Using libxml2 source code

.github/workflows/CI.yml

@@ -18,6 +18,8 @@ jobs:
         run: echo "LIBXML2=$(pkg-config libxml-2.0 --variable=libdir)/libxml2.so" >> "$GITHUB_ENV"
         if: ${{ matrix.with_default_bindings }}
       - uses: actions/checkout@v2
+        with:
+          submodules: 'true'
       - uses: actions-rs/toolchain@v1
         with:
           profile: minimal
@@ -40,6 +42,8 @@ jobs:
         with:
           packages: "libpython3-dev"
       - uses: actions/checkout@v2
+        with:
+          submodules: 'true'
       - name: Install libxml ${{ matrix.libxml_version }} by hand
         run: |
           wget https://download.gnome.org/sources/libxml2/$(echo ${{ matrix.libxml_version }} | sed -e 's/\.[0-9]*$//')/libxml2-${{ matrix.libxml_version }}.tar.xz

.github/workflows/gh-pages.yml

@@ -23,6 +23,8 @@ jobs:
           override: true
     - name: Checkout sources
       uses: actions/checkout@v4
+      with:
+        submodules: 'true'
     - name: Build Documentation
       uses: actions-rs/cargo@v1
       with:

.github/workflows/windows.yml

@@ -12,6 +12,8 @@ jobs:
       VCPKG_ROOT: C:\vcpkg
     steps:
       - uses: actions/checkout@v4
+        with:
+          submodules: 'true'
       - name: Setup vcpkg libxml2 Cache
         uses: actions/cache@v4
         id: vcpkg-cache
@@ -35,6 +37,8 @@ jobs:
         shell: msys2 {0}
     steps:
       - uses: actions/checkout@v4
+        with:
+          submodules: 'true'
       - uses: msys2/setup-msys2@v2
         with:
           path-type: minimal

.gitmodules

@@ -0,0 +1,4 @@
+[submodule "libxml-sys/libxml2"]
+	path = libxml-sys/libxml2
+	url = https://github.com/GNOME/libxml2.git
+	branch = 2.15

Cargo.toml

@@ -1,44 +1,42 @@
-[package]
-name = "libxml"
+[workspace]
+members = ['.', "libxml-sys"]
+[workspace.package]
 version = "0.3.8"
 edition = "2024"
-authors = ["Andreas Franzén <andreas@devil.se>", "Deyan Ginev <deyan.ginev@gmail.com>","Jan Frederik Schaefer <j.schaefer@jacobs-university.de>"]
+authors = [
+  "Andreas Franzén <andreas@devil.se>",
+  "Deyan Ginev <deyan.ginev@gmail.com>",
+  "Jan Frederik Schaefer <j.schaefer@jacobs-university.de>",
+]
 description = "A Rust wrapper for libxml2 - the XML C parser and toolkit developed for the Gnome project"
 repository = "https://github.com/KWARC/rust-libxml"
-documentation = "https://kwarc.github.io/rust-libxml/libxml/index.html"
 readme = "README.md"
 license = "MIT"
-keywords = ["xml", "libxml","xpath", "parser", "html"]
-build = "build.rs"
-exclude = [
-  "scripts/*"
-]
+keywords = ["xml", "libxml", "xpath", "parser", "html"]
+
+[package]
+name = "libxml"
+version.workspace = true
+edition.workspace = true
+authors.workspace = true
+description.workspace = true
+repository.workspace = true
+documentation = "https://kwarc.github.io/rust-libxml/libxml/index.html"
+readme.workspace = true
+license.workspace = true
+keywords.workspace = true
+exclude = ["scripts/*"]
 
 [lib]
 name = "libxml"
 
+[features]
+iconv = ["libxml-sys/iconv"]
+zlib = ["libxml-sys/zlib"]
+
 [dependencies]
 libc = "0.2"
-
-[target.'cfg(all(target_family = "windows", target_env = "msvc"))'.build-dependencies]
-vcpkg = "0.2"
-
-[target.'cfg(all(target_family = "windows", target_env = "gnu"))'.build-dependencies]
-pkg-config = "0.3.2"
-
-[target.'cfg(macos)'.build-dependencies]
-pkg-config = "0.3.2"
-
-[target.'cfg(unix)'.build-dependencies]
-pkg-config = "0.3.2"
-
-[build-dependencies.bindgen]
-version = "0.72"
-features = [
-    "runtime",
-]
-default-features = false
-
+libxml-sys = { path = "libxml-sys" }
 [dev-dependencies]
 rayon = "1.0.0"
 criterion = "0.8.0"

libxml-sys/Cargo.toml

@@ -0,0 +1,28 @@
+[package]
+name = "libxml-sys"
+version.workspace = true
+edition.workspace = true
+authors.workspace = true
+description.workspace = true
+repository.workspace = true
+readme.workspace = true
+license.workspace = true
+keywords.workspace = true
+
+[features]
+iconv = []
+zlib = []
+
+[dependencies]
+
+
+[build-dependencies]
+cmake = "0.1.57"
+bindgen = "0.72.1"
+regex = "1.12.3"
+
+[package.metadata.cross.target.x86_64-pc-windows-gnu]
+pre-build = [
+    "apt-get update",
+    "apt-get install -y win-iconv-mingw-w64-dev libz-mingw-w64-dev",
+]

libxml-sys/README.md

@@ -0,0 +1 @@
+* Build libxml2 from source

libxml-sys/build.rs

@@ -0,0 +1,74 @@
+use std::{env, fs, path::PathBuf};
+
+fn main() {
+  let out_dir = env::var("OUT_DIR").unwrap();
+
+  let iconv = if cfg!(feature = "iconv") { "ON" } else { "OFF" };
+  let zlib = if cfg!(feature = "zlib") { "ON" } else { "OFF" };
+
+  let path = cmake::Config::new("libxml2")
+    .define("BUILD_SHARED_LIBS", "OFF")
+    .define("LIBXML2_WITH_ICONV", iconv)
+    .define("LIBXML2_WITH_ZLIB", zlib)
+    .define("LIBXML2_WITH_C14N", "ON")
+    .build();
+
+  println!("cargo::rerun-if-changed=libxml2");
+
+  let host = env::var("HOST").unwrap();
+
+  let mut libs = std::process::Command::new("sh")
+    .arg(path.join("bin/xml2-config"))
+    .arg("--libs")
+    .output()
+    .map(|output| String::from_utf8_lossy(&output.stdout).to_string());
+  let mut cflags = std::process::Command::new("sh")
+    .arg(path.join("bin/xml2-config"))
+    .arg("--cflags")
+    .output()
+    .map(|output| String::from_utf8_lossy(&output.stdout).to_string());
+
+  if host.contains("windows") {
+    let reg = regex::Regex::new("-(.)/(.)/").expect("reg");
+    libs = libs.map(|v| reg.replace_all(&v, "-$1$2:/").to_string());
+    cflags = cflags.map(|v| reg.replace_all(&v, "-$1$2:/").to_string());
+  }
+  // NOTE: Manually specify
+  let mut libs = libs.unwrap_or_else(|_| format!("-L{} -lxml2", path.join("lib").display()));
+  if host.contains("msvc") {
+    let mut iters = fs::read_dir(path.join("lib"))
+      .expect("read_dir")
+      .filter_map(|p| {
+        p.ok().and_then(|p| {
+          let metadata = p.metadata().ok()?;
+          let file_name = p.file_name();
+          let name = file_name.to_string_lossy();
+          if metadata.is_file() && name.starts_with("libxml2") && name.ends_with(".lib") {
+            return Some(
+              name
+                .trim_end_matches(".lib")
+                .to_string(),
+            );
+          }
+          None
+        })
+      });
+    let name = iters.next().expect("xml name");
+    println!("cargo:rustc-link-lib=bcrypt");
+    libs = libs.replace("-lxml2", &format!("-l{}", name));
+  }
+  println!("cargo::rustc-flags={}", libs);
+
+  // Note: Manually specify
+  let cflags = cflags.unwrap_or_else(|_| format!("-I{}", path.join("include/libxml2").display()));
+  let bindings_path = PathBuf::from(out_dir).join("bindings.rs");
+  bindgen::builder()
+    .opaque_type("max_align_t")
+    .header("src/wrapper.h")
+    .clang_args(&["-DLIBXML_C14N_ENABLED", "-DLIBXML_OUTPUT_ENABLED"])
+    .clang_args(cflags.split_whitespace())
+    .generate()
+    .expect("failed to generate bindings with bindgen")
+    .write_to_file(bindings_path)
+    .expect("Failed to write bindings.rs");
+}