Examples and demos

Two kinds of runnable material live in the repository: small programs that show the API in use, and benchmark harnesses that measure it. Everything here builds from the same checkout; none of it downloads data.

Render a factor report

examples/render_factor_report.cpp builds a synthetic factor panel, analyses it and writes three artefacts: an English HTML report, a Chinese HTML report and an XLSX workbook. It shows the shortest path from prepared data to a rendered document.

cmake --build build --parallel --target fincore_report_render
./build/fincore_report_render report.html
# writes report.html, report.zh.html and report.xlsx

Pass --echarts to render the charts with the embedded ECharts build instead of inline SVG:

./build/fincore_report_render --echarts report.html

The report is deterministic: rendering the same document twice produces identical bytes, which is what makes the files safe to diff and hash.

Render one document through every backend

examples/render_iteration8_artifacts.cpp takes a directory and writes the same document as HTML (inline SVG and ECharts), XLSX, Plotly JSON, Bokeh JSON and PDF, then prints the document’s semantic digest.

cmake --build build --parallel --target fincore_i8_artifacts
./build/fincore_i8_artifacts ./out

The digest depends only on the document, not on the backend, so the six files can be compared against a recorded digest even though their formats differ.

An interactive ImPlot dashboard

examples/implot_dashboard.cpp draws the tear charts, frontier, heatmap and event panels in an ImGui/ImPlot window. It is built only when the GUI options are on and the GUI dependencies are present:

cmake -S . -B build-gui -DFINCORE_BUILD_IMPLOT=ON -DFINCORE_BUILD_IMPLOT_DEMO=ON
cmake --build build-gui --parallel --target fincore_implot_demo

Variable	Effect
`FINCORE_IMPLOT_VIEW`	`general`, `factor`, `groups`, `events` or `specialized`
`FINCORE_IMPLOT_CAPTURE`	Renders three frames headlessly and writes a PPM image

An unrecognized view name is rejected with exit code 2 rather than silently falling back to the general view.

Benchmarks

Target	What it does
`fincore_factor_benchmark`	Compute-once factor pipeline cases, one JSON record per case
`fincore_alpha003_research`	The Alpha003 A-share research workload used in the performance comparisons

cmake --build build --parallel --target fincore_factor_benchmark fincore_alpha003_research
python benchmarks/alpha003_a_stocks/run_benchmark.py --help

The Python driver runs the paired reference/candidate campaign, compares the research tables by digest, and writes one HTML comparison report. Read performance for what the resulting ratio does and does not mean, and keep generated results out of git.

Where the code lives

examples/        report rendering and dashboard demos
benchmarks/      factor and alpha003 workloads, plus the Python drivers
tests/           unit, consumer and oracle tests

Every example is a normal consumer of the installed library: it includes <fincore/...> headers and links fincore::fincore, so it also serves as a compile-time check that the public surface is usable from outside the build tree.