raise human-friendly Sparv errors when parsing of odt or docx fails

spraakbanken · Oct 3, 2023 · ccd06ef · ccd06ef
1 parent 1ceacf0
commit ccd06ef
Show file tree

Hide file tree

Showing 2 changed files with 11 additions and 4 deletions.
diff --git a/sparv/modules/docx_import/docx_import.py b/sparv/modules/docx_import/docx_import.py
@@ -6,7 +6,7 @@
 from docx2python import docx2python
 from docx2python.iterators import iter_at_depth
 
-from sparv.api import Config, SourceFilename, Output, Source, SourceStructure, Text, importer, util
+from sparv.api import Config, SourceFilename, Output, Source, SourceStructure, Text, importer, util, SparvErrorMessage
 
 
 @importer("docx import", file_extension="docx", outputs=["text"], text_annotation="text", config=[
@@ -41,7 +41,10 @@ def parse(source_file: SourceFilename = SourceFilename(),
             'NFC' is used by default.
     """
     source_file_path = source_dir.get_path(source_file, ".docx")
-    d = docx2python(source_file_path)
+    try:
+        d = docx2python(source_file_path)
+    except Exception as e:
+        raise SparvErrorMessage(f"Failed to parse docx file '{source_file}'. {type(e).__name__}: {e}")
 
     # Extract all text from the body, ignoring headers and footers
     text = "\n\n".join(iter_at_depth(d.body, 4))

diff --git a/sparv/modules/odt_import/odt_import.py b/sparv/modules/odt_import/odt_import.py
@@ -5,7 +5,8 @@
 import zipfile
 from typing import Optional
 
-from sparv.api import Config, SourceFilename, Output, Source, SourceStructure, Text, get_logger, importer, util
+from sparv.api import (Config, Output, Source, SourceFilename, SourceStructure, SparvErrorMessage, Text, get_logger,
+                       importer, util)
 
 logger = get_logger(__name__)
 
@@ -44,7 +45,10 @@ def parse(source_file: SourceFilename = SourceFilename(),
     source_file_path = str(source_dir.get_path(source_file, ".odt"))
 
     # Parse odt and extract all text content
-    text = OdtParser(source_file_path).text
+    try:
+        text = OdtParser(source_file_path).text
+    except Exception as e:
+        raise SparvErrorMessage(f"Failed to parse odt file '{source_file}'. {type(e).__name__}: {e}")
 
     if not keep_control_chars:
         text = util.misc.remove_control_characters(text)