PR comments

Author: hbmartin

.github/workflows/lint.yml

@@ -25,17 +25,16 @@ jobs:
       - run: pip install -r requirements.txt
       - run: pip install black ruff pytest pytest-cov
       - run: ruff check .
-      - name: Run mypy if on 3.12, pytype otherwise
+      - name: Run pytype if on 3.11, pyrefly if on 3.14, mypy otherwise
         run: |
           if [[ '${{ steps.cpython3.outputs.python-version }}' == 3.11* ]]; then
             pip install pytype
             pytype -j auto graphviz2drawio
           elif [[ '${{ steps.cpython3.outputs.python-version }}' == 3.14* ]]; then
             echo "Using pyrefly for Python 3.14"
             pip install pyrefly
-            pyrefly check
+            pyrefly check graphviz2drawio
           else
-            echo "Using mypy for Python 3.12"
             pip install mypy
             mypy graphviz2drawio --ignore-missing-imports
           fi

graphviz2drawio/__main__.py

@@ -1,6 +1,5 @@
 #!/usr/bin/env python3
 
-import pathlib
 import sys
 from argparse import Namespace
 from io import TextIOWrapper
@@ -15,7 +14,7 @@
 RED_TEXT = "\033[91m"
 BOLD = "\033[1m"
 
-utf8 = "utf-8"
+UTF8 = "utf-8"
 
 
 def _gv_filename_to_xml(filename: str) -> str:
@@ -45,25 +44,17 @@ def _convert_file(
             with to_convert.open(encoding=encoding) as contents:
                 output = convert(contents.read(), program)
     except UnicodeDecodeError:
-        if encoding != utf8:
-            # Attempt to automatically recover. Chinese Windows systems in particular
-            # typically use other encodings e.g. gbk, cp950, cp1252, etc. but the
-            # actual dot files are still UTF-8 encoded
+        if encoding != UTF8 and isinstance(to_convert, Path):
+            # Attempt to automatically recover for file. Chinese Windows systems in
+            # particular often use other encodings e.g. gbk, cp950, cp1252, etc. but
+            # the actual dot files are still UTF-8 encoded
             # https://github.com/hbmartin/graphviz2drawio/issues/105
-            return _convert_file(to_convert, program, utf8, outfile)
-
-        from locale import getpreferredencoding
-
-        error_message = f"Error decoding {to_convert} with {utf8}"
-        stderr.write(f"{RED_TEXT}{BOLD}{error_message}\n")
-        stderr.write("Try again by specifying your file encoding with the -e flag.\n")
-        if (sys_enc := getpreferredencoding(do_setlocale=False).lower()) != utf8:
-            stderr.write(f"e.g. using your system's default encoding -e {sys_enc})")
-            stderr.write(f"\n\n{DEFAULT_TEXT}")
-        else:
-            _write_stderr_message(str(to_convert))
-            raise
-    except BaseException:
+            return _convert_file(to_convert, program, UTF8, outfile)
+
+        _write_stderr_message(str(to_convert))
+        raise
+
+    except Exception:
         _write_stderr_message(str(to_convert))
         raise
 
@@ -75,7 +66,7 @@ def _convert_file(
         print(output)
         return None
 
-    out_path = pathlib.Path(outfile)
+    out_path = Path(outfile)
     out_path.parent.mkdir(parents=True, exist_ok=True)
     out_path.write_text(output)
     stderr.write("Converted file: " + outfile + "\n")

graphviz2drawio/graphviz2drawio.py

@@ -14,25 +14,7 @@ def convert(
     graph_to_convert: AGraph | str | TextIOBase | Path | TextIO,
     layout_prog: str = "dot",
 ) -> str:
-    if isinstance(graph_to_convert, AGraph):
-        graph = graph_to_convert
-    elif isinstance(graph_to_convert, str):
-        # This fixes a pygraphviz bug where a string beginning with a comment
-        # is mistakenly identified as a filename.
-        # https://github.com/pygraphviz/pygraphviz/issues/536
-        pattern = re.compile(
-            r"(?:\s*(?:/\*[^*]*\*+(?:[^/*][^*]*\*+)*/|//[^\r\n]*|\#[^\r\n]*)\s*)*\s*(strict)?\s*(graph|digraph).*?\{.*\}\s*",
-            re.DOTALL | re.MULTILINE | re.VERBOSE,
-        )
-        if pattern.match(graph_to_convert):
-            graph = AGraph(string=graph_to_convert)
-        else:
-            graph = AGraph(filename=graph_to_convert)
-    elif hasattr(graph_to_convert, "read"):
-        graph = AGraph(string=graph_to_convert.read())
-    else:
-        # Use builtin type detection which includes:  hasattr(thing, "open")
-        graph = AGraph(graph_to_convert)
+    graph = _load_pygraphviz_graph(graph_to_convert)
 
     graph_edges: dict[str, dict] = {
         f"{e[0]}->{e[1]}-"
@@ -65,3 +47,26 @@ def convert(
     # Put clusters first, so that nodes are drawn in front
     mx_graph = MxGraph(clusters, nodes, edges)
     return mx_graph.value()
+
+
+def _load_pygraphviz_graph(
+    graph_to_convert: AGraph | str | TextIOBase | Path | TextIO,
+) -> AGraph:
+    if isinstance(graph_to_convert, AGraph):
+        return graph_to_convert
+    if isinstance(graph_to_convert, str):
+        # This fixes a pygraphviz bug where a string beginning with a comment
+        # is mistakenly identified as a filename.
+        # https://github.com/pygraphviz/pygraphviz/issues/536
+        pattern = re.compile(
+            r"(?:\s*(?:/\*[^*]*\*+(?:[^/*][^*]*\*+)*/|//[^\r\n]*|\#[^\r\n]*)\s*)*\s*(strict)?\s*(graph|digraph).*?\{.*\}\s*",
+            re.DOTALL | re.MULTILINE | re.VERBOSE,
+        )
+        if pattern.match(graph_to_convert):
+            return AGraph(string=graph_to_convert)
+        return AGraph(filename=graph_to_convert)
+    # pyrefly: ignore  # missing-attribute
+    if hasattr(graph_to_convert, "read") and callable(graph_to_convert.read):
+        return AGraph(string=graph_to_convert.read())
+    # Use builtin type detection which includes:  hasattr(thing, "open")
+    return AGraph(graph_to_convert)

graphviz2drawio/models/Arguments.py

@@ -9,9 +9,7 @@
 class NonOpeningFileType:
     def __call__(self, string: str) -> TextIO | Path:
         # the special argument "-" means sys.std{in,out}
-        if string == "-":
-            return sys.stdin
-        return Path(string)
+        return sys.stdin if string == "-" else Path(string)
 
 
 class Arguments(ArgumentParser):