Long-awaited rewrite to pyparsing and AST

.flake8

@@ -4,11 +4,12 @@ max-complexity = 16
 # B = bugbear
 # B9 = bugbear opinionated (incl line length)
 select = C,E,F,W,B,B9
+# B905 `zip()` without an explicit `strict=` parameter
 # E203: whitespace before ':' (black behaviour)
 # E501: flake8 line length (covered by bugbear B950)
 # W503: line break before binary operator (black behaviour)
-ignore = E203,E501,W503
+ignore = B905,B907,E203,E501,W503
 classmethod-decorators =
     classmethod
     validator
-exclude = pytopas/lark_standalone.py
+per-file-ignores = __init__.py:F401

.github/workflows/linter.yml

@@ -12,7 +12,7 @@ jobs:
     runs-on: ubuntu-latest
     strategy:
       matrix:
-        python-version: ["3.8"]
+        python-version: ["3.11"]
 
     steps:
       - name: Checkout repository
@@ -50,5 +50,5 @@ jobs:
 
       - name: pyupgrade
         run: |
-          find -name "*.py" -and -not -name "lark_standalone.py" -print0 |\
-          xargs -0 -n1 pyupgrade --py38-plus
+          find -name "*.py" -print0 |\
+          xargs -0 -n1 pyupgrade --py38-plus --keep-runtime-typing

README.md

@@ -7,24 +7,31 @@ This is the parser for Bruker's TOPAS macro language. We have compared two canon
 Parse TOPAS input and convert it to JSON:
 
 ```python
+import json
 from pytopas import TOPASParser
 
-src = "a = a + 1 ; 0"
+src = "a(b,c)"
 
-parser = TOPASParser()
-tree = parser.parse(src)
-print(tree.to_json(compact=True))
+tree = TOPASParser.parse(src)
+print(json.dumps(tree))
 ```
 
 Convert parser's JSON-encoded TOPAS code back into the TOPAS input format:
 
 ```python
+import json
 from pytopas import TOPASParseTree
 
-input_json = '["topas", ["a = a + 1 ; 0"]]'
-
-tree = TOPASParseTree.from_json(input_json)
-print(tree.to_topas())
+input_json = """
+["topas",
+  ["formula",
+    ["func_call", "a",
+      ["formula", ["p", {"n": ["parameter_name", "b"]}]],
+      ["formula", ["p", {"n": ["parameter_name", "c"]}]]]]]
+"""
+serialized = json.loads(input_json)
+src = TOPASParser.reconstruct(serialized)
+print(src)
 
 ```
 
@@ -33,16 +40,16 @@ print(tree.to_topas())
 After installing the package, two command line utilities will be available.
 
 ```
-usage: topas2json [-h] [-c] file
+usage: topas2json [-h] [--ignore-warnings] file
 
 Parse TOPAS input and output JSON
 
 positional arguments:
-  file           Path to TOPAS file or '-' for stdin input
+  file               Path to TOPAS file or '-' for stdin input
 
 options:
-  -h, --help     show this help message and exit
-  -c, --compact  Use compact output
+  -h, --help         show this help message and exit
+  --ignore-warnings  Don't print parsing warnings
 ```
 
 ```
@@ -60,9 +67,7 @@ options:
 
 ## Development
 
-Install package with optional dependencies: `pip install -e .[dev,lint,test,release]`
-
-Regenerate standalone `lark` parser after `grammar.lark` change: `make parser`
+Install package with optional dependencies: `pip install -e .[lint,test,release]`
 
 ## License
 

pyproject.toml

@@ -5,7 +5,7 @@ build-backend = "flit_core.buildapi"
 
 [project]
 name = "pytopas"
-version = "0.0.1"
+version = "0.0.3"
 description = """Bruker's TOPAS macro language parser"""
 authors = [{name = "Sergei Korolev", email = "[email protected]"}]
 readme = "README.md"
@@ -30,12 +30,11 @@ classifiers = [
 keywords = ["TOPAS", "XRPD", "Rietveld refinement", "pair distribution function", "stacking-faults", "charge flipping", "structure solution", "deconvolution"]
 requires-python = ">=3.8"
 dependencies = [
+    "pyparsing >= 3",
+    "typing-extensions >= 4.2.0; python_version < '3.11'",
 ]
 
 [project.optional-dependencies]
-dev = [
-    "lark >= 1.1.7",
-]
 lint = [
      "autoflake",
      "black >= 23.1",
@@ -71,10 +70,8 @@ ignore-init-module-imports = true
 remove-all-unused-imports = true
 remove-duplicate-keys = true
 remove-unused-variables = true
-exclude = "pytopas/lark_standalone.py"
 
 [tool.black]
-exclude = "pytopas/lark_standalone.py"
 target-version = ['py38', 'py39', 'py310', 'py311']
 
 [tool.commitizen]
@@ -83,27 +80,22 @@ tag_format = "v$version"
 major_version_zero = true
 version_files = [
     "pyproject.toml:^version",
-    "pytopas/__init__.py:^__version__",
+    "pytopas/__init__.py:^__VERSION__",
 ]
 version_provider = "pep621"
 update_changelog_on_bump = true
 
 [tool.coverage.run]
-omit = ['pytopas/lark_standalone.py', 'tests/*']
+omit = ['tests/*']
 
 [tool.flit.module]
 name = "pytopas"
 
-[tool.flit.sdist]
-include = ["pytopas/grammar.lark"]
-
 [tool.isort]
 profile = "black"
 py_version = 38
-skip = ["pytopas/lark_standalone.py"]
 
 [tool.pylint.MASTER]
-ignore = "lark_standalone.py"
 load-plugins=[
     "pylint_per_file_ignores",
 ]
@@ -115,7 +107,7 @@ recursive = true
 suggestion-mode = true
 
 [tool.pylint.basic]
-good-names = ["id", "x", "xy", "y", "_", "__"]
+good-names = ["id", "kv", "op", "x", "xy", "y", "_", "__"]
 
 [tool.pylint.design]
 max-args = 7

pytopas/__init__.py

@@ -1,9 +1,5 @@
 "TOPAS parser"
 
-from .lark_standalone import UnexpectedToken
-from .parser import TOPASParser
-from .tree import TOPASParseTree
+from .parser import Parser as TOPASParser
 
-__VERSION__ = "0.0.1"
-
-__all__ = ["TOPASParser", "TOPASParseTree", "UnexpectedToken"]
+__VERSION__ = "0.0.3"