Spaces:
Sleeping
Sleeping
Upload engine.py
Browse files
engine.py
CHANGED
|
@@ -86,20 +86,34 @@ def resolve_join_path(start_table, end_table):
|
|
| 86 |
)
|
| 87 |
|
| 88 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 89 |
def resolve_field(field_name, module):
|
| 90 |
meta = load_metadata()
|
| 91 |
fields = meta["fields"]
|
| 92 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 93 |
if field_name not in fields:
|
| 94 |
raise ValueError(f"Unknown field: {field_name}")
|
| 95 |
|
| 96 |
field = fields[field_name]
|
| 97 |
|
|
|
|
| 98 |
if field["module"] != module:
|
| 99 |
raise ValueError(
|
| 100 |
f"Field '{field_name}' does not belong to module '{module}'"
|
| 101 |
)
|
| 102 |
|
|
|
|
| 103 |
if "table" not in field or "column" not in field:
|
| 104 |
raise ValueError(
|
| 105 |
f"Field '{field_name}' is missing table/column mapping"
|
|
@@ -108,6 +122,7 @@ def resolve_field(field_name, module):
|
|
| 108 |
return field
|
| 109 |
|
| 110 |
|
|
|
|
| 111 |
def build_join_sql(base_table, steps):
|
| 112 |
sql = []
|
| 113 |
prev_alias = base_table # alias == table name
|
|
@@ -128,8 +143,24 @@ def build_join_sql(base_table, steps):
|
|
| 128 |
# =========================
|
| 129 |
|
| 130 |
def parse_intent(question):
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 131 |
prompt = f"""
|
| 132 |
-
You are a query planner.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 133 |
|
| 134 |
Extract:
|
| 135 |
- module
|
|
@@ -142,7 +173,7 @@ Example:
|
|
| 142 |
{{
|
| 143 |
"module": "employees",
|
| 144 |
"filters": [
|
| 145 |
-
{{ "field": "
|
| 146 |
]
|
| 147 |
}}
|
| 148 |
|
|
@@ -158,6 +189,7 @@ User question:
|
|
| 158 |
|
| 159 |
return json.loads(res.choices[0].message.content)
|
| 160 |
|
|
|
|
| 161 |
# =========================
|
| 162 |
# SQL GENERATOR
|
| 163 |
# =========================
|
|
|
|
| 86 |
)
|
| 87 |
|
| 88 |
|
| 89 |
+
FIELD_ALIASES = {
|
| 90 |
+
"join_date": "start_date",
|
| 91 |
+
"joining_date": "start_date",
|
| 92 |
+
"joined": "start_date",
|
| 93 |
+
"hire_date": "start_date"
|
| 94 |
+
}
|
| 95 |
+
|
| 96 |
def resolve_field(field_name, module):
|
| 97 |
meta = load_metadata()
|
| 98 |
fields = meta["fields"]
|
| 99 |
|
| 100 |
+
# 🔹 Normalize field name
|
| 101 |
+
field_name = field_name.lower().strip()
|
| 102 |
+
field_name = FIELD_ALIASES.get(field_name, field_name)
|
| 103 |
+
|
| 104 |
+
# 🔹 Validate existence
|
| 105 |
if field_name not in fields:
|
| 106 |
raise ValueError(f"Unknown field: {field_name}")
|
| 107 |
|
| 108 |
field = fields[field_name]
|
| 109 |
|
| 110 |
+
# 🔹 Validate module
|
| 111 |
if field["module"] != module:
|
| 112 |
raise ValueError(
|
| 113 |
f"Field '{field_name}' does not belong to module '{module}'"
|
| 114 |
)
|
| 115 |
|
| 116 |
+
# 🔹 Validate mapping
|
| 117 |
if "table" not in field or "column" not in field:
|
| 118 |
raise ValueError(
|
| 119 |
f"Field '{field_name}' is missing table/column mapping"
|
|
|
|
| 122 |
return field
|
| 123 |
|
| 124 |
|
| 125 |
+
|
| 126 |
def build_join_sql(base_table, steps):
|
| 127 |
sql = []
|
| 128 |
prev_alias = base_table # alias == table name
|
|
|
|
| 143 |
# =========================
|
| 144 |
|
| 145 |
def parse_intent(question):
|
| 146 |
+
# 1️⃣ Load metadata
|
| 147 |
+
meta = load_metadata()
|
| 148 |
+
|
| 149 |
+
# 2️⃣ Build schema description for the LLM
|
| 150 |
+
schema_description = "\n".join([
|
| 151 |
+
f"{module}: {', '.join(meta['modules'][module]['fields'])}"
|
| 152 |
+
for module in meta["modules"]
|
| 153 |
+
])
|
| 154 |
+
|
| 155 |
+
# 3️⃣ Prompt with schema awareness
|
| 156 |
prompt = f"""
|
| 157 |
+
You are a SQL query planner.
|
| 158 |
+
|
| 159 |
+
You MUST only use fields listed below.
|
| 160 |
+
If a field does not exist, choose the closest valid one.
|
| 161 |
+
|
| 162 |
+
Available schema:
|
| 163 |
+
{schema_description}
|
| 164 |
|
| 165 |
Extract:
|
| 166 |
- module
|
|
|
|
| 173 |
{{
|
| 174 |
"module": "employees",
|
| 175 |
"filters": [
|
| 176 |
+
{{ "field": "start_date", "operator": "greater_or_equal", "value": "2024-01-01" }}
|
| 177 |
]
|
| 178 |
}}
|
| 179 |
|
|
|
|
| 189 |
|
| 190 |
return json.loads(res.choices[0].message.content)
|
| 191 |
|
| 192 |
+
|
| 193 |
# =========================
|
| 194 |
# SQL GENERATOR
|
| 195 |
# =========================
|