fix(common): Fix source map generation with inputSourceMap (#8546)

**Description:**

- This PR fixes the source map generation when `inputSourceMap` is specified.
- This PR fixes `minify()` not accepting parsed source map in the option.


**Related issue:**

 - Closes #8372.

Cargo.lock

@@ -3963,9 +3963,12 @@ dependencies = [
 name = "swc_config"
 version = "0.1.9"
 dependencies = [
+ "anyhow",
  "indexmap 2.1.0",
  "serde",
  "serde_json",
+ "sourcemap",
+ "swc_cached",
  "swc_config_macro",
 ]
 
@@ -4534,7 +4537,6 @@ dependencies = [
  "serde",
  "serde_json",
  "swc_atoms",
- "swc_cached",
  "swc_common",
  "swc_config",
  "swc_ecma_ast",

crates/swc/src/lib.rs

@@ -765,14 +765,8 @@ impl Compiler {
                 .source_map
                 .as_ref()
                 .map(|obj| -> Result<_, Error> {
-                    let orig = obj
-                        .content
-                        .as_ref()
-                        .map(|s| sourcemap::SourceMap::from_slice(s.as_bytes()));
-                    let orig = match orig {
-                        Some(v) => Some(v?),
-                        None => None,
-                    };
+                    let orig = obj.content.as_ref().map(|s| s.to_sourcemap()).transpose()?;
+
                     Ok((SourceMapsConfig::Bool(true), orig))
                 })
                 .unwrap_as_option(|v| {

crates/swc/tests/minify/issue-7475/issue-8372/1/config.json

@@ -0,0 +1,23 @@
+{
+    "compress": {
+        "negate_iife": false,
+        "sequences": 30
+    },
+    "mangle": {
+        "safari10": true
+    },
+    "output": {
+        "semicolons": false
+    },
+    "sourceMap": {
+        "filename": "input.js",
+        "url": "input.map",
+        "content": {
+            "version": 3,
+            "sources": ["input-preprocess.js"],
+            "sourcesContent": ["const a = {aób: 'ó'}\n\nconsole.log(a)"],
+            "names": ["console", "log", "aób"],
+            "mappings": "AAEAA,QAAQC,GAAG,CAFD;IAACC,UAAK;AAAG"
+        }
+    }
+}

crates/swc/tests/minify/issue-7475/issue-8372/1/input.js

@@ -0,0 +1,3 @@
+const a = { aób: 'ó' }
+
+console.log(a)

crates/swc/tests/minify/issue-7475/issue-8372/1/output.js

@@ -0,0 +1 @@
+let a={aób:"\xf3"};console.log(a);

crates/swc/tests/minify/issue-7475/issue-8372/1/output.map

@@ -0,0 +1,16 @@
+{
+  "mappings": "AAEAA,IAAAA,EAAQC,CAFEC,IAAA,MAAA,EAASF,QAAAC,GAAA,CAAAE",
+  "names": [
+    "console",
+    "log",
+    "aób",
+    "a"
+  ],
+  "sources": [
+    "input-preprocess.js"
+  ],
+  "sourcesContent": [
+    "const a = {aób: 'ó'}\n\nconsole.log(a)"
+  ],
+  "version": 3
+}

crates/swc/tests/minify/issue-7475/issue-8372/2/config.json

@@ -0,0 +1,12 @@
+{
+    "compress": {
+        "negate_iife": false,
+        "sequences": 30
+    },
+    "mangle": {
+        "safari10": true
+    },
+    "output": {
+        "semicolons": false
+    }
+}

crates/swc/tests/minify/issue-7475/issue-8372/2/input.js

@@ -0,0 +1,3 @@
+const a = { aób: 'ó' }
+
+console.log(a)

crates/swc/tests/minify/issue-7475/issue-8372/2/output.js

@@ -0,0 +1 @@
+let a={aób:"\xf3"};console.log(a);

crates/swc/tests/minify/issue-7475/issue-8372/2/output.map

@@ -0,0 +1,16 @@
+{
+  "mappings": "AAAA,IAAMA,EAAI,CAAEC,IAAK,MAAI,EAErBC,QAAQC,GAAG,CAACH",
+  "names": [
+    "a",
+    "aób",
+    "console",
+    "log"
+  ],
+  "sources": [
+    "$DIR/tests/minify/issue-7475/issue-8372/2/input.js"
+  ],
+  "sourcesContent": [
+    "const a = { aób: 'ó' }\n\nconsole.log(a)"
+  ],
+  "version": 3
+}

crates/swc/tests/projects.rs

@@ -1128,10 +1128,13 @@ fn minify(input_js: PathBuf) {
         let c = Compiler::new(cm);
         let fm = c.cm.load_file(&input_js).unwrap();
 
-        let mut config: JsMinifyOptions =
-            serde_json::from_str(&std::fs::read_to_string(&config_json_path).unwrap()).unwrap();
+        let config_str = std::fs::read_to_string(&config_json_path).unwrap();
+        let mut config: JsMinifyOptions = serde_json::from_str(&config_str).unwrap();
+
+        if config.source_map.inner().is_none() {
+            config.source_map = BoolOrDataConfig::from_bool(true);
+        }
 
-        config.source_map = BoolOrDataConfig::from_bool(true);
         let output = c.minify(fm, &handler, &config).unwrap();
 
         NormalizedOutput::from(output.code)

crates/swc_cached/src/lib.rs

@@ -4,3 +4,4 @@
 #![deny(warnings)]
 
 pub mod regex;
+pub use anyhow::Error;

crates/swc_cached/src/regex.rs

@@ -1,6 +1,6 @@
 //! Regex cache
 
-use std::{ops::Deref, sync::Arc};
+use std::{ops::Deref, str::FromStr, sync::Arc};
 
 pub use anyhow::Error;
 use anyhow::{Context, Result};
@@ -77,3 +77,11 @@ impl From<&'_ str> for CachedRegex {
         Self::new(s).unwrap()
     }
 }
+
+impl FromStr for CachedRegex {
+    type Err = Error;
+
+    fn from_str(s: &str) -> Result<Self, Self::Err> {
+        Self::new(s)
+    }
+}

crates/swc_common/src/source_map.rs

@@ -1269,11 +1269,13 @@ impl SourceMap {
                     if config.skip(&f.name) {
                         continue;
                     }
-                    src_id = builder.add_source(&config.file_name_to_source(&f.name));
+                    if orig.is_none() {
+                        src_id = builder.add_source(&config.file_name_to_source(&f.name));
 
-                    inline_sources_content = config.inline_sources_content(&f.name);
-                    if inline_sources_content && orig.is_none() {
-                        builder.set_source_contents(src_id, Some(&f.src));
+                        inline_sources_content = config.inline_sources_content(&f.name);
+                        if inline_sources_content && orig.is_none() {
+                            builder.set_source_contents(src_id, Some(&f.src));
+                        }
                     }
 
                     ch_state = ByteToCharPosState::default();

crates/swc_config/Cargo.toml

@@ -9,10 +9,13 @@ repository    = "https://github.com/swc-project/swc.git"
 version       = "0.1.9"
 
 [dependencies]
+anyhow     = "1"
 indexmap   = "2.0.0"
 serde      = { version = "1", features = ["derive"] }
 serde_json = "1"
+sourcemap  = { version = "6.4", optional = true }
 
+swc_cached       = { version = "0.3.18", path = "../swc_cached" }
 swc_config_macro = { version = "0.1.3", path = "../swc_config_macro" }
 
 [lib]

crates/swc_config/src/config_types/bool_or_data.rs

@@ -69,6 +69,14 @@ impl<T> BoolOrDataConfig<T> {
     pub fn into_inner(self) -> Option<BoolOr<T>> {
         self.0
     }
+
+    pub fn inner(&self) -> Option<BoolOr<&T>> {
+        match &self.0 {
+            Some(BoolOr::Data(v)) => Some(BoolOr::Data(v)),
+            Some(BoolOr::Bool(b)) => Some(BoolOr::Bool(*b)),
+            None => None,
+        }
+    }
 }
 
 impl<T> From<T> for BoolOrDataConfig<T> {

crates/swc_config/src/lib.rs

@@ -4,3 +4,10 @@
 mod macros;
 pub mod config_types;
 pub mod merge;
+#[cfg(feature = "sourcemap")]
+mod source_map;
+
+pub use swc_cached::{regex::CachedRegex, Error};
+
+#[cfg(feature = "sourcemap")]
+pub use crate::source_map::*;

crates/swc_config/src/source_map.rs

@@ -0,0 +1,118 @@
+use anyhow::{bail, Context, Result};
+use serde::{Deserialize, Serialize};
+use sourcemap::{vlq::parse_vlq_segment, RawToken, SourceMap};
+
+#[derive(Debug, Clone, Serialize, Deserialize)]
+#[serde(untagged)]
+pub enum SourceMapContent {
+    Json(String),
+    #[serde(rename_all = "camelCase")]
+    Parsed {
+        #[serde(default)]
+        sources: Vec<String>,
+        #[serde(default)]
+        names: Vec<String>,
+        #[serde(default)]
+        mappings: String,
+        #[serde(default)]
+        file: Option<String>,
+        #[serde(default)]
+        source_root: Option<String>,
+        #[serde(default)]
+        sources_content: Option<Vec<Option<String>>>,
+    },
+}
+
+impl SourceMapContent {
+    pub fn to_sourcemap(&self) -> Result<SourceMap> {
+        match self {
+            SourceMapContent::Json(s) => {
+                SourceMap::from_slice(s.as_bytes()).context("failed to parse sourcemap")
+            }
+            SourceMapContent::Parsed {
+                sources,
+                names,
+                mappings,
+                file,
+                source_root,
+                sources_content,
+            } => {
+                let mut dst_col;
+                let mut src_id = 0;
+                let mut src_line = 0;
+                let mut src_col = 0;
+                let mut name_id = 0;
+
+                let allocation_size = mappings.matches(&[',', ';'][..]).count() + 10;
+                let mut tokens = Vec::with_capacity(allocation_size);
+
+                let mut nums = Vec::with_capacity(6);
+
+                for (dst_line, line) in mappings.split(';').enumerate() {
+                    if line.is_empty() {
+                        continue;
+                    }
+
+                    dst_col = 0;
+
+                    for segment in line.split(',') {
+                        if segment.is_empty() {
+                            continue;
+                        }
+
+                        nums.clear();
+                        nums = parse_vlq_segment(segment)?;
+                        dst_col = (i64::from(dst_col) + nums[0]) as u32;
+
+                        let mut src = !0;
+                        let mut name = !0;
+
+                        if nums.len() > 1 {
+                            if nums.len() != 4 && nums.len() != 5 {
+                                bail!(
+                                    "invalid vlq segment size; expected 4 or 5, got {}",
+                                    nums.len()
+                                );
+                            }
+                            src_id = (i64::from(src_id) + nums[1]) as u32;
+                            if src_id >= sources.len() as u32 {
+                                bail!("invalid source reference: {}", src_id);
+                            }
+
+                            src = src_id;
+                            src_line = (i64::from(src_line) + nums[2]) as u32;
+                            src_col = (i64::from(src_col) + nums[3]) as u32;
+
+                            if nums.len() > 4 {
+                                name_id = (i64::from(name_id) + nums[4]) as u32;
+                                if name_id >= names.len() as u32 {
+                                    bail!("invalid name reference: {}", name_id);
+                                }
+                                name = name_id;
+                            }
+                        }
+
+                        tokens.push(RawToken {
+                            dst_line: dst_line as u32,
+                            dst_col,
+                            src_line,
+                            src_col,
+                            src_id: src,
+                            name_id: name,
+                        });
+                    }
+                }
+
+                let mut map = SourceMap::new(
+                    file.clone(),
+                    tokens,
+                    names.clone(),
+                    sources.clone(),
+                    sources_content.clone(),
+                );
+                map.set_source_root(source_root.clone());
+                Ok(map)
+            }
+        }
+    }
+}

crates/swc_ecma_minifier/Cargo.toml

@@ -51,9 +51,10 @@ serde_json        = "1.0.61"
 tracing           = "0.1.37"
 
 swc_atoms = { version = "0.6.5", path = "../swc_atoms" }
-swc_cached = { version = "0.3.18", path = "../swc_cached" }
 swc_common = { version = "0.33.15", path = "../swc_common" }
-swc_config = { version = "0.1.9", path = "../swc_config" }
+swc_config = { version = "0.1.9", path = "../swc_config", features = [
+  "sourcemap",
+] }
 swc_ecma_ast = { version = "0.111.1", path = "../swc_ecma_ast", features = [
   "serde",
 ] }

crates/swc_ecma_minifier/src/js.rs

@@ -1,7 +1,7 @@
 //! NOT A PUBLIC API
 
 use serde::{Deserialize, Serialize};
-use swc_config::config_types::BoolOrDataConfig;
+use swc_config::{config_types::BoolOrDataConfig, SourceMapContent};
 
 use crate::option::{
     terser::{TerserCompressorOptions, TerserEcmaVersion},
@@ -59,6 +59,8 @@ fn true_by_default() -> bool {
     true
 }
 
+/// `sourceMap` of `minify()`.`
+///
 /// `jsc.minify.sourceMap`
 #[derive(Debug, Clone, Default, Serialize, Deserialize)]
 #[serde(deny_unknown_fields, rename_all = "camelCase")]
@@ -73,7 +75,7 @@ pub struct TerserSourceMapOption {
     pub root: Option<String>,
 
     #[serde(default)]
-    pub content: Option<String>,
+    pub content: Option<SourceMapContent>,
 }
 
 /// Parser options for `minify()`, which should have the same API as terser.

crates/swc_ecma_minifier/src/option/mod.rs

@@ -2,9 +2,8 @@
 
 use serde::{Deserialize, Serialize};
 use swc_atoms::JsWord;
-use swc_cached::regex::CachedRegex;
 use swc_common::{collections::AHashMap, Mark};
-use swc_config::merge::Merge;
+use swc_config::{merge::Merge, CachedRegex};
 use swc_ecma_ast::{EsVersion, Expr};
 
 /// Implement default using serde.