unstructured-ingest 0.3.13__py3-none-any.whl

unstructured_ingest/v2/processes/connectors/notion/types/blocks/bulleted_list_item.py

@@ -0,0 +1,31 @@
+# https://developers.notion.com/reference/block#bulleted-list-item
+from dataclasses import dataclass, field
+from typing import List, Optional
+
+from htmlBuilder.tags import HtmlTag, Li
+
+from unstructured_ingest.connector.notion.interfaces import BlockBase
+from unstructured_ingest.connector.notion.types.rich_text import RichText
+
+
+@dataclass
+class BulletedListItem(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", [])
+        return cls(
+            color=data["color"],
+            children=data.get("children", []),
+            rich_text=[RichText.from_dict(rt) for rt in rich_text],
+        )
+
+    def get_html(self) -> Optional[HtmlTag]:
+        return Li([], [rt.get_html() for rt in self.rich_text])

unstructured_ingest/v2/processes/connectors/notion/types/blocks/callout.py

@@ -0,0 +1,94 @@
+# https://developers.notion.com/reference/block#callout
+from dataclasses import dataclass, field
+from typing import List, Optional, Union
+
+from htmlBuilder.attributes import Href, Style
+from htmlBuilder.tags import A, Div, HtmlTag, P
+
+from unstructured_ingest.connector.notion.interfaces import (
+    BlockBase,
+    FromJSONMixin,
+    GetHTMLMixin,
+)
+from unstructured_ingest.connector.notion.types.rich_text import RichText
+
+
+@dataclass
+class EmojiIcon(FromJSONMixin, GetHTMLMixin):
+    emoji: str
+    type: str = "emoji"
+
+    @classmethod
+    def from_dict(cls, data: dict):
+        return cls(**data)
+
+    def get_html(self) -> Optional[HtmlTag]:
+        return P([], self.emoji)
+
+
+@dataclass
+class ExternalIconContent(FromJSONMixin):
+    url: str
+
+    @classmethod
+    def from_dict(cls, data: dict):
+        return cls(**data)
+
+
+@dataclass
+class ExternalIcon(FromJSONMixin, GetHTMLMixin):
+    external: ExternalIconContent
+    type: str = "external"
+
+    @classmethod
+    def from_dict(cls, data: dict):
+        return cls(external=ExternalIconContent.from_dict(data=data.pop("external")), **data)
+
+    def get_html(self) -> Optional[HtmlTag]:
+        if self.external:
+            return A([Href(self.external.url)], [self.external.url])
+        else:
+            return None
+
+
+class Icon(FromJSONMixin):
+    @classmethod
+    def from_dict(cls, data: dict) -> Union[EmojiIcon, ExternalIcon]:
+        t = data.get("type")
+        if t == "emoji":
+            return EmojiIcon.from_dict(data)
+        elif t == "external":
+            return ExternalIcon.from_dict(data)
+        else:
+            raise ValueError(f"Unexpected icon type: {t} ({data})")
+
+
+@dataclass
+class Callout(BlockBase):
+    color: str
+    icon: Optional[Union[EmojiIcon, ExternalIcon]] = None
+    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", [])
+        return cls(
+            color=data["color"],
+            icon=Icon.from_dict(data.pop("icon")),
+            rich_text=[RichText.from_dict(rt) for rt in rich_text],
+        )
+
+    def get_html(self) -> Optional[HtmlTag]:
+        elements = []
+        if self.icon and self.icon.get_html():
+            elements.append(self.icon.get_html())
+        if self.rich_text:
+            elements.extend([rt.get_html() for rt in self.rich_text])
+        attributes = []
+        if self.color:
+            attributes.append(Style(f"color:{self.color}"))
+        return Div(attributes, elements)

unstructured_ingest/v2/processes/connectors/notion/types/blocks/child_database.py

@@ -0,0 +1,23 @@
+# https://developers.notion.com/reference/block#child-database
+from dataclasses import dataclass
+from typing import Optional
+
+from htmlBuilder.tags import HtmlTag, P
+
+from unstructured_ingest.connector.notion.interfaces import BlockBase
+
+
+@dataclass
+class ChildDatabase(BlockBase):
+    title: str
+
+    @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 P([], self.title)

unstructured_ingest/v2/processes/connectors/notion/types/blocks/child_page.py

@@ -0,0 +1,23 @@
+# https://developers.notion.com/reference/block#child-page
+from dataclasses import dataclass
+from typing import Optional
+
+from htmlBuilder.tags import HtmlTag, P
+
+from unstructured_ingest.connector.notion.interfaces import BlockBase, GetHTMLMixin
+
+
+@dataclass
+class ChildPage(BlockBase, GetHTMLMixin):
+    title: str
+
+    @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 P([], self.title)

unstructured_ingest/v2/processes/connectors/notion/types/blocks/code.py

@@ -0,0 +1,43 @@
+# https://developers.notion.com/reference/block#code
+from dataclasses import dataclass, field
+from typing import List, Optional
+
+from htmlBuilder.tags import Br, Div, HtmlTag
+from htmlBuilder.tags import Code as HtmlCode
+
+from unstructured_ingest.connector.notion.interfaces import BlockBase
+from unstructured_ingest.connector.notion.types.rich_text import RichText
+
+
+@dataclass
+class Code(BlockBase):
+    language: str
+    rich_text: List[RichText] = field(default_factory=list)
+    caption: 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", [])
+        caption = data.pop("caption", [])
+        return cls(
+            language=data["language"],
+            rich_text=[RichText.from_dict(rt) for rt in rich_text],
+            caption=[RichText.from_dict(c) for c in caption],
+        )
+
+    def get_html(self) -> Optional[HtmlTag]:
+        texts = []
+        if self.rich_text:
+            texts.append(HtmlCode([], [rt.get_html() for rt in self.rich_text]))
+        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/v2/processes/connectors/notion/types/blocks/column_list.py

@@ -0,0 +1,35 @@
+# https://developers.notion.com/reference/block#column-list-and-column
+from dataclasses import dataclass
+from typing import Optional
+
+from htmlBuilder.tags import HtmlTag
+
+from unstructured_ingest.connector.notion.interfaces import BlockBase
+
+
+@dataclass
+class ColumnList(BlockBase):
+    @staticmethod
+    def can_have_children() -> bool:
+        return True
+
+    @classmethod
+    def from_dict(cls, data: dict):
+        return cls()
+
+    def get_html(self) -> Optional[HtmlTag]:
+        return None
+
+
+@dataclass
+class Column(BlockBase):
+    @staticmethod
+    def can_have_children() -> bool:
+        return True
+
+    @classmethod
+    def from_dict(cls, data: dict):
+        return cls()
+
+    def get_html(self) -> Optional[HtmlTag]:
+        return None

unstructured_ingest/v2/processes/connectors/notion/types/blocks/divider.py

@@ -0,0 +1,22 @@
+# https://developers.notion.com/reference/block#divider
+from dataclasses import dataclass
+from typing import Optional
+
+from htmlBuilder.attributes import Style
+from htmlBuilder.tags import Hr, HtmlTag
+
+from unstructured_ingest.connector.notion.interfaces import BlockBase
+
+
+@dataclass
+class Divider(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 Hr([Style("border-top: 3px solid #bbb")])

unstructured_ingest/v2/processes/connectors/notion/types/blocks/embed.py

@@ -0,0 +1,36 @@
+# https://developers.notion.com/reference/block#embed
+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.connector.notion.interfaces import BlockBase
+from unstructured_ingest.connector.notion.types.rich_text import RichText
+
+
+@dataclass
+class Embed(BlockBase):
+    url: str
+    caption: List[RichText] = field(default_factory=list)
+
+    @staticmethod
+    def can_have_children() -> bool:
+        return False
+
+    @classmethod
+    def from_dict(cls, data: dict):
+        return cls(caption=[RichText.from_dict(d) for d in data.pop("caption", [])], **data)
+
+    def get_html(self) -> Optional[HtmlTag]:
+        texts = []
+        if self.url:
+            texts.append(A([Href(self.url)], self.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/v2/processes/connectors/notion/types/blocks/equation.py

@@ -0,0 +1,23 @@
+# https://developers.notion.com/reference/block#equation
+from dataclasses import dataclass
+from typing import Optional
+
+from htmlBuilder.tags import Div, HtmlTag
+
+from unstructured_ingest.connector.notion.interfaces import BlockBase
+
+
+@dataclass
+class Equation(BlockBase):
+    expression: 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 Div([], self.expression)

unstructured_ingest/v2/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.connector.notion.interfaces import BlockBase
+from unstructured_ingest.connector.notion.types.file import External
+from unstructured_ingest.connector.notion.types.file import File as FileContent
+from unstructured_ingest.connector.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/v2/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.connector.notion.interfaces import BlockBase
+from unstructured_ingest.connector.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/v2/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.connector.notion.interfaces import BlockBase
+from unstructured_ingest.connector.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/v2/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.connector.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/v2/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.connector.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/v2/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.connector.notion.interfaces import BlockBase
+from unstructured_ingest.connector.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/v2/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.connector.notion.interfaces import BlockBase
+from unstructured_ingest.connector.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/v2/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.connector.notion.interfaces import BlockBase
+from unstructured_ingest.connector.notion.types.file import External, File
+from unstructured_ingest.connector.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/v2/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.connector.notion.interfaces import BlockBase
+from unstructured_ingest.connector.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)