fix(sexp): Accept another digit after an \x.. or \... escape sequence in strings

shym · rgrinberg · commit a237caacb63c · 2022-12-15T16:43:11.000-06:00
The two cases that this modifies are meant to catch the issue of
"unterminated ... escape sequences" but, due to the use of '*', they
also caught the cases where the escape sequence was followed by another
digit

Signed-off-by: Samuel Hym &lt;samuel.hym@rustyne.lautre.net&gt;
diff --git a/CHANGES.md b/CHANGES.md
@@ -1,6 +1,9 @@
 Unreleased
 ----------
 
+- Fix the parsing of decimal and hexadecimal escape literals in `dune`,
+  `dune-package`, and other dune s-expression based files (#6710, @shym)
+
 - Report an error if `dune init ...` would create a "dune" file in a location
   which already contains a "dune" directory (#6705, @gridbugs)
 
diff --git a/src/dune_sexp/lexer.mll b/src/dune_sexp/lexer.mll
@@ -320,15 +320,15 @@ and escape_sequence = parse
   | digit digit digit
     { error lexbuf "escape sequence in quoted string out of range" ~delta:(-1);
     }
-  | digit*
+  | digit digit?
     { error lexbuf "unterminated decimal escape sequence" ~delta:(-1);
     }
   | 'x' (hexdigit as c1) (hexdigit as c2)
     { let v = eval_hex_escape c1 c2 in
       Template.Buffer.add_text_c (Char.chr v);
       Other
     }
-  | 'x' hexdigit*
+  | 'x' hexdigit?
     { error lexbuf "unterminated hexadecimal escape sequence" ~delta:(-1);
     }
   | _
diff --git a/test/expect-tests/dune_lang/sexp_tests.ml b/test/expect-tests/dune_lang/sexp_tests.ml
@@ -195,13 +195,13 @@ Ok [ "bar%foo" ]
 let%expect_test _ =
   parse {|"\0000"|};
   [%expect {|
-Error "unterminated decimal escape sequence"
+Ok [ "\0000" ]
 |}]
 
 let%expect_test _ =
   parse {|"\x000"|};
   [%expect {|
-Error "unterminated hexadecimal escape sequence"
+Ok [ "\0000" ]
 |}]
 
 (* Printing tests *)

Original file line number	Diff line number	Diff line change
`@@ -320,15 +320,15 @@ and escape_sequence = parse`
`320`	`320`	`\| digit digit digit`
`321`	`321`	`{ error lexbuf "escape sequence in quoted string out of range" ~delta:(-1);`
`322`	`322`	`}`
`323`		`- \| digit*`
	`323`	`+ \| digit digit?`
`324`	`324`	`{ error lexbuf "unterminated decimal escape sequence" ~delta:(-1);`
`325`	`325`	`}`
`326`	`326`	`\| 'x' (hexdigit as c1) (hexdigit as c2)`
`327`	`327`	`{ let v = eval_hex_escape c1 c2 in`
`328`	`328`	`Template.Buffer.add_text_c (Char.chr v);`
`329`	`329`	`Other`
`330`	`330`	`}`
`331`		`- \| 'x' hexdigit*`
	`331`	`+ \| 'x' hexdigit?`
`332`	`332`	`{ error lexbuf "unterminated hexadecimal escape sequence" ~delta:(-1);`
`333`	`333`	`}`
`334`	`334`	`\| _`