unsupported file format exception

5 files changed, 18 insertions, 3 deletions

diff --git a/bsie/lib/pipeline.py b/bsie/lib/pipeline.py
index e5ce1b7..02119bc 100644
--- a/bsie/lib/pipeline.py
+++ b/bsie/lib/pipeline.py
@@ -126,6 +126,8 @@ class Pipeline():
                 # get content
                 content = rdr(path) if rdr is not None else None
 
+                #logger.info('extracted %s from %s', rdr, path)
+
                 # apply extractors on this content
                 for ext in extrs:
                     try:
@@ -137,6 +139,11 @@ class Pipeline():
                         # critical extractor failure.
                         logger.error('%s failed to extract triples from content: %s', ext, err)
 
+            except errors.UnsupportedFileFormatError as err:
+                # failed to read the file format. skip.
+                #logger.warning('%s could not process the file format of %s', rdr, err)
+                pass
+
             except errors.ReaderError as err:
                 # failed to read any content. skip.
                 logger.error('%s failed to read content: %s', rdr, err)
diff --git a/bsie/reader/chain.py b/bsie/reader/chain.py
index 8e900e1..db7c2d5 100644
--- a/bsie/reader/chain.py
+++ b/bsie/reader/chain.py
@@ -73,13 +73,16 @@ class ReaderChain(base.Reader, typing.Generic[T_CONTENT]):
         return hash((super().__hash__(), self._children))
 
     def __call__(self, path: str) -> T_CONTENT:
+        raise_error = errors.UnsupportedFileFormatError
         for child in self._children:
             try:
                 return child(path)
+            except errors.UnsupportedFileFormatError:
+                pass
             except errors.ReaderError:
                 # child cannot read the file, skip.
-                pass
+                raise_error = errors.ReaderError # type: ignore [assignment] # mypy is confused
 
-        raise errors.ReaderError(path)
+        raise raise_error(path)
 
 ## EOF ##
diff --git a/bsie/reader/image/_pillow.py b/bsie/reader/image/_pillow.py
index ee0662d..3144509 100644
--- a/bsie/reader/image/_pillow.py
+++ b/bsie/reader/image/_pillow.py
@@ -31,6 +31,8 @@ class PillowImage(base.Reader):
         try:
             # open file with PIL
             return PIL.Image.open(path)
+        except PIL.UnidentifiedImageError as err:
+            raise errors.UnsupportedFileFormatError(path) from err
         except IOError as err:
             raise errors.ReaderError(path) from err
 
diff --git a/bsie/reader/image/_raw.py b/bsie/reader/image/_raw.py
index 77be357..cd60453 100644
--- a/bsie/reader/image/_raw.py
+++ b/bsie/reader/image/_raw.py
@@ -45,7 +45,7 @@ class RawImage(base.Reader):
     def __call__(self, path: str) -> PIL.Image:
         # perform quick checks first
         if not self._match(path):
-            raise errors.ReaderError(path)
+            raise errors.UnsupportedFileFormatError(path)
 
         try:
             # open file with rawpy
diff --git a/bsie/utils/errors.py b/bsie/utils/errors.py
index fbc16f7..8133cd4 100644
--- a/bsie/utils/errors.py
+++ b/bsie/utils/errors.py
@@ -42,4 +42,7 @@ class UnreachableError(ProgrammingError):
 class ParserError(_BSIEError):
     """Failed to parse due to invalid syntax or structures."""
 
+class UnsupportedFileFormatError(ReaderError):
+    """Failed to read a file format."""
+
 ## EOF ##