airflow-provider-google-sheets 0.1.5

Apache Airflow provider for Google Sheets — read, write, and smart merge

airflow-provider-google-sheets

Apache Airflow provider for Google Sheets API v4. Read, write, and manage Google Sheets spreadsheets from Airflow DAGs.

---

AI Disclosure: This provider was developed with the assistance of Claude Code (Anthropic, model Claude Opus 4.6). The code, tests, and documentation were co-authored by a human developer and an LLM. Please evaluate the code quality on its own merits and make informed decisions about whether to use it in your projects.

---

Features

• Read data from Google Sheets with chunked streaming, schema-based type conversion, and CSV/JSON/XCom output
• Write data in three modes: overwrite, append, and smart merge (upsert by key)
• Smart merge — update, insert, and delete rows based on a key column with correct index recalculation
• Manage spreadsheets — create new spreadsheets and sheets
• Large datasets — streaming read/write without loading everything into memory
• Schema support — automatic type conversion (date, int, float, bool) on read and write
• Header processing — deduplication, Cyrillic transliteration, snake_case normalization

Installation

pip install airflow-provider-google-sheets

With Cyrillic header transliteration support:

pip install airflow-provider-google-sheets[transliterate]

Requirements

• Python >= 3.10
• Apache Airflow 2.x (>= 2.7, tested on 2.9.1; Airflow 3.x not tested)
• Google service account with Sheets API access

Connection Setup

1. Create a Google Cloud service account with Google Sheets API enabled.
2. Download the JSON key file.
3. In Airflow UI, create a connection with one of the supported configurations:

Option A: Standard Google Cloud connection (recommended)

Use this if you already have a google_cloud_platform connection configured in Airflow.

• Conn Id: google_cloud_default
• Conn Type: Google Cloud
• Keyfile Path: /path/to/service-account.json
• Scopes: https://www.googleapis.com/auth/spreadsheets (add more if needed)

Option B: Inline JSON key

• Conn Id: google_cloud_default
• Conn Type: Google Cloud or google_sheets
• Keyfile JSON: paste the full service account JSON

Option C: JSON in Extra field

• Conn Id: google_cloud_default
• Conn Type: google_sheets
• Extra: paste the full JSON key, or use {"keyfile_dict": <JSON key>}

The hook checks credentials in this order: key_path / keyfile_path (file on disk) → keyfile_dict (inline JSON) → raw Extra JSON.

Operators

GoogleSheetsReadOperator

Read data from a spreadsheet.

from airflow_provider_google_sheets.operators.read import GoogleSheetsReadOperator

# Basic read — returns list[dict] via XCom
read = GoogleSheetsReadOperator(
    task_id="read_sheets",
    spreadsheet_id="your-spreadsheet-id",
    sheet_name="Sheet1",
)

# Stream large sheet to CSV file (no memory accumulation)
read_csv = GoogleSheetsReadOperator(
    task_id="read_to_csv",
    spreadsheet_id="your-spreadsheet-id",
    output_type="csv",
    output_path="/tmp/export.csv",
    chunk_size=10000,
)

# Read with type conversion
read_typed = GoogleSheetsReadOperator(
    task_id="read_typed",
    spreadsheet_id="your-spreadsheet-id",
    schema={
        "date": {"type": "date", "format": "%Y-%m-%d"},
        "revenue": {"type": "float", "required": True},
        "quantity": {"type": "int"},
    },
)

Parameters:

Parameter	Type	Default	Description
gcp_conn_id	str	"google_cloud_default"	Airflow Connection ID
spreadsheet_id	str	—	Spreadsheet ID
sheet_name	str	None	Sheet name (None = first sheet)
cell_range	str	None	A1-notation range (None = entire sheet)
has_headers	bool	True	First row contains headers
transliterate_headers	bool	False	Transliterate Cyrillic to Latin
normalize_headers	bool	False	Normalize to snake_case
schema	dict	None	Column type schema
chunk_size	int	5000	Rows per API request
output_type	str	"xcom"	"xcom", "csv", or "json"
output_path	str	None	File path for csv/json output
max_xcom_rows	int	50000	Max rows for XCom output

GoogleSheetsWriteOperator

Write data to a spreadsheet.

from airflow_provider_google_sheets.operators.write import GoogleSheetsWriteOperator

# Overwrite with list[dict]
write = GoogleSheetsWriteOperator(
    task_id="write_sheets",
    spreadsheet_id="your-spreadsheet-id",
    sheet_name="Output",
    write_mode="overwrite",
    data=[{"date": "2024-01-01", "value": 100}],
)

# Append rows
append = GoogleSheetsWriteOperator(
    task_id="append_sheets",
    spreadsheet_id="your-spreadsheet-id",
    write_mode="append",
    data=[{"event": "login", "user": "alice"}],
)

# Smart merge by key
merge = GoogleSheetsWriteOperator(
    task_id="smart_merge",
    spreadsheet_id="your-spreadsheet-id",
    write_mode="smart_merge",
    merge_key="date",
    data=[
        {"date": "2024-01-01", "value": 110},  # update existing
        {"date": "2024-01-03", "value": 200},  # append new
    ],
)

Parameters:

Parameter	Type	Default	Description
gcp_conn_id	str	"google_cloud_default"	Airflow Connection ID
spreadsheet_id	str	—	Spreadsheet ID
sheet_name	str	None	Sheet name
cell_range	str	None	Target A1 range (overwrite mode)
write_mode	str	"overwrite"	"overwrite", "append", "smart_merge"
data	Any	None	Data: list[list], list[dict], or file path
data_xcom_task_id	str	None	Pull data from this task's XCom
data_xcom_key	str	"return_value"	XCom key
has_headers	bool	True	Data contains headers
write_headers	bool	True	Write header row (overwrite mode)
schema	dict	None	Schema for formatting values
batch_size	int	1000	Rows per API request
pause_between_batches	float	1.0	Seconds between batches
merge_key	str	None	Key column for smart_merge

Data input formats:

• list[dict] — headers auto-detected from keys
• list[list] — raw rows (set has_headers=True if first row is header)
• str — path to CSV or JSON file
• XCom — set data_xcom_task_id

Smart Merge Algorithm

Smart merge reads the key column from the sheet, compares with incoming data, and generates minimal operations:

1. Read the key column to build an index {key_value: [row_numbers]}
2. Compare each key: same count → update, more incoming → insert, fewer → delete, new key → append
3. Sort structural operations bottom-up (descending row number) to prevent index corruption
4. Execute inserts/deletes via batchUpdate, then recalculate row indices for value updates
5. Write values via batch_update_values for efficiency

GoogleSheetsCreateSpreadsheetOperator

from airflow_provider_google_sheets.operators.manage import GoogleSheetsCreateSpreadsheetOperator

create = GoogleSheetsCreateSpreadsheetOperator(
    task_id="create_spreadsheet",
    title="Monthly Report",
    sheet_titles=["Summary", "Details"],
)
# Returns spreadsheet_id via XCom

GoogleSheetsCreateSheetOperator

from airflow_provider_google_sheets.operators.manage import GoogleSheetsCreateSheetOperator

add_sheet = GoogleSheetsCreateSheetOperator(
    task_id="add_sheet",
    spreadsheet_id="your-spreadsheet-id",
    sheet_title="NewSheet",
)

Schema

Define column types for automatic conversion on read/write:

schema = {
    "date": {"type": "date", "format": "%Y-%m-%d", "required": True},
    "revenue": {"type": "float", "required": True},
    "quantity": {"type": "int"},
    "comment": {"type": "str"},
    "is_active": {"type": "bool"},
}

Supported types: str, int, float, date, datetime, bool

Examples

See the examples/ directory for complete DAG examples:

• example_read.py — reading with various configurations
• example_write.py — overwrite and append modes
• example_smart_merge.py — smart merge scenarios
• example_manage.py — creating spreadsheets and sheets
• example_sheets_to_bigquery.py — Google Sheets → BigQuery (overwrite, append, date-range update)
• example_bigquery_to_sheets.py — BigQuery → Google Sheets (overwrite, smart merge by date)

License

MIT License