get_prompts_by_dataset

scribe_python_client/client.py

@@ -1268,7 +1268,7 @@ class ScribeClient:
 
         self.get_dataset_data(dataset_links)
 
-        #Load data from the CSV file (generated by get_dataset_data)
+        # Load data from the CSV file (generated by get_dataset_data)
         records = []
         try:
             with open("dataset_metadata.csv", mode="r", encoding="utf-8") as f:
@@ -1279,7 +1279,7 @@ class ScribeClient:
             logging.error(f"Failed to load dataset_metadata.csv: {e}")
             return {}
 
-        #Group column data by table name
+        # Group column data by table name
         table_dict = {}
         for row in records:
             table_name = row.get("table_name")
@@ -1296,6 +1296,33 @@ class ScribeClient:
                 "description": row.get("column_description", "")
             })
 
+        # Add markdown for each table
+        for table_name, cols in table_dict.items():
+            md_parts = []
+            md_parts.append(f"## `{table_name}` Columns\n\n")
+            md_parts.append("| Column Name | Label | Type | Description | Like-compliant |\n")
+            md_parts.append("|-------------|-------|------|-------------|----------------|\n")
+            for col in cols:
+                column_name = col["column_name"]
+                label = col["label"]
+                col_type = col["type"].upper()
+                description = col["description"]
+                expr = row.get("expression", "") or ""
+                if col_type == "STRING" and (" in (" in expr.lower() or " in(" in expr.lower()):
+                    like_compliant = False
+                else:
+                    like_compliant = True
+                md_parts.append(
+                    f"| `{column_name}` | {label} | {col_type} | {description} | {str(like_compliant)} |\n"
+                )
+            md_parts.append("\n")
+            md_str = "".join(md_parts)
+            # Add markdown string to the dict for this table
+            # If table_dict[table_name] is a list, convert to dict with 'columns' and 'md'
+            table_dict[table_name] = {
+                "columns": cols,
+                "md": md_str
+            }
         return table_dict
 
 
@@ -1371,4 +1398,41 @@ class ScribeClient:
             f.write(markdown_output)
 
         logging.info("Saved table metadata Markdown to dataset_tables.md")
-        return markdown_output
+        return markdown_output
+    
+    def get_dataset_prompt(self, dataset, pretable_tex_file="", posttable_text_file=""):
+        """
+        Returns a prompt string for the given dataset, optionally wrapped with pre/post text from files.
+        - dataset: table name to fetch from docs/dataset_tables.json
+        - pretable_tex_file: filename to prepend (optional)
+        - posttable_text_file: filename to append (optional)
+        """
+        import os
+        import json
+        # Load table data from docs/dataset_tables.json
+        table_json_path = os.path.join("docs", "dataset_tables.json")
+        if not os.path.exists(table_json_path):
+            raise FileNotFoundError(f"dataset_tables.json not found at {table_json_path}")
+        with open(table_json_path, "r", encoding="utf-8") as f:
+            tables = json.load(f)
+        if dataset not in tables:
+            raise ValueError(f"Dataset '{dataset}' not found in dataset_tables.json")
+        table_md = tables[dataset].get("md")
+        if not table_md:
+            raise ValueError(f"No markdown found for dataset '{dataset}' in dataset_tables.json")
+        # Read pre/post text if provided
+        pre_text = ""
+        post_text = ""
+        if pretable_tex_file:
+            if not os.path.exists(pretable_tex_file):
+                raise FileNotFoundError(f"Pre-table file '{pretable_tex_file}' not found.")
+            with open(pretable_tex_file, "r", encoding="utf-8") as f:
+                pre_text = f.read()
+        if posttable_text_file:
+            if not os.path.exists(posttable_text_file):
+                raise FileNotFoundError(f"Post-table file '{posttable_text_file}' not found.")
+            with open(posttable_text_file, "r", encoding="utf-8") as f:
+                post_text = f.read()
+        # Build prompt
+        prompt = f"{pre_text}{table_md}{post_text}"
+        return prompt