datagrunt 3.2.2__tar.gz → 3.2.4__tar.gz

{datagrunt-3.2.2/src/datagrunt.egg-info → datagrunt-3.2.4}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: datagrunt
-Version: 3.2.2
+Version: 3.2.4
 Summary: Read CSV files and convert to other file formats easily
 Author-email: Martin Graham <datagrunt@datagrunt.io>
 License: MIT License

{datagrunt-3.2.2 → datagrunt-3.2.4}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "datagrunt"
-version = "3.2.2"
+version = "3.2.4"
 description = "Read CSV files and convert to other file formats easily"
 readme = "README.md"
 authors = [{ name = "Martin Graham", email = "datagrunt@datagrunt.io" }]
@@ -56,7 +56,7 @@ include = ["datagrunt*"]
 exclude = ["tests*"]
 
 [tool.bumpver]
-current_version = "3.2.2"
+current_version = "3.2.4"
 version_pattern = "MAJOR.MINOR.PATCH"
 commit_message = "build: bump version {old_version} -> {new_version} [skip ci]"
 commit = true

{datagrunt-3.2.2 → datagrunt-3.2.4}/src/datagrunt/__init__.py

@@ -23,7 +23,7 @@ Attributes:
     __license__: The license under which the package is released.
 """
 
-__version__ = "3.2.2"
+__version__ = "3.2.4"
 __author__ = "Martin Graham"
 __license__ = "MIT"
 

{datagrunt-3.2.2 → datagrunt-3.2.4}/src/datagrunt/core/databases/databases.py

@@ -174,20 +174,27 @@ class DuckDBQueries:
     def _set_database_table_name(self):
         """Return a unique, deterministic table name for this file.
 
-        The name combines the sanitized file stem with a short hash of the
-        file's absolute path. The hash is deterministic, so every
-        DuckDBQueries instance created for the same file agrees on the same
-        table name (callers can therefore still reference ``db_table`` in
-        their queries). At the same time, two different files that happen to
-        share a stem (e.g. ``dirA/data.csv`` and ``dirB/data.csv``) resolve
-        to distinct tables and cannot overwrite each other.
+        The name combines a constant ``tbl_`` prefix with the sanitized file
+        stem and a short hash of the file's absolute path. The hash is
+        deterministic, so every DuckDBQueries instance created for the same
+        file agrees on the same table name (callers can therefore still
+        reference ``db_table`` in their queries). At the same time, two
+        different files that happen to share a stem (e.g. ``dirA/data.csv``
+        and ``dirB/data.csv``) resolve to distinct tables and cannot overwrite
+        each other.
+
+        The ``tbl_`` prefix guarantees the name starts with a letter. The table
+        name is interpolated unquoted into every query, and DuckDB rejects an
+        unquoted identifier that begins with a digit, so a numeric-leading
+        filename (e.g. ``2026_sales.csv``) would otherwise produce a
+        ParserException (issue #149).
 
         Returns:
             str: The unique table name.
         """
         stem = self._format_filename_string()
         path_hash = hashlib.sha1(str(self.filepath.resolve()).encode()).hexdigest()[:8]  # noqa: E501
-        return f"{stem}_{path_hash}"
+        return f"tbl_{stem}_{path_hash}"
 
     def set_export_filename(self, default_filename, export_filename=None):
         """

{datagrunt-3.2.2 → datagrunt-3.2.4}/src/datagrunt/core/file_io/fileproperties.py

@@ -1,6 +1,7 @@
 """Module for deriving and evaluating file properties."""
 
 # standard library
+import codecs
 from functools import cached_property
 from pathlib import Path
 
@@ -119,6 +120,7 @@ class BlankFile:
     """Class for checking if a file is blank."""
 
     FILE_SIZE_MB_FACTOR = 10.0
+    READ_CHUNK_SIZE = 65_536  # 64 KiB bounded reads; never the whole file at once
 
     def __init__(self, filepath):
         """Initialize the BlankFile object.
@@ -130,18 +132,44 @@ class BlankFile:
 
     @cached_property
     def is_blank(self):
-        """Check if the file is blank. Blank files contain only whitespace."""
-
+        """Check if the file is blank. Blank files contain only whitespace.
+
+        The file is read in bounded binary chunks and decoded strictly. Any
+        byte sequence that is not valid text - a binary PDF/Parquet/Excel body
+        or a run of invalid-UTF-8 bytes - counts as content, instead of being
+        silently dropped by ``errors="ignore"`` and misreported as blank
+        (issue #146). Only a file that decodes cleanly and holds nothing but
+        whitespace is blank.
+        """
         filestats = FileStatistics(self.filepath)
 
         # Very low probability of being blank if file is 10MB or larger in size
         if filestats.size_in_mb >= self.FILE_SIZE_MB_FACTOR:
             return False
-        with open(self.filepath, "r", encoding=DEFAULT_ENCODING, errors="ignore") as f:
-            for line in f:
-                if line.strip():
-                    return False
-        return True
+        return not self._has_content()
+
+    def _has_content(self):
+        """Return True if the file holds binary bytes or non-whitespace text.
+
+        Decoding is strict so undecodable bytes are treated as content rather
+        than dropped; an incremental decoder keeps multibyte characters that
+        straddle a chunk boundary intact.
+        """
+        decoder = codecs.getincrementaldecoder(DEFAULT_ENCODING)(errors="strict")
+        with open(self.filepath, "rb") as f:
+            while chunk := f.read(self.READ_CHUNK_SIZE):
+                try:
+                    text = decoder.decode(chunk)
+                except UnicodeDecodeError:
+                    return True  # undecodable bytes => binary content
+                if text.strip():
+                    return True
+        # Flush any buffered partial character; a multibyte sequence truncated
+        # at EOF is not clean text, so treat it as content.
+        try:
+            return bool(decoder.decode(b"", final=True).strip())
+        except UnicodeDecodeError:
+            return True
 
 
 class EmptyFile:

{datagrunt-3.2.2 → datagrunt-3.2.4/src/datagrunt.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: datagrunt
-Version: 3.2.2
+Version: 3.2.4
 Summary: Read CSV files and convert to other file formats easily
 Author-email: Martin Graham <datagrunt@datagrunt.io>
 License: MIT License