arch, AsSelf, CodeId, DebugId, Language, Name, NameManglin
 
 use crate::base::*;
 use crate::function_builder::FunctionBuilder;
+use crate::sourcebundle::SourceFileDescriptor;
 use crate::Parse;
 
 #[derive(Clone, Debug)]
@@ -1277,7 +1278,10 @@ impl<'data> BreakpadDebugSession<'data> {
     }
 
     /// See [DebugSession::source_by_path] for more information.
-    pub fn source_by_path(&self, _path: &str) -> Result<Option<SourceCode<'_>>, BreakpadError> {
+    pub fn source_by_path(
+        &self,
+        _path: &str,
+    ) -> Result<Option<SourceFileDescriptor<'_>>, BreakpadError> {
         Ok(None)
     }
 }
@@ -1295,7 +1299,7 @@ impl<'data, 'session> DebugSession<'session> for BreakpadDebugSession<'data> {
         self.files()
     }
 
-    fn source_by_path(&self, path: &str) -> Result<Option<SourceCode<'_>>, Self::Error> {
+    fn source_by_path(&self, path: &str) -> Result<Option<SourceFileDescriptor<'_>>, Self::Error> {
         self.source_by_path(path)
     }
 }

symbolic-debuginfo/src/dwarf.rs

@@ -29,6 +29,7 @@ use crate::base::*;
 use crate::function_builder::FunctionBuilder;
 #[cfg(feature = "macho")]
 use crate::macho::BcSymbolMap;
+use crate::sourcebundle::SourceFileDescriptor;
 
 /// This is a fake BcSymbolMap used when macho support is turned off since they are unfortunately
 /// part of the dwarf interface
@@ -1332,7 +1333,10 @@ impl<'data> DwarfDebugSession<'data> {
     }
 
     /// See [DebugSession::source_by_path] for more information.
-    pub fn source_by_path(&self, _path: &str) -> Result<Option<SourceCode<'_>>, DwarfError> {
+    pub fn source_by_path(
+        &self,
+        _path: &str,
+    ) -> Result<Option<SourceFileDescriptor<'_>>, DwarfError> {
         Ok(None)
     }
 }
@@ -1350,7 +1354,7 @@ impl<'data, 'session> DebugSession<'session> for DwarfDebugSession<'data> {
         self.files()
     }
 
-    fn source_by_path(&self, path: &str) -> Result<Option<SourceCode<'_>>, Self::Error> {
+    fn source_by_path(&self, path: &str) -> Result<Option<SourceFileDescriptor<'_>>, Self::Error> {
         self.source_by_path(path)
     }
 }

symbolic-debuginfo/src/object.rs

@@ -484,7 +484,10 @@ impl<'d> ObjectDebugSession<'d> {
 
     /// Looks up a file's source by its full canonicalized path.
     /// Returns either source contents, if it was embedded, or a source link.
-    pub fn source_by_path(&self, path: &str) -> Result<Option<SourceCode<'_>>, ObjectError> {
+    pub fn source_by_path(
+        &self,
+        path: &str,
+    ) -> Result<Option<SourceFileDescriptor<'_>>, ObjectError> {
         match *self {
             ObjectDebugSession::Breakpad(ref s) => {
                 s.source_by_path(path).map_err(ObjectError::transparent)
@@ -518,7 +521,7 @@ impl<'session> DebugSession<'session> for ObjectDebugSession<'_> {
         self.files()
     }
 
-    fn source_by_path(&self, path: &str) -> Result<Option<SourceCode<'_>>, Self::Error> {
+    fn source_by_path(&self, path: &str) -> Result<Option<SourceFileDescriptor<'_>>, Self::Error> {
         self.source_by_path(path)
     }
 }

symbolic-debuginfo/src/pdb.rs

@@ -24,6 +24,7 @@ use symbolic_common::{
 
 use crate::base::*;
 use crate::function_stack::FunctionStack;
+use crate::sourcebundle::SourceFileDescriptor;
 use crate::Parse;
 
 type Pdb<'data> = pdb::PDB<'data, Cursor<&'data [u8]>>;
@@ -638,7 +639,10 @@ impl<'d> PdbDebugSession<'d> {
     }
 
     /// See [DebugSession::source_by_path] for more information.
-    pub fn source_by_path(&self, _path: &str) -> Result<Option<SourceCode<'_>>, PdbError> {
+    pub fn source_by_path(
+        &self,
+        _path: &str,
+    ) -> Result<Option<SourceFileDescriptor<'_>>, PdbError> {
         Ok(None)
     }
 }
@@ -656,7 +660,7 @@ impl<'session> DebugSession<'session> for PdbDebugSession<'_> {
         self.files()
     }
 
-    fn source_by_path(&self, path: &str) -> Result<Option<SourceCode<'_>>, Self::Error> {
+    fn source_by_path(&self, path: &str) -> Result<Option<SourceFileDescriptor<'_>>, Self::Error> {
         self.source_by_path(path)
     }
 }

symbolic-debuginfo/src/ppdb.rs

@@ -10,6 +10,7 @@ use symbolic_ppdb::EmbeddedSource;
 use symbolic_ppdb::{Document, FormatError, PortablePdb};
 
 use crate::base::*;
+use crate::sourcebundle::SourceFileDescriptor;
 
 /// An iterator over symbols in a [`PortablePdbObject`].
 pub type PortablePdbSymbolIterator<'data> = iter::Empty<Symbol<'data>>;
@@ -191,16 +192,23 @@ impl<'data> PortablePdbDebugSession<'data> {
     }
 
     /// See [DebugSession::source_by_path] for more information.
-    pub fn source_by_path(&self, path: &str) -> Result<Option<SourceCode<'_>>, FormatError> {
+    pub fn source_by_path(
+        &self,
+        path: &str,
+    ) -> Result<Option<SourceFileDescriptor<'_>>, FormatError> {
         let sources = self.sources.borrow_with(|| self.init_sources());
         match sources.get(path) {
             None => Ok(None),
-            Some(PPDBSource::Embedded(source)) => source
-                .get_contents()
-                .map(|bytes| Some(SourceCode::Content(from_utf8_cow_lossy(&bytes)))),
-            Some(PPDBSource::Link(document)) => {
-                Ok(self.ppdb.get_source_link(document).map(SourceCode::Url))
-            }
+            Some(PPDBSource::Embedded(source)) => source.get_contents().map(|bytes| {
+                Some(SourceFileDescriptor::new_embedded(
+                    from_utf8_cow_lossy(&bytes),
+                    None,
+                ))
+            }),
+            Some(PPDBSource::Link(document)) => Ok(self
+                .ppdb
+                .get_source_link(document)
+                .map(SourceFileDescriptor::new_remote)),
         }
     }
 }
@@ -218,7 +226,7 @@ impl<'data, 'session> DebugSession<'session> for PortablePdbDebugSession<'data>
         self.files()
     }
 
-    fn source_by_path(&self, path: &str) -> Result<Option<SourceCode<'_>>, Self::Error> {
+    fn source_by_path(&self, path: &str) -> Result<Option<SourceFileDescriptor<'_>>, Self::Error> {
         self.source_by_path(path)
     }
 }