ref(uploads): Split build_artifact_bundle into separate module (#2922)

### Description

Move `build_artifact_bundle` (and functions only it calls) out of
`crate::utils::file_upload` and into a new module called
`crate::utils::source_bundle`, also renaming the function to `build`.

This change only moves code around; it is done in preparation for the
next PR.

### Issues
- See #2927
- See
[CLI-220](https://linear.app/getsentry/issue/CLI-220/dont-call-chunk-upload-for-bundle-jvm)

Author: szokeasaurusrex

src/utils/file_upload.rs

@@ -2,27 +2,22 @@
 use std::collections::{BTreeMap, HashMap};
 use std::ffi::OsStr;
 use std::fmt::{self, Display};
-use std::io::BufWriter;
 use std::path::PathBuf;
 use std::str;
 use std::sync::Arc;
 use std::time::{Duration, Instant};
 
 use anyhow::{anyhow, bail, Result};
 use console::style;
-use log::info;
 use parking_lot::RwLock;
 use rayon::prelude::*;
 use rayon::ThreadPoolBuilder;
 use sentry::types::DebugId;
 use sha1_smol::Digest;
 use symbolic::common::ByteView;
 use symbolic::debuginfo::js;
-use symbolic::debuginfo::sourcebundle::{
-    SourceBundleErrorKind, SourceBundleWriter, SourceFileInfo, SourceFileType,
-};
+use symbolic::debuginfo::sourcebundle::SourceFileType;
 use thiserror::Error;
-use url::Url;
 
 use crate::api::NewRelease;
 use crate::api::{Api, ChunkServerOptions, ChunkUploadCapability};
@@ -31,6 +26,7 @@ use crate::utils::chunks::{upload_chunks, Chunk, ASSEMBLE_POLL_INTERVAL};
 use crate::utils::fs::{get_sha1_checksums, TempFile};
 use crate::utils::non_empty::NonEmptySlice;
 use crate::utils::progress::{ProgressBar, ProgressBarMode, ProgressStyle};
+use crate::utils::source_bundle;
 
 use super::file_search::ReleaseFileMatch;
 
@@ -415,7 +411,7 @@ impl<'a> FileUpload<'a> {
     }
 
     pub fn build_jvm_bundle(&self, debug_id: Option<DebugId>) -> Result<TempFile> {
-        build_artifact_bundle(self.context, &self.files, debug_id)
+        source_bundle::build(self.context, &self.files, debug_id)
     }
 }
 
@@ -622,7 +618,7 @@ fn upload_files_chunked(
     files: &SourceFiles,
     options: &ChunkServerOptions,
 ) -> Result<()> {
-    let archive = build_artifact_bundle(context, files, None)?;
+    let archive = source_bundle::build(context, files, None)?;
 
     let progress_style =
         ProgressStyle::default_spinner().template("{spinner} Optimizing bundle for upload...");
@@ -678,137 +674,6 @@ fn upload_files_chunked(
     poll_assemble(checksum, &checksums, context, options)
 }
 
-/// Creates a debug id from a map of source files by hashing each file's
-/// URL, contents, type, and headers.
-fn build_debug_id(files: &SourceFiles) -> DebugId {
-    let mut hash = sha1_smol::Sha1::new();
-    for source_file in files.values() {
-        hash.update(source_file.url.as_bytes());
-        hash.update(&source_file.contents);
-        hash.update(format!("{:?}", source_file.ty).as_bytes());
-
-        for (key, value) in &source_file.headers {
-            hash.update(key.as_bytes());
-            hash.update(value.as_bytes());
-        }
-    }
-
-    let mut sha1_bytes = [0u8; 16];
-    sha1_bytes.copy_from_slice(&hash.digest().bytes()[..16]);
-    DebugId::from_uuid(uuid::Builder::from_sha1_bytes(sha1_bytes).into_uuid())
-}
-
-fn build_artifact_bundle(
-    context: &UploadContext,
-    files: &SourceFiles,
-    debug_id: Option<DebugId>,
-) -> Result<TempFile> {
-    let progress_style = ProgressStyle::default_bar().template(
-        "{prefix:.dim} Bundling files for upload... {msg:.dim}\
-       \n{wide_bar}  {pos}/{len}",
-    );
-
-    let pb = ProgressBar::new(files.len());
-    pb.set_style(progress_style);
-    pb.set_prefix(">");
-
-    let archive = TempFile::create()?;
-    let mut bundle = SourceBundleWriter::start(BufWriter::new(archive.open()?))?;
-
-    // artifact bundles get a random UUID as debug id
-    let debug_id = debug_id.unwrap_or_else(|| build_debug_id(files));
-    bundle.set_attribute("debug_id", debug_id.to_string());
-
-    if let Some(note) = context.note {
-        bundle.set_attribute("note", note.to_owned());
-    }
-
-    bundle.set_attribute("org".to_owned(), context.org.to_owned());
-    if let Some([project]) = context.projects.as_deref() {
-        // Only set project if there is exactly one project
-        bundle.set_attribute("project".to_owned(), project);
-    }
-    if let Some(release) = context.release {
-        bundle.set_attribute("release".to_owned(), release.to_owned());
-    }
-    if let Some(dist) = context.dist {
-        bundle.set_attribute("dist".to_owned(), dist.to_owned());
-    }
-
-    let mut bundle_file_count = 0;
-
-    for file in files.values() {
-        pb.inc(1);
-        pb.set_message(&file.url);
-
-        let mut info = SourceFileInfo::new();
-        info.set_ty(file.ty);
-        info.set_url(file.url.clone());
-        for (k, v) in &file.headers {
-            info.add_header(k.clone(), v.clone());
-        }
-
-        let bundle_path = url_to_bundle_path(&file.url)?;
-        if let Err(e) = bundle.add_file(bundle_path, file.contents.as_slice(), info) {
-            if e.kind() == SourceBundleErrorKind::ReadFailed {
-                info!(
-                    "Skipping {} because it is not valid UTF-8.",
-                    file.path.display()
-                );
-                continue;
-            } else {
-                return Err(e.into());
-            }
-        }
-        bundle_file_count += 1;
-    }
-
-    bundle.finish()?;
-
-    pb.finish_with_duration("Bundling");
-
-    println!(
-        "{} Bundled {} {} for upload",
-        style(">").dim(),
-        style(bundle_file_count).yellow(),
-        match bundle_file_count {
-            1 => "file",
-            _ => "files",
-        }
-    );
-
-    println!(
-        "{} Bundle ID: {}",
-        style(">").dim(),
-        style(debug_id).yellow(),
-    );
-
-    Ok(archive)
-}
-
-fn url_to_bundle_path(url: &str) -> Result<String> {
-    let base = Url::parse("http://~").expect("this url is valid");
-    let url = if let Some(rest) = url.strip_prefix("~/") {
-        base.join(rest)?
-    } else {
-        base.join(url)?
-    };
-
-    let mut path = url.path().to_owned();
-    if let Some(fragment) = url.fragment() {
-        path = format!("{path}#{fragment}");
-    }
-    if path.starts_with('/') {
-        path.remove(0);
-    }
-
-    Ok(match url.host_str() {
-        Some("~") => format!("_/_/{path}"),
-        Some(host) => format!("{}/{host}/{path}", url.scheme()),
-        None => format!("{}/_/{path}", url.scheme()),
-    })
-}
-
 fn print_upload_context_details(context: &UploadContext) {
     println!(
         "{} {}",
@@ -853,77 +718,3 @@ fn is_hermes_bytecode(slice: &[u8]) -> bool {
     const HERMES_MAGIC: [u8; 8] = [0xC6, 0x1F, 0xBC, 0x03, 0xC1, 0x03, 0x19, 0x1F];
     slice.starts_with(&HERMES_MAGIC)
 }
-
-#[cfg(test)]
-mod tests {
-    use sha1_smol::Sha1;
-
-    use super::*;
-
-    #[test]
-    fn test_url_to_bundle_path() {
-        assert_eq!(url_to_bundle_path("~/bar").unwrap(), "_/_/bar");
-        assert_eq!(url_to_bundle_path("~/foo/bar").unwrap(), "_/_/foo/bar");
-        assert_eq!(
-            url_to_bundle_path("~/dist/js/bundle.js.map").unwrap(),
-            "_/_/dist/js/bundle.js.map"
-        );
-        assert_eq!(
-            url_to_bundle_path("~/babel.config.js").unwrap(),
-            "_/_/babel.config.js"
-        );
-
-        assert_eq!(url_to_bundle_path("~/#/bar").unwrap(), "_/_/#/bar");
-        assert_eq!(url_to_bundle_path("~/foo/#/bar").unwrap(), "_/_/foo/#/bar");
-        assert_eq!(
-            url_to_bundle_path("~/dist/#js/bundle.js.map").unwrap(),
-            "_/_/dist/#js/bundle.js.map"
-        );
-        assert_eq!(
-            url_to_bundle_path("~/#foo/babel.config.js").unwrap(),
-            "_/_/#foo/babel.config.js"
-        );
-    }
-
-    #[test]
-    fn build_artifact_bundle_deterministic() {
-        let projects_slice = &["wat-project".into()];
-        let context = UploadContext {
-            org: "wat-org",
-            projects: Some(projects_slice.into()),
-            release: None,
-            dist: None,
-            note: None,
-            wait: false,
-            max_wait: DEFAULT_MAX_WAIT,
-            chunk_upload_options: None,
-        };
-
-        let source_files = ["bundle.min.js.map", "vendor.min.js.map"]
-            .into_iter()
-            .map(|name| {
-                let file = SourceFile {
-                    url: format!("~/{name}"),
-                    path: format!("tests/integration/_fixtures/{name}").into(),
-                    contents: std::fs::read(format!("tests/integration/_fixtures/{name}"))
-                        .unwrap()
-                        .into(),
-                    ty: SourceFileType::SourceMap,
-                    headers: Default::default(),
-                    messages: Default::default(),
-                    already_uploaded: false,
-                };
-                (format!("~/{name}"), file)
-            })
-            .collect();
-
-        let file = build_artifact_bundle(&context, &source_files, None).unwrap();
-
-        let buf = std::fs::read(file.path()).unwrap();
-        let hash = Sha1::from(buf);
-        assert_eq!(
-            hash.digest().to_string(),
-            "f0e25ae149b711c510148e022ebc883ad62c7c4c"
-        );
-    }
-}

src/utils/mod.rs

@@ -21,6 +21,7 @@ pub mod progress;
 pub mod proguard;
 pub mod releases;
 pub mod retry;
+pub mod source_bundle;
 pub mod sourcemaps;
 pub mod system;
 pub mod ui;