Add checks and unsafe fixes for mojibake

This detects whether text has likely been encoded in one encoding
and decoded in another, perhaps multiple times. This often results
in display of "mojibake" characters.

For example, a file encoded in UTF-8 is opened as CP-1252 (Windows
Latin codepage) in Microsoft Excel, and saved again as UTF-8. You
will see strings like this in the resulting file:

    - CIAT Publicaçao
    - CIAT Publicación

The correct version of these in UTF-8 would be:

    - CIAT Publicaçao
    - CIAT Publicación

I use a code snippet from Martijn Pieters on StackOverflow to de-
tect whether a string is "weird" as determined by the excellent
"fixes text for you" (ftfy) Python library, then check if a weird
string encodes as CP-1252 or not. If so, I can try to fix it.

See: https://stackoverflow.com/questions/29071995/identify-garbage-unicode-string-using-python

csv_metadata_quality/app.py

@@ -107,6 +107,13 @@ def run(argv):
         # Check: suspicious characters
         df[column].apply(check.suspicious_characters, field_name=column)
 
+        # Check: mojibake
+        df[column].apply(check.mojibake, field_name=column)
+
+        # Fix: mojibake
+        if args.unsafe_fixes:
+            df[column] = df[column].apply(fix.mojibake, field_name=column)
+
         # Fix: invalid and unnecessary multi-value separators
         df[column] = df[column].apply(fix.separators, field_name=column)
         # Run whitespace fix again after fixing invalid separators

csv_metadata_quality/check.py

@@ -11,6 +11,8 @@ from pycountry import languages
 from stdnum import isbn as stdnum_isbn
 from stdnum import issn as stdnum_issn
 
+from csv_metadata_quality.util import is_mojibake
+
 
 def issn(field):
     """Check if an ISSN is valid.
@@ -345,3 +347,22 @@ def duplicate_items(df):
                 )
             else:
                 items.append(item_title_type_date)
+
+
+def mojibake(field, field_name):
+    """Check for mojibake (text that was encoded in one encoding and decoded in
+    in another, perhaps multiple times). See util.py.
+
+    Prints the string if it contains suspected mojibake.
+    """
+
+    # Skip fields with missing values
+    if pd.isna(field):
+        return
+
+    if is_mojibake(field):
+        print(
+            f"{Fore.YELLOW}Possible encoding issue ({field_name}): {Fore.RESET}{field}"
+        )
+
+    return

csv_metadata_quality/fix.py

@@ -3,8 +3,9 @@ from unicodedata import normalize
 
 import pandas as pd
 from colorama import Fore
+from ftfy import fix_text
 
-from csv_metadata_quality.util import is_nfc
+from csv_metadata_quality.util import is_mojibake, is_nfc
 
 
 def whitespace(field, field_name):
@@ -253,3 +254,22 @@ def normalize_unicode(field, field_name):
         field = normalize("NFC", field)
 
     return field
+
+
+def mojibake(field, field_name):
+    """Attempts to fix mojibake (text that was encoded in one encoding and deco-
+    ded in another, perhaps multiple times). See util.py.
+
+    Return fixed string.
+    """
+
+    # Skip fields with missing values
+    if pd.isna(field):
+        return field
+
+    if is_mojibake(field):
+        print(f"{Fore.GREEN}Fixing encoding issue ({field_name}): {Fore.RESET}{field}")
+
+        return fix_text(field)
+    else:
+        return field

csv_metadata_quality/util.py

@@ -1,3 +1,6 @@
+from ftfy.badness import sequence_weirdness
+
+
 def is_nfc(field):
     """Utility function to check whether a string is using normalized Unicode.
     Python's built-in unicodedata library has the is_normalized() function, but
@@ -12,3 +15,35 @@ def is_nfc(field):
     from unicodedata import normalize
 
     return field == normalize("NFC", field)
+
+
+def is_mojibake(field):
+    """Determines whether a string contains mojibake.
+
+    We commonly deal with CSV files that were *encoded* in UTF-8, but decoded
+    as something else like CP-1252 (Windows Latin). This manifests in the form
+    of "mojibake", for example:
+
+        - CIAT Publicaçao
+        - CIAT Publicación
+
+    This uses the excellent "fixes text for you" (ftfy) library to determine
+    whether a string contains characters that have been encoded in one encoding
+    and decoded in another.
+
+    Inspired by this code snippet from Martijn Pieters on StackOverflow:
+    https://stackoverflow.com/questions/29071995/identify-garbage-unicode-string-using-python
+
+    Return boolean.
+    """
+    if not sequence_weirdness(field):
+        # Nothing weird, should be okay
+        return False
+    try:
+        field.encode("sloppy-windows-1252")
+    except UnicodeEncodeError:
+        # Not CP-1252 encodable, probably fine
+        return False
+    else:
+        # Encodable as CP-1252, Mojibake alert level high
+        return True

pyproject.toml

@@ -21,6 +21,7 @@ pycountry = "^19.8.18"
 langid = "^1.1.6"
 colorama = "^0.4.4"
 spdx-license-list = "^0.5.2"
+ftfy = "^5.9"
 
 [tool.poetry.dev-dependencies]
 pytest = "^6.1.1"