Merge pull request #73 from esl/cdata

Add CDATA escaping to exml

Author: JanuszJakubiec

c_src/exml.cpp

@@ -68,6 +68,8 @@ namespace {
   ERL_NIF_TERM atom_xmlstreamstart;
   ERL_NIF_TERM atom_xmlstreamend;
   ERL_NIF_TERM atom_pretty;
+  ERL_NIF_TERM atom_escaped;
+  ERL_NIF_TERM atom_cdata;
   ERL_NIF_TERM atom_true;
   constexpr const unsigned char EMPTY[1] = {0};
 
@@ -154,8 +156,9 @@ ERL_NIF_TERM make_attr_tuple(ParseCtx &ctx,
 
 ERL_NIF_TERM get_xmlcdata(ParseCtx &ctx,
                           rapidxml::xml_node<unsigned char> *node) {
-  return enif_make_tuple2(ctx.env, atom_xmlcdata,
-                          to_subbinary(ctx, node->value(), node->value_size()));
+  return enif_make_tuple3(ctx.env, atom_xmlcdata,
+                          to_subbinary(ctx, node->value(), node->value_size()),
+                          atom_escaped);
 }
 
 ERL_NIF_TERM merge_data_nodes(ParseCtx &ctx,
@@ -170,7 +173,7 @@ ERL_NIF_TERM merge_data_nodes(ParseCtx &ctx,
     node = node->next_sibling();
   }
 
-  return enif_make_tuple2(ctx.env, atom_xmlcdata, bin);
+  return enif_make_tuple3(ctx.env, atom_xmlcdata, bin, atom_escaped);
 }
 
 void append_pending_data_nodes(ParseCtx &ctx,
@@ -299,7 +302,15 @@ bool build_cdata(ErlNifEnv *env, xml_document &doc, const ERL_NIF_TERM elem[],
   if (!enif_inspect_iolist_as_binary(env, elem[1], &bin))
     return false;
 
-  auto child = doc.impl.allocate_node(rapidxml::node_data);
+  rapidxml::node_type cdata_type;
+  if (enif_compare(atom_escaped, elem[2]) == 0)
+    cdata_type = rapidxml::node_data;
+  else if (enif_compare(atom_cdata, elem[2]) == 0)
+    cdata_type = rapidxml::node_cdata;
+  else
+    return false;
+
+  auto child = doc.impl.allocate_node(cdata_type);
   child->value(bin.size > 0 ? bin.data : EMPTY, bin.size);
   node.append_node(child);
   return true;
@@ -336,7 +347,7 @@ bool build_attrs(ErlNifEnv *env, xml_document &doc, ERL_NIF_TERM attrs,
 bool build_el(ErlNifEnv *env, xml_document &doc, const ERL_NIF_TERM elem[],
               rapidxml::xml_node<unsigned char> &node) {
   ErlNifBinary name;
-  if (!enif_inspect_iolist_as_binary(env, elem[1], &name))
+  if (!enif_inspect_binary(env, elem[1], &name))
     return false;
 
   auto child = doc.impl.allocate_node(rapidxml::node_element);
@@ -358,7 +369,7 @@ bool build_child(ErlNifEnv *env, xml_document &doc, ERL_NIF_TERM child,
   if (!enif_get_tuple(env, child, &arity, &tuple))
     return false;
 
-  if (arity == 2 && enif_compare(atom_xmlcdata, tuple[0]) == 0) {
+  if (arity == 3 && enif_compare(atom_xmlcdata, tuple[0]) == 0) {
     if (!build_cdata(env, doc, tuple, node))
       return false;
   } else if (arity == 4 && enif_compare(atom_xmlel, tuple[0]) == 0) {
@@ -443,6 +454,8 @@ static int load(ErlNifEnv *env, void **priv_data, ERL_NIF_TERM load_info) {
   atom_xmlstreamstart = enif_make_atom(env, "xmlstreamstart");
   atom_xmlstreamend = enif_make_atom(env, "xmlstreamend");
   atom_pretty = enif_make_atom(env, "pretty");
+  atom_escaped = enif_make_atom(env, "escaped");
+  atom_cdata = enif_make_atom(env, "cdata");
   atom_true = enif_make_atom(env, "true");
 
   get_static_doc().impl.set_allocator(enif_alloc, enif_free);
@@ -608,7 +621,15 @@ static ERL_NIF_TERM escape_cdata(ErlNifEnv *env, int argc,
   if (!enif_inspect_iolist_as_binary(env, argv[0], &bin))
     return enif_make_badarg(env);
 
-  rapidxml::xml_node<unsigned char> node(rapidxml::node_data);
+  rapidxml::node_type cdata_type;
+  if (enif_compare(atom_escaped, argv[1]) == 0)
+    cdata_type = rapidxml::node_data;
+  else if (enif_compare(atom_cdata, argv[1]) == 0)
+    cdata_type = rapidxml::node_cdata;
+  else
+    return enif_make_badarg(env);
+
+  rapidxml::xml_node<unsigned char> node(cdata_type);
   node.value(bin.data, bin.size);
   return node_to_binary(env, node, rapidxml::print_no_indenting);
 }
@@ -647,7 +668,7 @@ static ERL_NIF_TERM reset_parser(ErlNifEnv *env, int argc,
 
 static ErlNifFunc nif_funcs[] = {
     {"create", 2, create},         {"parse", 1, parse},
-    {"parse_next", 2, parse_next}, {"escape_cdata", 1, escape_cdata},
+    {"parse_next", 2, parse_next}, {"escape_cdata", 2, escape_cdata},
     {"to_binary", 2, to_binary},   {"reset_parser", 1, reset_parser}};
 }
 

c_src/rapidxml_print.hpp

@@ -151,7 +151,7 @@ namespace rapidxml
         {
             assert(node->type() == node_data);
             if (!(flags & print_no_indenting))
-                out = fill_chars(out, indent, Ch('\t'));
+                out = fill_chars(out, indent, Ch(' '));
             out = copy_and_expand_chars(node->value(), node->value() + node->value_size(), Ch(0), out);
             return out;
         }
@@ -162,7 +162,7 @@ namespace rapidxml
         {
             assert(node->type() == node_cdata);
             if (!(flags & print_no_indenting))
-                out = fill_chars(out, indent, Ch('\t'));
+                out = fill_chars(out, indent, Ch(' '));
             *out = Ch('<'); ++out;
             *out = Ch('!'); ++out;
             *out = Ch('['); ++out;
@@ -187,7 +187,7 @@ namespace rapidxml
 
             // Print element name and attributes, if any
             if (!(flags & print_no_indenting))
-                out = fill_chars(out, indent, Ch('\t'));
+                out = fill_chars(out, indent, Ch(' '));
             *out = Ch('<'), ++out;
             out = copy_chars(node->name(), node->name() + node->name_size(), out);
             out = print_attributes(out, node, flags);
@@ -221,9 +221,9 @@ namespace rapidxml
                     // Print all children with full indenting
                     if (!(flags & print_no_indenting))
                         *out = Ch('\n'), ++out;
-                    out = print_children(out, node, flags, indent + 1);
+                    out = print_children(out, node, flags, indent + 2);
                     if (!(flags & print_no_indenting))
-                        out = fill_chars(out, indent, Ch('\t'));
+                        out = fill_chars(out, indent, Ch(' '));
                 }
 
                 // Print node end
@@ -241,7 +241,7 @@ namespace rapidxml
         {
             // Print declaration start
             if (!(flags & print_no_indenting))
-                out = fill_chars(out, indent, Ch('\t'));
+                out = fill_chars(out, indent, Ch(' '));
             *out = Ch('<'), ++out;
             *out = Ch('?'), ++out;
             *out = Ch('x'), ++out;
@@ -264,7 +264,7 @@ namespace rapidxml
         {
             assert(node->type() == node_comment);
             if (!(flags & print_no_indenting))
-                out = fill_chars(out, indent, Ch('\t'));
+                out = fill_chars(out, indent, Ch(' '));
             *out = Ch('<'), ++out;
             *out = Ch('!'), ++out;
             *out = Ch('-'), ++out;
@@ -282,7 +282,7 @@ namespace rapidxml
         {
             assert(node->type() == node_doctype);
             if (!(flags & print_no_indenting))
-                out = fill_chars(out, indent, Ch('\t'));
+                out = fill_chars(out, indent, Ch(' '));
             *out = Ch('<'), ++out;
             *out = Ch('!'), ++out;
             *out = Ch('D'), ++out;
@@ -304,7 +304,7 @@ namespace rapidxml
         {
             assert(node->type() == node_pi);
             if (!(flags & print_no_indenting))
-                out = fill_chars(out, indent, Ch('\t'));
+                out = fill_chars(out, indent, Ch(' '));
             *out = Ch('<'), ++out;
             *out = Ch('?'), ++out;
             out = copy_chars(node->name(), node->name() + node->name_size(), out);
@@ -321,7 +321,7 @@ namespace rapidxml
         {
             assert(node->type() == node_literal);
             if (!(flags & print_no_indenting))
-                out = fill_chars(out, indent, Ch('\t'));
+                out = fill_chars(out, indent, Ch(' '));
             out = copy_chars(node->value(), node->value() + node->value_size(), out);
             return out;
         }

include/exml.hrl

@@ -6,7 +6,8 @@
 -ifndef(EXML_HEADER).
 -define(EXML_HEADER, true).
 
--record(xmlcdata, {content = [] :: iodata()}).
+-record(xmlcdata, {content = [] :: iodata(),
+                   style = escaped :: escaped | cdata}).
 
 -record(xmlel, {name :: binary(),
                 attrs = [] :: [exml:attr()],

rebar.config

@@ -55,7 +55,9 @@
     {doc, #{provider => ex_doc}}
 ]}.
 {ex_doc, [
-     {extras, [<<"README.md">>, <<"LICENSE">>]},
+     {source_url, <<"https://github.com/esl/exml">>},
      {main, <<"readme">>},
-     {source_url, <<"https://github.com/esl/exml">>}
+     {extras, [{'README.md', #{title => <<"README">>}},
+               {'LICENSE', #{title => <<"License">>}}
+              ]}
 ]}.

src/exml.erl

@@ -27,18 +27,26 @@
 
 -type attr() :: {binary(), binary()}.
 -type cdata() :: #xmlcdata{}.
+%% CDATA record. Printing escaping rules defaults to escaping character-wise.
+%%
+%% Escaping rules:
+%% <ul>
+%%   <li>`escaped': escapes all characters by regular `&' control escaping.</li>
+%%   <li>`cdata': wraps the entire string into a `<![CDATA[]]>' section.</li>
+%% </ul>
 -type element() :: #xmlel{}.
 -type item() :: element() | attr() | cdata() | exml_stream:start() | exml_stream:stop().
 -type prettify() :: pretty | not_pretty.
+%% Printing indentation rule, see `to_iolist/2'.
 
 %% @doc Calculate the length of the original XML payload
 -spec xml_size(item() | [item()]) -> non_neg_integer().
 xml_size([]) ->
     0;
 xml_size([Elem | Rest]) ->
     xml_size(Elem) + xml_size(Rest);
-xml_size(#xmlcdata{ content = Content }) ->
-    iolist_size(exml_nif:escape_cdata(Content));
+xml_size(#xmlcdata{content = Content, style = Style}) ->
+    iolist_size(exml_nif:escape_cdata(Content, Style));
 xml_size(#xmlel{ name = Name, attrs = Attrs, children = [] }) ->
     3 % Self-closing: </>
     + byte_size(Name) + xml_size(Attrs);
@@ -56,7 +64,7 @@ xml_size({Key, Value}) when is_binary(Key) ->
     + 4 % ="" and whitespace before
     + byte_size(Value).
 
-%% @doc Sort in ascending order a list of xml `t:item()'.
+%% @doc Sort in ascending order a list of xml `t:item/0'.
 %%
 %% Sorting is defined as calling `lists:sort/1' at:
 %% <ul>
@@ -109,7 +117,7 @@ to_iolist(Element) ->
 to_pretty_iolist(Element) ->
     to_iolist(Element, pretty).
 
-%% @doc Parses a binary or a list of binaries into an XML `t:element()'.
+%% @doc Parses a binary or a list of binaries into an XML `t:element/0'.
 -spec parse(binary() | [binary()]) -> {ok, element()} | {error, any()}.
 parse(XML) ->
     exml_nif:parse(XML).
@@ -129,8 +137,8 @@ to_iolist(#xmlstreamstart{name = Name, attrs = Attrs}, _Pretty) ->
     [Front, $>];
 to_iolist(#xmlstreamend{name = Name}, _Pretty) ->
     [<<"</">>, Name, <<">">>];
-to_iolist(#xmlcdata{content = Content}, _Pretty) ->
-    exml_nif:escape_cdata(Content);
+to_iolist(#xmlcdata{content = Content, style = Style}, _Pretty) ->
+    exml_nif:escape_cdata(Content, Style);
 to_iolist([Element], Pretty) ->
     to_iolist(Element, Pretty);
 to_iolist([#xmlstreamstart{name = Name, attrs = Attrs} | Tail] = Elements, Pretty) ->

src/exml_nif.erl

@@ -5,14 +5,12 @@
 
 -module(exml_nif).
 
--nifs([create/2, escape_cdata/1, to_binary/2, parse/1, parse_next/2, reset_parser/1]).
+-nifs([create/2, escape_cdata/2, to_binary/2, parse/1, parse_next/2, reset_parser/1]).
 
 -type parser() :: term().
--type stream_element() :: exml:element() | exml_stream:start() | exml_stream:stop().
 
--export([create/2, parse/1, parse_next/2, escape_cdata/1,
+-export([create/2, parse/1, parse_next/2, escape_cdata/2,
          to_binary/2, reset_parser/1]).
--export_type([parser/0, stream_element/0]).
 
 -on_load(load/0).
 
@@ -40,12 +38,12 @@ load() ->
     erlang:load_nif(filename:join(PrivDir, ?MODULE_STRING), none).
 
 -spec create(MaxChildSize :: non_neg_integer(), InfiniteStream :: boolean()) ->
-                    {ok, parser()} | {error, Reason :: any()}.
+    {ok, parser()} | {error, Reason :: any()}.
 create(_, _) ->
     erlang:nif_error(not_loaded).
 
--spec escape_cdata(Bin :: iodata()) -> binary().
-escape_cdata(_Bin) ->
+-spec escape_cdata(Bin :: iodata(), atom()) -> binary().
+escape_cdata(_Bin, _Style) ->
      erlang:nif_error(not_loaded).
 
 -spec to_binary(Elem :: exml:element(), pretty | not_pretty) -> binary().
@@ -57,8 +55,8 @@ parse(_) ->
     erlang:nif_error(not_loaded).
 
 -spec parse_next(parser(), Data :: binary() | [binary()]) ->
-                        {ok, stream_element() | undefined, non_neg_integer()} |
-                        {error, Reason :: any()}.
+    {ok, exml_stream:element() | undefined, non_neg_integer()} |
+    {error, Reason :: any()}.
 parse_next(_, _) ->
     erlang:nif_error(not_loaded).
 

src/exml_stream.erl

@@ -21,7 +21,7 @@
               parser_opt/0]).
 
 -record(parser, {
-                 event_parser :: exml_nif:parser(),
+                 event_parser :: term(),
                  buffer :: [binary()]
                 }).
 
@@ -30,15 +30,15 @@
 -type stop() :: #xmlstreamend{}.
 %% `#xmlstreamend{}' record.
 -type parser() :: #parser{}.
-%% `#parser{}' record.
--type element() :: exml_nif:stream_element().
-%% One of `t:start()', `t:stop()' or `t:exml:element()'.
+%% `#parser{}' record. Keeps track of unparsed buffers.
+-type element() :: exml:element() | exml_stream:start() | exml_stream:stop().
+%% One of `t:exml:element/0', `t:start/0', or `t:stop/0'.
 
 -type parser_opt() :: {infinite_stream, boolean()} | {max_element_size, non_neg_integer()}.
 %% Parser options
 %%
 %% <ul>
-%%  <li>`infinite_stream': No distinct `t:start()' or `t:stop()', only `#xmlel{}' will be returned.</li>
+%%  <li>`infinite_stream': No distinct `t:start/0' or `t:stop/0', only `#xmlel{}' will be returned.</li>
 %%  <li>`max_element_size': Specifies maximum byte size of any parsed XML element.
 %%      The only exception is the "stream start" element,
 %%      for which only the size of the opening tag is limited.</li>
@@ -53,7 +53,7 @@
 new_parser() ->
     new_parser([]).
 
-%% @doc Creates a new parser
+%% @doc Creates a new parser. See `t:parser_opt/0' for configuration.
 -spec new_parser([parser_opt()]) -> {ok, parser()} | {error, any()}.
 new_parser(Opts)->
     MaxElementSize = proplists:get_value(max_element_size, Opts, 0),
@@ -65,7 +65,9 @@ new_parser(Opts)->
             Error
     end.
 
-%% @doc Makes a parser parse input
+%% @doc Makes a parser parse input.
+%%
+%% If successful, returns parsed elements and a new parser with updated buffers.
 -spec parse(parser(), binary()) ->
     {ok, parser(), [exml_stream:element()]} | {error, Reason :: any()}.
 parse(Parser, Input) when is_binary(Input) ->

test/exml_tests.erl

@@ -27,6 +27,12 @@ size_of_escaped_characters_test() ->
     Raw = <<"<a>&amp;</a>">>,
     ?assertEqual(iolist_size(Raw), exml:xml_size(parse(Raw))).
 
+cdata_size_of_escaped_characters_test() ->
+    Raw = <<"<a><![CDATA[some stuff]]></a>">>,
+    CData = #xmlcdata{content = <<"some stuff">>, style = cdata},
+    Final = #xmlel{name = <<"a">>, children = [CData]},
+    ?assertEqual(iolist_size(Raw), exml:xml_size(Final)).
+
 size_of_exml_with_cdata_test() ->
     Raw = <<"<a><![CDATA[ Within this Character Data block I can
             use double dashes as much as I want (along with <, &, ', and \")]]></a>">>,