fix: address max-effort review findings

- clamp LLM-derived page indices in _get_text_of_pages and
  get_text_of_pdf_pages_with_labels; dedupe get_text_of_pdf_pages
- guard _normalize_tree and folders/documents iterations against
  explicit nulls in cloud API responses
- coerce cloud OCR page numbers to int before filtering in
  get_page_content
- folder cache: raise on missing folder id instead of caching None;
  stop caching name-not-found so later lookups can succeed
- defang doc_id in agent doc-context prompt
- append api-key hint to 401 errors (request, legacy and streaming
  paths)
- remove stale legacy JSON workspace sample data unreadable by the
  SQLite storage

examples/workspace/_meta.json

@@ -1,9 +0,0 @@
-{
-  "12345678-abcd-4321-abcd-123456789abc": {
-    "type": "pdf",
-    "doc_name": "attention-residuals.pdf",
-    "doc_description": "This document introduces \"Attention Residuals\" (AttnRes) and its scalable variant \"Block AttnRes,\" novel mechanisms for replacing fixed residual accumulation in neural networks with learned, input-dependent depth-wise attention, addressing limitations of standard residual connections while optimizing memory, computation, and scalability for large-scale training and inference.",
-    "page_count": 21,
-    "path": "../documents/attention-residuals.pdf"
-  }
-}

pageindex/agent.py

@@ -65,7 +65,7 @@ def wrap_with_doc_context(docs: list[dict], question: str) -> str:
     """
     lines = []
     for d in docs:
-        line = f"- {d['doc_id']}: {_defang_delimiters(d.get('doc_name', ''))}"
+        line = f"- {_defang_delimiters(str(d['doc_id']))}: {_defang_delimiters(d.get('doc_name', ''))}"
         desc = d.get("doc_description") or ""
         if desc:
             line += f" — {_defang_delimiters(desc)}"

pageindex/backend/cloud.py

@@ -14,7 +14,7 @@ import requests
 from typing import AsyncIterator
 
 from ..cloud_api import API_BASE  # single source of truth for the cloud base URL
-from ..errors import CloudAPIError, DocumentNotFoundError, PageIndexError
+from ..errors import AUTH_HINT, CloudAPIError, DocumentNotFoundError, PageIndexError
 from ..events import QueryEvent
 
 logger = logging.getLogger(__name__)
@@ -22,6 +22,13 @@ logger = logging.getLogger(__name__)
 _INTERNAL_TOOLS = frozenset({"ToolSearch", "Read", "Grep", "Glob", "Bash", "Edit", "Write"})
 
 
+def _as_int(value):
+    try:
+        return int(value)
+    except (TypeError, ValueError):
+        return None
+
+
 class CloudBackend:
     def __init__(self, api_key: str):
         self._api_key = api_key
@@ -75,8 +82,10 @@ class CloudBackend:
                     continue
                 if resp.status_code != 200:
                     body = resp.text[:500] if resp.text else ""
-                    raise CloudAPIError(f"Cloud API error {resp.status_code}: {body}",
+                    msg = f"Cloud API error {resp.status_code}: {body}"
-                                        status_code=resp.status_code)
+                    if resp.status_code == 401:
+                        msg += f" — {AUTH_HINT}"
+                    raise CloudAPIError(msg, status_code=resp.status_code)
                 return resp.json() if resp.content else {}
             except requests.RequestException as e:
                 if attempt == retries - 1:
@@ -102,11 +111,21 @@ class CloudBackend:
 
     # ── Collection management (mapped to folders) ─────────────────────────
 
+    def _create_folder(self, name: str) -> str:
+        """POST /folder/ and return the new folder id, never a falsy value."""
+        resp = self._request("POST", "/folder/", json={"name": name})
+        folder_id = resp.get("folder", {}).get("id")
+        if not folder_id:
+            raise PageIndexError(
+                f"Cloud API returned no folder id when creating {name!r} "
+                f"(response keys: {list(resp)})"
+            )
+        return folder_id
+
     def create_collection(self, name: str) -> None:
         self._validate_collection_name(name)
         try:
-            resp = self._request("POST", "/folder/", json={"name": name})
+            self._folder_id_cache[name] = self._create_folder(name)
-            self._folder_id_cache[name] = resp.get("folder", {}).get("id")
         except CloudAPIError as e:
             if e.status_code in self._FOLDER_UNAVAILABLE:
                 self._warn_folder_upgrade()
@@ -118,12 +137,11 @@ class CloudBackend:
         self._validate_collection_name(name)
         try:
             data = self._request("GET", "/folders/")
-            for folder in data.get("folders", []):
+            for folder in data.get("folders", []) or []:
                 if folder.get("name") == name:
                     self._folder_id_cache[name] = folder["id"]
                     return
-            resp = self._request("POST", "/folder/", json={"name": name})
+            self._folder_id_cache[name] = self._create_folder(name)
-            self._folder_id_cache[name] = resp.get("folder", {}).get("id")
         except CloudAPIError as e:
             if e.status_code in self._FOLDER_UNAVAILABLE:
                 self._warn_folder_upgrade()
@@ -148,16 +166,15 @@ class CloudBackend:
                 self._folder_id_cache[name] = None
                 return None
             raise
-        for folder in data.get("folders", []):
+        for folder in data.get("folders", []) or []:
             if folder.get("name") == name:
                 self._folder_id_cache[name] = folder["id"]
                 return folder["id"]
-        self._folder_id_cache[name] = None
         return None
 
     def list_collections(self) -> list[str]:
         data = self._request("GET", "/folders/")
-        return [f["name"] for f in data.get("folders", [])]
+        return [f["name"] for f in data.get("folders", []) or []]
 
     def delete_collection(self, name: str) -> None:
         folder_id = self._get_folder_id(name)
@@ -243,23 +260,29 @@ class CloudBackend:
         from ..index.utils import parse_pages
         page_nums = set(parse_pages(pages))
         all_pages = resp.get("pages", resp.get("ocr", resp.get("result", [])))
-        if isinstance(all_pages, list):
+        if not isinstance(all_pages, list):
-            return [
+            return []
-                {"page": p.get("page", p.get("page_index")),
+        result = []
-                 "content": p.get("content", p.get("markdown", "")),
+        for p in all_pages:
-                 # Cloud OCR pages carry an `images` list (empty on text-only
+            page = _as_int(p.get("page", p.get("page_index")))
-                 # pages). Preserve it — omitting when empty, mirroring the local
+            if page not in page_nums:
-                 # backend — so cloud callers get the same PageContent shape and
+                continue
-                 # the SDK-prompted UI can render figures.
+            entry = {"page": page,
-                 **({"images": p["images"]} if p.get("images") else {})}
+                     "content": p.get("content", p.get("markdown", ""))}
-                for p in all_pages
+            # Cloud OCR pages carry an `images` list (empty on text-only
-                if p.get("page", p.get("page_index")) in page_nums
+            # pages). Preserve it — omitting when empty, mirroring the local
-            ]
+            # backend — so cloud callers get the same PageContent shape and
-        return []
+            # the SDK-prompted UI can render figures.
+            if p.get("images"):
+                entry["images"] = p["images"]
+            result.append(entry)
+        return result
 
     @staticmethod
-    def _normalize_tree(nodes: list) -> list:
+    def _normalize_tree(nodes: list | None) -> list:
         """Normalize cloud tree nodes to match local schema."""
+        if not nodes:
+            return []
         result = []
         for node in nodes:
             normalized = {
@@ -290,7 +313,7 @@ class CloudBackend:
             if folder_id:
                 params["folder_id"] = folder_id
             data = self._request("GET", "/docs/", params=params)
-            batch = data.get("documents", [])
+            batch = data.get("documents", []) or []
             docs.extend(
                 {
                     "doc_id": d.get("id", ""),
@@ -405,10 +428,10 @@ class CloudBackend:
                     return
                 if resp.status_code != 200:
                     body = resp.text[:500] if resp.text else ""
-                    raise CloudAPIError(
+                    msg = f"Cloud streaming error {resp.status_code}: {body}"
-                        f"Cloud streaming error {resp.status_code}: {body}",
+                    if resp.status_code == 401:
-                        status_code=resp.status_code,
+                        msg += f" — {AUTH_HINT}"
-                    )
+                    raise CloudAPIError(msg, status_code=resp.status_code)
 
                 current_tool_name = None
                 current_tool_args: list[str] = []

pageindex/cloud_api.py

@@ -6,7 +6,7 @@ from typing import Any, Iterator
 
 import requests
 
-from .errors import PageIndexAPIError
+from .errors import AUTH_HINT, PageIndexAPIError
 
 # Single source of truth for the cloud API base URL — imported by the modern
 # CloudBackend (as API_BASE) and PageIndexClient so a staging/migration change
@@ -47,7 +47,10 @@ class LegacyCloudAPI:
             raise PageIndexAPIError(f"{error_prefix}: {e}") from e
 
         if response.status_code != 200:
-            raise PageIndexAPIError(f"{error_prefix}: {response.text}")
+            msg = f"{error_prefix}: {response.text}"
+            if response.status_code == 401:
+                msg += f" — {AUTH_HINT}"
+            raise PageIndexAPIError(msg)
         return response
 
     def submit_document(

pageindex/errors.py

@@ -55,3 +55,10 @@ class FileTypeError(PageIndexError, ValueError):
     instead of) a bare ValueError one.
     """
     pass
+
+
+AUTH_HINT = (
+    "api_key must be a PageIndex cloud API key (https://dash.pageindex.ai/api-keys). "
+    "For local mode, omit api_key and set your LLM provider key "
+    "(e.g. OPENAI_API_KEY) in the environment."
+)

pageindex/index/utils.py

@@ -521,9 +521,9 @@ def create_clean_structure_for_description(structure):
 
 
 def _get_text_of_pages(page_list, start_page, end_page):
-    """Concatenate text from page_list for pages [start_page, end_page] (1-indexed)."""
+    """Concatenate text from page_list for pages [start_page, end_page] (1-indexed), clamped to the valid page range."""
     text = ""
-    for page_num in range(start_page - 1, end_page):
+    for page_num in range(max(start_page, 1) - 1, min(end_page, len(page_list))):
         text += page_list[page_num][0]
     return text
 
@@ -831,15 +831,12 @@ def get_page_tokens(pdf_path, model=None, pdf_parser="PyPDF2"):
 
 
 def get_text_of_pdf_pages(pdf_pages, start_page, end_page):
-    text = ""
+    return _get_text_of_pages(pdf_pages, start_page, end_page)
-    for page_num in range(start_page-1, end_page):
-        text += pdf_pages[page_num][0]
-    return text
 
 
 def get_text_of_pdf_pages_with_labels(pdf_pages, start_page, end_page):
     text = ""
-    for page_num in range(start_page-1, end_page):
+    for page_num in range(max(start_page, 1) - 1, min(end_page, len(pdf_pages))):
         text += f"<physical_index_{page_num+1}>\n{pdf_pages[page_num][0]}\n<physical_index_{page_num+1}>\n"
     return text