retab 0.0.87__py3-none-any.whl → 0.0.88__py3-none-any.whl

retab/types/documents/split.py

@@ -5,6 +5,7 @@ from ..mime import MIMEData
 class Category(BaseModel):
     name: str = Field(..., description="The name of the category")
     description: str = Field(..., description="The description of the category")
+    partition_key: str | None = Field(default=None, description="The key to partition the category")
 
 
 class SplitRequest(BaseModel):
@@ -13,20 +14,32 @@ class SplitRequest(BaseModel):
     model: str = Field(default="retab-small", description="The model to use to split the document")
 
 
+class Partition(BaseModel):
+    key: str = Field(..., description="The partition key value (e.g., property ID, invoice number)")
+    pages: list[int] = Field(..., description="The pages of the partition (1-indexed)")
+    first_page_y_start: float = Field(default=0.0, description="The y coordinate of the first page of the partition")
+    last_page_y_end: float = Field(default=1.0, description="The y coordinate of the last page of the partition")
+
 class SplitResult(BaseModel):
     name: str = Field(..., description="The name of the category")
-    start_page: int = Field(..., description="The start page of the category (1-indexed)")
-    end_page: int = Field(..., description="The end page of the category (1-indexed, inclusive)")
+    pages: list[int] = Field(..., description="The pages of the category (1-indexed)")
+    partitions: list[Partition] = Field(default_factory=list, description="The partitions of the category")
 
 
 class SplitResponse(BaseModel):
     splits: list[SplitResult] = Field(..., description="The list of document splits with their page ranges")
 
 
+class SplitOutputItem(BaseModel):
+    """Internal schema item for LLM structured output validation."""
+    name: str = Field(..., description="The name of the category")
+    start_page: int = Field(..., description="The start page of the category (1-indexed)")
+    end_page: int = Field(..., description="The end page of the category (1-indexed, inclusive)")
+
 
 class SplitOutputSchema(BaseModel):
     """Schema for LLM structured output."""
-    splits: list[SplitResult] = Field(
-        ..., 
+    splits: list[SplitOutputItem] = Field(
+        ...,
         description="List of document sections, each classified into one of the provided categories with their page ranges"
     )

{retab-0.0.87.dist-info → retab-0.0.88.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: retab
-Version: 0.0.87
+Version: 0.0.88
 Summary: Retab official python library
 Home-page: https://github.com/retab-dev/retab
 Author: Retab

{retab-0.0.87.dist-info → retab-0.0.88.dist-info}/RECORD

@@ -36,7 +36,7 @@ retab/types/documents/create_messages.py,sha256=Uym0SnVUGkyt1C5AOD37BsZ3puyeu_ig
 retab/types/documents/edit.py,sha256=b6UcYLOJkClpMu4QyYmdp-X4WtN8U_3oiMBc1KLklVY,5663
 retab/types/documents/extract.py,sha256=x_59fm69-icsxxGRgpFd0NN-SLRoMYqbvfCZuG7zyGc,18033
 retab/types/documents/parse.py,sha256=MXe7zh3DusWQhGe0Sr95nPy6cB8DRX8MA4Hmjj_AP7E,1300
-retab/types/documents/split.py,sha256=xRdJ6IpSRAPi_ZtAG2FNqg5A-v5tzfb1QQkW5UfO2pY,1246
+retab/types/documents/split.py,sha256=Bjk5iJdS3v7I3rCvqpFUPlzgO4HINqh3uMPQJg-MqPc,2166
 retab/types/edit/__init__.py,sha256=M8hF97h7fX8RP9IsB6qpkw0eyvO0DFQvP6FmWL8caCQ,331
 retab/types/edit/templates.py,sha256=RLRIMdXzU-5_3XPf0iMSozjRTAP5Tliq0nrjlZn0l8E,2412
 retab/types/extractions/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
@@ -59,7 +59,7 @@ retab/utils/hashing.py,sha256=_BMVUvftOcJav68QL0rLkH2dbhW9RRJPzeGC2akR0fc,757
 retab/utils/json_schema.py,sha256=zP4pQLpVHBKWo_abCjb_dU4kA0azhHopd-1TFUgVEvc,20655
 retab/utils/mime.py,sha256=mTP_lqSPttOP5DYJxopiWaeFXrUCPjhwd7y53nCVGO4,6189
 retab/utils/stream_context_managers.py,sha256=gI1gVQSj3nWz6Mvjz7Ix5AiY0g6vSL-c2tPfuP04izo,2314
-retab-0.0.87.dist-info/METADATA,sha256=Rz6B3ctJWOHF0hcaFxc2hEyBgpeBRgvScGxFNGjALMg,4532
-retab-0.0.87.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
-retab-0.0.87.dist-info/top_level.txt,sha256=waQR0EGdhLIQtztoE3AXg7ik5ONQ9q_bsKVpyFuJdq0,6
-retab-0.0.87.dist-info/RECORD,,
+retab-0.0.88.dist-info/METADATA,sha256=6Tzn_jnoHNc0-UkawkeLW5LQL-rGrfU-6A_mHEbROac,4532
+retab-0.0.88.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
+retab-0.0.88.dist-info/top_level.txt,sha256=waQR0EGdhLIQtztoE3AXg7ik5ONQ9q_bsKVpyFuJdq0,6
+retab-0.0.88.dist-info/RECORD,,