nmhit 0.2.1__tar.gz → 0.3.0__tar.gz

{nmhit-0.2.1 → nmhit-0.3.0}/.github/workflows/release.yml

@@ -19,12 +19,12 @@ jobs:
       - uses: actions/checkout@v4
 
       - name: Build wheels
-        uses: pypa/cibuildwheel@v2.23
+        uses: pypa/cibuildwheel@v3.4.1
         env:
           CIBW_BUILD: "cp39-* cp310-* cp311-* cp312-* cp313-* cp314-*"
           CIBW_SKIP:  "*-musllinux* *-manylinux_i686"
           CIBW_ARCHS_LINUX: "x86_64"
-          CIBW_ARCHS_MACOS: "x86_64"
+          CIBW_ARCHS_MACOS: "x86_64 arm64"
           CIBW_MANYLINUX_X86_64_IMAGE: manylinux_2_28
           CIBW_TEST_REQUIRES: "pytest"
           CIBW_TEST_COMMAND:  "pytest {project}/python/tests/ -v"
@@ -74,6 +74,8 @@ jobs:
 
       - name: Publish
         uses: pypa/gh-action-pypi-publish@release/v1
+        with:
+          skip-existing: true
         # No API token needed — configure Trusted Publishing once on pypi.org:
         # Publisher: GitHub Actions, repo: <owner>/nmhit, workflow: release.yml,
         # environment: pypi

{nmhit-0.2.1 → nmhit-0.3.0}/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 0.2.1 LANGUAGES CXX)
+project(neml2-hit VERSION 0.3.0 LANGUAGES CXX)
 
 set(CMAKE_CXX_STANDARD 17)
 set(CMAKE_CXX_STANDARD_REQUIRED ON)

{nmhit-0.2.1 → nmhit-0.3.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.2
 Name: nmhit
-Version: 0.2.1
+Version: 0.3.0
 Summary: Python bindings for the nmhit NEML2-flavored HIT parser
 License: MIT
 Classifier: Programming Language :: Python :: 3

{nmhit-0.2.1 → nmhit-0.3.0}/generated/Lexer.cpp

@@ -1004,8 +1004,8 @@ typedef nmhit_detail::ParseDriver HIT_Driver;
                       return nmhit_detail::Parser::token::t; } while(0)
 #define TOK_UNQUOTED do { DRIVER->flush_unquoted(); \
                           return nmhit_detail::Parser::token::TOK_UNQUOTED_STR; } while(0)
-#define TOK_TRIPLE do { \
-    DRIVER->set_verbatim_pending(DRIVER->_triple_str.c_str(), (int)DRIVER->_triple_str.size()); \
+#define TOK_TRIPLE(delim) do { \
+    DRIVER->set_verbatim_pending(DRIVER->_triple_str.c_str(), (int)DRIVER->_triple_str.size(), delim); \
     return nmhit_detail::Parser::token::TOK_UNQUOTED_STR; \
 } while(0)
 #define YY_NO_INPUT 1
@@ -1581,7 +1581,7 @@ YY_RULE_SETUP
 	YY_BREAK
 case 32:
 YY_RULE_SETUP
-{ BEGIN(SECTION_BODY); TOK_TRIPLE; }
+{ BEGIN(SECTION_BODY); TOK_TRIPLE('\''); }
 	YY_BREAK
 case 33:
 YY_RULE_SETUP
@@ -1598,7 +1598,7 @@ YY_RULE_SETUP
 	YY_BREAK
 case 36:
 YY_RULE_SETUP
-{ BEGIN(SECTION_BODY); TOK_TRIPLE; }
+{ BEGIN(SECTION_BODY); TOK_TRIPLE('"'); }
 	YY_BREAK
 case 37:
 YY_RULE_SETUP

{nmhit-0.2.1 → nmhit-0.3.0}/generated/Parser.h

@@ -32,7 +32,7 @@
 
 
 /**
- ** \file _build/Parser.h
+ ** \file /home/thu/projects/neml2-hit/build/Parser.h
  ** Define the nmhit_detail::parser class.
  */
 
@@ -42,8 +42,8 @@
 // especially those whose name start with YY_ or yy_.  They are
 // private implementation details that can be changed or removed.
 
-#ifndef YY_YY_BUILD_PARSER_H_INCLUDED
-# define YY_YY_BUILD_PARSER_H_INCLUDED
+#ifndef YY_YY_HOME_THU_PROJECTS_NEML2_HIT_BUILD_PARSER_H_INCLUDED
+# define YY_YY_HOME_THU_PROJECTS_NEML2_HIT_BUILD_PARSER_H_INCLUDED
 // "%code requires" blocks.
 
 #include <memory>
@@ -1515,4 +1515,4 @@ switch (yykind)
 
 
 
-#endif // !YY_YY_BUILD_PARSER_H_INCLUDED
+#endif // !YY_YY_HOME_THU_PROJECTS_NEML2_HIT_BUILD_PARSER_H_INCLUDED

{nmhit-0.2.1 → nmhit-0.3.0}/generated/location.hh

@@ -31,12 +31,12 @@
 // version 2.2 of Bison.
 
 /**
- ** \file _build/location.hh
+ ** \file /home/thu/projects/neml2-hit/build/location.hh
  ** Define the nmhit_detail::location class.
  */
 
-#ifndef YY_YY_BUILD_LOCATION_HH_INCLUDED
-# define YY_YY_BUILD_LOCATION_HH_INCLUDED
+#ifndef YY_YY_HOME_THU_PROJECTS_NEML2_HIT_BUILD_LOCATION_HH_INCLUDED
+# define YY_YY_HOME_THU_PROJECTS_NEML2_HIT_BUILD_LOCATION_HH_INCLUDED
 
 # include <iostream>
 # include <string>
@@ -299,4 +299,4 @@ namespace nmhit_detail {
 
 } // nmhit_detail
 
-#endif // !YY_YY_BUILD_LOCATION_HH_INCLUDED
+#endif // !YY_YY_HOME_THU_PROJECTS_NEML2_HIT_BUILD_LOCATION_HH_INCLUDED

{nmhit-0.2.1 → nmhit-0.3.0}/include/nmhit/Node.h

@@ -266,7 +266,10 @@ private:
 class Field : public Node
 {
 public:
-  Field(const std::string & name, const std::string & raw_value, bool verbatim = false);
+  Field(const std::string & name,
+        const std::string & raw_value,
+        bool verbatim = false,
+        char verbatim_delim = '\'');
 
   NodeType type() const override { return NodeType::Field; }
   std::string path() const override { return _name; }
@@ -283,10 +286,15 @@ public:
   /// Such fields can only be read via param<std::string>() / param_str().
   bool is_verbatim() const { return _verbatim; }
 
+  /// Which delimiter the verbatim body was wrapped in originally: '\'' or '"'.
+  /// Meaningful only when is_verbatim() is true.
+  char verbatim_delim() const { return _verbatim_delim; }
+
 private:
   std::string _name;
   std::string _raw;
   bool _verbatim = false;
+  char _verbatim_delim = '\'';
 };
 
 /// A comment, e.g. # some text.

{nmhit-0.2.1 → nmhit-0.3.0}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "scikit_build_core.build"
 
 [project]
 name = "nmhit"
-version = "0.2.1"  # Keep in sync with VERSION in CMakeLists.txt.
+version = "0.3.0"  # 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"
@@ -42,5 +42,5 @@ archs = ["x86_64"]
 manylinux-x86_64-image = "manylinux_2_28"
 
 [tool.cibuildwheel.macos]
-archs = ["x86_64"]
+archs = ["x86_64", "arm64"]
 environment = {MACOSX_DEPLOYMENT_TARGET = "10.15"}  # std::filesystem requires 10.15+

{nmhit-0.2.1 → nmhit-0.3.0}/python/tests/test_nmhit.py

@@ -625,3 +625,54 @@ def test_triple_quote_not_readable_as_list():
     assert root.param_str("k") == "1 2 3"
     with pytest.raises(nmhit.Error, match="verbatim"):
         root.param_list_int("k")
+
+
+# ── triple-quoted round-trip rendering ────────────────────────────────────────
+# Round-trip = parse_text(s).render() == s, byte-for-byte. The formatter
+# (nmhit-format) walks this exact pipeline, so any divergence here is a
+# direct formatter-breaking bug.
+
+
+def test_triple_single_quote_roundtrip_inline():
+    src = "k = '''hello world'''\n"
+    assert nmhit.parse_text(src).render() == src
+
+
+def test_triple_double_quote_roundtrip_inline():
+    src = 'k = """hello world"""\n'
+    assert nmhit.parse_text(src).render() == src
+
+
+def test_triple_single_quote_roundtrip_multiline():
+    src = "k = '''\n  line1\n  line2\n'''\n"
+    assert nmhit.parse_text(src).render() == src
+
+
+def test_triple_double_quote_roundtrip_multiline():
+    src = 'k = """\n  line1\n  line2\n"""\n'
+    assert nmhit.parse_text(src).render() == src
+
+
+def test_triple_quote_roundtrip_python_expression():
+    src = (
+        "[Tensors]\n"
+        "  [strain]\n"
+        "    type = Python\n"
+        "    expr = '''\n"
+        "torch.stack([\n"
+        "    torch.linspace(0, 1, 5),\n"
+        "    torch.linspace(1, 2, 5),\n"
+        "])\n"
+        "'''\n"
+        "  []\n"
+        "[]\n"
+    )
+    assert nmhit.parse_text(src).render() == src
+
+
+def test_triple_quote_roundtrip_preserves_delimiter_style():
+    """The renderer must wrap the body in the same delimiter the input used."""
+    src_sq = "k = '''\nbody\n'''\n"
+    src_dq = 'k = """\nbody\n"""\n'
+    assert nmhit.parse_text(src_sq).render() == src_sq
+    assert nmhit.parse_text(src_dq).render() == src_dq

{nmhit-0.2.1 → nmhit-0.3.0}/src/Lexer.l

@@ -16,8 +16,8 @@ typedef nmhit_detail::ParseDriver HIT_Driver;
                       return nmhit_detail::Parser::token::t; } while(0)
 #define TOK_UNQUOTED do { DRIVER->flush_unquoted(); \
                           return nmhit_detail::Parser::token::TOK_UNQUOTED_STR; } while(0)
-#define TOK_TRIPLE do { \
-    DRIVER->set_verbatim_pending(DRIVER->_triple_str.c_str(), (int)DRIVER->_triple_str.size()); \
+#define TOK_TRIPLE(delim) do { \
+    DRIVER->set_verbatim_pending(DRIVER->_triple_str.c_str(), (int)DRIVER->_triple_str.size(), delim); \
     return nmhit_detail::Parser::token::TOK_UNQUOTED_STR; \
 } while(0)
 %}
@@ -76,12 +76,12 @@ COMMENT    #[^\n]*
 <AFTER_EQ>[^ \t\n\r\[#$'"\\]+             { DRIVER->begin_unquoted(yytext, yyleng); BEGIN(IN_UNQUOTED); }
 <AFTER_EQ>.                                { DRIVER->lex_error(yytext, yylineno); BEGIN(SECTION_BODY); }
 
-<IN_TRIPLE_SQ>"'''"                        { BEGIN(SECTION_BODY); TOK_TRIPLE; }
+<IN_TRIPLE_SQ>"'''"                        { BEGIN(SECTION_BODY); TOK_TRIPLE('\''); }
 <IN_TRIPLE_SQ>"''"                         { DRIVER->_triple_str += yytext; }
 <IN_TRIPLE_SQ>"'"                          { DRIVER->_triple_str += yytext; }
 <IN_TRIPLE_SQ>[^']+                        { DRIVER->_triple_str += yytext; }
 
-<IN_TRIPLE_DQ>"\"\"\""                     { BEGIN(SECTION_BODY); TOK_TRIPLE; }
+<IN_TRIPLE_DQ>"\"\"\""                     { BEGIN(SECTION_BODY); TOK_TRIPLE('"'); }
 <IN_TRIPLE_DQ>"\"\""                       { DRIVER->_triple_str += yytext; }
 <IN_TRIPLE_DQ>"\""                         { DRIVER->_triple_str += yytext; }
 <IN_TRIPLE_DQ>[^\"]+                       { DRIVER->_triple_str += yytext; }

{nmhit-0.2.1 → nmhit-0.3.0}/src/Node.cpp

@@ -524,8 +524,11 @@ Section::clone() const
 // Field
 // ═══════════════════════════════════════════════════════════════════════════════
 
-Field::Field(const std::string & name, const std::string & raw_value, bool verbatim)
-  : _name(name), _raw(raw_value), _verbatim(verbatim)
+Field::Field(const std::string & name,
+             const std::string & raw_value,
+             bool verbatim,
+             char verbatim_delim)
+  : _name(name), _raw(raw_value), _verbatim(verbatim), _verbatim_delim(verbatim_delim)
 {}
 
 void
@@ -540,13 +543,22 @@ Field::render(int indent, const std::string & indent_text) const
   std::string pfx;
   for (int i = 0; i < indent; ++i)
     pfx += indent_text;
+  if (_verbatim)
+  {
+    // Wrap the body back in its original triple-quote delimiter so round-
+    // tripping is byte-exact. Body cannot contain three consecutive copies
+    // of the delimiter (the lexer would have closed the string), so the
+    // surrounding triple-quote is always a valid wrapping.
+    std::string triple(3, _verbatim_delim);
+    return pfx + _name + " = " + triple + _raw + triple + "\n";
+  }
   return pfx + _name + " = " + _raw + "\n";
 }
 
 std::unique_ptr<Node>
 Field::clone() const
 {
-  auto f = std::make_unique<Field>(_name, _raw);
+  auto f = std::make_unique<Field>(_name, _raw, _verbatim, _verbatim_delim);
   f->_set_location(filename(), line(), column());
   return f;
 }
@@ -937,8 +949,11 @@ ParseDriver::build_field(const std::string & name,
     segs.push_back(name);
 
   bool verbatim = _next_field_verbatim;
+  char verbatim_delim = _next_field_verbatim_delim;
   _next_field_verbatim = false;
-  auto field = std::make_unique<nmhit::Field>(segs.back(), raw_value, verbatim);
+  _next_field_verbatim_delim = '\'';
+  auto field =
+    std::make_unique<nmhit::Field>(segs.back(), raw_value, verbatim, verbatim_delim);
   field->_set_location(_fname, loc.begin.line, loc.begin.column);
   if (is_override)
     _override_fields.insert(field.get());

{nmhit-0.2.1 → nmhit-0.3.0}/src/ParseDriver.h

@@ -51,11 +51,14 @@ public:
 
   /// Like set_pending, but also marks the next field as verbatim (triple-quoted).
   /// Verbatim fields can only be retrieved as std::string via param_str().
-  void set_verbatim_pending(const char * text, int len)
+  /// `delim` records which character ('\'' or '"') the triple-quote used so
+  /// the renderer can wrap the body in the same delimiter on output.
+  void set_verbatim_pending(const char * text, int len, char delim)
   {
     _pending.assign(text, len);
     _has_pending = true;
     _next_field_verbatim = true;
+    _next_field_verbatim_delim = delim;
   }
 
   /// Report a lexer error and mark the parse as failed.
@@ -139,6 +142,9 @@ private:
   // True when the pending value came from a triple-quoted verbatim string.
   // Consumed and reset in build_field(); Field stores the flag permanently.
   bool _next_field_verbatim = false;
+  // Which character ('\'' or '"') wrapped the triple-quoted value; used by
+  // Field::render to round-trip with the same delimiter style.
+  char _next_field_verbatim_delim = '\'';
 
   // Brace expression nesting counter
   int _brace_depth = 0;

{nmhit-0.2.1 → nmhit-0.3.0}/tests/test_hit.cpp

@@ -298,6 +298,53 @@ main()
     EXPECT(rendered.find("# My comment") != std::string::npos);
   });
 
+  // Byte-exact round-trip for triple-quoted (verbatim) fields. The formatter
+  // (nmhit-format) walks this exact pipeline; before this was fixed, render
+  // emitted `name = <body>` and dropped the `'''...'''` delimiters entirely.
+
+  run("round_trip_triple_quoted_inline_single", []() {
+    std::string input = "k = '''hello world'''\n";
+    EXPECT(p(input)->render() == input);
+  });
+
+  run("round_trip_triple_quoted_inline_double", []() {
+    std::string input = "k = \"\"\"hello world\"\"\"\n";
+    EXPECT(p(input)->render() == input);
+  });
+
+  run("round_trip_triple_quoted_multiline", []() {
+    std::string input = "k = '''\n  line one\n  line two\n'''\n";
+    EXPECT(p(input)->render() == input);
+  });
+
+  run("round_trip_triple_quoted_python_expr", []() {
+    // The motivating use case: a Python expression for a [Tensors] block
+    // with intentional indentation. The leading newline + indented body +
+    // trailing newline must all round-trip byte-for-byte.
+    std::string input =
+      "[Tensors]\n"
+      "  [strain]\n"
+      "    type = Python\n"
+      "    expr = '''\n"
+      "torch.stack([\n"
+      "    torch.linspace(0, 1, 5),\n"
+      "    torch.linspace(1, 2, 5),\n"
+      "])\n"
+      "'''\n"
+      "  []\n"
+      "[]\n";
+    EXPECT(p(input)->render() == input);
+  });
+
+  run("round_trip_triple_quoted_delimiter_preserved", []() {
+    // The renderer must wrap the body in the same delimiter character the
+    // input used — '''...''' stays single-triple, """...""" stays double.
+    std::string sq = "k = '''\nbody\n'''\n";
+    std::string dq = "k = \"\"\"\nbody\n\"\"\"\n";
+    EXPECT(p(sq)->render() == sq);
+    EXPECT(p(dq)->render() == dq);
+  });
+
   // ── 7. Override assignment ────────────────────────────────────────────────
 
   run("override_assign", []() {