add c14n for node and document

Author: saks

src/tree/c14n.rs

@@ -0,0 +1,44 @@
+//! Shared canonicalization logic and types.
+//!
+use std::ffi::c_int;
+
+use crate::bindings::{
+  xmlC14NMode_XML_C14N_1_0, xmlC14NMode_XML_C14N_1_1, xmlC14NMode_XML_C14N_EXCLUSIVE_1_0,
+};
+
+/// Options for configuring how to canonicalize XML
+#[derive(Clone, Eq, PartialEq, Ord, PartialOrd, Hash, Debug, Default)]
+pub struct CanonicalizationOptions {
+  /// Canonicalization specification to use
+  pub mode: CanonicalizationMode,
+  /// If true, keep `<!-- ... -->` comments, otherwise remove
+  pub with_comments: bool,
+  /// Namespaces to keep even if they are unused. By default, in [CanonicalizationMode::ExclusiveCanonical1_0], unused namespaces are removed.
+  ///
+  /// Doesn't apply to other canonicalization modes.
+  pub inclusive_ns_prefixes: Vec<String>,
+}
+
+/// Canonicalization specification to use
+#[derive(Copy, Clone, Eq, PartialEq, Ord, PartialOrd, Hash, Debug, Default)]
+pub enum CanonicalizationMode {
+  /// Original C14N 1.0 spec
+  Canonical1_0,
+  /// Exclusive C14N 1.0 spec
+  #[default]
+  ExclusiveCanonical1_0,
+  /// C14N 1.1 spec
+  Canonical1_1,
+}
+
+impl From<CanonicalizationMode> for c_int {
+  fn from(mode: CanonicalizationMode) -> Self {
+    let c14n_mode = match mode {
+      CanonicalizationMode::Canonical1_0 => xmlC14NMode_XML_C14N_1_0,
+      CanonicalizationMode::ExclusiveCanonical1_0 => xmlC14NMode_XML_C14N_EXCLUSIVE_1_0,
+      CanonicalizationMode::Canonical1_1 => xmlC14NMode_XML_C14N_1_1,
+    };
+
+    c_int::from(c14n_mode as i32)
+  }
+}

src/tree/document.rs

@@ -346,3 +346,5 @@ impl Document {
     Ok(())
   }
 }
+
+mod c14n;

src/tree/document/c14n.rs

@@ -0,0 +1,111 @@
+//! Document canonicalization logic
+//!
+use std::ffi::{c_int, c_void, CString};
+use std::os::raw;
+use std::ptr::null_mut;
+
+use crate::tree::c14n::*;
+
+use super::{
+  xmlAllocOutputBuffer, xmlC14NExecute, xmlC14NIsVisibleCallback, xmlChar, xmlNodePtr,
+  xmlOutputBufferClose, xmlOutputBufferPtr, Document,
+};
+
+impl Document {
+  /// Canonicalize a document and return the results.
+  pub fn canonicalize(
+    &self,
+    options: CanonicalizationOptions,
+    callback: Option<(xmlNodePtr, xmlC14NIsVisibleCallback)>,
+  ) -> Result<String, ()> {
+    let document = (*self.0).borrow().doc_ptr;
+
+    let mut ns_list_c = to_xml_string_vec(options.inclusive_ns_prefixes);
+    let inclusive_ns_prefixes = ns_list_c.as_mut_ptr();
+    let with_comments = c_int::from(options.with_comments);
+
+    let (is_visible_callback, user_data) = if let Some((node_ptr, visibility_callback)) = callback {
+      (visibility_callback, node_ptr as *mut _)
+    } else {
+      (None, null_mut())
+    };
+
+    let mode = options.mode.into();
+    unsafe {
+      let c_obuf = create_output_buffer();
+
+      let status = xmlC14NExecute(
+        document,
+        is_visible_callback,
+        user_data,
+        mode,
+        inclusive_ns_prefixes,
+        with_comments,
+        c_obuf,
+      );
+
+      let res = c_obuf_into_output(c_obuf);
+
+      if status < 0 {
+        Err(())
+      } else {
+        Ok(res)
+      }
+    }
+  }
+}
+
+unsafe fn c_obuf_into_output(c_obuf: xmlOutputBufferPtr) -> String {
+  let ctx_ptr = (*c_obuf).context;
+  let output = Box::from_raw(ctx_ptr as *mut String);
+
+  (*c_obuf).context = std::ptr::null_mut::<c_void>();
+
+  xmlOutputBufferClose(c_obuf);
+
+  *output
+}
+
+unsafe fn create_output_buffer() -> xmlOutputBufferPtr {
+  let output = String::new();
+  let ctx_ptr = Box::into_raw(Box::new(output));
+  let encoder = std::ptr::null_mut();
+
+  let buf = xmlAllocOutputBuffer(encoder);
+
+  (*buf).writecallback = Some(xml_write_io);
+  (*buf).closecallback = Some(xml_close_io);
+  (*buf).context = ctx_ptr as _;
+
+  buf
+}
+
+unsafe extern "C" fn xml_close_io(_context: *mut raw::c_void) -> raw::c_int {
+  0
+}
+
+unsafe extern "C" fn xml_write_io(
+  io_ptr: *mut raw::c_void,
+  buffer: *const raw::c_char,
+  len: raw::c_int,
+) -> raw::c_int {
+  if io_ptr.is_null() {
+    0
+  } else {
+    let buf = std::slice::from_raw_parts_mut(buffer as *mut u8, len as usize);
+    let buf = String::from_utf8_lossy(buf);
+    let s2_ptr = io_ptr as *mut String;
+    String::push_str(&mut *s2_ptr, &buf);
+
+    len
+  }
+}
+
+/// Create a [Vec] of null-terminated [*mut xmlChar] strings
+fn to_xml_string_vec(vec: Vec<String>) -> Vec<*mut xmlChar> {
+  vec
+    .into_iter()
+    .map(|s| CString::new(s).unwrap().into_raw() as *mut xmlChar)
+    .chain(std::iter::once(std::ptr::null_mut()))
+    .collect()
+}

src/tree/mod.rs

@@ -1,6 +1,7 @@
 //! The tree functionality
 //!
 
+pub mod c14n;
 pub mod document;
 pub mod namespace;
 pub mod node;

src/tree/node.rs

@@ -1039,6 +1039,34 @@ impl Node {
     context.findnodes(xpath, Some(self))
   }
 
+  /// Search this node for XPath `path`, and return only the first match.
+  pub fn at_xpath(&self, path: &str, ns_binlings: &[(&str, &str)]) -> Result<Option<Node>, ()> {
+    let mut context = Context::from_node(self)?;
+    for (prefix, href) in ns_binlings {
+      context.register_namespace(prefix, href)?;
+    }
+    let nodes = context.findnodes(path, Some(self))?;
+
+    Ok(nodes.first().cloned())
+  }
+
+  /// Get a list of ancestor Node for this Node.
+  pub fn ancestors(&self) -> Vec<Node> {
+    let node_ptr = self.node_ptr();
+
+    let mut res = Vec::new();
+
+    let ancestor_ptrs = node_ancestors(node_ptr);
+
+    for ptr in ancestor_ptrs {
+      if let Some(node) = self.ptr_as_option(ptr) {
+        res.push(node)
+      }
+    }
+
+    res
+  }
+
   /// find String values via xpath, at a specified node or the document root
   pub fn findvalues(&self, xpath: &str) -> Result<Vec<String>, ()> {
     let mut context = Context::from_node(self)?;
@@ -1085,3 +1113,26 @@ impl Node {
     }
   }
 }
+
+fn node_ancestors(node_ptr: xmlNodePtr) -> Vec<xmlNodePtr> {
+  if node_ptr.is_null() {
+    return Vec::new();
+  }
+
+  let mut parent_ptr = xmlGetParent(node_ptr);
+
+  if parent_ptr.is_null() {
+    Vec::new()
+  } else {
+    let mut parents = vec![parent_ptr];
+
+    while !xmlGetParent(parent_ptr).is_null() {
+      parent_ptr = xmlGetParent(parent_ptr);
+      parents.push(parent_ptr);
+    }
+
+    parents
+  }
+}
+
+mod c14n;

src/tree/node/c14n.rs

@@ -0,0 +1,58 @@
+//! Node canonicalization logic
+//!
+use std::ffi::c_void;
+
+use crate::{
+  bindings::{xmlC14NIsVisibleCallback, xmlNodePtr},
+  c_helpers::xmlGetNodeType,
+  tree::{c14n::*, Node},
+};
+
+use super::node_ancestors;
+
+impl Node {
+  /// Canonicalize a document and return the results.
+  pub fn canonicalize(&mut self, options: CanonicalizationOptions) -> Result<String, ()> {
+    let doc_ref = self.get_docref().upgrade().unwrap();
+    let document = crate::tree::Document(doc_ref);
+
+    let user_data = self.node_ptr_mut().unwrap();
+    let callback: xmlC14NIsVisibleCallback = Some(callback_wrapper);
+
+    document.canonicalize(options, Some((user_data, callback)))
+  }
+}
+
+unsafe extern "C" fn callback_wrapper(
+  c14n_root_ptr: *mut c_void,
+  node_ptr: xmlNodePtr,
+  parent_ptr: xmlNodePtr,
+) -> ::std::os::raw::c_int {
+  let c14n_root_ptr = c14n_root_ptr as xmlNodePtr;
+  let node_type = xmlGetNodeType(node_ptr);
+
+  let tn_ptr = if NODE_TYPES.contains(&node_type) {
+    node_ptr
+  } else {
+    parent_ptr
+  };
+
+  let tn_ancestors = node_ancestors(tn_ptr);
+
+  let ret = (tn_ptr == c14n_root_ptr) || tn_ancestors.contains(&c14n_root_ptr);
+  if ret {
+    1
+  } else {
+    0
+  }
+}
+
+const NODE_TYPES: [u32; 7] = [
+  super::xmlElementType_XML_ELEMENT_NODE,
+  super::xmlElementType_XML_ATTRIBUTE_NODE,
+  super::xmlElementType_XML_DOCUMENT_TYPE_NODE,
+  super::xmlElementType_XML_TEXT_NODE,
+  super::xmlElementType_XML_DTD_NODE,
+  super::xmlElementType_XML_PI_NODE,
+  super::xmlElementType_XML_COMMENT_NODE,
+];