Datalys2 Reporting Python API

Version 0.6.0

A Python library to build and compile interactive HTML reports using the Datalys2 Reporting framework.

Note: Compatible with dl2 version 0.4.1 https://github.com/kameronbrooks/datalys2-reporting

Installation

pip install dl2-reports

Quick Start

import pandas as pd
from dl2_reports import DL2Report, KPI, Table

# Create a report
report = DL2Report(title="My Report")

# Add data
df = pd.DataFrame({"A": [1, 2], "B": [3, 4]})
report.add_df("my_data", df, compress=True)

# Add a page and visuals (typed component API)
page = report.add_page("Overview")
page.add_row(
    KPI("my_data", value_column="A", title="Metric A"),
    Table("my_data", page_size=10),
)

# Save to HTML or show in Jupyter
report.save("report.html")
report.show()

The Typed Component API (v2)

Since 0.5.0 the recommended way to build reports is with typed component classes — one class per visual, imported from the package root:

KPI, Table, Card, Pie, Bar, Line, Area, Scatter, Checklist, Histogram, Heatmap, Gauge, Boxplot, Tabs, Link, ModalButton

plus typed shapes for structured props:

Threshold, SortSpec, TotalRow, TotalColumn, GaugeRange, AggregateColumn, Tab, ColumnFormat, ConditionalFormat

from dl2_reports import DL2Report, Line, Table, Tabs, Tab, Threshold, TotalRow, SortSpec

page.add_row(
    Line("sales", x_column="Month", y_columns=["Revenue"],
         threshold=Threshold(value=5000, mode="above")),
)
row = page.add_row()
row.add(Table("sales",
              id="orders-table",
              group_by="Region",
              default_sort=[SortSpec("Amount", "desc")],
              total_row=TotalRow(fns={"Units": "sum", "Amount": "avg"})))
row.add(Tabs(id="views", tabs=[
    Tab("Chart", children=[Line("sales", x_column="Month", y_columns=["Revenue"])]),
    Tab("Data",  children=[Table("sales")]),
]))

Why it's better:

Typos fail fast. Table("sales", pagesize=20) raises TypeError: unknown prop 'pagesize' (did you mean 'page_size'?) at construction — previously it serialized silently and the viewer ignored it.
Autocomplete and type checking work everywhere (the package ships py.typed).
extra={...} passes unmodeled viewer props through explicitly when you need forward compatibility.
compile() lints legacy calls too: props the viewer doesn't know are reported as [dl2] warnings with suggestions (report.compile(strict=True) turns them into errors).

The legacy row.add_kpi(...) helpers keep working unchanged — they now delegate to the component classes, and their unknown kwargs still pass through (flagged by the compile lint). add() returns the component, so chaining (.add_trend(), .get_value()) works as before.

Migrating existing scripts

A codemod ships with the package (comments and formatting preserved; requires pip install dl2-reports[migrate]):

python -m dl2_reports.migrate my_report.py            # dry run: shows a diff
python -m dl2_reports.migrate my_report.py --write    # apply
python -m dl2_reports.migrate notebooks/ --write      # directories & .ipynb work too

Data Compression

Always use compression for production reports. The Python API provides automatic gzip compression for your datasets, which significantly reduces file size and improves browser performance.

Why Compression Matters

Large datasets will cause severe performance issues or fail to load entirely without compression
Compressed reports load faster and consume less memory in the browser
File sizes can be reduced by 80-90% or more
The browser automatically decompresses data on-the-fly using the built-in DecompressionStream API

Using Compression

Report-Level Default

When creating a DL2Report, you can set the default compression behavior:

from dl2_reports import DL2Report

# Enable compression by default (recommended)
report = DL2Report(
    title="My Report",
    compress_visuals=True  # This is the default
)

Per-Dataset Control

Control compression for individual datasets using the compress parameter in add_df():

import pandas as pd
from dl2_reports import DL2Report

report = DL2Report(title="Sales Report")

# Compress large datasets (recommended for most data)
large_df = pd.read_csv("sales_data.csv")
report.add_df("salesData", large_df, compress=True)

# Small datasets can be uncompressed for easier debugging
small_df = pd.DataFrame({"kpi": [100]})
report.add_df("kpiData", small_df, compress=False)

How It Works

When you set compress=True, the Python API automatically:

Serializes your data to JSON
Compresses it using gzip
Encodes it as a Base64 string
Stores it in a separate <script> tag in the HTML
Adds the gc-compressed-data meta tag for automatic memory cleanup

The browser then decompresses the data when the report loads.

Best Practices

✅ Always compress in production - essential for performance and reliability
✅ Compress any dataset with more than a few rows - the overhead is minimal
❌ Only disable compression when:
- Debugging and you need to inspect the raw JSON in the HTML file
- Working with extremely small datasets (single-row KPI values) during development

Features

Jupyter Notebook Support

You can render reports directly inside Jupyter Notebooks (including VS Code and JupyterLab).

report.show(height=800): Displays the report in an iframe.
Automatic Rendering: Simply placing the report object at the end of a cell will render it automatically.

Requirements:

IPython must be installed in your environment.

Available Visuals

All visuals are added to a layout row using row.add_<type>(...).

Common Visual Keyword Arguments

All Layout.add_* visual helpers accept **kwargs which are passed through to the viewer as visual properties.

Common properties supported by the viewer include:

padding: number (px)
margin: number (px)
border: bool or CSS border string (e.g. "2px dashed #f59e0b"; CSS strings honored by the viewer since dl2 0.4.1)
shadow: bool or CSS box-shadow string (CSS strings honored by the viewer since dl2 0.4.1)
flex: number (flex grow)
modal_id: string (global modal id opened via the expand icon)

The Python API accepts these in snake_case; the compiled report JSON uses camelCase.

Layout Visual Helper APIs

Below are the current convenience helpers available on Layout (rows are layouts).

Visual Types (Quick Summary)

These are the visual types you can add via the Layout helpers:

kpi (via add_kpi)
table (via add_table)
card (via add_card)
pie (via add_pie)
clusteredBar / stackedBar (via add_bar(stacked=...))
scatter (via add_scatter)
line (via add_line)
area (via add_area)
checklist (via add_checklist)
histogram (via add_histogram)
heatmap (via add_heatmap)
boxplot (via add_boxplot)
modal (via add_modal_button)
tabs (via add_tabs, dl2 0.3+)
link (via add_link, dl2 0.4+)

You can also add any viewer-supported visual type directly using add_visual(type=..., ...).

Generic Visual

Use this when you want to pass through viewer props that don't have a dedicated helper yet.

Parameter	Type	Default	Description
`type`	`str`	(required)	Visual type (e.g., `'kpi'`, `'table'`, `'line'`, `'scatter'`).
`dataset_id`	`str \| None`	`None`	Dataset id to bind to this visual (required for most chart/data visuals).
`**kwargs`	`Any`	—	Additional visual properties (serialized to JSON). Common ones include `padding`, `margin`, `border`, `shadow`, `flex`, `modal_id`.

KPI

Parameter	Type	Default	Description
`dataset_id`	`str`	(required)	The dataset id.
`value_column`	`str \| int`	(required)	Column for the main KPI value.
`title`	`str \| None`	`None`	Optional KPI card title.
`comparison_column`	`str \| int \| None`	`None`	Column for the comparison value.
`comparison_row_index`	`int \| None`	`None`	Row index to use for comparison (supports negative indices). If not provided, the viewer uses the same row as `row_index`.
`comparison_text`	`str`	The comparison text to show alongside the comparison value. Ex. ("Last Month", "Yesterday", etc.)
`row_index`	`int \| None`	`None`	Row index to display (supports negative indices).
`format`	`str \| None`	`None`	`'number'`, `'currency'`, `'percent'`, `'date'`, `'hms'`.
`
`currency_symbol`	`str \| None`	`None`	Currency symbol (viewer default is usually `'$'`).
`good_direction`	`str \| None`	`None`	Which direction is “good” (`'higher'` or `'lower'`).
`breach_value`	`float \| int \| None`	`None`	Value that triggers a breach indicator.
`warning_value`	`float \| int \| None`	`None`	Value that triggers a warning indicator.
`description`	`str \| None`	`None`	Optional description text.
`width`	`int \| None`	`None`	Optional width.
`height`	`int \| None`	`None`	Optional height.
`**kwargs`	`Any`	—	Additional common visual properties (e.g., `modal_id`, padding/margins, etc.).

Table

Parameter	Type	Default	Description
`dataset_id`	`str`	(required)	The dataset id.
`title`	`str \| None`	`None`	Optional table title.
`columns`	`list[str] \| None`	`None`	Optional list of columns to display.
`page_size`	`int \| None`	`None`	Rows per page (groups per page while grouped).
`table_style`	`str \| None`	`None`	`'plain'`, `'bordered'`, or `'alternating'`.
`show_search`	`bool \| None`	`None`	Whether to show the search box.
`sortable`	`bool \| None`	`None`	Type-aware sorting; Shift+click multi-sort (viewer default true).
`default_sort`	`list[dict] \| None`	`None`	Initial sort, e.g. `[{"column": "Amount", "direction": "desc"}]`.
`hidden_columns`	`list[str] \| None`	`None`	Columns hidden initially.
`allow_column_hiding`	`bool \| None`	`None`	Runtime Columns menu (viewer default true).
`group_by`	`str \| None`	`None`	Initial grouping column (collapsible groups).
`group_aggregates`	`list[dict] \| None`	`None`	Per-group aggregates, e.g. `[aggregates.agg("Amount", "sum")]`.
`groups_collapsed`	`bool \| None`	`None`	Whether groups start collapsed.
`enable_export`	`bool \| None`	`None`	CSV export / clipboard copy (viewer default true).
`export_file_name`	`str \| None`	`None`	File name for CSV export.
`context_menu`	`bool \| None`	`None`	Right-click context menus (viewer default true).
`max_height`	`int \| None`	`None`	Max body height in px (scrollable body + sticky header).
`sticky_header`	`bool \| None`	`None`	Viewer default: true when `max_height` is set.
`total_row`	`bool \| dict \| None`	`None`	`True` or `{"label": ..., "fns": {"<col>": "<fn>"}}` — grand-total row (dl2 0.4+).
`total_column`	`bool \| dict \| None`	`None`	`True` or `{"label": ..., "columns": [...]}` — per-row total column (dl2 0.4+).
`row_modal`	`bool \| None`	`None`	Built-in row detail modal on double-click (dl2 0.4+).
`row_modal_id`	`str \| None`	`None`	Open a custom modal instead; cards can use `{{ row.Col }}` templates (dl2 0.4+).
`row_modal_columns`	`list[str] \| None`	`None`	Columns listed in the built-in detail modal.
`row_modal_title`	`str \| None`	`None`	Title of the built-in detail modal.
`column_formats`	`dict \| None`	`None`	Per-column display formats (dl2 0.4.1+) — see Column Formatting.
`conditional_formats`	`list \| None`	`None`	Highlight rules (dl2 0.4.1+) — see Conditional Formatting.
`id`	`str \| None`	`None`	Stable element id (persistence + link targeting).
`persist_state`	`bool \| None`	`None`	Persist sort/columns/grouping (viewer default: true when `id` is set).
`**kwargs`	`Any`	—	Additional common visual properties (e.g. `filter=`, `aggregate=`).

Card

Parameter	Type	Default	Description
`title`	`str \| None`	(required)	Optional title (supports template syntax in the viewer).
`text`	`str`	(required)	Main card text (supports template syntax in the viewer).
`**kwargs`	`Any`	—	Additional common visual properties.

Pie / Donut

Parameter	Type	Default	Description
`dataset_id`	`str`	(required)	The dataset id.
`category_column`	`str \| int`	(required)	Column for slice labels.
`value_column`	`str \| int`	(required)	Column for slice values.
`inner_radius`	`int \| None`	`None`	Inner radius for donut styling.
`show_legend`	`bool \| None`	`None`	Whether to show the legend.
`**kwargs`	`Any`	—	Additional common visual properties.

Bar (Clustered / Stacked)

Parameter	Type	Default	Description
`dataset_id`	`str`	(required)	The dataset id.
`x_column`	`str \| int`	(required)	Column for X-axis categories.
`y_columns`	`list[str]`	(required)	Series columns for Y values.
`stacked`	`bool`	`False`	If `True`, uses stacked bars; otherwise clustered.
`threshold`	`dict \| None`	`None`	Optional pass/fail coloring (see Threshold Configuration).
`x_axis_label`	`str \| None`	`None`	Optional X-axis label.
`y_axis_label`	`str \| None`	`None`	Optional Y-axis label.
`show_legend`	`bool \| None`	`None`	Whether to show the legend.
`show_labels`	`bool \| None`	`None`	Whether to show value labels.
`horizontal`	`bool \| None`	`None`	Whether to render bars horizontally (viewer-dependent).
`**kwargs`	`Any`	—	Additional common visual properties.

Scatter

Parameter	Type	Default	Description
`dataset_id`	`str`	(required)	The dataset id.
`x_column`	`str \| int`	(required)	Column for numeric X values.
`y_column`	`str \| int`	(required)	Column for numeric Y values.
`category_column`	`str \| int \| None`	`None`	Optional column for coloring points by category.
`show_trendline`	`bool \| None`	`None`	Whether to show a trendline.
`show_correlation`	`bool \| None`	`None`	Whether to show correlation stats.
`point_size`	`int \| None`	`None`	Point size.
`x_axis_label`	`str \| None`	`None`	Optional X-axis label.
`y_axis_label`	`str \| None`	`None`	Optional Y-axis label.
`**kwargs`	`Any`	—	Additional common visual properties.

Line

Parameter	Type	Default	Description
`dataset_id`	`str`	(required)	The dataset id.
`x_column`	`str \| int`	(required)	Column for X values (time or category).
`y_columns`	`list[str] \| str`	(required)	Column(s) for Y series.
`smooth`	`bool \| None`	`None`	Whether to render smooth curves.
`show_legend`	`bool \| None`	`None`	Whether to show the legend.
`show_labels`	`bool \| None`	`None`	Whether to show value labels.
`min_y`	`float \| int \| None`	`None`	Optional minimum Y.
`max_y`	`float \| int \| None`	`None`	Optional maximum Y.
`colors`	`list[str] \| None`	`None`	Optional list of series colors.
`threshold`	`dict \| None`	`None`	Optional pass/fail coloring (see Threshold Configuration).
`x_axis_label`	`str \| None`	`None`	Optional X-axis label.
`y_axis_label`	`str \| None`	`None`	Optional Y-axis label.
`**kwargs`	`Any`	—	Additional common visual properties.

Area

Parameter	Type	Default	Description
`dataset_id`	`str`	(required)	The dataset id.
`x_column`	`str \| int`	(required)	Column for X values.
`y_columns`	`list[str] \| str`	(required)	Column(s) for Y series.
`smooth`	`bool \| None`	`None`	Whether to render smooth curves.
`show_line`	`bool \| None`	`True`	Show line stroke on top of fill.
`show_markers`	`bool \| None`	`True`	Show interactive marker points.
`fill_opacity`	`float \| None`	`0.3`	Area fill opacity (0-1).
`show_legend`	`bool \| None`	`None`	Whether to show the legend.
`show_labels`	`bool \| None`	`None`	Whether to show value labels.
`min_y`	`float \| int \| None`	`None`	Optional minimum Y.
`max_y`	`float \| int \| None`	`None`	Optional maximum Y.
`colors`	`list[str] \| None`	`None`	Optional list of series colors.
`threshold`	`dict \| None`	`None`	Optional pass/fail coloring (see Threshold Configuration).
`x_axis_label`	`str \| None`	`None`	Optional X-axis label.
`y_axis_label`	`str \| None`	`None`	Optional Y-axis label.
`**kwargs`	`Any`	—	Additional common visual properties.

Checklist

Since dl2 0.4.1 the checklist has full table parity (type-aware sorting, column hiding, CSV export, context menus, sticky header, row detail modals, persistent view state) plus status filter chips and a completion progress bar. It remains read-only by design — status always comes from the dataset.

Parameter	Type	Default	Description
`dataset_id`	`str`	(required)	The dataset id.
`status_column`	`str`	(required)	Column containing a truthy completion value.
`warning_column`	`str \| None`	`None`	Optional date column to evaluate for warnings.
`warning_threshold`	`int \| None`	`None`	Days before due date to trigger warning (viewer default 3).
`columns`	`list[str] \| None`	`None`	Optional subset of columns to display.
`page_size`	`int \| None`	`None`	Rows per page.
`show_search`	`bool \| None`	`None`	Whether to show the search box.
`sortable`	`bool \| None`	`None`	Type-aware sorting; Shift+click multi-sort (viewer default true). The Status header sorts by urgency (dl2 0.4.1+).
`default_sort`	`list \| None`	`None`	Initial sort — `SortSpec` or dicts. Accepts the special column `"status"` (urgency: overdue → due soon → pending → complete). Viewer default: urgency, then due date (dl2 0.4.1+).
`hidden_columns`	`list[str] \| None`	`None`	Columns hidden initially (dl2 0.4.1+).
`allow_column_hiding`	`bool \| None`	`None`	Runtime Columns menu (viewer default true) (dl2 0.4.1+).
`enable_export`	`bool \| None`	`None`	CSV export / clipboard copy; exports include a derived `Status` column (viewer default true) (dl2 0.4.1+).
`export_file_name`	`str \| None`	`None`	File name for CSV export (dl2 0.4.1+).
`context_menu`	`bool \| None`	`None`	Right-click context menus (viewer default true) (dl2 0.4.1+).
`max_height`	`int \| None`	`None`	Max body height in px (scrollable body + sticky header) (dl2 0.4.1+).
`sticky_header`	`bool \| None`	`None`	Viewer default: true when `max_height` is set (dl2 0.4.1+).
`row_modal`	`bool \| None`	`None`	Built-in row detail modal on double-click; leads with the status (dl2 0.4.1+).
`row_modal_id`	`str \| None`	`None`	Open a custom modal instead; cards can use `{{ row.Col }}` templates (dl2 0.4.1+).
`row_modal_columns`	`list[str] \| None`	`None`	Columns listed in the built-in detail modal (dl2 0.4.1+).
`row_modal_title`	`str \| None`	`None`	Title of the built-in detail modal (dl2 0.4.1+).
`show_status_filter`	`bool \| None`	`None`	Status filter chips with counts — All / Pending / Due Soon / Overdue / Complete (viewer default true) (dl2 0.4.1+).
`show_progress`	`bool \| None`	`None`	Completion progress bar next to the "X / Y Completed" summary (viewer default true) (dl2 0.4.1+).
`hide_completed`	`bool \| None`	`None`	Start with completed tasks hidden — the Complete chip toggled off (dl2 0.4.1+).
`column_formats`	`dict \| None`	`None`	Per-column display formats — see Column Formatting (dl2 0.4.1+).
`conditional_formats`	`list \| None`	`None`	Highlight rules — see Conditional Formatting (dl2 0.4.1+).
`id`	`str \| None`	`None`	Stable element id (persistence + link targeting).
`persist_state`	`bool \| None`	`None`	Persist sort/columns/status chips (viewer default: true when `id` is set).
`**kwargs`	`Any`	—	Additional common visual properties.

Histogram

Parameter	Type	Default	Description
`dataset_id`	`str`	(required)	The dataset id.
`column`	`str \| int`	(required)	Numeric column to bin.
`bins`	`int \| None`	`None`	Number of bins.
`color`	`str \| None`	`None`	Bar color.
`show_labels`	`bool \| None`	`None`	Whether to show count labels.
`x_axis_label`	`str \| None`	`None`	Optional X-axis label.
`y_axis_label`	`str \| None`	`None`	Optional Y-axis label.
`**kwargs`	`Any`	—	Additional common visual properties.

Heatmap

Parameter	Type	Default	Description
`dataset_id`	`str`	(required)	The dataset id.
`x_column`	`str \| int`	(required)	Column for X categories.
`y_column`	`str \| int`	(required)	Column for Y categories.
`value_column`	`str \| int`	(required)	Column for cell values.
`show_cell_labels`	`bool \| None`	`None`	Whether to show values inside cells.
`min_value`	`float \| int \| None`	`None`	Optional minimum for the color scale.
`max_value`	`float \| int \| None`	`None`	Optional maximum for the color scale.
`color`	`str \| list[str] \| None`	`None`	D3 interpolator name (e.g., `'Viridis'`) or custom colors.
`x_axis_label`	`str \| None`	`None`	Optional X-axis label.
`y_axis_label`	`str \| None`	`None`	Optional Y-axis label.
`**kwargs`	`Any`	—	Additional common visual properties.

Boxplot

Supports two modes:

Data mode: provide data_column (and optional category_column)
Pre-calculated mode: provide min_column, q1_column, median_column, q3_column, max_column (and optional mean_column)

Parameter	Type	Default	Description
`dataset_id`	`str`	(required)	The dataset id.
`data_column`	`str \| int \| None`	`None`	Raw values column (data mode).
`category_column`	`str \| int \| None`	`None`	Grouping/label column.
`min_column`	`str \| int \| None`	`None`	Pre-calculated minimum column.
`q1_column`	`str \| int \| None`	`None`	Pre-calculated Q1 column.
`median_column`	`str \| int \| None`	`None`	Pre-calculated median column.
`q3_column`	`str \| int \| None`	`None`	Pre-calculated Q3 column.
`max_column`	`str \| int \| None`	`None`	Pre-calculated maximum column.
`mean_column`	`str \| int \| None`	`None`	Pre-calculated mean column (optional).
`direction`	`str \| None`	`None`	`'vertical'` or `'horizontal'`.
`show_outliers`	`bool \| None`	`None`	Whether to show outliers.
`color`	`str \| list[str] \| None`	`None`	Fill color or scheme.
`x_axis_label`	`str \| None`	`None`	Optional X-axis label.
`y_axis_label`	`str \| None`	`None`	Optional Y-axis label.
`**kwargs`	`Any`	—	Additional common visual properties.

Modal Button

Parameter	Type	Default	Description
`modal_id`	`str`	(required)	The global modal id to open.
`button_label`	`str`	(required)	Button label text.
`**kwargs`	`Any`	—	Additional common visual properties.

Note: Most visuals also support modal_id as a keyword argument to enable an "expand" icon that opens a modal on click.

Threshold Configuration

Color chart elements based on whether values pass or fail a threshold. Applies to Line, Area, and Clustered Bar.

Property	Type	Default	Description
`value`	`number`	(required)	The threshold value to compare against.
`pass_color`	`str`	`#22c55e`	Color for passing values.
`fail_color`	`str`	`#ef4444`	Color for failing values.
`mode`	`'above'`\|`'below'`\|`'equals'`	`'above'`	How to determine pass/fail.
`show_line`	`bool`	`True`	Show reference line at threshold.
`line_style`	`'solid'`\|`'dashed'`\|`'dotted'`	`'dashed'`	Threshold line style.
`blend_width`	`number`	`5`	Gradient blend zone width (line/area only).
`apply_to`	`'both'`\|`'markers'`\|`'lines'`	`'both'`	Which elements get threshold colors.

Mode Options:

'above' - Values >= threshold pass
'below' - Values <= threshold pass
'equals' - Only exact matches pass

Tabs (dl2 0.3+)

row.add_tabs() returns a Tabs container; tabs.add_tab(title) returns a full Layout, so every add_* helper works inside a tab. Tabs can be nested.

tabs = page.add_row().add_tabs(id="sales-tabs", default_tab=0, title="Sales Views")
tabs.add_tab("Chart").add_line("sales", x_column="Month", y_columns=["Revenue"])
tabs.add_tab("Data", direction="column").add_table("sales", id="sales-table")

Parameter	Type	Default	Description
`id`	`str \| None`	auto	Stable id — enables active-tab persistence (dl2 0.4+) and link targeting.
`default_tab`	`int \| None`	`None`	Index of the initially active tab (viewer default 0).
`title`	`str \| None`	`None`	Optional title above the tab strip.
`**kwargs`	`Any`	—	Container props (`padding`, `border`, `shadow`, `flex`, `persist_state`, ...).

add_tab(title, direction="column", **layout_kwargs) — the kwargs are layout props (gap, wrap, columns, ...).

Link (dl2 0.4+)

Navigate to any visual with an id (switches page, activates containing tabs, scrolls, flashes) or to an external URL. Exactly one of target_id / href is required.

row.add_link(target_id="sales-table", label="Jump to data", link_style="button")
row.add_link(href="https://example.com", label="Docs")

Filtering & Aggregation (dl2 0.3+)

Any visual accepts filter= and aggregate= — several visuals can show different client-side slices of one shared dataset, with no extra data embedded in the HTML. Use the filters and aggregates builder modules (plain dicts work too); invalid operators/functions raise ValueError at build time.

from dl2_reports import DL2Report, aggregates as A, filters as F

# Filter: leaf conditions + and_/or_/not_ composition
row.add_table(
    "sales",
    filter=F.and_(F.gte("Amount", 200), F.isin("Region", ["South", "West"])),
)

# Aggregate: groupBy + aggregate fns (applied after the filter)
row.add_bar(
    "sales",
    x_column="Region",
    y_columns=["sum_Amount"],   # default output name is "{fn}_{column}"
    aggregate=A.aggregate("Region", A.agg("Amount", "sum")),
)

Filter shortcuts: eq, neq, gt, gte, lt, lte, isin, notin, contains, starts_with, ends_with, between, is_null, not_null (plus generic where(column, op, ...)). Aggregate fns: sum, avg, min, max, count, countDistinct, first, last. Use A.agg("Amount", "sum", as_="Total") to name the output column.

Derived Datasets (dl2 0.3+)

Declare a dataset computed in the browser from another dataset — filtered and/or aggregated at load time. Chains are supported and declaration order doesn't matter (sources are checked at compile()).

report.add_derived_dataset(
    "north_by_category",
    source="sales",
    filter=F.eq("Region", "North"),
    aggregate=A.aggregate("Category", A.agg("Amount", "sum", as_="Total")),
)
page.add_row().add_table("north_by_category")

Note: derived values are not available to report.get_value() at compile time (they only exist in the browser) — compute with pandas if you need them while building.

Table Totals & Row Detail Modals (dl2 0.4+)

page.add_row().add_table(
    "orders",
    id="orders-table",
    total_row={"label": "Totals", "fns": {"Units": "sum", "Amount": "avg"}},
    total_column={"columns": ["Units", "Amount"]},
    row_modal_id="order-detail",       # or row_modal=True for the built-in modal
)

modal = report.add_modal("order-detail", "Order Details")
modal.add_row().add_card(
    title="Order — {{ row.Region }}",
    text="**Rep:** {{ row.Rep }}\n**Amount:** {{ formatCurrency(row.Amount) }}",
    content_type="md",
)

The column names in total_row["fns"] are preserved exactly as written (they are not snake_case→camelCase converted). For your own passthrough props whose dict keys are column names, wrap them in dl2_reports.RawDict to get the same protection.

Column Formatting (dl2 0.4.1+)

Give Table and Checklist columns display formats with column_formats= — a mapping of column name → ColumnFormat, dict, or shorthand kind string:

from dl2_reports import ColumnFormat

page.add_row().add_table(
    "orders",
    column_formats={
        "Amount": ColumnFormat("currency", digits=0),   # or {"format": "currency", "digits": 0}
        "Growth": ColumnFormat("percent", digits=1),    # raw values are ratios (0.42 → 42.0%)
        "Due": "date",                                  # shorthand string
        "Runtime": "hms",                               # seconds → HH:MM:SS
    },
)

Kinds: 'number', 'currency', 'percent', 'date', 'hms'; options: digits (currency default 2, percent default 1) and symbol (currency only, default '$').
Applies to cells, total row/column, group aggregates (matched by the aggregate's as name), and row detail modals.
Display-only: CSV export keeps raw values; clipboard copy matches the formatted view.
Column names used as keys are preserved verbatim (no snake_case→camelCase conversion).

Conditional Formatting (dl2 0.4.1+)

Highlight cells or whole rows with conditional_formats= — a list of rules evaluated per row using the standard filter grammar:

from dl2_reports import ConditionalFormat, filters as F

page.add_row().add_table(
    "orders",
    conditional_formats=[
        ConditionalFormat(when=F.gte("Amount", 300), style="success"),
        ConditionalFormat(when=F.lt("Amount", 100), target="row", style="error"),
        ConditionalFormat(
            when=F.and_(F.eq("Region", "West"), F.gt("Units", 10)),
            columns=["Units"],                       # required for compound `when`
            css={"font_weight": 600, "background_color": "#fef3c7"},
        ),
    ],
)

style is a named theme-aware preset: 'success', 'warning', 'error', 'info', or 'muted' (note: the field is style, not preset). css layers inline overrides on top; snake_case keys become camelCase React style names.
target is 'cell' (default — styles the matching cell(s)) or 'row'. For cell targets, columns defaults to the when condition's own column; compound (and/or/not) conditions must set columns explicitly (the Python API enforces this).
First matching rule wins per target; one row rule and one cell rule can compose. Totals/aggregate rows are exempt. Rules see raw values (before column_formats).
Every rule needs style and/or css — ConditionalFormat raises ValueError otherwise, and validates when at construction time.

Both props work identically on Checklist (dl2 0.4.1 rebuilt it on the table infrastructure).

Persistent View State (dl2 0.4+)

Runtime view changes (table sort/hidden columns/grouping, active tabs) are saved to localStorage per report + visual id and restored on reload.

Give tables/tabs a stable id= and the viewer persists them automatically; opt out per visual with persist_state=False.
Set a stable report identity so state survives title changes: DL2Report(title, report_id="my-report") or report.set_report_id("my-report") (emits <meta name="report-id">).
Users can reset via right-click → Reset view, or the report-wide Reset view button.

Layout Options (dl2 0.3+)

Layouts own spacing now (gap defaults to 10px; visuals default to margin: 0). Rows/columns accept wrap=True, align=..., justify=...; grids accept min_child_width=250 for responsive auto-fit columns. flex=0 and padding=0/margin=0 are respected.

page.add_row(wrap=True, gap=16, justify="space-between")
page.add_row(direction="grid", min_child_width=250)

Modals

Create detailed overlays that can be triggered from any element.

# Define a modal
modal = report.add_modal("details", "Detailed View")
modal.add_row().add_table("my_data")

# Trigger from a visual
page.add_row().add_kpi("my_data", "A", "Metric", modal_id="details")

# Or add a dedicated button
page.add_row().add_modal_button("details", "Open Details")

Visual Elements (Annotations)

Add trend lines, markers, and custom axes to your charts.

Trend Lines

You can add a trend line using the .add_trend() method. It can automatically calculate linear or polynomial regression if you don't provide coefficients.

Supported on line, area, scatter, clusteredBar, stackedBar, and histogram visuals (before dl2 0.4.1 the viewer only rendered trends on scatter plots).

chart = page.add_row().add_scatter("my_data", "A", "B")

# Auto-calculate linear trend (degree 1)
chart.add_trend(color="red")

# Auto-calculate polynomial trend (e.g., degree 2)
chart.add_trend(coefficients=2, color="blue", line_style="dashed")

# Manually provide coefficients [intercept, slope, ...]
chart.add_trend(coefficients=[0, 1.5], color="green")

Units: on categorical X axes (line, area, bars) the viewer evaluates coefficients against the 0-based category index; numeric axes (scatter, histogram) use real axis units. Auto-calculation needs x_column and y_column props, so histograms (binned counts) and multi-series charts (y_columns) require explicit coefficients.

Other Elements

Use .add_element(type, **kwargs) for other annotations.

Element Type	Description	Key Arguments
`xAxis`	Vertical line at a specific X value.	`value`, `color`, `label`, `line_style`
`yAxis`	Horizontal line at a specific Y value.	`value`, `color`, `label`, `line_style`
`marker`	A point marker at a specific value.	`value`, `size`, `shape` (`circle`, `square`, `triangle`), `color`
`label`	A text label at a specific value.	`value`, `label`, `font_size`, `font_weight`

chart.add_element("yAxis", value=100, label="Target", color="green")

Tree Traversal

All components (Pages, Rows, Layouts, Visuals) are part of a tree. You can access the root report from any component using .get_report().

visual = layout.add_visual("line", "my_data")
report = visual.get_report()
print(report.title)

Reading Values

The API provides two ways to read scalar values back from your data after the report is built. These are useful for conditional layout logic, threshold checks, or computing derived metrics without re-querying the original DataFrame.

`report.get_value()`

Query a value from any registered dataset by name — no visual reference required.

report.get_value(data_source_name, column_name, row_index=-1)

Parameter	Type	Default	Description
`data_source_name`	`str`	(required)	The dataset name passed to `add_df()`.
`column_name`	`str`	(required)	The column to read from.
`row_index`	`int`	`-1`	Row index. Negative indices count from the end (e.g. `-1` = last row).

This is the right tool when you need to inspect a value before or without creating a visual — for example, deciding whether to add a row at all:

report = DL2Report(title="Sales Report")
report.add_df("sales", sales_df, format="records", compress=False)

page = report.add_page("Overview")

# Add a bar chart
bar_row = page.add_row()
bar_row.add_bar(dataset_id="sales", x_column="region", y_columns=["revenue"])

# Only add a warning card if the worst region is below target
TARGET = 120_000
revenues = [report.get_value("sales", "revenue", i) for i in range(len(sales_df))]
worst = min(revenues)

if worst < TARGET:
    warning_row = page.add_row()
    warning_row.add_card(
        title="Warning: underperforming region detected",
        text=f"Lowest revenue is ${worst:,} — below the ${TARGET:,} target.",
        content_type="md",
    )

`visual.get_value()`

Read the scalar value that a specific visual represents, directly from its backing DataFrame. The visual must be part of the report tree (i.e. already added to a row) and its props must include row_index and value_column.

value = visual.get_value()

This is the right tool when you already have a visual reference and want to inspect or act on its value:

kpi = page.add_row().add_kpi(
    dataset_id="sales",
    value_column="revenue",
    row_index=0,
    title="Revenue – North",
    format="currency",
)

north_revenue = kpi.get_value()
print(f"North revenue: {north_revenue:,}")

`visual.copy()`

Create a duplicate of a visual with the same type, dataset_id, props, and annotations, but a new unique ID. Use this to stamp the same visual configuration into multiple rows without re-specifying every argument.

copied_visual = visual.copy()

After copying, add the copy back into any layout row using row.add_visual(copy.type, visual=copy). You can then mutate copy.props to override only what differs:

# Build a prototype KPI once
proto = row.add_kpi(
    dataset_id="sales",
    value_column="revenue",
    row_index=0,
    title="Revenue – North",
    format="currency",
)

# Stamp copies for remaining regions
for i, region in enumerate(["South", "East", "West"], start=1):
    copy = proto.copy()
    copy.props["row_index"] = i
    copy.props["title"] = f"Revenue – {region}"
    row.add_visual(copy.type, visual=copy)

# Each copy exposes get_value() once it is in the tree
total = proto.get_value() + sum(
    row.children[i].get_value() for i in range(1, 4)
)

Conditional Layout

Use layout.on_condition() to conditionally add visuals to a row at report-build time. This is a compile-time guard — it evaluates a plain Python bool and either delegates the add_* call to the real layout or silently discards it.

layout.on_condition(condition).add_<visual>(...)

Parameter	Type	Description
`condition`	`bool`	If `True`, the visual is added normally and returned. If `False`, nothing is added and `None` is returned.

How It Works

When condition is True, the call is forwarded to the parent layout exactly as if you had called layout.add_<visual>(...) directly.
When condition is False, no visual is created, no element ID is consumed, and None is returned.
The wrapper is not a tree node — it never occupies a slot in the report tree regardless of the condition.

Example: Show a warning card only when a threshold is breached

report = DL2Report(title="Sales Report")
report.add_df("sales", sales_df, format="records", compress=False)

page = report.add_page("Overview")
row = page.add_row()
row.add_bar(dataset_id="sales", x_column="region", y_columns=["revenue"])

TARGET = 120_000
worst = min(report.get_value("sales", "revenue", i) for i in range(len(sales_df)))

warning_row = page.add_row()
warning_row.on_condition(worst < TARGET).add_card(
    title="Warning: underperforming region detected",
    text=f"Lowest revenue is ${worst:,} — below the ${TARGET:,} target.",
)

Example: Toggle a chart based on a flag

show_details = True  # could come from any Python logic

detail_row = page.add_row()
detail_row.on_condition(show_details).add_table("sales", title="Detail View")

Tip: Since 0.5.0, on_condition(False) returns a falsy NullComponent instead of None — chained calls like .add_trend() are silently absorbed, so no guard is needed. Truthiness checks (if result:) keep working; is None checks should become truthiness checks.

Datalys2 Documentation

For detailed information on available visuals and configuration, see DOCUMENTATION.md.

Or see the github repo at https://github.com/kameronbrooks/datalys2-reporting

Projects → Datalys2 Reports → Python API

Datalys2 Reports - Python API

Datalys2 Reporting Python API

Installation

Quick Start

The Typed Component API (v2)

Migrating existing scripts

Data Compression

Why Compression Matters

Using Compression

Report-Level Default

Per-Dataset Control

How It Works

Best Practices

Features

Jupyter Notebook Support

Available Visuals

Common Visual Keyword Arguments

Layout Visual Helper APIs

Visual Types (Quick Summary)

Generic Visual

KPI

Table

Card

Pie / Donut

Bar (Clustered / Stacked)

Scatter

Line

Area

Checklist

Histogram

Heatmap

Boxplot

Modal Button

Threshold Configuration

Tabs (dl2 0.3+)

Link (dl2 0.4+)

Filtering & Aggregation (dl2 0.3+)

Derived Datasets (dl2 0.3+)

Table Totals & Row Detail Modals (dl2 0.4+)

Column Formatting (dl2 0.4.1+)

Conditional Formatting (dl2 0.4.1+)

Persistent View State (dl2 0.4+)

Layout Options (dl2 0.3+)

Modals

Visual Elements (Annotations)

Trend Lines

Other Elements

Tree Traversal

Reading Values

report.get_value()

visual.get_value()

visual.copy()

Conditional Layout

How It Works

Example: Show a warning card only when a threshold is breached

Example: Toggle a chart based on a flag

Datalys2 Documentation

`report.get_value()`

`visual.get_value()`

`visual.copy()`