chunkr-ai 0.0.37__py3-none-any.whl → 0.0.39__py3-none-any.whl

chunkr_ai/api/configuration.py

@@ -19,20 +19,18 @@ class GenerationConfig(BaseModel):
 class SegmentProcessing(BaseModel):
     model_config = ConfigDict(populate_by_name=True, alias_generator=str.title)
 
-    title: Optional[GenerationConfig] = Field(default=None, alias="Title")
-    section_header: Optional[GenerationConfig] = Field(
-        default=None, alias="SectionHeader"
-    )
-    text: Optional[GenerationConfig] = Field(default=None, alias="Text")
-    list_item: Optional[GenerationConfig] = Field(default=None, alias="ListItem")
-    table: Optional[GenerationConfig] = Field(default=None, alias="Table")
-    picture: Optional[GenerationConfig] = Field(default=None, alias="Picture")
     caption: Optional[GenerationConfig] = Field(default=None, alias="Caption")
-    formula: Optional[GenerationConfig] = Field(default=None, alias="Formula")
     footnote: Optional[GenerationConfig] = Field(default=None, alias="Footnote")
-    page_header: Optional[GenerationConfig] = Field(default=None, alias="PageHeader")
-    page_footer: Optional[GenerationConfig] = Field(default=None, alias="PageFooter")
+    formula: Optional[GenerationConfig] = Field(default=None, alias="Formula")
+    list_item: Optional[GenerationConfig] = Field(default=None, alias="ListItem")
     page: Optional[GenerationConfig] = Field(default=None, alias="Page")
+    page_footer: Optional[GenerationConfig] = Field(default=None, alias="PageFooter")
+    page_header: Optional[GenerationConfig] = Field(default=None, alias="PageHeader")
+    picture: Optional[GenerationConfig] = Field(default=None, alias="Picture")
+    section_header: Optional[GenerationConfig] = Field(default=None, alias="SectionHeader")
+    table: Optional[GenerationConfig] = Field(default=None, alias="Table")
+    text: Optional[GenerationConfig] = Field(default=None, alias="Text")
+    title: Optional[GenerationConfig] = Field(default=None, alias="Title")
 
 class ChunkProcessing(BaseModel):
     ignore_headers_and_footers: Optional[bool] = None
@@ -84,11 +82,13 @@ class Segment(BaseModel):
     page_width: float
     segment_id: str
     segment_type: SegmentType
+    confidence: Optional[float]
 
 class Chunk(BaseModel):
     chunk_id: str
     chunk_length: int
     segments: List[Segment]
+    embed: Optional[str] = None
 
 class OutputResponse(BaseModel):
     chunks: List[Chunk]

chunkr_ai/api/task_response.py

@@ -144,7 +144,7 @@ class TaskResponse(BaseModel, Generic[T]):
         Args:
             output_file (str, optional): Path to save the markdown content. Defaults to None.
         """
-        content = self._get_content("markdown")
+        content = self._get_content("markdown", separator="\n\n")
         self._write_to_file(content, output_file)
         return content
 
@@ -168,7 +168,7 @@ class TaskResponse(BaseModel, Generic[T]):
         self._write_to_file(data, output_file, is_json=True)
         return data
 
-    def _get_content(self, t: str) -> str:
+    def _get_content(self, t: str, separator: str = "\n") -> str:
         if not self.output:
             return ""
         parts = []
@@ -177,4 +177,4 @@ class TaskResponse(BaseModel, Generic[T]):
                 v = getattr(s, t)
                 if v:
                     parts.append(v)
-        return "\n".join(parts)
+        return separator.join(parts)

{chunkr_ai-0.0.37.dist-info → chunkr_ai-0.0.39.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.2
 Name: chunkr-ai
-Version: 0.0.37
+Version: 0.0.39
 Summary: Python client for Chunkr: open source document intelligence
 Author-email: Ishaan Kapoor <ishaan@lumina.sh>
 License: MIT License

{chunkr_ai-0.0.37.dist-info → chunkr_ai-0.0.39.dist-info}/RECORD

@@ -4,13 +4,13 @@ chunkr_ai/api/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 chunkr_ai/api/auth.py,sha256=hlv0GiUmlsbFO1wLL9sslqOnsBSoBqkL_6Mk2SDvxgE,413
 chunkr_ai/api/chunkr.py,sha256=BzwcKNCuLfVR-HzgY8tKStsW4pIDVVjBgnEqPLyUUMM,3292
 chunkr_ai/api/chunkr_base.py,sha256=FDl0Ew8eOY4hur5FFqPENZiq9YQy0G3XWEqcKPeCO-U,6130
-chunkr_ai/api/configuration.py,sha256=2Bfw_c8eQVijb0EvsexiuRbF1pZUspYFBMuZ-ErJHvs,3835
+chunkr_ai/api/configuration.py,sha256=KrXKcC1Yd7wfK8JMfihlWjNxlDyzKydr1Pe1_r1DTZw,3885
 chunkr_ai/api/decorators.py,sha256=VJX4qGBIL00K2zY8bh5KAMWv7SltJ38TvPJH06FnFss,4415
 chunkr_ai/api/misc.py,sha256=QN-2YWQ8e3VvvK63Ua-e8jsx6gxVxkO88Z96yWOofu0,3653
 chunkr_ai/api/protocol.py,sha256=LjPrYSq52m1afIlAo0yVGXlGZxPRh8J6g7S4PAit3Zo,388
-chunkr_ai/api/task_response.py,sha256=E1H5Cmn9GSYHX60f3Iz8hcTItPv6DpBEEO2vP2vcKDM,6282
-chunkr_ai-0.0.37.dist-info/LICENSE,sha256=w3R12yNDyZpMiy2lxy_hvNbsldC75ww79sF0u11rkho,1069
-chunkr_ai-0.0.37.dist-info/METADATA,sha256=eKMu3F3yYMO2v1i-PAgUKEz1LyU_ManhLOpZqY7RiPo,7031
-chunkr_ai-0.0.37.dist-info/WHEEL,sha256=In9FTNxeP60KnTkGw7wk6mJPYd_dQSjEZmXdBdMCI-8,91
-chunkr_ai-0.0.37.dist-info/top_level.txt,sha256=0IZY7PZIiS8bw5r4NUQRUQ-ATi-L_3vLQVq3ZLouOW8,10
-chunkr_ai-0.0.37.dist-info/RECORD,,
+chunkr_ai/api/task_response.py,sha256=ti_2VTYtYS9FWyW-QIm16rp6qhs8RVy4vvgCZUkI2wA,6328
+chunkr_ai-0.0.39.dist-info/LICENSE,sha256=w3R12yNDyZpMiy2lxy_hvNbsldC75ww79sF0u11rkho,1069
+chunkr_ai-0.0.39.dist-info/METADATA,sha256=Q99T7lpBksl0FuWYaEb6k9Qblp3gk8fnq9-yHLN2i9Y,7031
+chunkr_ai-0.0.39.dist-info/WHEEL,sha256=In9FTNxeP60KnTkGw7wk6mJPYd_dQSjEZmXdBdMCI-8,91
+chunkr_ai-0.0.39.dist-info/top_level.txt,sha256=0IZY7PZIiS8bw5r4NUQRUQ-ATi-L_3vLQVq3ZLouOW8,10
+chunkr_ai-0.0.39.dist-info/RECORD,,