Add descriptive error for malformed TOC entry (#10574)

* create TocParseError and throw error when toc values don't match their expected type

---------

Co-authored-by: Tomas Rodriguez <[email protected]>
Co-authored-by: Tomas Rodriguez <[email protected]>
Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>
Co-authored-by: Drini Cami <[email protected]>

Author: 5 people

openlibrary/plugins/upstream/addbook.py

@@ -23,6 +23,7 @@
 from openlibrary.plugins.upstream import spamcheck, utils
 from openlibrary.plugins.upstream.account import as_admin
 from openlibrary.plugins.upstream.models import Author, Edition, Work
+from openlibrary.plugins.upstream.table_of_contents import TocParseError
 from openlibrary.plugins.upstream.utils import fuzzy_find, render_template
 from openlibrary.plugins.worksearch.search import get_solr
 
@@ -644,7 +645,12 @@ def save(self, formdata: web.Storage) -> None:
                 edition_data.pop('physical_dimensions', None)
             )
             self.edition.set_weight(edition_data.pop('weight', None))
-            self.edition.set_toc_text(edition_data.pop('table_of_contents', None))
+            try:
+                self.edition.set_toc_text(edition_data.pop('table_of_contents', None))
+            except TocParseError as e:
+                raise ClientException(
+                    "400 Bad Request", f"Table of contents parse error: {e}"
+                )
 
             if edition_data.pop('translation', None) != 'yes':
                 edition_data.translation_of = None

openlibrary/plugins/upstream/table_of_contents.py

@@ -64,6 +64,10 @@ class AuthorRecord(TypedDict, total=False):
     author: ThingReferenceDict | None
 
 
+class TocParseError(BaseException):
+    pass
+
+
 @dataclass
 class TocEntry:
     level: int
@@ -120,6 +124,7 @@ def from_markdown(line: str) -> 'TocEntry':
         >>> f("1.1 | Apple")
         (0, '1.1', 'Apple', None)
         """
+
         RE_LEVEL = web.re_compile(r"(\**)(.*)")
         level, text = RE_LEVEL.match(line.strip()).groups()
 
@@ -130,13 +135,20 @@ def from_markdown(line: str) -> 'TocEntry':
             title = text
             label = page = ""
             extra_fields = ''
+        try:
+            extra_fields = json.loads(extra_fields or '{}')
+        except json.JSONDecodeError as e:
+            raise TocParseError(f"Invalid JSON: {e}")
+
+        if not isinstance(extra_fields, dict):
+            raise TocParseError("Invalid formatting!")
 
         return TocEntry(
             level=len(level),
             label=label.strip() or None,
             title=title.strip() or None,
             pagenum=page.strip() or None,
-            **json.loads(extra_fields or '{}'),
+            **extra_fields,
         )
 
     def to_markdown(self) -> str: