Revert "Revert "Rust: allow to specify more cargo configuration options""

Cargo.lock

@@ -389,10 +389,12 @@ dependencies = [
  "log",
  "num-traits",
  "ra_ap_base_db",
+ "ra_ap_cfg",
  "ra_ap_hir",
  "ra_ap_hir_def",
  "ra_ap_hir_expand",
  "ra_ap_ide_db",
+ "ra_ap_intern",
  "ra_ap_load-cargo",
  "ra_ap_parser",
  "ra_ap_paths",

rust/codeql-extractor.yml

@@ -35,3 +35,22 @@ options:
       reduce execution time of consecutive extractor runs. By default, a new scratch
       directory is used for each run.
     type: string
+  cargo_target:
+    title: Target architecture
+    description: >
+      Target architecture to use for analysis, analogous to `cargo --target`. By
+      default the host architecture is used.
+    type: string
+  cargo_features:
+    title: Cargo features to turn on
+    description: >
+      Comma-separated list of features to turn on. If any value is `*` all features
+      are turned on. By default only default cargo features are enabled. Can be
+      repeated.
+    type: array
+  cargo_cfg_overrides:
+    title: Cargo cfg overrides
+    description: >
+      Comma-separated list of cfg settings to enable, or disable if prefixed with `-`.
+      Can be repeated.
+    type: array

rust/extractor/Cargo.toml

@@ -22,6 +22,8 @@ ra_ap_syntax = "0.0.232"
 ra_ap_vfs = "0.0.232"
 ra_ap_parser = "0.0.232"
 ra_ap_span = "0.0.232"
+ra_ap_cfg = "0.0.232"
+ra_ap_intern = "0.0.232"
 serde = "1.0.209"
 serde_with = "3.9.0"
 stderrlog = "0.6.0"

rust/extractor/macros/src/lib.rs

@@ -1,11 +1,36 @@
 use proc_macro::TokenStream;
 use quote::{format_ident, quote};
+use syn::{Ident, Type};
+
+fn get_type_tip(t: &Type) -> Option<&Ident> {
+    let syn::Type::Path(path) = t else {
+        return None;
+    };
+    let segment = path.path.segments.last()?;
+    Some(&segment.ident)
+}
 
 /// Allow all fields in the extractor config to be also overrideable by extractor CLI flags
 #[proc_macro_attribute]
 pub fn extractor_cli_config(_attr: TokenStream, item: TokenStream) -> TokenStream {
     let ast = syn::parse_macro_input!(item as syn::ItemStruct);
     let name = &ast.ident;
+    let fields = ast
+        .fields
+        .iter()
+        .map(|f| {
+            if f.ident.as_ref().is_some_and(|i| i != "inputs")
+                && get_type_tip(&f.ty).is_some_and(|i| i == "Vec")
+            {
+                quote! {
+                    #[serde(deserialize_with="deserialize_newline_or_comma_separated")]
+                    #f
+                }
+            } else {
+                quote! { #f }
+            }
+        })
+        .collect::<Vec<_>>();
     let cli_name = format_ident!("Cli{}", name);
     let cli_fields = ast
         .fields
@@ -13,35 +38,37 @@ pub fn extractor_cli_config(_attr: TokenStream, item: TokenStream) -> TokenStrea
         .map(|f| {
             let id = f.ident.as_ref().unwrap();
             let ty = &f.ty;
-            if let syn::Type::Path(p) = ty {
+            let type_tip = get_type_tip(ty);
-                if p.path.is_ident(&format_ident!("bool")) {
+            if type_tip.is_some_and(|i| i == "bool") {
-                    return quote! {
+                quote! {
-                        #[arg(long)]
+                    #[arg(long)]
-                        #[serde(skip_serializing_if="<&bool>::not")]
+                    #[serde(skip_serializing_if="<&bool>::not")]
-                        #id: bool,
+                    #id: bool
-                    };
                 }
-                if p.path.segments.len() == 1 && p.path.segments[0].ident == "Option" {
+            } else if type_tip.is_some_and(|i| i == "Option") {
-                    return quote! {
+                quote! {
-                        #[arg(long)]
+                    #[arg(long)]
-                        #id: #ty,
+                    #f
-                    };
                 }
-            }
+            } else if id == &format_ident!("verbose") {
-            if id == &format_ident!("verbose") {
                 quote! {
                     #[arg(long, short, action=clap::ArgAction::Count)]
                     #[serde(skip_serializing_if="u8::is_zero")]
-                    #id: u8,
+                    #id: u8
                 }
             } else if id == &format_ident!("inputs") {
                 quote! {
-                    #id: #ty,
+                    #f
+                }
+            } else if type_tip.is_some_and(|i| i == "Vec") {
+                quote! {
+                    #[arg(long)]
+                    #id: Option<String>
                 }
             } else {
                 quote! {
                     #[arg(long)]
-                    #id: Option<#ty>,
+                    #id: Option<#ty>
                 }
             }
         })
@@ -66,7 +93,9 @@ pub fn extractor_cli_config(_attr: TokenStream, item: TokenStream) -> TokenStrea
     let gen = quote! {
         #[serde_with::apply(_ => #[serde(default)])]
         #[derive(Deserialize, Default)]
-        #ast
+        pub struct #name {
+            #(#fields),*
+        }
 
         impl Debug for #name {
             fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
@@ -80,7 +109,7 @@ pub fn extractor_cli_config(_attr: TokenStream, item: TokenStream) -> TokenStrea
         #[derive(clap::Parser, Serialize)]
         #[command(about, long_about = None)]
         struct #cli_name {
-            #(#cli_fields)*
+            #(#cli_fields),*
         }
     };
     gen.into()

rust/extractor/src/config.rs

@@ -7,9 +7,14 @@ use figment::{
     Figment,
 };
 use itertools::Itertools;
+use log::warn;
 use num_traits::Zero;
+use ra_ap_cfg::{CfgAtom, CfgDiff};
+use ra_ap_intern::Symbol;
+use ra_ap_paths::Utf8PathBuf;
+use ra_ap_project_model::{CargoConfig, CargoFeatures, CfgOverrides, RustLibSource};
 use rust_extractor_macros::extractor_cli_config;
-use serde::{Deserialize, Serialize};
+use serde::{Deserialize, Deserializer, Serialize};
 use std::fmt::Debug;
 use std::ops::Not;
 use std::path::PathBuf;
@@ -32,12 +37,23 @@ impl From<Compression> for trap::Compression {
     }
 }
 
+// required by the extractor_cli_config macro.
+fn deserialize_newline_or_comma_separated<'a, D: Deserializer<'a>, T: for<'b> From<&'b str>>(
+    deserializer: D,
+) -> Result<Vec<T>, D::Error> {
+    let value = String::deserialize(deserializer)?;
+    Ok(value.split(['\n', ',']).map(T::from).collect())
+}
+
 #[extractor_cli_config]
 pub struct Config {
     pub scratch_dir: PathBuf,
     pub trap_dir: PathBuf,
     pub source_archive_dir: PathBuf,
     pub cargo_target_dir: Option<PathBuf>,
+    pub cargo_target: Option<String>,
+    pub cargo_features: Vec<String>,
+    pub cargo_cfg_overrides: Vec<String>,
     pub verbose: u8,
     pub compression: Compression,
     pub inputs: Vec<PathBuf>,
@@ -52,7 +68,7 @@ impl Config {
             .context("expanding parameter files")?;
         let cli_args = CliConfig::parse_from(args);
         let mut figment = Figment::new()
-            .merge(Env::prefixed("CODEQL_"))
+            .merge(Env::raw().only(["CODEQL_VERBOSE"].as_slice()))
             .merge(Env::prefixed("CODEQL_EXTRACTOR_RUST_"))
             .merge(Env::prefixed("CODEQL_EXTRACTOR_RUST_OPTION_"))
             .merge(Serialized::defaults(cli_args));
@@ -78,4 +94,81 @@ impl Config {
         }
         figment.extract().context("loading configuration")
     }
+
+    pub fn to_cargo_config(&self) -> CargoConfig {
+        let sysroot = Some(RustLibSource::Discover);
+
+        let target_dir = self
+            .cargo_target_dir
+            .clone()
+            .unwrap_or_else(|| self.scratch_dir.join("target"));
+        let target_dir = Utf8PathBuf::from_path_buf(target_dir).ok();
+
+        let features = if self.cargo_features.is_empty() {
+            Default::default()
+        } else if self.cargo_features.contains(&"*".to_string()) {
+            CargoFeatures::All
+        } else {
+            CargoFeatures::Selected {
+                features: self.cargo_features.clone(),
+                no_default_features: false,
+            }
+        };
+
+        let target = self.cargo_target.clone();
+
+        let cfg_overrides = to_cfg_overrides(&self.cargo_cfg_overrides);
+
+        CargoConfig {
+            sysroot,
+            target_dir,
+            features,
+            target,
+            cfg_overrides,
+            ..Default::default()
+        }
+    }
+}
+
+fn to_cfg_override(spec: &str) -> CfgAtom {
+    if let Some((key, value)) = spec.split_once("=") {
+        CfgAtom::KeyValue {
+            key: Symbol::intern(key),
+            value: Symbol::intern(value),
+        }
+    } else {
+        CfgAtom::Flag(Symbol::intern(spec))
+    }
+}
+
+fn to_cfg_overrides(specs: &Vec<String>) -> CfgOverrides {
+    let mut enabled_cfgs = Vec::new();
+    let mut disabled_cfgs = Vec::new();
+    let mut has_test_explicitly_enabled = false;
+    for spec in specs {
+        if spec.starts_with("-") {
+            disabled_cfgs.push(to_cfg_override(&spec[1..]));
+        } else {
+            enabled_cfgs.push(to_cfg_override(spec));
+            if spec == "test" {
+                has_test_explicitly_enabled = true;
+            }
+        }
+    }
+    if !has_test_explicitly_enabled {
+        disabled_cfgs.push(to_cfg_override("test"));
+    }
+    if let Some(global) = CfgDiff::new(enabled_cfgs, disabled_cfgs) {
+        CfgOverrides {
+            global,
+            ..Default::default()
+        }
+    } else {
+        warn!("non-disjoint cfg overrides, ignoring: {}", specs.join(", "));
+        CfgOverrides {
+            global: CfgDiff::new(Vec::new(), vec![to_cfg_override("test")])
+                .expect("disabling one cfg should always succeed"),
+            ..Default::default()
+        }
+    }
 }

rust/extractor/src/main.rs

@@ -130,11 +130,9 @@ fn main() -> anyhow::Result<()> {
         }
         extractor.extract_without_semantics(file, "no manifest found");
     }
-    let target_dir = &cfg
+    let cargo_config = cfg.to_cargo_config();
-        .cargo_target_dir
-        .unwrap_or_else(|| cfg.scratch_dir.join("target"));
     for (manifest, files) in map.values().filter(|(_, files)| !files.is_empty()) {
-        if let Some((ref db, ref vfs)) = RustAnalyzer::load_workspace(manifest, target_dir) {
+        if let Some((ref db, ref vfs)) = RustAnalyzer::load_workspace(manifest, &cargo_config) {
             let semantics = Semantics::new(db);
             for file in files {
                 let Some(id) = path_to_file_id(file, vfs) else {

rust/extractor/src/rust_analyzer.rs

@@ -7,7 +7,6 @@ use ra_ap_load_cargo::{load_workspace_at, LoadCargoConfig, ProcMacroServerChoice
 use ra_ap_paths::Utf8PathBuf;
 use ra_ap_project_model::CargoConfig;
 use ra_ap_project_model::ProjectManifest;
-use ra_ap_project_model::RustLibSource;
 use ra_ap_span::Edition;
 use ra_ap_span::EditionedFileId;
 use ra_ap_span::TextRange;
@@ -20,6 +19,7 @@ use ra_ap_vfs::{AbsPathBuf, FileId};
 use std::borrow::Cow;
 use std::path::{Path, PathBuf};
 use triomphe::Arc;
+
 pub enum RustAnalyzer<'a> {
     WithSemantics {
         vfs: &'a Vfs,
@@ -45,13 +45,8 @@ pub struct ParseResult<'a> {
 impl<'a> RustAnalyzer<'a> {
     pub fn load_workspace(
         project: &ProjectManifest,
-        target_dir: &Path,
+        config: &CargoConfig,
     ) -> Option<(RootDatabase, Vfs)> {
-        let config = CargoConfig {
-            sysroot: Some(RustLibSource::Discover),
-            target_dir: ra_ap_paths::Utf8PathBuf::from_path_buf(target_dir.to_path_buf()).ok(),
-            ..Default::default()
-        };
         let progress = |t| (log::trace!("progress: {}", t));
         let load_config = LoadCargoConfig {
             load_out_dirs_from_check: true,
@@ -60,7 +55,7 @@ impl<'a> RustAnalyzer<'a> {
         };
         let manifest = project.manifest_path();
 
-        match load_workspace_at(manifest.as_ref(), &config, &load_config, &progress) {
+        match load_workspace_at(manifest.as_ref(), config, &load_config, &progress) {
             Ok((db, vfs, _macro_server)) => Some((db, vfs)),
             Err(err) => {
                 log::error!("failed to load workspace for {}: {}", manifest, err);

rust/ql/integration-tests/options/cfg/Cargo.toml

@@ -0,0 +1,5 @@
+[workspace]
+[package]
+name = "cfg"
+version = "0.1.0"
+edition = "2021"

rust/ql/integration-tests/options/cfg/functions.expected

@@ -0,0 +1,3 @@
+| src/lib.rs:7:1:8:19 | cfg_no_flag |
+| src/lib.rs:10:1:11:18 | cfg_no_key |
+| src/lib.rs:16:1:17:24 | pointer_width_64 |

rust/ql/integration-tests/options/cfg/functions.override.expected

@@ -0,0 +1,4 @@
+| src/lib.rs:1:1:2:16 | cfg_flag |
+| src/lib.rs:4:1:5:15 | cfg_key |
+| src/lib.rs:13:1:14:12 | test |
+| src/lib.rs:19:1:20:24 | pointer_width_32 |

rust/ql/integration-tests/options/cfg/functions.ql

@@ -0,0 +1,5 @@
+import rust
+
+from Function f
+where f.hasExtendedCanonicalPath()
+select f

rust/ql/integration-tests/options/cfg/src/lib.rs

@@ -0,0 +1,20 @@
+#[cfg(cfg_flag)]
+fn cfg_flag() {}
+
+#[cfg(cfg_key = "value")]
+fn cfg_key() {}
+
+#[cfg(not(cfg_flag))]
+fn cfg_no_flag() {}
+
+#[cfg(not(cfg_key = "value"))]
+fn cfg_no_key() {}
+
+#[cfg(test)]
+fn test() {}
+
+#[cfg(target_pointer_width = "64")]
+fn pointer_width_64() {}
+
+#[cfg(target_pointer_width = "32")]
+fn pointer_width_32() {}

rust/ql/integration-tests/options/cfg/test_cfg_overrides_option.py

@@ -0,0 +1,19 @@
+import pytest
+import platform
+import os
+
+
+def test_default(codeql, rust):
+    codeql.database.create()
+
+
+@pytest.mark.ql_test(expected=".override.expected")
+def test_cfg_overrides(codeql, rust):
+    overrides = ",".join((
+        "cfg_flag",
+        "cfg_key=value",
+        "-target_pointer_width=64",
+        "target_pointer_width=32",
+        "test",
+    ))
+    codeql.database.create(extractor_option=f"cargo_cfg_overrides={overrides}")

rust/ql/integration-tests/options/features/Cargo.toml

@@ -0,0 +1,9 @@
+[workspace]
+[package]
+name = "features"
+version = "0.1.0"
+edition = "2021"
+
+[features]
+foo = []
+bar = []

rust/ql/integration-tests/options/features/functions.all.expected

@@ -0,0 +1,3 @@
+| src/lib.rs:1:1:2:11 | foo |
+| src/lib.rs:4:1:5:11 | bar |
+| src/lib.rs:7:1:7:14 | always |

rust/ql/integration-tests/options/features/functions.bar.expected

@@ -0,0 +1,2 @@
+| src/lib.rs:4:1:5:11 | bar |
+| src/lib.rs:7:1:7:14 | always |

rust/ql/integration-tests/options/features/functions.expected

@@ -0,0 +1 @@
+| src/lib.rs:7:1:7:14 | always |

rust/ql/integration-tests/options/features/functions.foo.expected

@@ -0,0 +1,2 @@
+| src/lib.rs:1:1:2:11 | foo |
+| src/lib.rs:7:1:7:14 | always |

rust/ql/integration-tests/options/features/functions.ql

@@ -0,0 +1,5 @@
+import rust
+
+from Function f
+where f.hasExtendedCanonicalPath()
+select f

rust/ql/integration-tests/options/features/src/lib.rs

@@ -0,0 +1,7 @@
+#[cfg(feature = "foo")]
+fn foo() {}
+
+#[cfg(feature = "bar")]
+fn bar() {}
+
+fn always() {}

rust/ql/integration-tests/options/features/test_features_option.py

@@ -0,0 +1,17 @@
+import pytest
+
+def test_default(codeql, rust):
+    codeql.database.create()
+
+@pytest.mark.parametrize("features",
+                         [
+                             pytest.param(p,
+                                          marks=pytest.mark.ql_test(expected=f".{e}.expected"))
+                             for p, e in (
+                                 ("foo", "foo"),
+                                 ("bar", "bar"),
+                                 ("*", "all"),
+                                 ("foo,bar", "all"))
+                         ])
+def test_features(codeql, rust, features):
+    codeql.database.create(extractor_option=f"cargo_features={features}")

rust/ql/integration-tests/options/target/Cargo.toml

@@ -0,0 +1,5 @@
+[workspace]
+[package]
+name = "target"
+version = "0.1.0"
+edition = "2021"

rust/ql/integration-tests/options/target/functions.Darwin.expected

@@ -0,0 +1 @@
+| src/lib.rs:7:1:8:13 | macos |

rust/ql/integration-tests/options/target/functions.Linux.expected

@@ -0,0 +1 @@
+| src/lib.rs:1:1:2:13 | linux |

rust/ql/integration-tests/options/target/functions.Windows.expected

@@ -0,0 +1 @@
+| src/lib.rs:4:1:5:15 | windows |

rust/ql/integration-tests/options/target/functions.ql

@@ -0,0 +1,5 @@
+import rust
+
+from Function f
+where f.hasExtendedCanonicalPath()
+select f

rust/ql/integration-tests/options/target/src/lib.rs

@@ -0,0 +1,8 @@
+#[cfg(target_os = "linux")]
+fn linux() {}
+
+#[cfg(target_os = "windows")]
+fn windows() {}
+
+#[cfg(target_os = "macos")]
+fn macos() {}

rust/ql/integration-tests/options/target/test_target_option.py

@@ -0,0 +1,19 @@
+import pytest
+import platform
+
+
+@pytest.mark.ql_test(expected=f".{platform.system()}.expected")
+def test_default(codeql, rust):
+    codeql.database.create()
+
+@pytest.mark.ql_test(expected=".Windows.expected")
+def test_target_windows(codeql, rust):
+    codeql.database.create(extractor_option="cargo_target=x86_64-pc-windows-msvc")
+
+@pytest.mark.ql_test(expected=".Darwin.expected")
+def test_target_macos(codeql, rust):
+    codeql.database.create(extractor_option="cargo_target=aarch64-apple-darwin")
+
+@pytest.mark.ql_test(expected=".Linux.expected")
+def test_target_linux(codeql, rust):
+    codeql.database.create(extractor_option="cargo_target=x86_64-unknown-linux-gnu")