ContentPDF now retains url information (if any) and translates to Inspect

chatlas/_content.py

@@ -665,12 +665,17 @@ class ContentPDF(Content):
 
     Parameters
     ----------
-    value
+    data
         The PDF data extracted
+    filename
+        The name of the PDF file
+    url
+        An optional URL where the PDF can be accessed
     """
 
     data: bytes
     filename: str
+    url: Optional[str] = None
 
     content_type: ContentTypeEnum = "pdf"
 

chatlas/_content_pdf.py

@@ -77,7 +77,7 @@ def content_pdf_url(url: str) -> ContentPDF:
     # apis where they exist. Might need some kind of mutable state so can
     # record point to uploaded file.
     data = download_pdf_bytes(url)
-    return ContentPDF(data=data, filename=unique_pdf_name())
+    return ContentPDF(data=data, filename=unique_pdf_name(), url=url)
 
 
 def parse_data_url(url: str) -> tuple[str, str]:

chatlas/_inspect.py

@@ -172,8 +172,11 @@ def chatlas_content_as_inspect(content: ContentUnion) -> InspectContent:
         data_url = f"data:{content.image_content_type};base64,{content.data or ''}"
         return itool.ContentImage(image=data_url, detail="auto")
     elif isinstance(content, ContentPDF):
+        doc = content.url
+        if doc is None:
+            doc = f"data:application/pdf;base64,{base64.b64encode(content.data).decode('ascii')}"
         return itool.ContentDocument(
-            document=base64.b64encode(content.data).decode("ascii"),
+            document=doc,
             mime_type="application/pdf",
             filename=content.filename,
         )
@@ -211,13 +214,17 @@ def inspect_content_as_chatlas(content: str | InspectContent) -> Content:
                 image_content_type=content_type,  # type: ignore
             )
     if isinstance(content, itool.ContentDocument):
+        doc = content.document
         if content.mime_type == "application/pdf":
-            return ContentPDF(
-                data=base64.b64decode(content.document),
-                filename=content.filename,
-            )
+            url = None
+            if doc.startswith("http://") or doc.startswith("https://"):
+                url = doc
+                data = None
+            else:
+                data = base64.b64decode(doc.split(",", 1)[1])
+            return ContentPDF(data=data, url=url, filename=content.filename)
         else:
-            return ContentText(text=content.document)
+            return ContentText(text=doc)
     if isinstance(content, itool.ContentData):
         return ContentJson(value=content.data)
     raise ValueError(

chatlas/_provider_openai_completions.py

@@ -56,18 +56,22 @@
 
 def ChatOpenAICompletions(
     *,
+    base_url: str = "https://api.openai.com/v1",
     system_prompt: Optional[str] = None,
     model: "Optional[ChatModel | str]" = None,
     api_key: Optional[str] = None,
-    base_url: str = "https://api.openai.com/v1",
     seed: int | None | MISSING_TYPE = MISSING,
     kwargs: Optional["ChatClientArgs"] = None,
 ) -> Chat["SubmitInputArgs", ChatCompletion]:
     """
-    Chat with an OpenAI model via the Completions API.
+    Chat with an OpenAI-compatible model (via the Completions API).
 
     This function exists mainly for historical reasons; new code should
     prefer `ChatOpenAI()`, which uses the newer Responses API.
+
+    This function may also be useful for using an "OpenAI-compatible model"
+    hosted by another provider (e.g., vLLM, Ollama, etc.) that supports the
+    OpenAI Completions API.
     """
     if isinstance(seed, MISSING_TYPE):
         seed = 1014 if is_testing() else None