Merge pull request #173949 from jacoblambda/fix-toInt-zero-padding

lib: add strings.toIntBase10 to parse zero-padded strings

lib/default.nix

@@ -103,7 +103,7 @@ let
       getName getVersion
       nameFromURL enableFeature enableFeatureAs withFeature
       withFeatureAs fixedWidthString fixedWidthNumber isStorePath
-      toInt readPathsFromFile fileContents;
+      toInt toIntBase10 readPathsFromFile fileContents;
     inherit (self.stringsWithDeps) textClosureList textClosureMap
       noDepEntry fullDepEntry packEntry stringAfter;
     inherit (self.customisation) overrideDerivation makeOverridable

lib/strings.nix

@@ -783,24 +783,105 @@ rec {
     else
       false;
 
-  /* Parse a string as an int.
+  /* Parse a string as an int. Does not support parsing of integers with preceding zero due to
+  ambiguity between zero-padded and octal numbers. See toIntBase10.
 
      Type: string -> int
 
      Example:
+
        toInt "1337"
        => 1337
+
        toInt "-4"
        => -4
+
+       toInt " 123 "
+       => 123
+
+       toInt "00024"
+       => error: Ambiguity in interpretation of 00024 between octal and zero padded integer.
+
        toInt "3.14"
        => error: floating point JSON numbers are not supported
   */
-  # Obviously, it is a bit hacky to use fromJSON this way.
   toInt = str:
-    let may_be_int = fromJSON str; in
-    if isInt may_be_int
-    then may_be_int
-    else throw "Could not convert ${str} to int.";
+    let
+      # RegEx: Match any leading whitespace, then any digits, and finally match any trailing
+      # whitespace.
+      strippedInput = match "[[:space:]]*([[:digit:]]+)[[:space:]]*" str;
+
+      # RegEx: Match a leading '0' then one or more digits.
+      isLeadingZero = match "0[[:digit:]]+" (head strippedInput) == [];
+
+      # Attempt to parse input
+      parsedInput = fromJSON (head strippedInput);
+
+      generalError = "toInt: Could not convert ${escapeNixString str} to int.";
+
+      octalAmbigError = "toInt: Ambiguity in interpretation of ${escapeNixString str}"
+      + " between octal and zero padded integer.";
+
+    in
+      # Error on presence of non digit characters.
+      if strippedInput == null
+      then throw generalError
+      # Error on presence of leading zero/octal ambiguity.
+      else if isLeadingZero
+      then throw octalAmbigError
+      # Error if parse function fails.
+      else if !isInt parsedInput
+      then throw generalError
+      # Return result.
+      else parsedInput;
+
+
+  /* Parse a string as a base 10 int. This supports parsing of zero-padded integers.
+
+     Type: string -> int
+
+     Example:
+       toIntBase10 "1337"
+       => 1337
+
+       toIntBase10 "-4"
+       => -4
+
+       toIntBase10 " 123 "
+       => 123
+
+       toIntBase10 "00024"
+       => 24
+
+       toIntBase10 "3.14"
+       => error: floating point JSON numbers are not supported
+  */
+  toIntBase10 = str:
+    let
+      # RegEx: Match any leading whitespace, then match any zero padding, capture any remaining
+      # digits after that, and finally match any trailing whitespace.
+      strippedInput = match "[[:space:]]*0*([[:digit:]]+)[[:space:]]*" str;
+
+      # RegEx: Match at least one '0'.
+      isZero = match "0+" (head strippedInput) == [];
+
+      # Attempt to parse input
+      parsedInput = fromJSON (head strippedInput);
+
+      generalError = "toIntBase10: Could not convert ${escapeNixString str} to int.";
+
+    in
+      # Error on presence of non digit characters.
+      if strippedInput == null
+      then throw generalError
+      # In the special case zero-padded zero (00000), return early.
+      else if isZero
+      then 0
+      # Error if parse function fails.
+      else if !isInt parsedInput
+      then throw generalError
+      # Return result.
+      else parsedInput;
 
   /* Read a list of paths from `file`, relative to the `rootPath`.
      Lines beginning with `#` are treated as comments and ignored.

lib/tests/misc.nix

@@ -327,6 +327,77 @@ runTests {
     expected = "Hello\\x20World";
   };
 
+  testToInt = testAllTrue [
+    # Naive
+    (123 == toInt "123")
+    (0 == toInt "0")
+    # Whitespace Padding
+    (123 == toInt " 123")
+    (123 == toInt "123 ")
+    (123 == toInt " 123 ")
+    (123 == toInt "   123   ")
+    (0 == toInt " 0")
+    (0 == toInt "0 ")
+    (0 == toInt " 0 ")
+  ];
+
+  testToIntFails = testAllTrue [
+    ( builtins.tryEval (toInt "") == { success = false; value = false; } )
+    ( builtins.tryEval (toInt "123 123") == { success = false; value = false; } )
+    ( builtins.tryEval (toInt "0 123") == { success = false; value = false; } )
+    ( builtins.tryEval (toInt " 0d ") == { success = false; value = false; } )
+    ( builtins.tryEval (toInt " 1d ") == { success = false; value = false; } )
+    ( builtins.tryEval (toInt " d0 ") == { success = false; value = false; } )
+    ( builtins.tryEval (toInt "00") == { success = false; value = false; } )
+    ( builtins.tryEval (toInt "01") == { success = false; value = false; } )
+    ( builtins.tryEval (toInt "002") == { success = false; value = false; } )
+    ( builtins.tryEval (toInt " 002 ") == { success = false; value = false; } )
+    ( builtins.tryEval (toInt " foo ") == { success = false; value = false; } )
+    ( builtins.tryEval (toInt " foo 123 ") == { success = false; value = false; } )
+    ( builtins.tryEval (toInt " foo123 ") == { success = false; value = false; } )
+  ];
+
+  testToIntBase10 = testAllTrue [
+    # Naive
+    (123 == toIntBase10 "123")
+    (0 == toIntBase10 "0")
+    # Whitespace Padding
+    (123 == toIntBase10 " 123")
+    (123 == toIntBase10 "123 ")
+    (123 == toIntBase10 " 123 ")
+    (123 == toIntBase10 "   123   ")
+    (0 == toIntBase10 " 0")
+    (0 == toIntBase10 "0 ")
+    (0 == toIntBase10 " 0 ")
+    # Zero Padding
+    (123 == toIntBase10 "0123")
+    (123 == toIntBase10 "0000123")
+    (0 == toIntBase10 "000000")
+    # Whitespace and Zero Padding
+    (123 == toIntBase10 " 0123")
+    (123 == toIntBase10 "0123 ")
+    (123 == toIntBase10 " 0123 ")
+    (123 == toIntBase10 " 0000123")
+    (123 == toIntBase10 "0000123 ")
+    (123 == toIntBase10 " 0000123 ")
+    (0 == toIntBase10 " 000000")
+    (0 == toIntBase10 "000000 ")
+    (0 == toIntBase10 " 000000 ")
+  ];
+
+  testToIntBase10Fails = testAllTrue [
+    ( builtins.tryEval (toIntBase10 "") == { success = false; value = false; } )
+    ( builtins.tryEval (toIntBase10 "123 123") == { success = false; value = false; } )
+    ( builtins.tryEval (toIntBase10 "0 123") == { success = false; value = false; } )
+    ( builtins.tryEval (toIntBase10 " 0d ") == { success = false; value = false; } )
+    ( builtins.tryEval (toIntBase10 " 1d ") == { success = false; value = false; } )
+    ( builtins.tryEval (toIntBase10 " d0 ") == { success = false; value = false; } )
+    ( builtins.tryEval (toIntBase10 " foo ") == { success = false; value = false; } )
+    ( builtins.tryEval (toIntBase10 " foo 123 ") == { success = false; value = false; } )
+    ( builtins.tryEval (toIntBase10 " foo 00123 ") == { success = false; value = false; } )
+    ( builtins.tryEval (toIntBase10 " foo00123 ") == { success = false; value = false; } )
+  ];
+
 # LISTS
 
   testFilter = {

lib/tests/modules.sh

@@ -162,7 +162,7 @@ checkConfigError 'A definition for option .* is not.*string or signed integer co
 # Check coerced value with unsound coercion
 checkConfigOutput '^12$' config.value ./declare-coerced-value-unsound.nix
 checkConfigError 'A definition for option .* is not of type .*. Definition values:\n\s*- In .*: "1000"' config.value ./declare-coerced-value-unsound.nix ./define-value-string-bigint.nix
-checkConfigError 'json.exception.parse_error' config.value ./declare-coerced-value-unsound.nix ./define-value-string-arbitrary.nix
+checkConfigError 'toInt: Could not convert .* to int' config.value ./declare-coerced-value-unsound.nix ./define-value-string-arbitrary.nix
 
 # Check mkAliasOptionModule.
 checkConfigOutput '^true$' config.enable ./alias-with-priority.nix