Merge pull request #50 from ollama/mxyng/base64-encode-image

fix: encode base64 inputs

ollama/_client.py

@@ -2,11 +2,12 @@ import os
 import io
 import json
 import httpx
+import binascii
 import urllib.parse
 from os import PathLike
 from pathlib import Path
 from hashlib import sha256
-from base64 import b64encode
+from base64 import b64encode, b64decode
 
 from typing import Any, AnyStr, Union, Optional, Sequence, Mapping, Literal
 
@@ -589,19 +590,43 @@ class AsyncClient(BaseClient):
 
 
 def _encode_image(image) -> str:
-  if p := _as_path(image):
-    b64 = b64encode(p.read_bytes())
-  elif b := _as_bytesio(image):
-    b64 = b64encode(b.read())
-  else:
-    raise RequestError('images must be a list of bytes, path-like objects, or file-like objects')
+  """
+  >>> _encode_image(b'ollama')
+  'b2xsYW1h'
+  >>> _encode_image(io.BytesIO(b'ollama'))
+  'b2xsYW1h'
+  >>> _encode_image('LICENSE')
+  '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'
+  >>> _encode_image(Path('LICENSE'))
+  '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'
+  >>> _encode_image('YWJj')
+  'YWJj'
+  >>> _encode_image(b'YWJj')
+  'YWJj'
+  """
 
-  return b64.decode('utf-8')
+  if p := _as_path(image):
+    return b64encode(p.read_bytes()).decode('utf-8')
+
+  try:
+    b64decode(image, validate=True)
+    return image if isinstance(image, str) else image.decode('utf-8')
+  except (binascii.Error, TypeError):
+    ...
+
+  if b := _as_bytesio(image):
+    return b64encode(b.read()).decode('utf-8')
+
+  raise RequestError('image must be bytes, path-like object, or file-like object')
 
 
 def _as_path(s: Optional[Union[str, PathLike]]) -> Union[Path, None]:
   if isinstance(s, str) or isinstance(s, Path):
-    return Path(s)
+    try:
+      if (p := Path(s)).exists():
+        return p
+    except Exception:
+      ...
   return None