XML Validation

lib/elixlsx/compiler.ex

@@ -25,7 +25,7 @@ defmodule Elixlsx.Compiler do
 
   def compinfo_cell_pass_value wci, value do
     cond do
-      is_binary(value) && String.valid?(value)
+      is_binary(value) && XML.valid?(value)
         -> update_in wci.stringdb, &StringDB.register_string(&1, value)
       true -> wci
     end

lib/elixlsx/compiler/string_db.ex

@@ -23,8 +23,14 @@ defmodule Elixlsx.Compiler.StringDB do
   def get_id(stringdb, s) do
     case Map.fetch(stringdb.strings, s) do
       :error ->
-        raise %ArgumentError{
-          message: "Invalid key provided for StringDB.get_id: " <> inspect(s)}
+        if XML.valid?(s) do
+          raise %ArgumentError{
+            message: "Invalid key provided for StringDB.get_id: " <> inspect(s)}
+        else
+          # if the xml is invalid, then we never wanted it in the stringdb to
+          # begin with
+          -1
+        end
       {:ok, id} ->
         id
     end

lib/elixlsx/style/num_fmt.ex

@@ -12,7 +12,7 @@ defmodule Elixlsx.Style.NumFmt do
       props[:yyyymmdd] -> date_yyyy_mm_dd()
       props[:datetime] -> date_datetime()
       props[:num_format] ->
-        if String.valid? props[:num_format] do
+        if XML.valid? props[:num_format] do
           %NumFmt{format: props[:num_format]}
         else
           raise %ArgumentError{

lib/elixlsx/util.ex

@@ -16,7 +16,7 @@ defmodule Elixlsx.Util do
   @spec encode_col(non_neg_integer) :: String.t
   def encode_col(0), do: ""
   def encode_col(num) when num <= 26, do: <<num + 64>>
-  
+
   def encode_col(num, suffix \\ "")
   def encode_col(num, suffix) when num <= 26, do: <<num + 64>> <> suffix
   def encode_col(num, suffix) do
@@ -47,9 +47,9 @@ defmodule Elixlsx.Util do
   def decode_col(""), do: 0
   def decode_col(s) when is_binary(s) do
     case String.match? s, ~r/^[A-Z]*$/ do
-      false -> 
+      false ->
         raise %ArgumentError{message: "Invalid column string: " <> inspect s}
-  
+
       true ->
         # translate list of strings to the base-26 value they represent
         Enum.map(String.to_charlist(s), (fn x -> :string.chr(@col_alphabet, x) end)) |>
@@ -163,7 +163,7 @@ defmodule Elixlsx.Util do
         iso_from_datetime(:calendar.now_to_universal_time({div(input, 1000000), rem(input, 1000000), 0}))
       # TODO this case should parse the string i guess
       # TODO also prominently absent: [char].
-      String.valid? input ->
+      XML.valid? input ->
         input
       true -> raise "Invalid input to iso_timestamp." <> (inspect input)
     end

lib/elixlsx/xml.ex

@@ -0,0 +1,19 @@
+defmodule XML do
+  @xml_chars_block_1 [9, 10, 13]
+  @xml_chars_block_2 32..55_295
+  @xml_chars_block_3 57_344..65_533
+  @xml_chars_block_4 65_536..1_114_111
+
+  # From the xml spec 1.0: https://www.w3.org/TR/REC-xml/#charsets
+  # Character Range
+  #   any Unicode character, excluding the surrogate blocks, FFFE, and FFFF.
+  #   Char ::= #x9 | #xA | #xD | [#x20-#xD7FF] | [#xE000-#xFFFD] | [#x10000-#x10FFFF]
+  def valid?(<<h::utf8, t::binary>>)
+    when h in @xml_chars_block_1 or h in @xml_chars_block_2
+      or h in @xml_chars_block_3 or h in @xml_chars_block_4 do
+    valid?(t)
+  end
+
+  def valid?(<<>>), do: true
+  def valid?(_), do: false
+end

lib/elixlsx/xml_templates.ex

@@ -165,7 +165,13 @@ defmodule Elixlsx.XMLTemplates do
       x when is_number(x) ->
         {"n", to_string(x)}
       x when is_binary(x) ->
-        {"s", to_string(StringDB.get_id wci.stringdb, x)}
+        id = StringDB.get_id(wci.stringdb, x)
+
+        if id == -1 do
+          {:empty, :empty}
+        else
+          {"s", to_string(id)}
+        end
       x when is_boolean(x) ->
         {"b", if x do "1" else "0" end}
       :empty ->
@@ -178,59 +184,76 @@ defmodule Elixlsx.XMLTemplates do
 
   # TODO i know now about string interpolation, i should probably clean this up. ;)
   defp xl_sheet_cols(row, rowidx, wci) do
-    Enum.zip(row, 1 .. length row) |>
-    Enum.map(
-      fn {cell, colidx} ->
+    {updated_row, _id} =
+      row
+      |> List.foldl({"", 1}, fn cell, {acc, colidx} ->
         {content, styleID, cellstyle} = split_into_content_style(cell, wci)
+
         if is_nil(content) do
-          ""
+          {acc, colidx + 1}
         else
-          content = if CellStyle.is_date? cellstyle do
-            U.to_excel_datetime content
-          else
-            content
-          end
+          content =
+            if CellStyle.is_date?(cellstyle) do
+              U.to_excel_datetime(content)
+            else
+              content
+            end
 
           cv = get_content_type_value(content, wci)
-          {content_type, content_value, content_opts} =
-          case cv do
-            {t, v} -> {t, v, []}
-            {t, v, opts} -> {t, v, opts}
-            :error -> raise %ArgumentError{
-                        message: "Invalid column content at " <>
-                                    U.to_excel_coords(rowidx, colidx) <> ": "
-                                    <> (inspect content)
-                                    }
-          end
 
-          case content_type do
-            :formula ->
-              value = if not is_nil(content_opts[:value]), do: "<v>#{content_opts[:value]}</v>", else: ""
+          {content_type, content_value, content_opts} =
+            case cv do
+              {t, v} ->
+                {t, v, []}
+
+              {t, v, opts} ->
+                {t, v, opts}
+
+              :error ->
+                raise %ArgumentError{
+                  message:
+                    "Invalid column content at " <>
+                      U.to_excel_coords(rowidx, colidx) <> ": " <> inspect(content)
+                }
+            end
+
+          cell_xml =
+            case content_type do
+              :formula ->
+                value =
+                  if not is_nil(content_opts[:value]),
+                    do: "<v>#{content_opts[:value]}</v>",
+                    else: ""
+
+                """
+                <c r="#{U.to_excel_coords(rowidx, colidx)}"
+                s="#{styleID}">
+                <f>#{content_value}</f>
+                #{value}
+                </c>
+                """
+
+              :empty ->
+                """
+                <c r="#{U.to_excel_coords(rowidx, colidx)}"
+                s="#{styleID}">
+                </c>
+                """
+
+              type ->
+                """
+                <c r="#{U.to_excel_coords(rowidx, colidx)}"
+                s="#{styleID}" t="#{type}">
+                <v>#{content_value}</v>
+                </c>
+                """
+            end
+
+          {acc <> cell_xml, colidx + 1}
+        end
+      end)
 
-              """
-              <c r="#{U.to_excel_coords(rowidx, colidx)}"
-              s="#{styleID}">
-              <f>#{content_value}</f>
-              #{value}
-              </c>
-              """
-            :empty ->
-              """
-              <c r="#{U.to_excel_coords(rowidx, colidx)}"
-              s="#{styleID}">
-              </c>
-              """
-            type ->
-              """
-              <c r="#{U.to_excel_coords(rowidx, colidx)}"
-              s="#{styleID}" t="#{type}">
-              <v>#{content_value}</v>
-              </c>
-              """
-          end
-          end
-        end) |>
-    List.foldr("", &<>/2)
+    updated_row
   end
 
   defp xl_merge_cells([]) do

test/xml_test.exs

@@ -0,0 +1,11 @@
+defmodule XMLTest do
+  use ExUnit.Case
+
+  test "valid? allows normal string" do
+    assert XML.valid?("Hello World & Goodbye Cruel World")
+  end
+
+  test "valid? rejects invalid xml characters" do
+    refute XML.valid?(<<31>>)
+  end
+end