chunkr-ai 0.1.0a7__py3-none-any.whl → 0.1.0a9__py3-none-any.whl

chunkr_ai/types/file_info.py

@@ -19,3 +19,6 @@ class FileInfo(BaseModel):
 
     page_count: Optional[int] = None
     """The number of pages in the file."""
+
+    ss_cell_count: Optional[int] = None
+    """The number of cells in the file. Only used for spreadsheets."""

chunkr_ai/types/ocr_result.py

@@ -15,14 +15,14 @@ class OcrResult(BaseModel):
     text: str
     """The recognized text of the OCR result."""
 
-    cell_ref: Optional[str] = None
-    """
-    Excel-style cell reference (e.g., "A1" or "A1:B2") when OCR originates from a
-    spreadsheet cell
-    """
-
     confidence: Optional[float] = None
     """The confidence score of the recognized text."""
 
     ocr_id: Optional[str] = None
     """The unique identifier for the OCR result."""
+
+    ss_cell_ref: Optional[str] = None
+    """
+    Excel-style cell reference (e.g., "A1" or "A1:B2") when OCR originates from a
+    spreadsheet cell
+    """

chunkr_ai/types/parse_configuration.py

@@ -4,7 +4,6 @@ from typing import Optional
 from typing_extensions import Literal
 
 from .._models import BaseModel
-from .llm_processing import LlmProcessing
 from .chunk_processing import ChunkProcessing
 from .segment_processing import SegmentProcessing
 
@@ -23,9 +22,6 @@ class ParseConfiguration(BaseModel):
       LLM refusals etc.)
     """
 
-    llm_processing: Optional[LlmProcessing] = None
-    """Controls the LLM used for the task."""
-
     ocr_strategy: Optional[Literal["All", "Auto"]] = None
     """Controls the Optical Character Recognition (OCR) strategy.
 

chunkr_ai/types/parse_configuration_param.py

@@ -5,7 +5,6 @@ from __future__ import annotations
 from typing import Optional
 from typing_extensions import Literal, TypedDict
 
-from .llm_processing_param import LlmProcessingParam
 from .chunk_processing_param import ChunkProcessingParam
 from .segment_processing_param import SegmentProcessingParam
 
@@ -24,9 +23,6 @@ class ParseConfigurationParam(TypedDict, total=False):
       LLM refusals etc.)
     """
 
-    llm_processing: LlmProcessingParam
-    """Controls the LLM used for the task."""
-
     ocr_strategy: Literal["All", "Auto"]
     """Controls the Optical Character Recognition (OCR) strategy.
 

chunkr_ai/types/segment.py

@@ -31,20 +31,23 @@ class Segment(BaseModel):
         "Caption",
         "Footnote",
         "Formula",
+        "FormRegion",
+        "GraphicalItem",
+        "Legend",
+        "LineNumber",
         "ListItem",
         "Page",
         "PageFooter",
         "PageHeader",
+        "PageNumber",
         "Picture",
-        "SectionHeader",
         "Table",
         "Text",
         "Title",
+        "Unknown",
+        "SectionHeader",
     ]
-    """
-    All the possible types for a segment. Note: Different configurations will
-    produce different types. Please refer to the documentation for more information.
-    """
+    """All the possible types for a segment."""
 
     confidence: Optional[float] = None
     """Confidence score of the layout analysis model"""

chunkr_ai/types/segment_processing.py

@@ -47,6 +47,24 @@ class SegmentProcessing(BaseModel):
     - `extended_context` uses the full page image as context for LLM generation.
     """
 
+    form_region: Optional[GenerationConfig] = FieldInfo(alias="FormRegion", default=None)
+    """Controls the processing and generation for the segment.
+
+    - `crop_image` controls whether to crop the file's images to the segment's
+      bounding box. The cropped image will be stored in the segment's `image` field.
+      Use `All` to always crop, or `Auto` to only crop when needed for
+      post-processing.
+    - `format` specifies the output format: `Html` or `Markdown`
+    - `strategy` determines how the content is generated: `Auto`, `LLM`, or `Ignore`
+      - `Auto`: Process content automatically
+      - `LLM`: Use large language models for processing
+      - `Ignore`: Exclude segments from final output
+    - `description` enables LLM-generated descriptions for segments. **Note:** This
+      uses chunkr's own VLM models and is not configurable via LLM processing
+      configuration.
+    - `extended_context` uses the full page image as context for LLM generation.
+    """
+
     formula: Optional[GenerationConfig] = FieldInfo(alias="Formula", default=None)
     """Controls the processing and generation for the segment.
 
@@ -65,6 +83,60 @@ class SegmentProcessing(BaseModel):
     - `extended_context` uses the full page image as context for LLM generation.
     """
 
+    graphical_item: Optional[GenerationConfig] = FieldInfo(alias="GraphicalItem", default=None)
+    """Controls the processing and generation for the segment.
+
+    - `crop_image` controls whether to crop the file's images to the segment's
+      bounding box. The cropped image will be stored in the segment's `image` field.
+      Use `All` to always crop, or `Auto` to only crop when needed for
+      post-processing.
+    - `format` specifies the output format: `Html` or `Markdown`
+    - `strategy` determines how the content is generated: `Auto`, `LLM`, or `Ignore`
+      - `Auto`: Process content automatically
+      - `LLM`: Use large language models for processing
+      - `Ignore`: Exclude segments from final output
+    - `description` enables LLM-generated descriptions for segments. **Note:** This
+      uses chunkr's own VLM models and is not configurable via LLM processing
+      configuration.
+    - `extended_context` uses the full page image as context for LLM generation.
+    """
+
+    legend: Optional[GenerationConfig] = FieldInfo(alias="Legend", default=None)
+    """Controls the processing and generation for the segment.
+
+    - `crop_image` controls whether to crop the file's images to the segment's
+      bounding box. The cropped image will be stored in the segment's `image` field.
+      Use `All` to always crop, or `Auto` to only crop when needed for
+      post-processing.
+    - `format` specifies the output format: `Html` or `Markdown`
+    - `strategy` determines how the content is generated: `Auto`, `LLM`, or `Ignore`
+      - `Auto`: Process content automatically
+      - `LLM`: Use large language models for processing
+      - `Ignore`: Exclude segments from final output
+    - `description` enables LLM-generated descriptions for segments. **Note:** This
+      uses chunkr's own VLM models and is not configurable via LLM processing
+      configuration.
+    - `extended_context` uses the full page image as context for LLM generation.
+    """
+
+    line_number: Optional[GenerationConfig] = FieldInfo(alias="LineNumber", default=None)
+    """Controls the processing and generation for the segment.
+
+    - `crop_image` controls whether to crop the file's images to the segment's
+      bounding box. The cropped image will be stored in the segment's `image` field.
+      Use `All` to always crop, or `Auto` to only crop when needed for
+      post-processing.
+    - `format` specifies the output format: `Html` or `Markdown`
+    - `strategy` determines how the content is generated: `Auto`, `LLM`, or `Ignore`
+      - `Auto`: Process content automatically
+      - `LLM`: Use large language models for processing
+      - `Ignore`: Exclude segments from final output
+    - `description` enables LLM-generated descriptions for segments. **Note:** This
+      uses chunkr's own VLM models and is not configurable via LLM processing
+      configuration.
+    - `extended_context` uses the full page image as context for LLM generation.
+    """
+
     list_item: Optional[GenerationConfig] = FieldInfo(alias="ListItem", default=None)
     """Controls the processing and generation for the segment.
 
@@ -137,7 +209,7 @@ class SegmentProcessing(BaseModel):
     - `extended_context` uses the full page image as context for LLM generation.
     """
 
-    picture: Optional[GenerationConfig] = FieldInfo(alias="Picture", default=None)
+    page_number: Optional[GenerationConfig] = FieldInfo(alias="PageNumber", default=None)
     """Controls the processing and generation for the segment.
 
     - `crop_image` controls whether to crop the file's images to the segment's
@@ -155,7 +227,7 @@ class SegmentProcessing(BaseModel):
     - `extended_context` uses the full page image as context for LLM generation.
     """
 
-    section_header: Optional[GenerationConfig] = FieldInfo(alias="SectionHeader", default=None)
+    picture: Optional[GenerationConfig] = FieldInfo(alias="Picture", default=None)
     """Controls the processing and generation for the segment.
 
     - `crop_image` controls whether to crop the file's images to the segment's
@@ -226,3 +298,21 @@ class SegmentProcessing(BaseModel):
       configuration.
     - `extended_context` uses the full page image as context for LLM generation.
     """
+
+    unknown: Optional[GenerationConfig] = FieldInfo(alias="Unknown", default=None)
+    """Controls the processing and generation for the segment.
+
+    - `crop_image` controls whether to crop the file's images to the segment's
+      bounding box. The cropped image will be stored in the segment's `image` field.
+      Use `All` to always crop, or `Auto` to only crop when needed for
+      post-processing.
+    - `format` specifies the output format: `Html` or `Markdown`
+    - `strategy` determines how the content is generated: `Auto`, `LLM`, or `Ignore`
+      - `Auto`: Process content automatically
+      - `LLM`: Use large language models for processing
+      - `Ignore`: Exclude segments from final output
+    - `description` enables LLM-generated descriptions for segments. **Note:** This
+      uses chunkr's own VLM models and is not configurable via LLM processing
+      configuration.
+    - `extended_context` uses the full page image as context for LLM generation.
+    """

chunkr_ai/types/segment_processing_param.py

@@ -48,6 +48,24 @@ class SegmentProcessingParam(TypedDict, total=False):
     - `extended_context` uses the full page image as context for LLM generation.
     """
 
+    form_region: Annotated[Optional[GenerationConfigParam], PropertyInfo(alias="FormRegion")]
+    """Controls the processing and generation for the segment.
+
+    - `crop_image` controls whether to crop the file's images to the segment's
+      bounding box. The cropped image will be stored in the segment's `image` field.
+      Use `All` to always crop, or `Auto` to only crop when needed for
+      post-processing.
+    - `format` specifies the output format: `Html` or `Markdown`
+    - `strategy` determines how the content is generated: `Auto`, `LLM`, or `Ignore`
+      - `Auto`: Process content automatically
+      - `LLM`: Use large language models for processing
+      - `Ignore`: Exclude segments from final output
+    - `description` enables LLM-generated descriptions for segments. **Note:** This
+      uses chunkr's own VLM models and is not configurable via LLM processing
+      configuration.
+    - `extended_context` uses the full page image as context for LLM generation.
+    """
+
     formula: Annotated[Optional[GenerationConfigParam], PropertyInfo(alias="Formula")]
     """Controls the processing and generation for the segment.
 
@@ -66,6 +84,60 @@ class SegmentProcessingParam(TypedDict, total=False):
     - `extended_context` uses the full page image as context for LLM generation.
     """
 
+    graphical_item: Annotated[Optional[GenerationConfigParam], PropertyInfo(alias="GraphicalItem")]
+    """Controls the processing and generation for the segment.
+
+    - `crop_image` controls whether to crop the file's images to the segment's
+      bounding box. The cropped image will be stored in the segment's `image` field.
+      Use `All` to always crop, or `Auto` to only crop when needed for
+      post-processing.
+    - `format` specifies the output format: `Html` or `Markdown`
+    - `strategy` determines how the content is generated: `Auto`, `LLM`, or `Ignore`
+      - `Auto`: Process content automatically
+      - `LLM`: Use large language models for processing
+      - `Ignore`: Exclude segments from final output
+    - `description` enables LLM-generated descriptions for segments. **Note:** This
+      uses chunkr's own VLM models and is not configurable via LLM processing
+      configuration.
+    - `extended_context` uses the full page image as context for LLM generation.
+    """
+
+    legend: Annotated[Optional[GenerationConfigParam], PropertyInfo(alias="Legend")]
+    """Controls the processing and generation for the segment.
+
+    - `crop_image` controls whether to crop the file's images to the segment's
+      bounding box. The cropped image will be stored in the segment's `image` field.
+      Use `All` to always crop, or `Auto` to only crop when needed for
+      post-processing.
+    - `format` specifies the output format: `Html` or `Markdown`
+    - `strategy` determines how the content is generated: `Auto`, `LLM`, or `Ignore`
+      - `Auto`: Process content automatically
+      - `LLM`: Use large language models for processing
+      - `Ignore`: Exclude segments from final output
+    - `description` enables LLM-generated descriptions for segments. **Note:** This
+      uses chunkr's own VLM models and is not configurable via LLM processing
+      configuration.
+    - `extended_context` uses the full page image as context for LLM generation.
+    """
+
+    line_number: Annotated[Optional[GenerationConfigParam], PropertyInfo(alias="LineNumber")]
+    """Controls the processing and generation for the segment.
+
+    - `crop_image` controls whether to crop the file's images to the segment's
+      bounding box. The cropped image will be stored in the segment's `image` field.
+      Use `All` to always crop, or `Auto` to only crop when needed for
+      post-processing.
+    - `format` specifies the output format: `Html` or `Markdown`
+    - `strategy` determines how the content is generated: `Auto`, `LLM`, or `Ignore`
+      - `Auto`: Process content automatically
+      - `LLM`: Use large language models for processing
+      - `Ignore`: Exclude segments from final output
+    - `description` enables LLM-generated descriptions for segments. **Note:** This
+      uses chunkr's own VLM models and is not configurable via LLM processing
+      configuration.
+    - `extended_context` uses the full page image as context for LLM generation.
+    """
+
     list_item: Annotated[Optional[GenerationConfigParam], PropertyInfo(alias="ListItem")]
     """Controls the processing and generation for the segment.
 
@@ -138,7 +210,7 @@ class SegmentProcessingParam(TypedDict, total=False):
     - `extended_context` uses the full page image as context for LLM generation.
     """
 
-    picture: Annotated[Optional[GenerationConfigParam], PropertyInfo(alias="Picture")]
+    page_number: Annotated[Optional[GenerationConfigParam], PropertyInfo(alias="PageNumber")]
     """Controls the processing and generation for the segment.
 
     - `crop_image` controls whether to crop the file's images to the segment's
@@ -156,7 +228,7 @@ class SegmentProcessingParam(TypedDict, total=False):
     - `extended_context` uses the full page image as context for LLM generation.
     """
 
-    section_header: Annotated[Optional[GenerationConfigParam], PropertyInfo(alias="SectionHeader")]
+    picture: Annotated[Optional[GenerationConfigParam], PropertyInfo(alias="Picture")]
     """Controls the processing and generation for the segment.
 
     - `crop_image` controls whether to crop the file's images to the segment's
@@ -227,3 +299,21 @@ class SegmentProcessingParam(TypedDict, total=False):
       configuration.
     - `extended_context` uses the full page image as context for LLM generation.
     """
+
+    unknown: Annotated[Optional[GenerationConfigParam], PropertyInfo(alias="Unknown")]
+    """Controls the processing and generation for the segment.
+
+    - `crop_image` controls whether to crop the file's images to the segment's
+      bounding box. The cropped image will be stored in the segment's `image` field.
+      Use `All` to always crop, or `Auto` to only crop when needed for
+      post-processing.
+    - `format` specifies the output format: `Html` or `Markdown`
+    - `strategy` determines how the content is generated: `Auto`, `LLM`, or `Ignore`
+      - `Auto`: Process content automatically
+      - `LLM`: Use large language models for processing
+      - `Ignore`: Exclude segments from final output
+    - `description` enables LLM-generated descriptions for segments. **Note:** This
+      uses chunkr's own VLM models and is not configurable via LLM processing
+      configuration.
+    - `extended_context` uses the full page image as context for LLM generation.
+    """

chunkr_ai/types/task_get_params.py

@@ -16,6 +16,3 @@ class TaskGetParams(TypedDict, total=False):
 
     include_chunks: bool
     """Whether to include chunks in the output response"""
-
-    wait_for_completion: bool
-    """Whether to wait for the task to complete"""

chunkr_ai/types/tasks/extract_create_response.py

@@ -55,153 +55,6 @@ class ExtractCreateResponse(BaseModel):
       array-of-primitives) contain a `Vec<Citation>` supporting that field.
     - `metrics`: mirror of `results`; only leaf positions contain a `Metrics` object
       for that field.
-
-    Detailed shape:
-
-    - Shared structure: `results`, `citations`, and `metrics` have the same
-      object/array shape as the user schema. Non-leaf nodes (objects, arrays of
-      objects) are mirrored; only leaves carry values.
-    - Leaf definition:
-      - A leaf is either a JSON primitive (string, number, bool, or null) or an
-        array of primitives (including empty).
-      - Arrays of objects are not leaves; recurse into their elements (`items`
-        mirror index-by-index).
-    - Null handling:
-      - If a leaf in `results` is null, the corresponding position in `citations`
-        and `metrics` remains null.
-    - Arrays:
-      - Array of objects: `citations`/`metrics` are arrays whose elements mirror
-        each object and carry values at their own leaves.
-      - Array of primitives: treated as a single leaf. `citations[path]` is a list
-        of `Citation` supporting the array as a whole. `metrics[path]` is a
-        `Metrics` object for the array as a whole.
-    - Citations leaves:
-      - Type: JSON array of `Citation` objects.
-      - Each `Citation` has: `citation_id: string`, `citation_type: Segment|Word`,
-        `bbox: BoundingBox[]`, `content: string`, `segment_id?: string`,
-        `segment_type: SegmentType`, `ss_range?: string[]`.
-        - Segment citation: represents a full parsed segment; `segment_id` set,
-          `bbox` has one entry (segment box), `content` is the segment text. If the
-          segment is from a spreadsheet, `ss_range` contains the table range
-          (single-element array) or the underlying cell refs if available.
-        - Word citation: represents selected OCR words within a segment;
-          `segment_id` is null, `bbox` has one entry per word, `content` is the
-          whitespace-joined text of those words; `segment_type` is `Text`. If OCR
-          words came from spreadsheet cells, `ss_range` lists those cell refs.
-    - Metrics leaves:
-      - Type: `Metrics` object with `confidence: "High" | "Low"`, indicating whether
-        citations sufficiently support the item.
-
-    Example:
-
-    results
-
-    ```json
-    {
-      "invoice_id": "INV-001",
-      "seller": { "name": "Acme" },
-      "line_items": [{ "sku": "A1", "qty": 2 }],
-      "tags": ["urgent", "paid"],
-      "notes": null
-    }
-    ```
-
-    citations
-
-    ```json
-    {
-      "invoice_id": [
-        {
-          "citation_id": "abc1234",
-          "citation_type": "Segment",
-          "bbox": [{ "left": 10, "top": 20, "width": 100, "height": 18 }],
-          "content": "Invoice INV-001",
-          "segment_id": "seg_001",
-          "segment_type": "Text",
-          "ss_range": ["A1:C10"]
-        },
-        {
-          "citation_id": "pqr2345",
-          "citation_type": "Word",
-          "bbox": [
-            { "left": 12, "top": 24, "width": 36, "height": 18 },
-            { "left": 52, "top": 24, "width": 48, "height": 18 }
-          ],
-          "content": "INV-001",
-          "segment_id": null,
-          "segment_type": "Text",
-          "ss_range": ["B3", "C3"]
-        }
-      ],
-      "seller": {
-        "name": [
-          {
-            "citation_id": "def5678",
-            "citation_type": "Word",
-            "bbox": [
-              { "left": 45, "top": 80, "width": 30, "height": 12 },
-              { "left": 80, "top": 80, "width": 40, "height": 12 }
-            ],
-            "content": "Acme",
-            "segment_id": null,
-            "segment_type": "Text"
-          }
-        ]
-      },
-      "line_items": [
-        {
-          "sku": [
-            {
-              "citation_id": "ghi9012",
-              "citation_type": "Segment",
-              "bbox": [{ "left": 12, "top": 140, "width": 60, "height": 16 }],
-              "content": "A1",
-              "segment_id": "seg_010",
-              "segment_type": "Text",
-              "ss_range": ["D5:E12"]
-            }
-          ],
-          "qty": [
-            {
-              "citation_id": "jkl3456",
-              "citation_type": "Word",
-              "bbox": [{ "left": 85, "top": 140, "width": 12, "height": 16 }],
-              "content": "2",
-              "segment_id": null,
-              "segment_type": "Text",
-              "ss_range": ["E12"]
-            }
-          ]
-        }
-      ],
-      "tags": [
-        {
-          "citation_id": "mno7890",
-          "citation_type": "Segment",
-          "bbox": [{ "left": 12, "top": 200, "width": 100, "height": 16 }],
-          "content": "urgent paid",
-          "segment_id": "seg_020",
-          "segment_type": "Text",
-          "ss_range": ["A20:C25"]
-        }
-      ],
-      "notes": null
-    }
-    ```
-
-    metrics
-
-    ```json
-    {
-      "invoice_id": { "confidence": "High" },
-      "seller": { "name": { "confidence": "Low" } },
-      "line_items": [
-        { "sku": { "confidence": "High" }, "qty": { "confidence": "High" } }
-      ],
-      "tags": { "confidence": "Low" },
-      "notes": null
-    }
-    ```
     """
 
     source_task_id: Optional[str] = None

chunkr_ai/types/tasks/extract_get_params.py

@@ -16,6 +16,3 @@ class ExtractGetParams(TypedDict, total=False):
 
     include_chunks: bool
     """Whether to include chunks in the output response"""
-
-    wait_for_completion: bool
-    """Whether to wait for the task to complete"""

chunkr_ai/types/tasks/extract_get_response.py

@@ -55,153 +55,6 @@ class ExtractGetResponse(BaseModel):
       array-of-primitives) contain a `Vec<Citation>` supporting that field.
     - `metrics`: mirror of `results`; only leaf positions contain a `Metrics` object
       for that field.
-
-    Detailed shape:
-
-    - Shared structure: `results`, `citations`, and `metrics` have the same
-      object/array shape as the user schema. Non-leaf nodes (objects, arrays of
-      objects) are mirrored; only leaves carry values.
-    - Leaf definition:
-      - A leaf is either a JSON primitive (string, number, bool, or null) or an
-        array of primitives (including empty).
-      - Arrays of objects are not leaves; recurse into their elements (`items`
-        mirror index-by-index).
-    - Null handling:
-      - If a leaf in `results` is null, the corresponding position in `citations`
-        and `metrics` remains null.
-    - Arrays:
-      - Array of objects: `citations`/`metrics` are arrays whose elements mirror
-        each object and carry values at their own leaves.
-      - Array of primitives: treated as a single leaf. `citations[path]` is a list
-        of `Citation` supporting the array as a whole. `metrics[path]` is a
-        `Metrics` object for the array as a whole.
-    - Citations leaves:
-      - Type: JSON array of `Citation` objects.
-      - Each `Citation` has: `citation_id: string`, `citation_type: Segment|Word`,
-        `bbox: BoundingBox[]`, `content: string`, `segment_id?: string`,
-        `segment_type: SegmentType`, `ss_range?: string[]`.
-        - Segment citation: represents a full parsed segment; `segment_id` set,
-          `bbox` has one entry (segment box), `content` is the segment text. If the
-          segment is from a spreadsheet, `ss_range` contains the table range
-          (single-element array) or the underlying cell refs if available.
-        - Word citation: represents selected OCR words within a segment;
-          `segment_id` is null, `bbox` has one entry per word, `content` is the
-          whitespace-joined text of those words; `segment_type` is `Text`. If OCR
-          words came from spreadsheet cells, `ss_range` lists those cell refs.
-    - Metrics leaves:
-      - Type: `Metrics` object with `confidence: "High" | "Low"`, indicating whether
-        citations sufficiently support the item.
-
-    Example:
-
-    results
-
-    ```json
-    {
-      "invoice_id": "INV-001",
-      "seller": { "name": "Acme" },
-      "line_items": [{ "sku": "A1", "qty": 2 }],
-      "tags": ["urgent", "paid"],
-      "notes": null
-    }
-    ```
-
-    citations
-
-    ```json
-    {
-      "invoice_id": [
-        {
-          "citation_id": "abc1234",
-          "citation_type": "Segment",
-          "bbox": [{ "left": 10, "top": 20, "width": 100, "height": 18 }],
-          "content": "Invoice INV-001",
-          "segment_id": "seg_001",
-          "segment_type": "Text",
-          "ss_range": ["A1:C10"]
-        },
-        {
-          "citation_id": "pqr2345",
-          "citation_type": "Word",
-          "bbox": [
-            { "left": 12, "top": 24, "width": 36, "height": 18 },
-            { "left": 52, "top": 24, "width": 48, "height": 18 }
-          ],
-          "content": "INV-001",
-          "segment_id": null,
-          "segment_type": "Text",
-          "ss_range": ["B3", "C3"]
-        }
-      ],
-      "seller": {
-        "name": [
-          {
-            "citation_id": "def5678",
-            "citation_type": "Word",
-            "bbox": [
-              { "left": 45, "top": 80, "width": 30, "height": 12 },
-              { "left": 80, "top": 80, "width": 40, "height": 12 }
-            ],
-            "content": "Acme",
-            "segment_id": null,
-            "segment_type": "Text"
-          }
-        ]
-      },
-      "line_items": [
-        {
-          "sku": [
-            {
-              "citation_id": "ghi9012",
-              "citation_type": "Segment",
-              "bbox": [{ "left": 12, "top": 140, "width": 60, "height": 16 }],
-              "content": "A1",
-              "segment_id": "seg_010",
-              "segment_type": "Text",
-              "ss_range": ["D5:E12"]
-            }
-          ],
-          "qty": [
-            {
-              "citation_id": "jkl3456",
-              "citation_type": "Word",
-              "bbox": [{ "left": 85, "top": 140, "width": 12, "height": 16 }],
-              "content": "2",
-              "segment_id": null,
-              "segment_type": "Text",
-              "ss_range": ["E12"]
-            }
-          ]
-        }
-      ],
-      "tags": [
-        {
-          "citation_id": "mno7890",
-          "citation_type": "Segment",
-          "bbox": [{ "left": 12, "top": 200, "width": 100, "height": 16 }],
-          "content": "urgent paid",
-          "segment_id": "seg_020",
-          "segment_type": "Text",
-          "ss_range": ["A20:C25"]
-        }
-      ],
-      "notes": null
-    }
-    ```
-
-    metrics
-
-    ```json
-    {
-      "invoice_id": { "confidence": "High" },
-      "seller": { "name": { "confidence": "Low" } },
-      "line_items": [
-        { "sku": { "confidence": "High" }, "qty": { "confidence": "High" } }
-      ],
-      "tags": { "confidence": "Low" },
-      "notes": null
-    }
-    ```
     """
 
     source_task_id: Optional[str] = None