feat(c14n): add canonicalization functionality

Author: Tieske

xmlua.rockspec

@@ -37,6 +37,7 @@ build = {
     ["xmlua"] = "xmlua.lua",
     ["xmlua.attribute"] = "xmlua/attribute.lua",
     ["xmlua.attribute-declaration"] = "xmlua/attribute-declaration.lua",
+    ["xmlua.c14n"] = "xmlua/c14n.lua",
     ["xmlua.cdata-section"] = "xmlua/cdata-section.lua",
     ["xmlua.converter"] = "xmlua/converter.lua",
     ["xmlua.comment"] = "xmlua/comment.lua",
@@ -51,6 +52,7 @@ build = {
     ["xmlua.html"] = "xmlua/html.lua",
     ["xmlua.html-sax-parser"] = "xmlua/html-sax-parser.lua",
     ["xmlua.libxml2"] = "xmlua/libxml2.lua",
+    ["xmlua.libxml2.c14n"] = "xmlua/libxml2/c14n.lua",
     ["xmlua.libxml2.dict"] = "xmlua/libxml2/dict.lua",
     ["xmlua.libxml2.encoding"] = "xmlua/libxml2/encoding.lua",
     ["xmlua.libxml2.entities"] = "xmlua/libxml2/entities.lua",

xmlua/c14n.lua

@@ -0,0 +1,105 @@
+local C14n = {}
+
+local libxml2 = require("xmlua.libxml2")
+local ffi = require("ffi")
+
+
+
+local C14N_MODES = {
+  C14N_1_0            = ffi.C.XML_C14N_1_0,           -- Original C14N 1.0 spec
+  C14N_EXCLUSIVE_1_0  = ffi.C.XML_C14N_EXCLUSIVE_1_0, -- Exclusive C14N 1.0 spec
+  C14N_1_1            = ffi.C.XML_C14N_1_1,           -- C14N 1.1 spec
+}
+
+
+
+local C14N_MODES_LOOKUP = {} -- lookup by name or number, returns the number
+for name, number in pairs(C14N_MODES) do
+  C14N_MODES_LOOKUP[name] = number
+  C14N_MODES_LOOKUP[number] = number
+end
+
+
+
+-- list can be a string (space separated), an array of strings, or nil
+local function getNamespacePrefixArray(list)
+  local list = list or {}
+
+  if type(list) == "string" then
+    -- list is a string, assume it is the space separated PrefixList attribute, split it
+    local list_str = list
+    list = {}
+    list_str:gsub("([^%s]+)", function(cap) list[#list+1] = cap end)
+  end
+
+  if #list == 0 then
+    return nil
+  end
+
+  local result = ffi.new('xmlChar*[?]', #list+1)
+  local refs = {}
+  for i, prefix in ipairs(list) do
+    local c_ns = ffi.new("unsigned char[?]", #prefix+1, prefix)
+    ffi.copy(c_ns, prefix)
+    result[i-1] = c_ns
+    refs[i] = c_ns -- hold on to refs to prevent GC while in use
+  end
+  result[#list] = nil
+
+  return ffi.gc(result, function(ptr)
+    refs = nil -- release references, so they can be GC'ed
+  end)
+end
+
+
+
+local function getNodesList(nodes)
+  if (not nodes) or #nodes == 0 then
+    return nil
+  end
+
+  local nodeTab = ffi.new("xmlNodePtr[?]", #nodes)
+  for i = 1, #nodes do
+      nodeTab[i - 1] = nodes[i] -- FFI side is 0 indexed
+  end
+
+  local set = ffi.new("xmlNodeSet")
+  set.nodeNr = #nodes
+  set.nodeMax = #nodes
+  set.nodeTab = nodeTab
+
+  return set
+end
+
+
+
+--- Canonicalise an xmlDocument or set of elements.
+-- @param self xmlDoc from which to canonicalize elements
+-- @param nodes list of nodes to include when canonicalizing, if 'nil' entire doc will be canonicalized
+-- @param mode any of C14N_1_0, C14N_EXCLUSIVE_1_0 (default), C14N_1_1
+-- @param inclusive_ns_prefixes array, or space-separated string, of namespace prefixes to include
+-- @param with_comments if truthy, comments will be included (default: false)
+-- @return string containing canonicalized xml
+function C14n:c14n(nodes, mode, inclusive_ns_prefixes, with_comments)
+  if mode == nil then -- default is exclusive 1.0
+    mode = "C14N_EXCLUSIVE_1_0"
+  end
+  with_comments = with_comments and 1 or 0 -- default = not including comments
+
+  mode = assert(C14N_MODES_LOOKUP[mode], "mode must be a valid C14N mode constant")
+  local prefixes = getNamespacePrefixArray(inclusive_ns_prefixes)
+  local nodeSet = getNodesList(nodes)
+  local buffer = libxml2.xmlBufferCreate()
+  local output_buffer = libxml2.xmlOutputBufferCreate(buffer)
+
+  local success = libxml2.xmlC14NDocSaveTo(self.document, nodeSet, mode,
+                                            prefixes, with_comments, output_buffer)
+
+  if success < 0 then
+    return nil, "failed to generate C14N string"
+  end
+  return libxml2.xmlBufferGetContent(buffer)
+end
+
+
+return C14n

xmlua/document.lua

@@ -7,6 +7,7 @@ local to_string = converter.to_string
 
 local Serializable = require("xmlua.serializable")
 local Searchable = require("xmlua.searchable")
+local C14n = require("xmlua.c14n")
 
 
 local CDATASection
@@ -36,7 +37,8 @@ local metatable = {}
 function metatable.__index(document, key)
   return methods[key] or
     Serializable[key] or
-    Searchable[key]
+    Searchable[key] or
+    C14n[key]
 end
 
 function methods:root()

xmlua/libxml2.lua

@@ -15,6 +15,7 @@ require("xmlua.libxml2.html-tree")
 require("xmlua.libxml2.xmlsave")
 require("xmlua.libxml2.xpath")
 require("xmlua.libxml2.entities")
+require("xmlua.libxml2.c14n")
 
 local ffi = require("ffi")
 local loaded, xml2 = pcall(ffi.load, "xml2")
@@ -68,6 +69,17 @@ else
   libxml2.xmlFree = xml2.xmlFree
 end
 
+local function __xmlC14NDocSaveToIsAvailable()
+  local success, err = pcall(function()
+    local func = xml2.xmlC14NDocSaveTo
+  end)
+  return success, err
+end
+
+if __xmlC14NDocSaveToIsAvailable() then
+  libxml2.xmlC14NDocSaveTo = xml2.xmlC14NDocSaveTo
+end
+
 libxml2.xmlInitParser = xml2.xmlInitParser
 libxml2.xmlCleanupParser = xml2.xmlCleanupParser
 
@@ -581,6 +593,10 @@ function libxml2.xmlBufferCreate()
   return ffi.gc(xml2.xmlBufferCreate(), xml2.xmlBufferFree)
 end
 
+function libxml2.xmlOutputBufferCreate(buffer)
+  return ffi.gc(xml2.xmlOutputBufferCreateBuffer(buffer, nil), xml2.xmlOutputBufferClose)
+end
+
 function libxml2.xmlBufferGetContent(buffer)
   return ffi.string(buffer.content, buffer.use)
 end

xmlua/libxml2/c14n.lua

@@ -0,0 +1,46 @@
+local ffi = require("ffi")
+
+ffi.cdef[[
+  typedef enum {
+    XML_C14N_1_0 = 0, /* Original C14N 1.0 spec */
+    XML_C14N_EXCLUSIVE_1_0 = 1, /* Exclusive C14N 1.0 spec */
+    XML_C14N_1_1 = 2, /*  C14N 1.1 spec */
+  } xmlC14NMode;
+
+  typedef unsigned char xmlChar;
+  typedef struct _xmlBuffer xmlBuffer;
+  typedef xmlBuffer *xmlBufferPtr;
+
+  typedef struct _xmlCharEncodingHandler xmlCharEncodingHandler;
+  typedef xmlCharEncodingHandler *xmlCharEncodingHandlerPtr;
+
+  typedef int (*xmlOutputWriteCallback) (void * context, const char * buffer, int len);
+  typedef int (*xmlOutputCloseCallback) (void * context);
+
+  struct _xmlOutputBuffer {
+     void*                   context;
+     xmlOutputWriteCallback  writecallback;
+     xmlOutputCloseCallback  closecallback;
+
+     xmlCharEncodingHandlerPtr encoder; /* I18N conversions to UTF-8 */
+
+     xmlBufferPtr buffer;    /* Local buffer encoded in UTF-8 or ISOLatin */
+     xmlBufferPtr conv;      /* if encoder != NULL buffer for output */
+     int written;            /* total number of byte written */
+     int error;
+ };
+
+  typedef struct _xmlOutputBuffer xmlOutputBuffer;
+  typedef xmlOutputBuffer *xmlOutputBufferPtr;
+
+  xmlOutputBufferPtr xmlAllocOutputBuffer (xmlCharEncodingHandlerPtr encoder);
+  xmlOutputBufferPtr xmlOutputBufferCreateBuffer	(xmlBufferPtr buffer, xmlCharEncodingHandlerPtr encoder);
+  int xmlOutputBufferClose	(xmlOutputBufferPtr out);
+
+  int xmlC14NDocSaveTo	(xmlDocPtr doc,
+          xmlNodeSetPtr nodes,
+          int mode, /* a xmlC14NMode */
+          xmlChar **inclusive_ns_prefixes,
+          int with_comments,
+          xmlOutputBufferPtr buf);
+]]