Fix escaping of long strings in Turtle/TriG encoder

Author: marcelotto

CHANGELOG.md

@@ -6,6 +6,19 @@ This project adheres to [Semantic Versioning](http://semver.org/) and
 
 
 
+## Unreleased
+
+### Fixed
+
+- The Turtle/TriG encoder didn't escape strings properly when using the long
+  literal form, i.e. when the encoded string contains newlines, which could 
+  result in invalid output in edge-cases.
+
+
+[Compare v2.0.0...HEAD](https://github.com/rdf-elixir/rdf-ex/compare/v2.0.0...HEAD)
+
+
+
 ## 2.0.0 - 2024-08-07
 
 Elixir versions < 1.13 and OTP version < 23 are no longer supported

VERSION

@@ -1 +1 @@
-2.0.0
+2.0.1-pre

lib/rdf/serializations/ntriples/encoder.ex

@@ -125,11 +125,11 @@ defmodule RDF.NTriples.Encoder do
   def escape_string(string) do
     string
     |> String.replace("\\", "\\\\")
-    |> String.replace("\b", "\\b")
-    |> String.replace("\f", "\\f")
     |> String.replace("\t", "\\t")
+    |> String.replace("\b", "\\b")
     |> String.replace("\n", "\\n")
     |> String.replace("\r", "\\r")
+    |> String.replace("\f", "\\f")
     |> String.replace("\"", ~S[\"])
   end
 end

lib/rdf/serializations/turtle_trig/encoder.ex

@@ -4,8 +4,6 @@ defmodule RDF.TurtleTriG.Encoder do
   alias RDF.TurtleTriG.Encoder.{State, Sequencer, CompactStarGraph}
   alias RDF.{BlankNode, Description, Graph, Dataset, IRI, XSD, Literal, LangString, PrefixMap}
 
-  import RDF.NTriples.Encoder, only: [escape_string: 1]
-
   import State,
     only: [
       line_prefixed: 4,
@@ -483,9 +481,54 @@ defmodule RDF.TurtleTriG.Encoder do
 
   defp quoted(string) do
     if String.contains?(string, ["\n", "\r"]) do
-      [~s["""], string, ~s["""]]
+      [~s["""], escape_string(string, :double, true), ~s["""]]
+    else
+      [~s["], escape_string(string, :double), ~s["]]
+    end
+  end
+
+  @doc false
+  def escape_string(string, quotes \\ :double, long \\ false) do
+    string
+    |> String.replace("\\", "\\\\")
+    |> String.replace("\b", "\\b")
+    |> String.replace("\f", "\\f")
+    |> escape_tabs(long)
+    |> escape_newlines(long)
+    |> escape_quotes(quotes, long)
+  end
+
+  defp escape_tabs(string, true), do: string
+  defp escape_tabs(string, false), do: String.replace(string, "\t", "\\t")
+
+  defp escape_newlines(string, true), do: string
+
+  defp escape_newlines(string, false) do
+    string
+    |> String.replace("\n", "\\n")
+    |> String.replace("\r", "\\r")
+  end
+
+  defp escape_quotes(string, :double, false), do: String.replace(string, ~S["], ~S[\"])
+  defp escape_quotes(string, :single, false), do: String.replace(string, ~S['], ~S[\'])
+
+  defp escape_quotes(string, :double, true) do
+    string
+    |> String.replace(~S["""], ~S[\"""])
+    |> escape_ending_quotes(~S["])
+  end
+
+  defp escape_quotes(string, :single, true) do
+    string
+    |> String.replace(~S['''], ~S[\'''])
+    |> escape_ending_quotes(~S['])
+  end
+
+  defp escape_ending_quotes(string, quote) do
+    if String.ends_with?(string, quote) do
+      String.slice(string, 0..-2//1) <> "\\" <> quote
     else
-      [~s["], escape_string(string), ~s["]]
+      string
     end
   end
 end

test/unit/serializations/turtle_encoder_test.exs

@@ -1138,7 +1138,7 @@ defmodule RDF.Turtle.EncoderTest do
       )
     end
 
-    test "backslash-escaping" do
+    test "backslash escaping" do
       EX.S
       |> EX.p("\\")
       |> assert_serialization(matches: [~s["\\\\"]])
@@ -1148,6 +1148,45 @@ defmodule RDF.Turtle.EncoderTest do
       |> assert_serialization(matches: [~s["\\\\\\\\"]])
     end
 
+    test "double quote escaping" do
+      EX.S
+      |> EX.p(~s["foo"bar"])
+      |> assert_serialization(matches: [~s["\\"foo\\"bar\\""]])
+
+      #      EX.S
+      #      |> EX.p(~s[foo\n"])
+      #      |> assert_serialization(matches: [~s["""foo\n\\""""]])
+      EX.S
+      |> EX.p(
+        """
+        "foo"
+        \"""
+        "bar""
+        """
+        |> String.trim_trailing()
+      )
+      |> assert_serialization(
+        matches: [
+          """
+          \""""foo"
+          \\\"""
+          "bar"\\\""""
+          """
+          |> String.trim_trailing()
+        ]
+      )
+    end
+
+    test "tab escaping" do
+      EX.S
+      |> EX.p(~s[\tbar])
+      |> assert_serialization(matches: [~s["\\tbar"]])
+
+      EX.S
+      |> EX.p(~s[foo\n\tbar])
+      |> assert_serialization(matches: [~s["""foo\n\tbar"""]])
+    end
+
     test "language-tagged literals with newlines embedded are encoded with long quotes" do
       Turtle.read_string!(~s[<http://a> <http:/b> """testing string parsing in Turtle.
            """@en .])