Fixed some banana seeds edge case & error catching

bananalyzer/__main__.py

@@ -8,6 +8,7 @@
 from pathlib import Path
 from typing import List
 from urllib.parse import urlparse
+from botocore.exceptions import ClientError
 
 from bananalyzer import AgentRunner
 from bananalyzer.data.banana_seeds import download_mhtml
@@ -322,14 +323,22 @@ def main() -> int:
         print("=======================================================================")
         return 0
 
-    for example in examples:
+    missing_mhtml = []
+    for i, example in enumerate(examples):
         if example.mhtml_url is not None:
             mhtml_path = get_examples_path() / example.id / "index.mhtml"
             if not mhtml_path.exists():
-                mhtml_str = download_mhtml(example.mhtml_url)
-                mhtml_path.parent.mkdir(parents=True, exist_ok=False)
-                with open(mhtml_path, "w") as file:
-                    file.write(mhtml_str)
+                try:
+                    mhtml_str = download_mhtml(example.mhtml_url)
+                except ClientError:
+                    missing_mhtml.append(i)
+                    continue
+                else:
+                    mhtml_path.parent.mkdir(parents=True, exist_ok=False)
+                    with open(mhtml_path, "w") as file:
+                        file.write(mhtml_str)
+
+    examples = [e for i, e in enumerate(examples) if i not in missing_mhtml]
 
     # Load the desired tests
     generator = PytestTestGenerator()

bananalyzer/data/banana_seeds.py

@@ -20,12 +20,13 @@ def download_examples_from_s3(examples_bucket: str) -> List[Dict[str, Any]]:
             if example["fetch_id"] == "":
                 del example["fetch_id"]
 
-            for row in example["evals"][0]["expected"]:
-                row = {k: v for k, v in row.items() if not k.startswith("__")}
-                if "context" in row:
-                    for key, value in row["context"].items():
-                        row[key] = value
-                    del row["context"]
+            if isinstance(example["evals"][0]["expected"], list):
+                for row in example["evals"][0]["expected"]:
+                    row = {k: v for k, v in row.items() if not k.startswith("__")}
+                    if "context" in row:
+                        for key, value in row["context"].items():
+                            row[key] = value
+                        del row["context"]
 
             examples.append(example)
 

bananalyzer/data/fetch_schemas.py

@@ -38,22 +38,37 @@ class File(BaseModel):
 class GovernmentContractSchema(BaseModel):
     id: str = Field(description="Unique identifier for the contract")
     title: str = Field(description="Title of the contract")
-    description: Optional[str] = Field(default=None,
-                                       description="Description or synopsis field. Combine the solicitation summary and additional instructuions section / process.")
-    location: Optional[str] = Field(default=None,
-                                    description="Location of the issuer. May be a combination of city and state")
-    type: Optional[str] = Field(default=None,
-                                description="Type of contract. May be placed under `Solicitation Type`, `Opportunity Type`, `Market Type`, etc")
-    category: Optional[str] = Field(default=None, description="Category the contract falls under if given")
+    description: Optional[str] = Field(
+        default=None,
+        description="Description or synopsis field. Combine the solicitation summary and additional instructuions section / process.",
+    )
+    location: Optional[str] = Field(
+        default=None,
+        description="Location of the issuer. May be a combination of city and state",
+    )
+    type: Optional[str] = Field(
+        default=None,
+        description="Type of contract. May be placed under `Solicitation Type`, `Opportunity Type`, `Market Type`, etc",
+    )
+    category: Optional[str] = Field(
+        default=None, description="Category the contract falls under if given"
+    )
 
     posted_date: Optional[datetime] = Field(default=None)
     due_date: Optional[datetime] = Field(default=None)
 
-    buyer_name: str = Field(description="Name of the company, organization, or agency that issued the contract")
+    buyer_name: str = Field(
+        description="Name of the company, organization, or agency that issued the contract"
+    )
     buyer_contact_name: str = Field(
-        description="Name of the specific individual that is championing the contract, if available")
-    buyer_contact_number: Optional[str] = Field(default=None, description="Contact number of the issuer")
-    buyer_contact_email: Optional[str] = Field(default=None, description="Contact email of the issuer")
+        description="Name of the specific individual that is championing the contract, if available"
+    )
+    buyer_contact_number: Optional[str] = Field(
+        default=None, description="Contact number of the issuer"
+    )
+    buyer_contact_email: Optional[str] = Field(
+        default=None, description="Contact email of the issuer"
+    )
 
     attachments: List[File] = Field(
         default_factory=list,