unstructured-ingest 1.2.32__py3-none-any.whl

unstructured_ingest/processes/connectors/notion/types/blocks/file.py

@@ -0,0 +1,49 @@
+# https://developers.notion.com/reference/block#file
+from dataclasses import dataclass, field
+from typing import List, Optional
+
+from htmlBuilder.attributes import Href
+from htmlBuilder.tags import A, Br, Div, HtmlTag
+
+from unstructured_ingest.processes.connectors.notion.interfaces import BlockBase
+from unstructured_ingest.processes.connectors.notion.types.file import External
+from unstructured_ingest.processes.connectors.notion.types.file import File as FileContent
+from unstructured_ingest.processes.connectors.notion.types.rich_text import RichText
+
+
+@dataclass
+class File(BlockBase):
+    type: str
+    external: Optional[External] = None
+    file: Optional[FileContent] = None
+    caption: List[RichText] = field(default_factory=list)
+
+    @staticmethod
+    def can_have_children() -> bool:
+        return False
+
+    @classmethod
+    def from_dict(cls, data: dict):
+        caption = [RichText.from_dict(rt) for rt in data.pop("caption", [])]
+        t = data["type"]
+        file = cls(type=t, caption=caption)
+        if t == "external":
+            file.external = External.from_dict(data["external"])
+        elif t == "file":
+            file.file = FileContent.from_dict(data["file"])
+        return file
+
+    def get_html(self) -> Optional[HtmlTag]:
+        texts = []
+        if self.file:
+            texts.append(A([Href(self.file.url)], self.file.url))
+        if self.external:
+            texts.append(A([Href(self.external.url)], self.external.url))
+        if self.caption:
+            texts.append(Div([], [rt.get_html() for rt in self.caption]))
+        if not texts:
+            return None
+        joined = [Br()] * (len(texts) * 2 - 1)
+        joined[0::2] = texts
+
+        return Div([], joined)

unstructured_ingest/processes/connectors/notion/types/blocks/heading.py

@@ -0,0 +1,37 @@
+# https://developers.notion.com/reference/block#headings
+from dataclasses import dataclass, field
+from typing import List, Optional
+
+from htmlBuilder.attributes import Style
+from htmlBuilder.tags import Div, HtmlTag
+
+from unstructured_ingest.processes.connectors.notion.interfaces import BlockBase
+from unstructured_ingest.processes.connectors.notion.types.rich_text import RichText
+
+
+@dataclass
+class Heading(BlockBase):
+    color: str
+    is_toggleable: bool
+    rich_text: List[RichText] = field(default_factory=list)
+
+    @staticmethod
+    def can_have_children() -> bool:
+        return False
+
+    @classmethod
+    def from_dict(cls, data: dict):
+        rich_text = data.pop("rich_text", [])
+        heading = cls(**data)
+        heading.rich_text = [RichText.from_dict(rt) for rt in rich_text]
+        return heading
+
+    def get_html(self) -> Optional[HtmlTag]:
+        if not self.rich_text:
+            return None
+
+        texts = [rt.get_html() for rt in self.rich_text]
+        attributes = []
+        if self.color and self.color != "default":
+            attributes.append(Style(f"color: {self.color}"))
+        return Div(attributes, texts)

unstructured_ingest/processes/connectors/notion/types/blocks/image.py

@@ -0,0 +1,21 @@
+# https://developers.notion.com/reference/block#image
+from typing import Optional
+
+from htmlBuilder.attributes import Src
+from htmlBuilder.tags import HtmlTag, Img
+
+from unstructured_ingest.processes.connectors.notion.interfaces import BlockBase
+from unstructured_ingest.processes.connectors.notion.types.file import FileObject
+
+
+class Image(BlockBase, FileObject):
+    @staticmethod
+    def can_have_children() -> bool:
+        return False
+
+    def get_html(self) -> Optional[HtmlTag]:
+        if self.external:
+            return Img([Src(self.external.url)], [])
+        if self.file:
+            return Img([Src(self.file.url)], [])
+        return None

unstructured_ingest/processes/connectors/notion/types/blocks/link_preview.py

@@ -0,0 +1,24 @@
+# https://developers.notion.com/reference/block#link-preview
+from dataclasses import dataclass
+from typing import Optional
+
+from htmlBuilder.attributes import Href
+from htmlBuilder.tags import A, HtmlTag
+
+from unstructured_ingest.processes.connectors.notion.interfaces import BlockBase
+
+
+@dataclass
+class LinkPreview(BlockBase):
+    url: str
+
+    @staticmethod
+    def can_have_children() -> bool:
+        return False
+
+    @classmethod
+    def from_dict(cls, data: dict):
+        return cls(**data)
+
+    def get_html(self) -> Optional[HtmlTag]:
+        return A([Href(self.url)], self.url)

unstructured_ingest/processes/connectors/notion/types/blocks/link_to_page.py

@@ -0,0 +1,29 @@
+# https://developers.notion.com/reference/block#link-to-page
+from dataclasses import dataclass
+from typing import Optional
+
+from htmlBuilder.tags import Div, HtmlTag
+
+from unstructured_ingest.processes.connectors.notion.interfaces import BlockBase
+
+
+@dataclass
+class LinkToPage(BlockBase):
+    type: str
+    page_id: Optional[str] = None
+    database_id: Optional[str] = None
+
+    @staticmethod
+    def can_have_children() -> bool:
+        return False
+
+    @classmethod
+    def from_dict(cls, data: dict):
+        return cls(**data)
+
+    def get_html(self) -> Optional[HtmlTag]:
+        if page_id := self.page_id:
+            return Div([], page_id)
+        if database_id := self.database_id:
+            return Div([], database_id)
+        return None

unstructured_ingest/processes/connectors/notion/types/blocks/numbered_list.py

@@ -0,0 +1,29 @@
+# https://developers.notion.com/reference/block#numbered-list-item
+from dataclasses import dataclass, field
+from typing import List, Optional
+
+from htmlBuilder.tags import HtmlTag, Li
+
+from unstructured_ingest.processes.connectors.notion.interfaces import BlockBase
+from unstructured_ingest.processes.connectors.notion.types.rich_text import RichText
+
+
+@dataclass
+class NumberedListItem(BlockBase):
+    color: str
+    children: List[dict] = field(default_factory=list)
+    rich_text: List[RichText] = field(default_factory=list)
+
+    @staticmethod
+    def can_have_children() -> bool:
+        return True
+
+    @classmethod
+    def from_dict(cls, data: dict):
+        rich_text = data.pop("rich_text", [])
+        numbered_list = cls(**data)
+        numbered_list.rich_text = [RichText.from_dict(rt) for rt in rich_text]
+        return numbered_list
+
+    def get_html(self) -> Optional[HtmlTag]:
+        return Li([], [rt.get_html() for rt in self.rich_text])

unstructured_ingest/processes/connectors/notion/types/blocks/paragraph.py

@@ -0,0 +1,31 @@
+# https://developers.notion.com/reference/block#paragraph
+from dataclasses import dataclass, field
+from typing import List, Optional
+
+from htmlBuilder.tags import Br, Div, HtmlTag
+
+from unstructured_ingest.processes.connectors.notion.interfaces import BlockBase
+from unstructured_ingest.processes.connectors.notion.types.rich_text import RichText
+
+
+@dataclass
+class Paragraph(BlockBase):
+    color: str
+    children: List[dict] = field(default_factory=list)
+    rich_text: List[RichText] = field(default_factory=list)
+
+    @staticmethod
+    def can_have_children() -> bool:
+        return True
+
+    @classmethod
+    def from_dict(cls, data: dict):
+        rich_text = data.pop("rich_text", [])
+        paragraph = cls(**data)
+        paragraph.rich_text = [RichText.from_dict(rt) for rt in rich_text]
+        return paragraph
+
+    def get_html(self) -> Optional[HtmlTag]:
+        if not self.rich_text:
+            return Br()
+        return Div([], [rt.get_html() for rt in self.rich_text])

unstructured_ingest/processes/connectors/notion/types/blocks/pdf.py

@@ -0,0 +1,49 @@
+# https://developers.notion.com/reference/block#pdf
+from dataclasses import dataclass, field
+from typing import List, Optional
+
+from htmlBuilder.attributes import Href
+from htmlBuilder.tags import A, Br, Div, HtmlTag
+
+from unstructured_ingest.processes.connectors.notion.interfaces import BlockBase
+from unstructured_ingest.processes.connectors.notion.types.file import External, File
+from unstructured_ingest.processes.connectors.notion.types.rich_text import RichText
+
+
+@dataclass
+class PDF(BlockBase):
+    type: str
+    caption: List[RichText] = field(default_factory=list)
+    external: Optional[External] = None
+    file: Optional[File] = None
+
+    @staticmethod
+    def can_have_children() -> bool:
+        return False
+
+    @classmethod
+    def from_dict(cls, data: dict):
+        caption = data.pop("caption", [])
+        t = data["type"]
+        paragraph = cls(type=t)
+        paragraph.caption = [RichText.from_dict(c) for c in caption]
+        if t == "external":
+            paragraph.external = External.from_dict(data["external"])
+        elif t == "file":
+            paragraph.file = File.from_dict(data["file"])
+        return paragraph
+
+    def get_html(self) -> Optional[HtmlTag]:
+        texts = []
+        if self.external:
+            texts.append(A([Href(self.external.url)], self.external.url))
+        if self.file:
+            texts.append(A([Href(self.file.url)], self.file.url))
+        if self.caption:
+            texts.append(Div([], [rt.get_html() for rt in self.caption]))
+        if not texts:
+            return None
+        joined = [Br()] * (len(texts) * 2 - 1)
+        joined[0::2] = texts
+
+        return Div([], joined)

unstructured_ingest/processes/connectors/notion/types/blocks/quote.py

@@ -0,0 +1,37 @@
+# https://developers.notion.com/reference/block#quote
+from dataclasses import dataclass, field
+from typing import List, Optional
+
+from htmlBuilder.attributes import Style
+from htmlBuilder.tags import Div, HtmlTag
+
+from unstructured_ingest.processes.connectors.notion.interfaces import BlockBase
+from unstructured_ingest.processes.connectors.notion.types.rich_text import RichText
+
+
+@dataclass
+class Quote(BlockBase):
+    color: str
+    children: List[dict] = field(default_factory=list)
+    rich_text: List[RichText] = field(default_factory=list)
+
+    @staticmethod
+    def can_have_children() -> bool:
+        return True
+
+    @classmethod
+    def from_dict(cls, data: dict):
+        rich_text = data.pop("rich_text", [])
+        quote = cls(**data)
+        quote.rich_text = [RichText.from_dict(rt) for rt in rich_text]
+        return quote
+
+    def get_html(self) -> Optional[HtmlTag]:
+        if not self.rich_text:
+            return None
+
+        texts = [rt.get_html() for rt in self.rich_text]
+        attributes = []
+        if self.color and self.color != "default":
+            attributes.append(Style(f"color: {self.color}"))
+        return Div(attributes, texts)

unstructured_ingest/processes/connectors/notion/types/blocks/synced_block.py

@@ -0,0 +1,109 @@
+# https://developers.notion.com/reference/block#synced-block
+from dataclasses import dataclass, field
+from typing import List, Optional
+
+from htmlBuilder.tags import HtmlTag
+
+from unstructured_ingest.processes.connectors.notion.interfaces import BlockBase
+
+
+@dataclass
+class OriginalSyncedBlock(BlockBase):
+    synced_from: Optional[str] = None
+    children: List[dict] = field(default_factory=list)
+
+    @staticmethod
+    def can_have_children() -> bool:
+        return True
+
+    @classmethod
+    def from_dict(cls, data: dict):
+        """Create OriginalSyncedBlock from dictionary data.
+
+        Original blocks contain children content.
+        """
+        if "children" not in data:
+            raise ValueError(f"OriginalSyncedBlock data missing 'children': {data}")
+        return cls(children=data["children"])
+
+    def get_html(self) -> Optional[HtmlTag]:
+        return None
+
+
+@dataclass
+class DuplicateSyncedBlock(BlockBase):
+    type: str
+    block_id: str
+
+    @staticmethod
+    def can_have_children() -> bool:
+        """Check if duplicate synced blocks can have children.
+
+        Duplicate blocks themselves don't have children directly fetched here,
+        but they represent content that does, so Notion API might report has_children=True
+        on the parent block object. The actual children are fetched from the original block.
+        """
+        return True
+
+    @classmethod
+    def from_dict(cls, data: dict):
+        """Create DuplicateSyncedBlock from dictionary data.
+
+        Duplicate blocks contain a 'synced_from' reference.
+        """
+        synced_from_data = data.get("synced_from")
+        if not synced_from_data or not isinstance(synced_from_data, dict):
+            raise ValueError(f"Invalid data structure for DuplicateSyncedBlock: {data}")
+        # Ensure required keys are present in the nested dictionary
+        if "type" not in synced_from_data or "block_id" not in synced_from_data:
+            raise ValueError(
+                f"Missing 'type' or 'block_id' in synced_from data: {synced_from_data}"
+            )
+        return cls(type=synced_from_data["type"], block_id=synced_from_data["block_id"])
+
+    def get_html(self) -> Optional[HtmlTag]:
+        """Get HTML representation of the duplicate synced block.
+
+        HTML representation might need fetching the original block's content,
+        which is outside the scope of this simple data class.
+        """
+        return None
+
+
+class SyncBlock(BlockBase):
+    @staticmethod
+    def can_have_children() -> bool:
+        """Check if synced blocks can have children.
+
+        Synced blocks (both original and duplicate) can conceptually have children.
+        """
+        return True
+
+    @classmethod
+    def from_dict(cls, data: dict):
+        """Create appropriate SyncedBlock subclass from dictionary data.
+
+        Determine if it's a duplicate (has 'synced_from') or original (has 'children').
+        """
+        if data.get("synced_from") is not None:
+            # It's a duplicate block containing a reference
+            return DuplicateSyncedBlock.from_dict(data)
+        elif "children" in data:
+            # It's an original block containing children
+            return OriginalSyncedBlock.from_dict(data)
+        else:
+            # Handle cases where neither 'synced_from' nor 'children' are present.
+            # Notion API might return this for an empty original synced block.
+            # Let's treat it as an empty OriginalSyncedBlock.
+            # If this assumption is wrong, errors might occur later.
+            # Consider logging a warning here if strictness is needed.
+            return OriginalSyncedBlock(children=[])
+
+    def get_html(self) -> Optional[HtmlTag]:
+        """Get HTML representation of the synced block.
+
+        The specific instance returned by from_dict (Original or Duplicate)
+        will handle its own get_html logic.
+        This method on the base SyncBlock might not be directly called.
+        """
+        return None

unstructured_ingest/processes/connectors/notion/types/blocks/table.py

@@ -0,0 +1,60 @@
+# https://developers.notion.com/reference/block#table
+from dataclasses import dataclass, field
+from typing import List, Optional
+
+from htmlBuilder.tags import HtmlTag, Td, Th, Tr
+
+from unstructured_ingest.processes.connectors.notion.interfaces import BlockBase, FromJSONMixin
+from unstructured_ingest.processes.connectors.notion.types.rich_text import RichText
+
+
+@dataclass
+class Table(BlockBase):
+    table_width: int
+    has_column_header: bool
+    has_row_header: bool
+
+    @staticmethod
+    def can_have_children() -> bool:
+        return True
+
+    @classmethod
+    def from_dict(cls, data: dict):
+        return cls(**data)
+
+    def get_html(self) -> Optional[HtmlTag]:
+        return None
+
+
+@dataclass
+class TableCell(FromJSONMixin):
+    rich_texts: List[RichText]
+
+    @classmethod
+    def from_dict(cls, data: dict):
+        return cls(rich_texts=[RichText.from_dict(rt) for rt in data.pop("rich_texts", [])])
+
+    def get_html(self, is_header: bool) -> Optional[HtmlTag]:
+        if is_header:
+            return Th([], [rt.get_html() for rt in self.rich_texts])
+        else:
+            return Td([], [rt.get_html() for rt in self.rich_texts])
+
+
+# https://developers.notion.com/reference/block#table-rows
+@dataclass
+class TableRow(BlockBase):
+    is_header: bool = False
+    cells: List[TableCell] = field(default_factory=list)
+
+    @classmethod
+    def from_dict(cls, data: dict):
+        cells = data.get("cells", [])
+        return cls(cells=[TableCell.from_dict({"rich_texts": c}) for c in cells])
+
+    @staticmethod
+    def can_have_children() -> bool:
+        return False
+
+    def get_html(self) -> Optional[HtmlTag]:
+        return Tr([], [cell.get_html(is_header=self.is_header) for cell in self.cells])

unstructured_ingest/processes/connectors/notion/types/blocks/table_of_contents.py

@@ -0,0 +1,23 @@
+# https://developers.notion.com/reference/block#table-of-contents
+from dataclasses import dataclass
+from typing import Optional
+
+from htmlBuilder.tags import HtmlTag
+
+from unstructured_ingest.processes.connectors.notion.interfaces import BlockBase
+
+
+@dataclass
+class TableOfContents(BlockBase):
+    color: str
+
+    @staticmethod
+    def can_have_children() -> bool:
+        return False
+
+    @classmethod
+    def from_dict(cls, data: dict):
+        return cls(**data)
+
+    def get_html(self) -> Optional[HtmlTag]:
+        return None

unstructured_ingest/processes/connectors/notion/types/blocks/template.py

@@ -0,0 +1,30 @@
+# https://developers.notion.com/reference/block#template
+from dataclasses import dataclass, field
+from typing import List, Optional
+
+from htmlBuilder.tags import Div, HtmlTag
+
+from unstructured_ingest.processes.connectors.notion.interfaces import BlockBase
+from unstructured_ingest.processes.connectors.notion.types.rich_text import RichText
+
+
+@dataclass
+class Template(BlockBase):
+    children: List[dict] = field(default_factory=list)
+    rich_text: List[RichText] = field(default_factory=list)
+
+    @staticmethod
+    def can_have_children() -> bool:
+        return True
+
+    @classmethod
+    def from_dict(cls, data: dict):
+        rich_text = data.pop("rich_text", [])
+        template = cls(**data)
+        template.rich_text = [RichText.from_dict(rt) for rt in rich_text]
+        return template
+
+    def get_html(self) -> Optional[HtmlTag]:
+        if not self.rich_text:
+            return None
+        return Div([], [rt.get_html() for rt in self.rich_text])

unstructured_ingest/processes/connectors/notion/types/blocks/todo.py

@@ -0,0 +1,42 @@
+# https://developers.notion.com/reference/block#to-do
+from dataclasses import dataclass, field
+from typing import List, Optional
+
+from htmlBuilder.attributes import Checked, Style, Type
+from htmlBuilder.tags import Div, HtmlTag, Input
+
+from unstructured_ingest.processes.connectors.notion.interfaces import BlockBase
+from unstructured_ingest.processes.connectors.notion.types.rich_text import RichText
+
+
+@dataclass
+class ToDo(BlockBase):
+    color: str
+    checked: bool = False
+    rich_text: List[RichText] = field(default_factory=list)
+
+    @staticmethod
+    def can_have_children() -> bool:
+        return True
+
+    @classmethod
+    def from_dict(cls, data: dict):
+        rich_text = data.pop("rich_text", [])
+        todo = cls(**data)
+        todo.rich_text = [RichText.from_dict(rt) for rt in rich_text]
+        return todo
+
+    def get_html(self) -> Optional[HtmlTag]:
+        if not self.rich_text:
+            return None
+
+        elements = []
+        check_input_attributes = [Type("checkbox")]
+        if self.checked:
+            check_input_attributes.append(Checked(""))
+        elements.append(Input(check_input_attributes))
+        elements.extend([rt.get_html() for rt in self.rich_text])
+        attributes = []
+        if self.color and self.color != "default":
+            attributes.append(Style(f"color: {self.color}"))
+        return Div(attributes, elements)

unstructured_ingest/processes/connectors/notion/types/blocks/toggle.py

@@ -0,0 +1,37 @@
+# https://developers.notion.com/reference/block#toggle-blocks
+from dataclasses import dataclass, field
+from typing import List, Optional
+
+from htmlBuilder.attributes import Style
+from htmlBuilder.tags import Div, HtmlTag
+
+from unstructured_ingest.processes.connectors.notion.interfaces import BlockBase
+from unstructured_ingest.processes.connectors.notion.types.rich_text import RichText
+
+
+@dataclass
+class Toggle(BlockBase):
+    color: str
+    children: List[dict] = field(default_factory=list)
+    rich_text: List[RichText] = field(default_factory=list)
+
+    @staticmethod
+    def can_have_children() -> bool:
+        return True
+
+    @classmethod
+    def from_dict(cls, data: dict):
+        rich_text = data.pop("rich_text", [])
+        toggle = cls(**data)
+        toggle.rich_text = [RichText.from_dict(rt) for rt in rich_text]
+        return toggle
+
+    def get_html(self) -> Optional[HtmlTag]:
+        if not self.rich_text:
+            return None
+
+        texts = [rt.get_html() for rt in self.rich_text]
+        attributes = []
+        if self.color and self.color != "default":
+            attributes.append(Style(f"color: {self.color}"))
+        return Div(attributes, texts)

unstructured_ingest/processes/connectors/notion/types/blocks/unsupported.py

@@ -0,0 +1,20 @@
+from dataclasses import dataclass
+from typing import Optional
+
+from htmlBuilder.tags import HtmlTag
+
+from unstructured_ingest.processes.connectors.notion.interfaces import BlockBase
+
+
+@dataclass
+class Unsupported(BlockBase):
+    @staticmethod
+    def can_have_children() -> bool:
+        return False
+
+    @classmethod
+    def from_dict(cls, data: dict):
+        return cls()
+
+    def get_html(self) -> Optional[HtmlTag]:
+        return None

unstructured_ingest/processes/connectors/notion/types/blocks/video.py

@@ -0,0 +1,22 @@
+# https://developers.notion.com/reference/block#image
+from typing import Optional
+
+from htmlBuilder.attributes import Src
+from htmlBuilder.tags import HtmlTag, Source
+from htmlBuilder.tags import Video as VideoHtml
+
+from unstructured_ingest.processes.connectors.notion.interfaces import BlockBase
+from unstructured_ingest.processes.connectors.notion.types.file import FileObject
+
+
+class Video(BlockBase, FileObject):
+    @staticmethod
+    def can_have_children() -> bool:
+        return False
+
+    def get_html(self) -> Optional[HtmlTag]:
+        if self.external:
+            return VideoHtml([], [Source([Src(self.external.url)], [self.external.url])])
+        if self.file:
+            return VideoHtml([], [Source([Src(self.file.url)], [self.file.url])])
+        return None