pyconverters-openai_vision 0.5.32__py3-none-any.whl → 0.5.36__py3-none-any.whl

pyconverters_openai_vision/__init__.py

@@ -1,2 +1,2 @@
 """OpenAIVision converter"""
-__version__ = "0.5.32"
+__version__ = "0.5.36"

pyconverters_openai_vision/openai_vision.py

@@ -147,6 +147,10 @@ class OpenAIVisionConverterBase(ConverterBase):
     PREFIX: str = ""
     oauth_token: OAuthToken = OAuthToken()
 
+    @classmethod
+    def is_model_gpt_5_model(cls, model: str) -> bool:
+        return "gpt-5" in model
+
     def compute_args(self, params: OpenAIVisionBaseParameters, source: UploadFile, kind
                      ) -> Dict[str, Any]:
         data = source.file.read()
@@ -155,7 +159,7 @@ class OpenAIVisionConverterBase(ConverterBase):
             binary_block = {
                 "type": "image_url",
                 "image_url": {
-                    "url": f"data:image/jpeg;base64,{rv.decode('utf-8')}"
+                    "url": f"data:{kind.mime};base64,{rv.decode('utf-8')}"
                 }
             }
         messages = [{"role": "system", "content": params.system_prompt}] if params.system_prompt is not None else []
@@ -167,10 +171,11 @@ class OpenAIVisionConverterBase(ConverterBase):
                              },
                              binary_block
                          ]})
+        max_tokens_param = 'max_completion_tokens' if self.is_model_gpt_5_model(params.model_str) else 'max_tokens'
         kwargs = {
             'model': params.model_str,
             'messages': messages,
-            'max_tokens': params.max_tokens,
+            max_tokens_param: params.max_tokens,
             'temperature': params.temperature,
             'top_p': params.top_p,
             'n': params.n,
@@ -299,12 +304,12 @@ class OpenAIVisionProcessorBaseParameters(ProcessorParameters):
         extra="advanced"
     )
     system_prompt: str = Field(
-        "Generate a textual description of the image",
+        None,
         description="""Contains the system prompt""",
         extra="multiline,advanced",
     )
     prompt: str = Field(
-        None,
+        "Generate a textual description of the image",
         description="""Contains the prompt""",
         extra="multiline",
     )
@@ -357,6 +362,10 @@ class OpenAIVisionProcessorBase(ProcessorBase):
     PREFIX: str = ""
     oauth_token: OAuthToken = OAuthToken()
 
+    @classmethod
+    def is_model_gpt_5_model(cls, model: str) -> bool:
+        return "gpt-5" in model
+
     def compute_args(self, params: OpenAIVisionProcessorBaseParameters, source: str, kind
                      ) -> Dict[str, Any]:
         if kind.mime.startswith("image"):
@@ -375,10 +384,11 @@ class OpenAIVisionProcessorBase(ProcessorBase):
                              },
                              binary_block
                          ]})
+        max_tokens_param = 'max_completion_tokens' if self.is_model_gpt_5_model(params.model_str) else 'max_tokens'
         kwargs = {
             'model': params.model_str,
             'messages': messages,
-            'max_tokens': params.max_tokens,
+            max_tokens_param: params.max_tokens,
             'temperature': params.temperature,
             'top_p': params.top_p,
             'n': params.n,
@@ -446,8 +456,9 @@ class OpenAIVisionProcessorBase(ProcessorBase):
                                     m = matchobj.group(0)
                                     m_id = matchobj.group(1)
                                     if m_id in alts:
-                                        m_desc = alts[m_id]
-                                        return f"{m}\n___\n{m_desc}\n___\n"
+                                        # markdown blockquote
+                                        m_desc = "\n".join(["> " + li for li in alts[m_id].splitlines()])
+                                        return f"{m}\n{m_desc}\n"
                                     return m
 
                                 ptext = re.sub(link_regex, convert_links, text, 0,

{pyconverters_openai_vision-0.5.32.dist-info → pyconverters_openai_vision-0.5.36.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: pyconverters-openai_vision
-Version: 0.5.32
+Version: 0.5.36
 Summary: OpenAIVision converter
 Home-page: https://kairntech.com/
 Author: Olivier Terrier

pyconverters_openai_vision-0.5.36.dist-info/RECORD

@@ -0,0 +1,7 @@
+pyconverters_openai_vision/__init__.py,sha256=-BX6FHKC8AX-7vCq_-SC81FmeXDVkY1SIy4vU9f9-gU,52
+pyconverters_openai_vision/openai_utils.py,sha256=XI4WYZ-EAVG0Vxd5yUDuZNDgEzqHJeriScxTUusi1oo,7740
+pyconverters_openai_vision/openai_vision.py,sha256=Lp8g8KbTvRznOrC7dPRM-C9qmZcgzs-Kq6QidHtZwA4,23780
+pyconverters_openai_vision-0.5.36.dist-info/entry_points.txt,sha256=NR0re-yebKKyhApky1I6nDQzjJQfEyfOkJlJju0Ngzo,404
+pyconverters_openai_vision-0.5.36.dist-info/WHEEL,sha256=G2gURzTEtmeR8nrdXUJfNiB3VYVxigPQ-bEQujpNiNs,82
+pyconverters_openai_vision-0.5.36.dist-info/METADATA,sha256=govETANuIkSFp_1AAH1t2AxwryMgupNpZ2VERpWhbyI,2662
+pyconverters_openai_vision-0.5.36.dist-info/RECORD,,

pyconverters_openai_vision-0.5.32.dist-info/RECORD

@@ -1,7 +0,0 @@
-pyconverters_openai_vision/__init__.py,sha256=u1qzNNU3oXNQEdCLbHC2tSiBYPkFG6jkkixct0t9Zwo,52
-pyconverters_openai_vision/openai_utils.py,sha256=XI4WYZ-EAVG0Vxd5yUDuZNDgEzqHJeriScxTUusi1oo,7740
-pyconverters_openai_vision/openai_vision.py,sha256=OGr1veVTxgWxxClMNXeNCiuVrBhGA5ZpjbsoThRMXn8,23231
-pyconverters_openai_vision-0.5.32.dist-info/entry_points.txt,sha256=NR0re-yebKKyhApky1I6nDQzjJQfEyfOkJlJju0Ngzo,404
-pyconverters_openai_vision-0.5.32.dist-info/WHEEL,sha256=G2gURzTEtmeR8nrdXUJfNiB3VYVxigPQ-bEQujpNiNs,82
-pyconverters_openai_vision-0.5.32.dist-info/METADATA,sha256=spGRRfpd6RtKo3OJWmsXRv90ENoJpraJ9UtYS8NbYqg,2662
-pyconverters_openai_vision-0.5.32.dist-info/RECORD,,