updated generation process

1 files changed, 27 insertions, 2 deletions

diff --git a/src/gap_pipeline/surface.py b/src/gap_pipeline/surface.py
index 914d847..aae34a8 100644
--- a/src/gap_pipeline/surface.py
+++ b/src/gap_pipeline/surface.py
@@ -29,6 +29,24 @@ SURFACE_FAMILIES: tuple[SurfaceFamily, ...] = (
 IDENTIFIER_RE = re.compile(r"^[a-z]{8,}$")
 
 
+def apply_rename_map(text: str, rename_map: dict[str, str]) -> str:
+    """Apply symbol renames without replacing letters inside other tokens."""
+
+    rendered = text
+    for source in sorted(rename_map, key=len, reverse=True):
+        pattern = re.escape(source)
+        if source[0].isalnum():
+            pattern = rf"(?<![A-Za-z0-9_]){pattern}"
+        if source[-1].isalnum():
+            pattern = rf"{pattern}(?![A-Za-z0-9_])"
+        rendered = re.sub(
+            pattern,
+            lambda _match, replacement=rename_map[source]: replacement,
+            rendered,
+        )
+    return rendered
+
+
 @dataclass(frozen=True)
 class SurfaceVariant:
     family: SurfaceFamily
@@ -69,6 +87,13 @@ def validate_surface_variant(
             )
     if not question.strip() or not solution.strip():
         raise ValueError("surface question and solution must be non-empty")
+    expected_question = apply_rename_map(item.problem, rename_map)
+    expected_solution = apply_rename_map(item.solution, rename_map)
+    if question != expected_question or solution != expected_solution:
+        raise ValueError(
+            "surface text must be the exact deterministic rename of the "
+            "canonical question and solution"
+        )
 
 
 class SurfacePipeline:
@@ -105,8 +130,8 @@ class SurfacePipeline:
             str(old): str(new)
             for old, new in dict(response.data["map"]).items()
         }
-        question = str(response.data["question"])
-        solution = str(response.data["solution"])
+        question = apply_rename_map(item.problem, rename_map)
+        solution = apply_rename_map(item.solution, rename_map)
         validate_surface_variant(
             item,
             rename_map=rename_map,