Allow backslashes in quoted strings (Windows paths); v0.3.6

CMakeLists.txt

@@ -1,6 +1,6 @@
 cmake_minimum_required(VERSION 3.20)
 # Keep this version in sync with [project] version in pyproject.toml.
-project(neml2-hit VERSION 0.3.5 LANGUAGES CXX)
+project(neml2-hit VERSION 0.3.6 LANGUAGES CXX)
 
 set(CMAKE_CXX_STANDARD 17)
 set(CMAKE_CXX_STANDARD_REQUIRED ON)

README.md

@@ -214,7 +214,9 @@ Only `param_str()` (and its `param_optional_str` variant) is allowed.
 
 A whitespace-delimited sequence of elements enclosed in single quotes **or double quotes** —
 both delimiters are completely equivalent.  Elements may be integers, floating-point numbers, or unquoted tokens
-(none of which may contain `;`, `#`, `$`, `'`, `"`, or `\`).
+(none of which may contain `;`, `#`, `$`, `'`, or `"`).  Backslashes **are** permitted inside
+quoted values, so a single-element quoted string can hold a Windows path, e.g.
+`path = 'C:\Users\me\model'`.  (Unquoted values still exclude `\` — quote such paths.)
 
 ```
 vals   = '1 2 3'
@@ -613,6 +615,9 @@ Pre-generated parser/lexer sources are committed to `generated/` and used
 automatically for non-Debug build types, so end users and CI release builds do
 not need flex or bison installed.
 
+nmhit builds and is tested on **Linux, macOS, and Windows**, with GCC, Clang,
+or MSVC. Windows uses the Visual Studio generator (no developer shell needed).
+
 ### Configure and build
 
 **Release build** (no flex or bison required — uses committed generated sources):

pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "scikit_build_core.build"
 
 [project]
 name = "nmhit"
-version = "0.3.5"  # Keep in sync with VERSION in CMakeLists.txt.
+version = "0.3.6"  # Keep in sync with VERSION in CMakeLists.txt.
 description = "Python bindings for the nmhit NEML2-flavored HIT parser"
 readme = "README.md"
 requires-python = ">=3.9"

python/tests/test_nmhit.py

@@ -92,6 +92,17 @@ def test_parse_file_pre_kwarg(tmp_path):
     assert root.param_int("a") == 1
 
 
+def test_backslash_path_in_quoted_string():
+    """Backslashes survive inside quoted strings, so Windows paths parse.
+
+    A single backslash in the HIT text is written here with a raw string. The
+    value has no spaces because both quote styles tokenise as arrays and
+    whitespace still separates elements.
+    """
+    root = nmhit.parse_text(r"path = 'C:\Users\me\model'")
+    assert root.param_str("path") == r"C:\Users\me\model"
+
+
 # ── find / children ───────────────────────────────────────────────────────────
 
 

src/Lexer.l

@@ -103,7 +103,7 @@ COMMENT    #[^\n]*
 <IN_ARRAY>"${"                             { DRIVER->begin_brace(); DRIVER->set_yymore(); yymore(); BEGIN(IN_ARRAY_BRACE); }
 <IN_ARRAY>{FLOAT}                           { TOK_S(TOK_FLOAT); }
 <IN_ARRAY>{INTEGER}                        { TOK_S(TOK_INTEGER); }
-<IN_ARRAY>[^ \t\n\r;#$'"\\]+              { TOK_S(TOK_ARRAY_ELEM); }
+<IN_ARRAY>[^ \t\n\r;#$'"]+              { TOK_S(TOK_ARRAY_ELEM); }
 <IN_ARRAY>.                                { DRIVER->lex_error(yytext, yylineno); }
 
 <IN_ARRAY_BRACE>"${"                       { DRIVER->begin_brace(); DRIVER->set_yymore(); yymore(); }
@@ -124,7 +124,7 @@ COMMENT    #[^\n]*
 <IN_DARRAY>"${"                            { DRIVER->begin_brace(); DRIVER->set_yymore(); yymore(); BEGIN(IN_DARRAY_BRACE); }
 <IN_DARRAY>{FLOAT}                          { TOK_S(TOK_FLOAT); }
 <IN_DARRAY>{INTEGER}                       { TOK_S(TOK_INTEGER); }
-<IN_DARRAY>[^ \t\n\r;#$'"\\]+             { TOK_S(TOK_ARRAY_ELEM); }
+<IN_DARRAY>[^ \t\n\r;#$'"]+             { TOK_S(TOK_ARRAY_ELEM); }
 <IN_DARRAY>.                               { DRIVER->lex_error(yytext, yylineno); }
 
 <IN_DARRAY_BRACE>"${"                      { DRIVER->begin_brace(); DRIVER->set_yymore(); yymore(); }

tests/test_hit.cpp

@@ -161,6 +161,18 @@ main()
     EXPECT(a1 == a2);
   });
 
+  run("backslash_path_in_quotes", []() {
+    // Backslashes must survive inside quoted strings so Windows paths are
+    // representable. Both quote styles tokenise as arrays, so the value has no
+    // spaces (whitespace still separates array elements). The C++ literals
+    // below use \\ per source escaping; the HIT text and parsed value each hold
+    // single backslashes.
+    auto sq = p("path = 'C:\\Users\\me\\model'");
+    EXPECT(sq->param<std::string>("path") == "C:\\Users\\me\\model");
+    auto dq = p("path = \"C:\\Users\\me\\model\"");
+    EXPECT(dq->param<std::string>("path") == "C:\\Users\\me\\model");
+  });
+
   // ── 2. Bool values ────────────────────────────────────────────────────────
 
   run("bool_true", []() {