Home

Awesome

notion-py

Unofficial Python 3 client for Notion.so API v3.

Read more about Notion and Notion-py on Jamie's blog

Usage

Quickstart

Note: the latest version of notion-py requires Python 3.5 or greater.

pip install notion

from notion.client import NotionClient

# Obtain the `token_v2` value by inspecting your browser cookies on a logged-in (non-guest) session on Notion.so
client = NotionClient(token_v2="<token_v2>")

# Replace this URL with the URL of the page you want to edit
page = client.get_block("https://www.notion.so/myorg/Test-c0d20a71c0944985ae96e661ccc99821")

print("The old title is:", page.title)

# Note: You can use Markdown! We convert on-the-fly to Notion's internal formatted text data structure.
page.title = "The title has now changed, and has *live-updated* in the browser!"

Concepts and notes

Updating records

We keep a local cache of all data that passes through. When you reference an attribute on a Record, we first look to that cache to retrieve the value. If it doesn't find it, it retrieves it from the server. You can also manually refresh the data for a Record by calling the refresh method on it. By default (unless we instantiate NotionClient with monitor=False), we also subscribe to long-polling updates for any instantiated Record, so the local cache data for these Records should be automatically live-updated shortly after any data changes on the server. The long-polling happens in a background daemon thread.

Example: Traversing the block tree

for child in page.children:
    print(child.title)

print("Parent of {} is {}".format(page.id, page.parent.id))

Example: Adding a new node

from notion.block import TodoBlock

newchild = page.children.add_new(TodoBlock, title="Something to get done")
newchild.checked = True

Example: Deleting nodes

# soft-delete
page.remove()

# hard-delete
page.remove(permanently=True)

Example: Create an embedded content type (iframe, video, etc)

from notion.block import VideoBlock

video = page.children.add_new(VideoBlock, width=200)
# sets "property.source" to the URL, and "format.display_source" to the embedly-converted URL
video.set_source_url("https://www.youtube.com/watch?v=oHg5SJYRHA0")

Example: Create a new embedded collection view block

collection = client.get_collection(COLLECTION_ID) # get an existing collection
cvb = page.children.add_new(CollectionViewBlock, collection=collection)
view = cvb.views.add_new(view_type="table")

# Before the view can be browsed in Notion, 
# the filters and format options on the view should be set as desired.
# 
# for example:
#   view.set("query", ...)
#   view.set("format.board_groups", ...)
#   view.set("format.board_properties", ...)

Example: Moving blocks around

# move my block to after the video
my_block.move_to(video, "after")

# move my block to the end of otherblock's children
my_block.move_to(otherblock, "last-child")

# (you can also use "before" and "first-child")

Example: Subscribing to updates

(Note: Notion->Python automatic updating is currently broken and hence disabled by default; call my_block.refresh() to update, in the meantime, while monitoring is being fixed)

We can "watch" a Record so that we get a callback whenever it changes. Combined with the live-updating of records based on long-polling, this allows for a "reactive" design, where actions in our local application can be triggered in response to interactions with the Notion interface.


# define a callback (note: all arguments are optional, just include the ones you care about)
def my_callback(record, difference):
    print("The record's title is now:" record.title)
    print("Here's what was changed:")
    print(difference)

# move my block to after the video
my_block.add_callback(my_callback)

Example: Working with databases, aka "collections" (tables, boards, etc)

Here's how things fit together:

Note: For convenience, we automatically map the database "columns" (aka properties), based on the schema defined in the Collection, into getter/setter attributes on the CollectionRowBlock instances. The attribute name is a "slugified" version of the name of the column. So if you have a column named "Estimated value", you can read and write it via myrowblock.estimated_value. Some basic validation may be conducted, and it will be converted into the appropriate internal format. For columns of type "Person", we expect a User instance, or a list of them, and for a "Relation" we expect a singular/list of instances of a subclass of Block.

# Access a database using the URL of the database page or the inline block
cv = client.get_collection_view("https://www.notion.so/myorg/8511b9fc522249f79b90768b832599cc?v=8dee2a54f6b64cb296c83328adba78e1")

# List all the records with "Bob" in them
for row in cv.collection.get_rows(search="Bob"):
    print("We estimate the value of '{}' at {}".format(row.name, row.estimated_value))

# Add a new record
row = cv.collection.add_row()
row.name = "Just some data"
row.is_confirmed = True
row.estimated_value = 399
row.files = ["https://www.birdlife.org/sites/default/files/styles/1600/public/slide.jpg"]
row.person = client.current_user
row.tags = ["A", "C"]
row.where_to = "https://learningequality.org"

# Run a filtered/sorted query using a view's default parameters
result = cv.default_query().execute()
for row in result:
    print(row)

# Run an "aggregation" query
aggregations = [{
    "property": "estimated_value",
    "aggregator": "sum",
    "id": "total_value",
}]
result = cv.build_query(aggregate=aggregate_params).execute()
print("Total estimated value:", result.get_aggregate("total_value"))

# Run a "filtered" query (inspect network tab in browser for examples, on queryCollection calls)
filter_params = {
    "filters": [{
        "filter": {
            "value": {
                "type": "exact",
                "value": {"table": "notion_user", "id": client.current_user.id}
            },
            "operator": "person_contains"
        },
        "property": "assigned_to"
    }],
    "operator": "and"
}
result = cv.build_query(filter=filter_params).execute()
print("Things assigned to me:", result)

# Run a "sorted" query
sort_params = [{
    "direction": "descending",
    "property": "estimated_value",
}]
result = cv.build_query(sort=sort_params).execute()
print("Sorted results, showing most valuable first:", result)

Note: You can combine filter, aggregate, and sort. See more examples of queries by setting up complex views in Notion, and then inspecting the full query: cv.get("query2").

You can also see more examples in action in the smoke test runner. Run it using:

python run_smoke_test.py --page [YOUR_NOTION_PAGE_URL] --token [YOUR_NOTION_TOKEN_V2]

Example: Lock/Unlock A Page

from notion.client import NotionClient

# Obtain the `token_v2` value by inspecting your browser cookies on a logged-in session on Notion.so
client = NotionClient(token_v2="<token_v2>")

# Replace this URL with the URL of the page or database you want to edit
page = client.get_block("https://www.notion.so/myorg/Test-c0d20a71c0944985ae96e661ccc99821")

# The "locked" property is available on PageBlock and CollectionViewBlock objects
# Set it to True to lock the page/database
page.locked = True
# and False to unlock it again
page.locked = False

Example: Set the current user for multi-account user

from notion.client import NotionClient
client = NotionClient(token_v2="<token_v2>")

# The initial current_user of a multi-account user may be an unwanted user
print(client.current_user.email) # → not_the_desired@email.co.jp

# Set current_user to the desired user
client.set_user_by_email('desired@email.com')
print(client.current_user.email) # → desired@email.com

# You can also set the current_user by uid.
client.set_user_by_uid('<uid>')
print(client.current_user.email) # → desired@email.com

Quick plug: Learning Equality needs your support!

If you'd like to support notion-py development, please consider donating to my open-source nonprofit, Learning Equality, since when I'm not working on notion-py, it probably means I'm heads-down fundraising for our global education work (bringing resources like Khan Academy to communities with no Internet). COVID has further amplified needs, with over a billion kids stuck at home, and over half of them without the connectivity they need for distance learning. You can now also support our work via GitHub Sponsors!

Related Projects

TODO