aetherdialect 0.1.4__tar.gz → 0.1.6__tar.gz

{aetherdialect-0.1.4/src/aetherdialect.egg-info → aetherdialect-0.1.6}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: aetherdialect
-Version: 0.1.4
+Version: 0.1.6
 Summary: Deterministic, validation-first Text-to-SQL system for business databases
 Author-email: Akul Ameya <akul.ameya@gmail.com>
 License: MIT

{aetherdialect-0.1.4 → aetherdialect-0.1.6}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "aetherdialect"
-version = "0.1.4"
+version = "0.1.6"
 description = "Deterministic, validation-first Text-to-SQL system for business databases"
 readme = "README.md"
 requires-python = ">=3.10"

{aetherdialect-0.1.4 → aetherdialect-0.1.6/src/aetherdialect.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: aetherdialect
-Version: 0.1.4
+Version: 0.1.6
 Summary: Deterministic, validation-first Text-to-SQL system for business databases
 Author-email: Akul Ameya <akul.ameya@gmail.com>
 License: MIT

{aetherdialect-0.1.4 → aetherdialect-0.1.6}/src/text2sql/_config.py

@@ -197,11 +197,13 @@ VALID_SCALAR_FUNCTIONS = {
     "date_part",
     "extract",
     "coalesce",
+    "concat",
     "year",
     "month",
     "day",
 }
-SCALAR_FUNCTIONS_STRING = {"upper", "lower", "trim", "ltrim", "rtrim", "length"}
+SCALAR_FUNCTIONS_STRING = {"upper", "lower", "trim", "ltrim", "rtrim", "length", "concat"}
+SCALAR_FUNCTIONS_VARIADIC = frozenset({"concat"})
 SCALAR_FUNCTIONS_NUMERIC = {"abs", "round", "floor", "ceil"}
 SCALAR_FUNCTIONS_TEMPORAL = {
     "date_trunc",
@@ -215,6 +217,7 @@ SCALAR_FUNCTIONS_LEADING_ARG = {"date_trunc", "date_part", "extract"}
 DISALLOWED_EXTRACT_UNITS = {"epoch"}
 VALID_GRAINS = {"scalar", "grouped", "row_level"}
 VALID_EXPECTED_ROWS = {"one", "few", "many"}
+REGISTRY_TOKEN_PATTERN = r"^[wc]\d+$"
 VALID_HAVING_OPS = {"=", "!=", "<", "<=", ">", ">=", "in", "not in", "between"}
 DATABRICKS_TABLE_QUALIFY_SKIP_IDENTIFIERS: frozenset[str] = frozenset(
     {
@@ -1292,6 +1295,7 @@ INTENT_CRITICAL_RULES: tuple[str, ...] = (
     "Do not embed COUNT(*) inside arithmetic subexpressions—use COUNT(*) only as a top-level aggregate where appropriate.",
     "Arithmetic combines expressions with +, -, *, /; aggregations may wrap arithmetic (e.g. SUM(tbl_a.col_a * tbl_a.col_b)). "
     "Subtract date columns directly (tbl_a.date_a - tbl_a.date_b) for day differences.",
+    "String concatenation uses CONCAT(expr1, ' ', expr2, ...) in expr strings; do not use the SQL || operator (pipe-pipe).",
     "Apply scalar functions such as ROUND after aggregates when needed (e.g. ROUND(SUM(tbl_a.col_a), 2)).",
     "Use exact identifiers from the provided schema text; never leave synthetic shape tokens from this prompt "
     "(tbl_a, tbl_b, col_a, date_a, date_b), generic instructional tokens (table_N, column_N), or angle-bracket markup in expressions.",
@@ -1734,7 +1738,7 @@ CASE_RESULT_BARE_LABEL_RE = re.compile(r"^[A-Za-z_][A-Za-z0-9_]*$")
 CASE_RESULT_REGISTRY_TOKEN_RE = re.compile(r"^[wc]\d{2}$")
 
 ROLE_VALUE_TYPE_COMPAT: dict[str, frozenset[str]] = {
-    "boolean": frozenset({"boolean"}),
+    "boolean": frozenset({"boolean", "integer", "string"}),
     "numeric_measure": frozenset({"integer", "number"}),
     "numeric_categorical": frozenset({"integer", "number"}),
     "temporal": frozenset({"date"}),

{aetherdialect-0.1.4 → aetherdialect-0.1.6}/src/text2sql/_contracts_base.py

@@ -782,6 +782,42 @@ class FKEdge:
             self.inference_tag = coerce_inference_tag(self.inference_tag)
 
 
+@dataclass
+class CatalogTableStructuralConstraints:
+    """
+    Catalog-sourced primary-key column names, foreign-key edges, and single-column unique names for one table.
+
+    Each :class:`FKEdge` carries ``src_table`` equal to the referencing table so the bundle can be converted into ``tables_meta`` foreign-key dicts without losing the child table identity.
+    """
+
+    primary_keys: list[str] = field(default_factory=list)
+    foreign_keys: list[FKEdge] = field(default_factory=list)
+    unique_columns: list[str] = field(default_factory=list)
+
+
+@dataclass
+class CatalogStructuralConstraintsIndex:
+    """
+    Per-table structural constraint bundles keyed by lowercased relation name within one catalog schema.
+
+    When ``tables`` is empty the caller should treat catalog reflection as unavailable and continue with DDL-based parsing.
+    """
+
+    tables: dict[str, CatalogTableStructuralConstraints] = field(default_factory=dict)
+
+    @classmethod
+    def empty(cls) -> CatalogStructuralConstraintsIndex:
+        """
+        Construct an empty index for failed information_schema queries.
+
+        Returns:
+
+            Empty :class:`CatalogStructuralConstraintsIndex` instance.
+        """
+
+        return cls(tables={})
+
+
 @dataclass
 class ValueDomain:
     """Value domain for sampling concrete values during question generation."""
@@ -1521,6 +1557,7 @@ class SchemaGraph:
                     continue
                 pk_marker = " [PK]" if c.is_primary_key else ""
                 fk_marker = f" [FK->{c.fk_target[0]}.{c.fk_target[1]}]" if c.fk_target else ""
+                unique_marker = " [UNIQUE]" if (c.is_unique and not c.is_primary_key) else ""
                 role_tag = ""
                 if c.role and c.role not in _SKIP_ROLE_TAGS:
                     role_tag = f" [{c.role}]"
@@ -1534,7 +1571,7 @@ class SchemaGraph:
                 if vt.lower() == "boolean" and c.boolean_truth_value:
                     truth_tag = f" truth_value={c.boolean_truth_value}"
                 filter_tag = " [filter]" if c.is_filterable else ""
-                out.append(f"  {c.name}: {vt}{truth_tag}{pk_marker}{fk_marker}{role_tag}{hint_tag}{filter_tag}")
+                out.append(f"  {c.name}: {vt}{truth_tag}{pk_marker}{fk_marker}{unique_marker}{role_tag}{hint_tag}{filter_tag}")
         if self.enum_values:
             out.append("")
             out.append("ENUM TYPES:")

{aetherdialect-0.1.4 → aetherdialect-0.1.6}/src/text2sql/_contracts_core.py

@@ -2526,9 +2526,6 @@ class RuntimeIntent:
     distinct_select_index: int = -1
     extra_tables: set[str] = field(default_factory=set)
     sql_param: str = ""
-    sql_display_param: str = ""
-    sql_substituted: str = ""
-    deterministic_sql: str = ""
     sql_shape: SQLShape | None = None
     schema_invalid: bool = False
 
@@ -2614,9 +2611,6 @@ class RuntimeIntent:
             distinct_select_index=int(d.get("distinct_select_index", -1)),
             extra_tables=set(d.get("extra_tables", [])),
             sql_param=d.get("sql_param", ""),
-            sql_display_param=d.get("sql_display_param", ""),
-            sql_substituted=d.get("sql_substituted", ""),
-            deterministic_sql=d.get("deterministic_sql", ""),
             sql_shape=(SQLShape.from_dict(d["sql_shape"]) if d.get("sql_shape") else None),
             schema_invalid=d.get("schema_invalid", False),
         )
@@ -2651,9 +2645,6 @@ class RuntimeIntent:
             "distinct_select_index": self.distinct_select_index,
             "extra_tables": sorted(self.extra_tables),
             "sql_param": self.sql_param,
-            "sql_display_param": self.sql_display_param,
-            "sql_substituted": self.sql_substituted,
-            "deterministic_sql": self.deterministic_sql,
             "sql_shape": self.sql_shape.to_dict() if self.sql_shape else None,
             "schema_invalid": self.schema_invalid,
         }
@@ -3510,7 +3501,6 @@ class Template:
     intent_key: str
     tables_used: list[str]
     sql_param: str
-    sql_display_param: str
     sql_fp: str
     shape: SQLShape
     colmap_sig: str
@@ -3520,9 +3510,6 @@ class Template:
     source: str = "human"
     trust_level: int = 1
     structural_defaults: dict[str, str | int | float] = field(default_factory=dict)
-    deterministic_sql: str = ""
-    aliased_sql: str = ""
-    execution_sql: str = ""
     display_alias_map: dict[str, str] = field(default_factory=dict)
     feedback_by_question: dict[str, FeedbackCounts] = field(default_factory=dict)
 
@@ -3591,7 +3578,6 @@ class Template:
             intent_key=d.get("intent_key", ""),
             tables_used=d.get("tables_used", []),
             sql_param=d.get("sql_param", ""),
-            sql_display_param=d.get("sql_display_param", ""),
             sql_fp=d.get("sql_fp", ""),
             shape=shape,
             colmap_sig=d.get("colmap_sig", ""),
@@ -3601,9 +3587,6 @@ class Template:
             source=d.get("source", "human"),
             trust_level=d.get("trust_level", 1),
             structural_defaults=d.get("structural_defaults", {}),
-            deterministic_sql=d.get("deterministic_sql", ""),
-            aliased_sql=d.get("aliased_sql", ""),
-            execution_sql=d.get("execution_sql") or d.get("spark_sql_param", ""),
             display_alias_map=dict(d.get("display_alias_map") or {}),
             feedback_by_question=feedback_by_question,
         )
@@ -3628,7 +3611,6 @@ class Template:
             "intent_key": self.intent_key,
             "tables_used": self.tables_used,
             "sql_param": self.sql_param,
-            "sql_display_param": self.sql_display_param,
             "sql_fp": self.sql_fp,
             "shape": self.shape.to_dict(),
             "colmap_sig": self.colmap_sig,
@@ -3638,9 +3620,6 @@ class Template:
             "source": self.source,
             "trust_level": self.trust_level,
             "structural_defaults": self.structural_defaults,
-            "deterministic_sql": self.deterministic_sql,
-            "aliased_sql": self.aliased_sql,
-            "execution_sql": self.execution_sql,
             "display_alias_map": self.display_alias_map,
             "feedback_by_question": {q: c.to_dict() for q, c in self.feedback_by_question.items()},
         }

{aetherdialect-0.1.4 → aetherdialect-0.1.6}/src/text2sql/_core_utils.py

@@ -80,6 +80,29 @@ def notify(message: str) -> None:
     print(message, file=sys.stdout, flush=True)
 
 
+_progress_depth = 0
+
+
+def progress(message: str) -> None:
+    """Print a status line only when :func:`progress_enabled` is active (interactive CLI)."""
+
+    if _progress_depth <= 0:
+        return
+    print(message, file=sys.stdout, flush=True)
+
+
+@contextmanager
+def progress_enabled() -> Iterator[None]:
+    """Enable :func:`progress` writes for the duration of the block (supports nesting)."""
+
+    global _progress_depth
+    _progress_depth += 1
+    try:
+        yield
+    finally:
+        _progress_depth -= 1
+
+
 def result(message: str) -> None:
     """Print a query-result line to stdout with no prefix."""
 
@@ -1182,7 +1205,7 @@ def ask_user_choice(prompt: str, options: list[str], silent_no: bool = False) ->
         ``"y"``, ``"n"``, or ``None`` on EOF/invalid.
     """
     options_display = "/".join(options)
-    print(f"{prompt} [{options_display}]: ", end="")
+    print(f"{prompt} ({options_display}): ", end="", flush=True)
     try:
         user_input = input().strip()
     except (EOFError, KeyboardInterrupt):