From c9aa59d87c654d72267c0295cf46faed683e0446 Mon Sep 17 00:00:00 2001
From: Mathias Kuhring <mathias.kuhring@bihealth.de>
Date: Mon, 3 May 2021 11:02:40 +0200
Subject: [PATCH 1/2] Add exception for invalid encodings

---
 altamisa/isatab/parse_assay_study.py   | 8 ++++++++
 altamisa/isatab/parse_investigation.py | 4 ++++
 2 files changed, 12 insertions(+)

diff --git a/altamisa/isatab/parse_assay_study.py b/altamisa/isatab/parse_assay_study.py
index bae41b8..e777375 100644
--- a/altamisa/isatab/parse_assay_study.py
+++ b/altamisa/isatab/parse_assay_study.py
@@ -768,6 +768,10 @@ def _read_next_line(self):
                 self.unique_rows.add("\t".join(self._line))
         except StopIteration:
             self._line = None
+        except UnicodeDecodeError as e:  # pragma: no cover
+            tpl = "Invalid encoding after line {} of study file '{}' (use Unicode/UTF-8)."
+            msg = tpl.format(self._reader.line_num, self.filename)
+            raise ParseIsatabException(msg) from e
         return prev_line
 
     def read(self):
@@ -888,6 +892,10 @@ def _read_next_line(self):
                 self.unique_rows.add("\t".join(self._line))
         except StopIteration:
             self._line = None
+        except UnicodeDecodeError as e:  # pragma: no cover
+            tpl = "Invalid encoding after line {} of assay file '{}' (use Unicode/UTF-8)."
+            msg = tpl.format(self._reader.line_num, self.filename)
+            raise ParseIsatabException(msg) from e
         return prev_line
 
     def read(self):
diff --git a/altamisa/isatab/parse_investigation.py b/altamisa/isatab/parse_investigation.py
index 8ddba3f..3d9a8ab 100644
--- a/altamisa/isatab/parse_investigation.py
+++ b/altamisa/isatab/parse_investigation.py
@@ -132,6 +132,10 @@ def _read_next_line(self):
                 self._line = list_strip(next(self._reader))
         except StopIteration:
             self._line = None
+        except UnicodeDecodeError as e:  # pragma: no cover
+            tpl = "Invalid encoding after line {} of investigation file '{}' (use Unicode/UTF-8)."
+            msg = tpl.format(self._reader.line_num, self._filename)
+            raise ParseIsatabException(msg) from e
         return prev_line
 
     def _next_line_startswith_comment(self):

From 811d254a440bfceb2b7101b6d06a366379fc3bc1 Mon Sep 17 00:00:00 2001
From: Thomas Sell <thomas.sell@charite.de>
Date: Thu, 2 Nov 2023 11:56:45 +0100
Subject: [PATCH 2/2] do not include line no in the error msg

- it doesn't work
- 'invalid encoding after line x' makes no sense
- line no with invalid chars is already reported by the csv reader exception
---
 altamisa/isatab/parse_assay_study.py   | 6 ++----
 altamisa/isatab/parse_investigation.py | 3 +--
 2 files changed, 3 insertions(+), 6 deletions(-)

diff --git a/altamisa/isatab/parse_assay_study.py b/altamisa/isatab/parse_assay_study.py
index 0a9c446..a49ac97 100644
--- a/altamisa/isatab/parse_assay_study.py
+++ b/altamisa/isatab/parse_assay_study.py
@@ -769,8 +769,7 @@ def _read_next_line(self):
         except StopIteration:
             self._line = None
         except UnicodeDecodeError as e:  # pragma: no cover
-            tpl = "Invalid encoding after line {} of study file '{}' (use Unicode/UTF-8)."
-            msg = tpl.format(self._reader.line_num, self.filename)
+            msg = f"Invalid encoding of study file '{self._filename}' (use Unicode/UTF-8)."
             raise ParseIsatabException(msg) from e
         return prev_line
 
@@ -893,8 +892,7 @@ def _read_next_line(self):
         except StopIteration:
             self._line = None
         except UnicodeDecodeError as e:  # pragma: no cover
-            tpl = "Invalid encoding after line {} of assay file '{}' (use Unicode/UTF-8)."
-            msg = tpl.format(self._reader.line_num, self.filename)
+            msg = f"Invalid encoding of assay file '{self._filename}' (use Unicode/UTF-8)."
             raise ParseIsatabException(msg) from e
         return prev_line
 
diff --git a/altamisa/isatab/parse_investigation.py b/altamisa/isatab/parse_investigation.py
index 2db728e..a365cbf 100644
--- a/altamisa/isatab/parse_investigation.py
+++ b/altamisa/isatab/parse_investigation.py
@@ -133,8 +133,7 @@ def _read_next_line(self):
         except StopIteration:
             self._line = None
         except UnicodeDecodeError as e:  # pragma: no cover
-            tpl = "Invalid encoding after line {} of investigation file '{}' (use Unicode/UTF-8)."
-            msg = tpl.format(self._reader.line_num, self._filename)
+            msg = f"Invalid encoding of investigation file '{self._filename}' (use Unicode/UTF-8)."
             raise ParseIsatabException(msg) from e
         return prev_line