nativegate 0.1.4__tar.gz → 0.1.5__tar.gz

{nativegate-0.1.4 → nativegate-0.1.5}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: nativegate
-Version: 0.1.4
+Version: 0.1.5
 Summary: Make legacy C++/C/Fortran AI-ready and microservice-ready: expose native code as an MCP server, REST API and container
 Requires-Python: >=3.10
 Description-Content-Type: text/markdown

{nativegate-0.1.4 → nativegate-0.1.5}/nativegate/parsers/dialect.py

@@ -41,7 +41,15 @@ import re
 
 DIALECT_CHOICES = ("cs", "cd")
 
-_PREFIX_RE = re.compile(r"^(CS|CD)(?=[ \t])", re.IGNORECASE | re.MULTILINE)
+# Statement labels are dialect-marked too ("CD900 DGAMMA = RES"), so no
+# space lookahead: any line starting with CS/CD is prefix-marked, with a
+# following space, tab, digit or line end allowed out to avoid tripping
+# on prose comments like "CSIMPLER". A live statement never starts with
+# literal letters in column 1 — the statement field begins in column 7 —
+# so there is no legitimate Fortran that begins "CS"/"CD" here. A comment
+# that merely starts with those letters is, at worst, re-marked; the
+# only real hazard would be more labels lost to a stricter pattern.
+_PREFIX_RE = re.compile(r"^(CS|CD)(?=[ \t\n]|\d|$)", re.IGNORECASE | re.MULTILINE)
 
 
 def resolve_dialect(raw: str | None) -> str:

{nativegate-0.1.4 → nativegate-0.1.5}/nativegate.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: nativegate
-Version: 0.1.4
+Version: 0.1.5
 Summary: Make legacy C++/C/Fortran AI-ready and microservice-ready: expose native code as an MCP server, REST API and container
 Requires-Python: >=3.10
 Description-Content-Type: text/markdown

{nativegate-0.1.4 → nativegate-0.1.5}/pyproject.toml

@@ -13,7 +13,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "nativegate"
-version = "0.1.4"
+version = "0.1.5"
 description = "Make legacy C++/C/Fortran AI-ready and microservice-ready: expose native code as an MCP server, REST API and container"
 readme = "README.md"
 requires-python = ">=3.10"

{nativegate-0.1.4 → nativegate-0.1.5}/tests/test_dialect.py

@@ -45,10 +45,42 @@ CD    DOUBLE PRECISION
      1    C,CONV,EPS
       DGAMMA = 24.0D0
       RETURN
+C the labelled assignment is dialect-marked as well: label in cols 1-5,
+C prefix occupying cols 1-2, so the 9 lands in column 3 — still inside
+C the label field (columns 1-5) after two blanks are substituted.
+900   DGAMMA = 24.0D0
+      RETURN
+      END
+""")
+
+DUAL_WITH_MARKED_LABEL = textwrap.dedent("""\
+CD    DOUBLE PRECISION FUNCTION DGAMMA(X)
+CS    REAL FUNCTION GAMMA(X)
+      GO TO 900
+CD900 DGAMMA = 24.0D0
+CS900 GAMMA = 24.0E0
+      RETURN
+900   CONTINUE
+      RETURN
       END
 """)
 
 
+def test_apply_dialect_marks_statement_labels_too():
+    """`CD900 DGAMMA = ...` is the line that exposed the strict space-lookahead
+    regex: labels start at column 1 and the prefix replaces its first two
+    characters, so the following character is a digit, not a space. A
+    space-lookahead regex would leave the label line inert and the resolved
+    file compiles with "label 900 referenced but never defined"."""
+    out = apply_dialect(DUAL_WITH_MARKED_LABEL, "cd")
+    lines = out.splitlines()
+    # label survives in the label field (digits starting column 3 is legal)
+    assert "  900 DGAMMA = 24.0D0" in lines
+    # the CS half is commented, label included
+    assert "C 900 GAMMA = 24.0E0" in lines
+    assert "C     REAL FUNCTION GAMMA(X)" in lines
+
+
 def test_apply_dialect_selects_cd():
     out = apply_dialect(DUAL_DIALECT, "cd")
     lines = out.splitlines()