From 97dee892069e2f019d9f1320545563f7b4e23444 Mon Sep 17 00:00:00 2001
From: Joe Eli McIlvain <joe.eli.mac@gmail.com>
Date: Thu, 8 Sep 2022 11:31:52 -0700
Subject: [PATCH] Add a compilation error for invalid escapes in string
 literals.

This helps to prevent bugs wherein the user believes they have
correctly written an escape code.

Prior to this commit, invalid escape code were treated as if
they were literal characters, including the backslash character.

See discussion in #348
---
 spec/parser_spec.cr              | 19 +++++++++++++++++++
 src/savi/parser/builder/state.cr | 30 +++++++++++++++++++-----------
 2 files changed, 38 insertions(+), 11 deletions(-)

diff --git a/spec/parser_spec.cr b/spec/parser_spec.cr
index ea2a7a73..ce2a5504 100644
--- a/spec/parser_spec.cr
+++ b/spec/parser_spec.cr
@@ -185,6 +185,25 @@ describe Savi::Parser do
     end
   end
 
+  it "complains when a string literal has an unknown escape character" do
+    source = Savi::Source.new_example <<-SOURCE
+    :actor Main
+      :new
+        greeting = "Hello, World\\?"
+    SOURCE
+
+    expected = <<-MSG
+    This is an invalid escape character:
+    from (example):3:
+        greeting = "Hello, World\\?"
+                                 ^
+    MSG
+
+    expect_raises Savi::Error, expected do
+      Savi::Parser.parse(source)
+    end
+  end
+
   it "handles nifty heredoc string literals" do
     source = Savi::Source.new_example <<-SOURCE
     :actor Main
diff --git a/src/savi/parser/builder/state.cr b/src/savi/parser/builder/state.cr
index c2c5a5e2..c6920b60 100644
--- a/src/savi/parser/builder/state.cr
+++ b/src/savi/parser/builder/state.cr
@@ -48,6 +48,18 @@ module Savi::Parser::Builder
       )
     end
 
+    def pos_single_with_offset(
+      token : Pegmatite::Token,
+      offset : Int32,
+    ) : Source::Pos
+      kind, start, finish = token
+
+      start = start + offset
+      finish = start + 1
+
+      pos({kind, start, finish})
+    end
+
     def slice(token : Pegmatite::Token)
       kind, start, finish = token
       slice(start...finish)
@@ -58,11 +70,7 @@ module Savi::Parser::Builder
     end
 
     def slice_with_escapes(token : Pegmatite::Token)
-      kind, start, finish = token
-      slice_with_escapes(start...finish)
-    end
-
-    def slice_with_escapes(range : Range)
+      range = token[1]...token[2]
       string = slice(range)
       reader = Char::Reader.new(string)
 
@@ -91,7 +99,8 @@ module Savi::Parser::Builder
                   elsif 'A' <= hex_char <= 'F'
                     10 + (hex_char - 'A')
                   else
-                    raise "invalid escape hex character: #{hex_char}"
+                    Error.at pos_single_with_offset(token, reader.pos),
+                      "This is an invalid escape hex character"
                   end
                 byte_value = 16 * byte_value + hex_value
               end
@@ -108,7 +117,8 @@ module Savi::Parser::Builder
                   elsif 'A' <= hex_char <= 'F'
                     10 + (hex_char - 'A')
                   else
-                    raise "invalid unicode escape hex character: #{hex_char}"
+                    Error.at pos_single_with_offset(token, reader.pos),
+                      "This is an invalid unicode escape hex character"
                   end
                 codepoint = 16 * codepoint + hex_value
               end
@@ -126,10 +136,8 @@ module Savi::Parser::Builder
                 reader.next_char
               end
             else
-              # Not a valid escape character - pass it on as a literal slash
-              # followed by that literal character, as if not an escape.
-              result << '\\'
-              result << reader.current_char
+              Error.at pos_single_with_offset(token, reader.pos),
+                "This is an invalid escape character"
             end
           else
             result << reader.current_char