Results (#20)

* result test

* ta2 schemas, docs, results changes

* UUID removal

* fix

* more tries

* better

* final

* really

* build fix

---------

Co-authored-by: Justin Gawrilow <[email protected]>

cdr_schemas/area_extraction.py

@@ -7,15 +7,15 @@
 
 
 class AreaType(str, Enum):
-    Map_Area = "map_area"
-    Legend_Area = "legend_area"
-    CrossSection = "cross_section"
-    OCR = "ocr"
-    Polygon_Legend_Area = "polygon_legend_area"
-    Line_Point_Legend_Area = "line_point_legend_area"
-    Line_Legend_Area = "line_legend_area"
-    Point_Legend_Area = "point_legend_area"
-    Correlation_Diagram = "correlation_diagram"
+    Map_Area = "Map_Area"
+    Legend_Area = "Legend_Area"
+    CrossSection = "CrossSection"
+    OCR = "OCR"
+    Polygon_Legend_Area = "Polygon_Legend_Area"
+    Line_Point_Legend_Area = "Line_Point_Legend_Area"
+    Line_Legend_Area = "Line_Legend_Area"
+    Point_Legend_Area = "Point_Legend_Area"
+    Correlation_Diagram = "Correlation_Diagram"
 
 
 class Area_Extraction(BaseModel):

cdr_schemas/document.py

@@ -0,0 +1,80 @@
+from typing import Optional
+
+from pydantic import BaseModel, Field
+
+
+class DocumentProvenance(BaseModel):
+    """JSON model for Document Provenance"""
+
+    external_system_name: str = Field(
+        ..., description="Name of system storing document"
+    )
+    external_system_id: str = Field(None, description="The system ID of the document")
+    external_system_url: str = Field(
+        None, description="Name of system storing document"
+    )
+
+
+class Document(BaseModel):
+    """JSON model for user-facing document metadata"""
+
+    id: str = Field(..., description="The internal ID of the document")
+    title: str = Field(..., description="Title of the document")
+    is_open: bool = Field(
+        ...,
+        description="Whether document is open or not.",
+    )
+
+    pages: Optional[int] = Field(None, description="Document page count")
+
+    provenance: Optional[list[DocumentProvenance]] = Field(
+        None, description="provenance list"
+    )
+
+    system: str = Field(
+        ...,
+        description="""
+            The name of the system used.
+        """,
+    )
+    system_version: str = Field(
+        ...,
+        description="""
+            The version of the system used.
+        """,
+    )
+
+
+class DocumentExtraction(BaseModel):
+    """JSON model for user-facing document metadata"""
+
+    id: str | None = Field(None, description="The internal ID of the xtraction")
+    document_id: str = Field(None, description="The internal ID of the source document")
+    extraction_type: str = Field(
+        ..., description="The type of model that produced the extraction"
+    )
+    extraction_label: str = Field(
+        ..., description="The classification of the extraction within its model"
+    )
+    score: float | None = Field(None, description="The confidence of the extraction")
+    bbox: tuple[float, float, float, float] | None = Field(
+        None, description="The bounding box of the extraction"
+    )
+    page_num: int | None = Field(None, description="The page number of the extraction")
+    external_link: str | None = Field(None, description="A link to the extraction")
+    data: dict | None = Field(
+        None, description="Extra information about the extraction"
+    )
+
+    system: str = Field(
+        ...,
+        description="""
+            The name of the system used.
+        """,
+    )
+    system_version: str = Field(
+        ...,
+        description="""
+            The version of the system used.
+        """,
+    )

cdr_schemas/feature_results.py

@@ -38,8 +38,18 @@ class FeatureResults(BaseModel):
             A list of legend extractions with associated polygon feature results.
         """,
     )
-    cog_area_extractions: Optional[List[Area_Extraction]]
-    cog_metadata_extractions: Optional[List[CogMetaData]]
+    cog_area_extractions: Optional[List[Area_Extraction]] = Field(
+        ...,
+        description="""
+            Higher level extraction pulled off a cog - legend area, map area, ocr text area, etc.
+        """,
+    )
+    cog_metadata_extractions: Optional[List[CogMetaData]] = Field(
+        ...,
+        description="""
+            Metadata extractions pulled off a cog.
+        """,
+    )
     system: str = Field(
         ...,
         description="""

cdr_schemas/map.py

@@ -0,0 +1,37 @@
+from typing import Optional
+
+from pydantic import BaseModel, Field
+
+
+class MapProvenance(BaseModel):
+    """JSON model for Document Provenance"""
+
+    system_name: str = Field(..., description="Name of system storing map")
+    id: str = Field(None, description="The system ID of the map")
+    url: str = Field(None, description="URL of map at system storing map")
+
+
+class Map(BaseModel):
+    """JSON model for 'Map''"""
+
+    id: str = Field(..., description="The CDR ID of the Map")
+    provenance: Optional[list[MapProvenance]] = Field(
+        None, description="provenance list"
+    )
+    is_open: bool = Field(
+        ...,
+        description="Whether map is open or not.",
+    )
+
+    system: str = Field(
+        ...,
+        description="""
+            The name of the system used.
+        """,
+    )
+    system_version: str = Field(
+        ...,
+        description="""
+            The version of the system used.
+        """,
+    )

cdr_schemas/map_results.py

@@ -0,0 +1,32 @@
+from typing import List, Optional
+
+from pydantic import BaseModel, Field
+
+from cdr_schemas.feature_results import FeatureResults
+from cdr_schemas.georeference import GeoreferenceResults
+
+
+class MapResults(BaseModel):
+    """
+    All results for map.
+    """
+
+    cog_id: str = Field(
+        ...,
+        description="""
+            Cog id.
+        """,
+    )
+
+    georef_results: Optional[List[GeoreferenceResults]] = Field(
+        ...,
+        description="""
+            A list of georef results from systems.
+        """,
+    )
+    extraction_results: Optional[List[FeatureResults]] = Field(
+        ...,
+        description="""
+            A list of feature extraction results from systems.
+        """,
+    )

cdr_schemas/mineral.py

@@ -0,0 +1,162 @@
+from datetime import datetime
+from enum import Enum
+from typing import List, Optional
+
+from pydantic import BaseModel, Field
+
+from cdr_schemas.document import Document
+
+
+class Geometry(Enum):
+    Point = "Point"
+    Polygon = "Polygon"
+
+
+class ResourceReserveCategory(Enum):
+    INFERRED = "Inferred Mineral Resource"
+    INDICATED = "Indicated Mineral Resource"
+    MEASURED = "Measured Mineral Resource"
+    PROBABLE = "Probable Mineral Reserve"
+    PROVEN = "Proven Mineral Reserve"
+
+
+class GeologyInfo(BaseModel):
+    age: Optional[str] = Field(description="Age of the geologic unit or event")
+    unit_name: Optional[str] = Field(description="Name of the geologic unit")
+    description: Optional[str]
+    lithology: Optional[list[str]]
+    process: Optional[list[str]]
+    environment: Optional[list[str]]
+    comments: Optional[str]
+
+
+class Ore(BaseModel):
+    ore_unit: str = Field(
+        description="The unit in which ore quantity is measured, eg, metric tonnes"
+    )
+    ore_value: float = Field(description="The value of ore quantity")
+
+
+class DepositType(BaseModel):
+    name: str = Field(description="Deposit type name")
+    environment: str = Field(description="Deposit type environment")
+    group: str = Field(description="Deposit type group")
+
+
+class DepositTypeCandidate(BaseModel):
+    observed_name: str = Field(
+        description="Source dataset that the site info is retrieved from. e.g., MRDS"
+    )
+    normalized_uri: DepositType = Field(
+        description="The deposit type of an inventory item"
+    )
+    confidence: float = Field(description="Score deposit type of an inventory item")
+    source: str = Field(
+        description="Source of the classification (automated model version / SME / etc...)"
+    )
+
+
+class BoundingBox(BaseModel):
+    x_min: float
+    x_max: float
+    y_min: float
+    y_max: float
+
+
+class PageInfo(BaseModel):
+    page: int
+    bounding_box: Optional[BoundingBox] = Field(
+        description="Coordinates of the document where reference is found"
+    )
+
+
+class Reference(BaseModel):
+    document: Document
+    page_info: List[PageInfo] = Field(
+        description="List of pages and their respective bounding boxes where the reference is found"
+    )
+
+
+class EvidenceLayer(BaseModel):
+    name: str
+    relevance_score: float
+
+
+class MappableCriteria(BaseModel):
+    criteria: str
+    theoretical: Optional[str]
+    potential_dataset: Optional[list[EvidenceLayer]]
+    supporting_references: list[Reference]
+
+
+class MineralSystem(BaseModel):
+    deposit_type: list[DepositType]
+    source: list[MappableCriteria]
+    pathway: list[MappableCriteria]
+    trap: Optional[list[MappableCriteria]]
+    preservation: Optional[list[MappableCriteria]]
+    energy: Optional[list[MappableCriteria]]
+    outflow: Optional[list[MappableCriteria]]
+
+
+class Commodity(BaseModel):
+    name: str
+
+
+class Grade(BaseModel):
+    grade_unit: str = Field(
+        description="The unit in which grade is measured, eg, percent"
+    )
+    grade_value: float = Field(description="The value of grade")
+
+
+class MineralInventory(BaseModel):
+    commodity: Commodity = Field(description="The commodity of an inventory item")
+    observed_commodity: Optional[str] = Field(
+        description="The observed commodity in the source data (textual format)"
+    )
+    category: Optional[ResourceReserveCategory] = Field(
+        description="The category of an inventory item"
+    )
+    ore: Optional[Ore] = Field(description="The ore of an inventory item")
+    grade: Optional[Grade] = Field(description="The grade of an inventory item")
+    cutoff_grade: Optional[Grade] = Field(
+        description="The cutoff grade of the observed inventory item"
+    )
+    contained_metal: Optional[float] = Field(
+        description="The quantity of a contained metal in an inventory item"
+    )
+    reference: Reference = Field(description="The reference of an inventory item")
+    date: Optional[datetime] = Field(
+        description="When in the point of time mineral inventory valid"
+    )
+    zone: Optional[str] = Field(
+        description="zone of mineral site where inventory item was discovered"
+    )
+
+
+class LocationInfo(BaseModel):
+    location: Geometry = Field(
+        description="Type: Polygon or Point, value indicates the geolocation of the site"
+    )
+    crs: str = Field(
+        description="The Coordinate Reference System (CRS) of the location"
+    )
+    country: Optional[str] = Field(description="Country that the mine site resides in")
+    state_or_province: Optional[str] = Field(
+        description="State or province that the mine site resides in"
+    )
+
+
+class MineralSite(BaseModel):
+    source_id: str = Field(
+        description="Source dataset that the site info is retrieved from. e.g., MRDS"
+    )
+    record_id: str = Field(
+        description="Unique ID of the record that the info is retrieved from e.g., 10022920"
+    )
+    name: Optional[str] = Field(description="Name of the mine, e.g., Tungsten Jim")
+    mineral_inventory: list[MineralInventory]
+    location_info: LocationInfo
+    geology_info: Optional[GeologyInfo]
+    deposit_type_candidate: list[DepositTypeCandidate]

dev/docs.py

@@ -5,13 +5,18 @@
 from jinja2 import Template
 from pydantic_mermaid import MermaidGenerator
 
+import cdr_schemas.area_extraction
+import cdr_schemas.document
 import cdr_schemas.events
 import cdr_schemas.feature_results
 import cdr_schemas.features.line_features
 import cdr_schemas.features.point_features
 import cdr_schemas.features.polygon_features
 import cdr_schemas.georeference
+import cdr_schemas.map
+import cdr_schemas.map_results
 import cdr_schemas.metadata
+import cdr_schemas.mineral
 
 
 @dataclass
@@ -47,13 +52,18 @@ def run():
     template = Template(Path("docs/schemas.md.j2").read_text())
 
     modules = [
+        Module(title="area extraction", ref=cdr_schemas.area_extraction),
         Module(title="georeference", ref=cdr_schemas.georeference),
         Module(title="metadata", ref=cdr_schemas.metadata),
         Module(title="feature results", ref=cdr_schemas.feature_results),
         Module(title="point feature", ref=cdr_schemas.features.point_features),
         Module(title="line feature", ref=cdr_schemas.features.line_features),
         Module(title="polygon feature", ref=cdr_schemas.features.polygon_features),
         Module(title="cog metadata", ref=cdr_schemas.metadata),
+        Module(title="document", ref=cdr_schemas.document),
+        Module(title="mineral", ref=cdr_schemas.mineral),
+        Module(title="map results", ref=cdr_schemas.map_results),
+        Module(title="map", ref=cdr_schemas.map),
     ]
 
     for m in modules: