Merge pull request #62 from datarootsio/add-tryparse-for-html

add exception for html tables and try-except when parsing outputs (fa…

databooks/data_models/cell.py

@@ -11,7 +11,7 @@
 from rich.text import Text
 
 from databooks.data_models.base import DatabooksBase
-from databooks.data_models.rich_helpers import HtmlTable
+from databooks.data_models.rich_helpers import HtmlTable, RichHtmlTableError
 from databooks.logging import get_logger
 
 logger = get_logger(__file__)
@@ -146,9 +146,17 @@ class CellDisplayDataOutput(DatabooksBase):
  @property
  def rich_output(self) -> Sequence[ConsoleRenderable]:
  """Dynamically compute the rich output - also in `CellExecuteResultOutput`."""
+
+ def _try_parse_html(s: str) -> Optional[ConsoleRenderable]:
+ """Try to parse HTML table, return `None` if any errors are raised."""
+ try:
+ return HtmlTable("".join(s)).rich()
+ except RichHtmlTableError:
+ logger.debug("Could not generate rich HTML table.")
+ return None
+
  mime_func: Dict[str, Callable[[str], Optional[ConsoleRenderable]]] = {
- "image/png": lambda s: None,
- "text/html": lambda s: HtmlTable("".join(s)).rich(),
+ "text/html": lambda s: _try_parse_html(s),
  "text/plain": lambda s: Text("".join(s)),
  }
  _rich = {

databooks/data_models/rich_helpers.py

@@ -8,6 +8,16 @@
 HtmlAttr = Tuple[str, Optional[str]]
 
 
+class RichHtmlTableError(Exception):
+ """Could not parse HTML table."""
+
+ def __init__(self, msg: str = "", *args: Any):
+ """Use class docstring as error 'prefix'."""
+ if self.__doc__ is None:
+ raise ValueError("Exception docstring required - used in error message.")
+ super().__init__(" ".join((self.__doc__, msg)), *args)
+
+
 class HtmlTable(HTMLParser):
  """Rich table from HTML string."""
 
@@ -23,13 +33,13 @@ def __init__(self, html: str, *args: Any, **kwargs: Any) -> None:
  def handle_starttag(self, tag: str, attrs: List[HtmlAttr]) -> None:
  """Active tags are indicated via instance boolean properties."""
  if getattr(self, tag, None):
- raise ValueError(f"Already in `{tag}`.")
+ raise RichHtmlTableError(f"Already in `{tag}`.")
  setattr(self, tag, True)
 
  def handle_endtag(self, tag: str) -> None:
  """Write table properties when closing tags."""
  if not getattr(self, tag):
- raise ValueError(f"Cannot end unopened `{tag}`.")
+ raise RichHtmlTableError(f"Cannot end unopened `{tag}`.")
 
  # If we are ending a row, either set a table header or row
  if tag == "tr":
@@ -53,7 +63,7 @@ def rich(self, **tbl_kwargs: Any) -> Optional[Table]:
  _ncols = len(self.rows[0])
  _headers = [""] * (_ncols - len(self.headers)) + self.headers
  if any(len(row) != _ncols for row in self.rows):
- raise ValueError(f"Expected all rows to have {_ncols} columns.")
+ raise RichHtmlTableError(f"Expected all rows to have {_ncols} columns.")
 
  _box = tbl_kwargs.pop("box", box.SIMPLE_HEAVY)
  _row_styles = tbl_kwargs.pop("row_styles", ["on bright_black", ""])