fix: Don't error on piped data 1db7277 #779 https://til.simonwillison.net/python/io-bufferedreader

Author: jpmckinney

.github/workflows/ci.yml

@@ -31,6 +31,10 @@ jobs:
           PYTHONIOENCODING: utf-8
           PYTHONUTF8: 1
         run: pytest --cov agate
+      - name: Read from stdin
+        run: python -c 'import sys; import agate; agate.Table.from_csv(sys.stdin, sniff_limit=1)' < examples/test.csv
+      - name: Read from pipe
+        run: printf 'a,b,c\n1,2,3' | python -c 'import sys; import agate; agate.Table.from_csv(sys.stdin, sniff_limit=1)'
       - run: python charts.py
       - env:
           GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}

CHANGELOG.rst

@@ -1,3 +1,8 @@
+Unreleased
+----------
+
+-  fix: Version 1.10.0 errors on piped data.
+
 1.10.1 - April 28, 2024
 -----------------------
 

agate/table/from_csv.py

@@ -1,5 +1,6 @@
+import io
 import itertools
-from io import StringIO
+import sys
 
 
 @classmethod
@@ -63,14 +64,20 @@ def from_csv(cls, path, column_names=None, column_types=None, row_names=None, sk
 
         if sniff_limit is None:
             # Reads to the end of the tile, but avoid reading the file twice.
-            handle = StringIO(f.read())
-            kwargs['dialect'] = csv.Sniffer().sniff(handle.getvalue())
+            handle = io.StringIO(f.read())
+            sample = handle.getvalue()
         elif sniff_limit > 0:
-            offset = f.tell()
+            if f == sys.stdin:
+                buffered = io.BufferedReader(sys.stdin.buffer)
+                handle = io.TextIOWrapper(buffered, encoding=encoding)
+                sample = buffered.peek(sniff_limit).decode(encoding)  # reads *bytes*
+            else:
+                offset = f.tell()
+                sample = f.read(sniff_limit)  # reads *characters*
+                f.seek(offset)  # can't do f.seek(-sniff_limit, os.SEEK_CUR) unless file was opened in binary mode
 
-            # Reads only the start of the file.
-            kwargs['dialect'] = csv.Sniffer().sniff(f.read(sniff_limit))
-            f.seek(offset)
+        if sniff_limit is None or sniff_limit > 0:
+            kwargs['dialect'] = csv.Sniffer().sniff(sample)
 
         reader = csv.reader(handle, header=header, **kwargs)