[gbc] Handle forward- and backslashes in paths

Since .bond files can be authored on platforms with different directory
separator characters, the paths in build scripts and Bond `import`
statements may use a forwardslash when a backslash is expected or vice
versa.

In order to handle this, gbc now normalizes anything that is a
path (file paths, response file items, import directories, import
statement paths, output directories, &c.) to use the platform's
preferred directory separator character.

Tests for C++, C#, and Java have been updated to use some imports with
mixed slashes.

gbc doesn't use a dedicate type for file paths. `FilePath` is just a
type synonym for `String`, so some places may have been inadvertently
missed.

This does now mean that gbc cannot process files, on, say, Linux, with
backslashes in their names. This is expected to be rare, and is deemed
an acceptable compromise for a cross-platform tool.

Fixes #869

CHANGELOG.md

@@ -42,7 +42,10 @@ different versioning scheme, following the Haskell community's
 * C++ codegen now applies the `--export-attribute` to the `ToString`,
   `FromString`, `ToEnum` and `FromEnum` functions.
 * `import` statements can now end with an optional semicolon.
-
+* File and directory paths on the command line, in response files, or in
+  `import` statements can now use a mix of forward and backslashes. [Issue
+  #869](https://github.com/Microsoft/bond/issues/869)
+
 ### C++ ###
 
 * **Breaking change** The deprecated Bond Comm functionality has been

compiler/IO.hs

@@ -7,6 +7,7 @@ module IO
     , parseASTFile
     , parseNamespaceMappings
     , parseAliasMappings
+    , slashNormalize
     )
     where
 
@@ -54,9 +55,10 @@ parseBondFile importDirs file = do
             Just path' -> do
                 content <- readFileUtf8 path'
                 return (path', content)
-            Nothing -> fail $ "Can't find import file " ++ importFile
+            Nothing -> fail $ "Can't find import file " ++ importFile'
       where
-        findFilePath dirs = fmap (</> importFile) <$> firstM (doesFileExist . (</> importFile)) dirs
+        importFile' = slashNormalize importFile
+        findFilePath dirs = fmap (</> importFile') <$> firstM (doesFileExist . (</> importFile')) dirs
 
     readFileUtf8 name = do
         h <- openFile name ReadMode
@@ -102,3 +104,15 @@ combinedMessage err = id $ T.unpack $ T.intercalate (T.pack ", ") messages
         -- parseErrorPretty returns a multi-line String.
         -- We need to break it up to make a useful one-line message.
         messages = T.splitOn (T.pack "\n") $ T.strip $ T.pack $ parseErrorTextPretty err
+
+-- | Normalizes a file path to only use the current platform's preferred
+-- directory separator.
+--
+-- Bond doesn't support files or directories with backslashes in their
+-- names, so backslashes are always converted to the platform's preferred
+-- separator.
+slashNormalize :: FilePath -> FilePath
+slashNormalize path = map replaceSlashes path
+  where replaceSlashes '/'  = pathSeparator
+        replaceSlashes '\\' = pathSeparator
+        replaceSlashes c    = c

compiler/Options.hs

@@ -1,7 +1,7 @@
 -- Copyright (c) Microsoft. All rights reserved.
 -- Licensed under the MIT license. See LICENSE file in the project root for full license information.
 
-{-# LANGUAGE DeriveDataTypeable #-}
+{-# LANGUAGE DeriveDataTypeable, RecordWildCards #-}
 {-# OPTIONS_GHC -fno-warn-missing-fields #-}
 {-# OPTIONS_GHC -fno-cse #-}
 
@@ -16,6 +16,7 @@ import Paths_bond (version)
 import Data.Version (showVersion)
 import System.Console.CmdArgs
 import System.Console.CmdArgs.Explicit (processValue)
+import IO (slashNormalize)
 
 data ApplyOptions =
     Compact |
@@ -128,14 +129,30 @@ schema = Schema
     name "schema" &=
     help "Output the JSON representation of the schema"
 
+slashNormalizeOption :: Options -> Options
+slashNormalizeOption Options = Options
+slashNormalizeOption o@Cpp{..}    = o { files = map slashNormalize files,
+                                        import_dir = map slashNormalize import_dir,
+                                        output_dir = slashNormalize output_dir }
+slashNormalizeOption o@Cs{..}     = o { files = map slashNormalize files,
+                                        import_dir = map slashNormalize import_dir,
+                                        output_dir = slashNormalize output_dir }
+slashNormalizeOption o@Java{..}   = o { files = map slashNormalize files,
+                                        import_dir = map slashNormalize import_dir,
+                                        output_dir = slashNormalize output_dir }
+slashNormalizeOption o@Schema{..} = o { files = map slashNormalize files,
+                                        import_dir = map slashNormalize import_dir,
+                                        output_dir = slashNormalize output_dir }
+
+
 mode :: Mode (CmdArgs Options)
 mode = cmdArgsMode $ modes [cpp, cs, java, schema] &=
     program "gbc" &=
     help "Compile Bond schema file(s) and generate specified output. The schema file(s) can be in one of two formats: Bond IDL or JSON representation of the schema abstract syntax tree as produced by `gbc schema`. Multiple schema files can be specified either directly on the command line or by listing them in a text file passed to gbc via @listfile syntax." &=
     summary ("Bond Compiler " ++ showVersion version ++ ", (C) Microsoft")
 
 getOptions :: IO Options
-getOptions = cmdArgsRun mode
+getOptions = slashNormalizeOption <$> cmdArgsRun mode
 
 processOptions :: [String] -> Options
 processOptions = cmdArgsValue . processValue mode

cpp/test/core/CMakeLists.txt

@@ -36,7 +36,7 @@ add_library (core_test_common
     "${CMAKE_CURRENT_BINARY_DIR}/${CMAKE_CFG_INTDIR}/allocator_test_types.cpp"
     "${CMAKE_CURRENT_BINARY_DIR}/${CMAKE_CFG_INTDIR}/import_test1_types.cpp"
     "${CMAKE_CURRENT_BINARY_DIR}/${CMAKE_CFG_INTDIR}/import_test1_apply.cpp"
-    "${CMAKE_CURRENT_BINARY_DIR}/${CMAKE_CFG_INTDIR}/import_test2_types.cpp"
+    "${CMAKE_CURRENT_BINARY_DIR}/${CMAKE_CFG_INTDIR}/dir1/dir2/import_test2_types.cpp"
     "${CMAKE_CURRENT_BINARY_DIR}/${CMAKE_CFG_INTDIR}/apply_test_types.cpp"
     "${CMAKE_CURRENT_BINARY_DIR}/${CMAKE_CFG_INTDIR}/apply_test_apply.cpp"
     "${CMAKE_CURRENT_BINARY_DIR}/${CMAKE_CFG_INTDIR}/validation_types.cpp"
@@ -50,7 +50,8 @@ add_dependencies(core_test_common
     unit_test_codegen1
     unit_test_codegen2
     unit_test_codegen3
-    unit_test_codegen4)
+    unit_test_codegen4
+    unit_test_codegen_import2)
 target_include_directories (core_test_common PUBLIC
     ${CMAKE_CURRENT_SOURCE_DIR}
     ${CMAKE_CURRENT_BINARY_DIR}/${CMAKE_CFG_INTDIR})
@@ -64,10 +65,10 @@ target_link_libraries (core_test_common PUBLIC
     bond
     ${Boost_UNIT_TEST_FRAMEWORK_LIBRARY})
 
-
 add_bond_codegen (TARGET unit_test_codegen1
     unit_test.bond
     OPTIONS
+        --import-dir=imports
         --using=\"static_string=std::array<char, {0}>\"
         --using=\"static_wstring=std::array<wchar_t, {0}>\"
         --using=\"simple_list=SimpleList<{0}>\"
@@ -78,13 +79,13 @@ add_bond_codegen (TARGET unit_test_codegen2
     unit_test_core.bond
     apply_test.bond
     import_test1.bond
-    import_test2.bond
     scope_test1.bond
     scope_test2.bond
     validation.bond
     cmdargs.bond
     OPTIONS
-        --header=\\\"custom_protocols.h\\\")
+      --import-dir=imports
+      --header=\\\"custom_protocols.h\\\")
 
 add_bond_codegen (TARGET unit_test_codegen3
     allocator_test.bond
@@ -103,6 +104,13 @@ add_bond_codegen (TARGET unit_test_codegen4
         --allocator=\"bond::capped_allocator<>\"
         --namespace=\"allocator_test=capped_allocator_tests\")
 
+add_bond_codegen (TARGET unit_test_codegen_import2
+    imports/dir1/dir2/import_test2.bond
+    # Need a custom output path so the generated #include paths line up
+    OUTPUT_DIR ${CMAKE_CURRENT_BINARY_DIR}/${CMAKE_CFG_INTDIR}/dir1/dir2
+    OPTIONS
+        --header=\\\"custom_protocols.h\\\")
+
 # Full unit test is too big, .pdb files exceed compiler limit;
 # Since debugging such large executable is not practical anyways,
 # we disable .pdb generation.

cpp/test/core/apply_test.bond

@@ -4,7 +4,7 @@
 ///
 
 import "import_test1.bond"
-import "import_test2.bond"
+import "dir1/dir2/import_test2.bond"
 
 namespace unittest.apply;
 

cpp/test/core/import_test1.bond

@@ -1,4 +1,5 @@
-import "import_test2.bond"
+// Uses mixed slashes to test gbc can deal with that
+import "dir1/dir2\import_test2.bond"
 
 namespace import1
 namespace csharp Microsoft.Import1
@@ -8,4 +9,3 @@ struct ImportedStruct
     10: required_optional int32 x;
     20: required_optional import2.ImportedEnum e = ImportedEnum2;
 };
-

cpp/test/core/unit_test.bond

@@ -4,7 +4,7 @@
 ///
 
 import "import_test1.bond"
-import "import_test2.bond"
+import "dir1/dir2/import_test2.bond"
 import "unit_test_core.bond"
 
 namespace unittest;

cs/dnc/test/core.tests/core.tests.codegen.proj

@@ -24,7 +24,7 @@
     <BondCodegen Include="..\..\..\test\core\UnitTest.bond">
       <Options>$(BondOptions) --using="DateTime=System.DateTime"</Options>
     </BondCodegen>
-    <BondCodegen Include="..\..\..\test\core\import dir with spaces\Bond File With Spaces.bond" />
+    <BondCodegen Include="..\..\..\test\core\import dir with spaces\dir1\dir2\Bond File With Spaces.bond" />
   </ItemGroup>
   <Import Project="$(MSBuildThisFileDirectory)\..\..\..\build\nuget\Common.targets" />
 </Project>

cs/test/core/Core.csproj

@@ -1,4 +1,4 @@
-<?xml version="1.0" encoding="utf-8"?>
+<?xml version="1.0" encoding="utf-8"?>
 <Project ToolsVersion="12.0" DefaultTargets="Build" xmlns="http://schemas.microsoft.com/developer/msbuild/2003">
   <Import Project="$(MSBuildExtensionsPath32)\$(MSBuildToolsVersion)\Microsoft.Common.props" Condition="Exists('$(MSBuildExtensionsPath32)\$(MSBuildToolsVersion)\Microsoft.Common.props')" />
   <PropertyGroup Condition="'$(BuildNonportable)' == 'true'">
@@ -75,7 +75,8 @@
     <BondCodegen Include="UnitTest.bond">
       <Options>$(BondOptions) --using="DateTime=System.DateTime"</Options>
     </BondCodegen>
-    <BondCodegen Include="import dir with spaces\Bond File With Spaces.bond" />
+    <!-- uses mixed slashes to test the targets can deal with that -->
+    <BondCodegen Include="import dir with spaces\dir1/dir2/Bond File With Spaces.bond" />
     <!-- Resharper Workaround -->
     <Compile Include="$(IntermediateOutputPath)\Aliases_types.cs" Condition="False" />
     <Compile Include="$(IntermediateOutputPath)\Bond File With Spaces_types.cs" Condition="False" />

cs/test/core/UnitTest.bond

@@ -1,6 +1,7 @@
 import "bond/core/bond.bond"
 import "Aliases.bond"
-import "Bond File With Spaces.bond"
+// Uses mixed slashes to test gbc can deal with that
+import "dir1\dir2/Bond File With Spaces.bond"
 
 namespace UnitTest
 

java/core/build.gradle

@@ -63,3 +63,7 @@ compileBond {
               '../../idl/bond/core/bond_const.bond'
     options '-n', 'bond=org.bondlib'
 }
+
+compileTestBond {
+    options '--import-dir', 'src/test/bond/imports/'
+}

java/core/src/test/bond/bonded.bond

@@ -1,4 +1,5 @@
-import "common.bond"
+// Uses mixed slashes to test gbc can deal with that
+import "dir1\dir2/common.bond"
 
 namespace org.bondlib.test
 

java/core/src/test/bond/equality.bond

@@ -1,4 +1,4 @@
-import "common.bond"
+import "dir1/dir2/common.bond"
 
 namespace org.bondlib.test
 

java/core/src/test/bond/generic.bond

@@ -1,4 +1,4 @@
-import "common.bond"
+import "dir1/dir2/common.bond"
 
 namespace org.bondlib.test