Add support for UTF-32 matching, and other fixes

.github/workflows/ci.yml

@@ -50,14 +50,14 @@ jobs:
       with:
         toolchain: ${{ matrix.rust }}
     - name: Build
-      run: cargo build --verbose --all
+      run: cargo build --verbose --all --features utf32
     - name: Build docs
-      run: cargo doc --verbose --all
+      run: cargo doc --verbose --all --features utf32
     - name: Run tests
-      run: cargo test --verbose --all
+      run: cargo test --verbose --all --features utf32
     - name: Run tests with static build
       shell: bash
-      run: PCRE2_SYS_STATIC=1 cargo test --verbose --all
+      run: PCRE2_SYS_STATIC=1 cargo test --verbose --all  --features utf32
 
   rustfmt:
     name: rustfmt

Cargo.toml

@@ -19,3 +19,12 @@ members = ["pcre2-sys"]
 libc = "0.2.146"
 log = "0.4.19"
 pcre2-sys = { version = "0.2.7", path = "pcre2-sys" }
+
+[features]
+default = ["jit"]
+
+# Enable matching on UTF-32 strings
+utf32 = ["pcre2-sys/utf32"]
+
+# Enable the PCRE2 JIT
+jit = ["pcre2-sys/jit"]

pcre2-sys/Cargo.toml

@@ -18,3 +18,7 @@ libc = "0.2.146"
 [build-dependencies]
 cc = { version = "1.0.73", features = ["parallel"] }
 pkg-config = "0.3.27"
+
+[features]
+utf32 = []
+jit = []

pcre2-sys/build.rs

@@ -21,35 +21,29 @@
 
 use std::path::PathBuf;
 
-fn main() {
-    println!("cargo:rerun-if-env-changed=PCRE2_SYS_STATIC");
-
+// Build and link against a PCRE2 library with the given code unit width,
+// which should be "8" or "32".
+fn build_1_pcre2_lib(code_unit_width: &str) {
     let target = std::env::var("TARGET").unwrap();
-    // let out = PathBuf::from(std::env::var_os("OUT_DIR").unwrap());
     let upstream = PathBuf::from("upstream");
-
-    // Don't link to a system library if we want a static build.
-    let want_static = pcre2_sys_static().unwrap_or(target.contains("musl"));
-    if !want_static && pkg_config::probe_library("libpcre2-8").is_ok() {
-        return;
-    }
-
     // Set some config options. We mostly just use the default values. We do
     // this in lieu of patching config.h since it's easier.
     let mut builder = cc::Build::new();
     builder
-        .define("PCRE2_CODE_UNIT_WIDTH", "8")
+        .define("PCRE2_CODE_UNIT_WIDTH", code_unit_width)
         .define("HAVE_STDLIB_H", "1")
         .define("HAVE_MEMMOVE", "1")
         .define("HAVE_CONFIG_H", "1")
         .define("PCRE2_STATIC", "1")
         .define("STDC_HEADERS", "1")
-        .define("SUPPORT_PCRE2_8", "1")
+        .define(&format!("SUPPORT_PCRE2_{}", code_unit_width), "1")
         .define("SUPPORT_UNICODE", "1");
     if target.contains("windows") {
         builder.define("HAVE_WINDOWS_H", "1");
     }
-    enable_jit(&target, &mut builder);
+    if feature_enabled("JIT") {
+        enable_jit(&target, &mut builder);
+    }
 
     builder.include(upstream.join("src")).include(upstream.join("include"));
     for result in std::fs::read_dir(upstream.join("src")).unwrap() {
@@ -78,7 +72,34 @@ fn main() {
     {
         builder.debug(true);
     }
-    builder.compile("libpcre2.a");
+    builder.compile(&format!("libpcre2-{}.a", code_unit_width));
+}
+
+fn main() {
+    println!("cargo:rerun-if-env-changed=PCRE2_SYS_STATIC");
+
+    let target = std::env::var("TARGET").unwrap();
+    let do_utf32 = feature_enabled("UTF32");
+
+    // Don't link to a system library if we want a static build.
+    let want_static = pcre2_sys_static().unwrap_or(target.contains("musl"));
+    if want_static || pkg_config::probe_library("libpcre2-8").is_err() {
+        build_1_pcre2_lib("8");
+    }
+    if do_utf32
+        && (want_static || pkg_config::probe_library("libpcre2-32").is_err())
+    {
+        build_1_pcre2_lib("32");
+    }
+}
+
+// Return whether a given feature is enabled.
+fn feature_enabled(feature: &str) -> bool {
+    let env_var_name = format!("CARGO_FEATURE_{}", feature);
+    match std::env::var(&env_var_name) {
+        Ok(s) => s == "1",
+        Err(_) => false,
+    }
 }
 
 fn pcre2_sys_static() -> Option<bool> {

pcre2-sys/generate-bindings

@@ -5,7 +5,7 @@
 
 if ! command -V bindgen > /dev/null 2>&1; then
     echo "bindgen must be installed" >&2
-    echo "to install: cargo install bindgen" >&2
+    echo "to install: cargo install bindgen-cli" >&2
     exit 1
 fi
 if ! [ -f "$PCRE2SYS_HEADER" ]; then
@@ -14,6 +14,12 @@ if ! [ -f "$PCRE2SYS_HEADER" ]; then
     exit 1
 fi
 
+if [ -z "$PCRE2_CODE_UNIT_WIDTH" ]; then
+    echo "The PCRE2_CODE_UNIT_WIDTH environment variable must be set" >&2
+    echo "Valid values are 8, 16, and 32" >&2
+    exit 1
+fi
+
 bindgen \
     "$PCRE2SYS_HEADER" \
     --ctypes-prefix '::libc' \
@@ -22,4 +28,4 @@ bindgen \
     --allowlist-var '^PCRE2_.*' \
     --blocklist-function '^.*_callout_.*' \
     --blocklist-type '^.*_callout_.*' \
-    -- -DPCRE2_CODE_UNIT_WIDTH=8 > "$PCRE2SYS_BINDINGS"
+    -- -DPCRE2_CODE_UNIT_WIDTH=${PCRE2_CODE_UNIT_WIDTH} > "$PCRE2SYS_BINDINGS"