---
title: "Sending R results to Mellio"
output: rmarkdown::html_vignette
vignette: >
  %\VignetteIndexEntry{Sending R results to Mellio}
  %\VignetteEngine{knitr::rmarkdown}
  %\VignetteEncoding{UTF-8}
---

```{r setup, include = FALSE}
knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>",
  eval = FALSE,
  purl = FALSE
)
```

The Mellio R bridge moves statistical results from R into Mellio's
Stats workspace as structured Result Cards that can be saved, formatted,
and used in a manuscript.

This vignette walks through the common workflow.

## Setup

By default, `mellio_open()` opens the hosted Mellio app at
`https://www.mellioapp.com`.

```{r, eval=FALSE}
# Advanced users can point this at a trusted Mellio deployment:
# options(mellio.editor_url = "https://www.mellioapp.com")
```

The R payload is encoded in the URL fragment. URL fragments are not sent as
HTTP requests to the server, but the full URL can still be visible to the
browser, the opened web app, browser history, extensions, and anyone the URL
is shared with.

## Opening a result in Mellio

`mellio_open()` builds a Result Card from a supported R object and opens
it in your browser.

### t-tests, correlations, non-parametric tests

```{r, eval=FALSE}
library(mellio)

# Welch's two-sample t-test
mellio_open(t.test(extra ~ group, data = sleep))

# Pearson correlation
mellio_open(cor.test(mtcars$mpg, mtcars$wt))

# Chi-squared
M <- matrix(c(10, 20, 30, 40), nrow = 2)
mellio_open(chisq.test(M))
```

The card shows the test statistic, df, p-value, and effect when
available, formatted in APA 7 style.

### Regression and ANOVA

```{r, eval=FALSE}
m1 <- lm(mpg ~ wt + cyl, data = mtcars)
mellio_open(m1)

m2 <- lm(mpg ~ wt + cyl + hp, data = mtcars)
mellio_open(anova(m1, m2))  # nested model comparison
```

### Mixed models

```{r, eval=FALSE}
library(lme4)
mellio_open(lmer(Reaction ~ Days + (Days | Subject), data = sleepstudy))
```

`lmer` and `glmer` cards show AIC, logLik, and N inline (lavaan
gets full fit indices below).

### CFA / SEM / mediation (lavaan)

```{r, eval=FALSE}
library(lavaan)
HS.model <- "visual  =~ x1 + x2 + x3
              textual =~ x4 + x5 + x6
              speed   =~ x7 + x8 + x9"
fit <- cfa(HS.model, data = HolzingerSwineford1939)
mellio_open(fit)
```

You'll see a **structural** card with two zones:

- **Report zone** — fit indices (χ², CFI, TLI, RMSEA + CI, SRMR,
  AIC, BIC) with quality badges
- **Inspection zone** — full parameter table, grouped by lavaan
  operator (`=~` loadings, `~` regressions, `~~` (co)variances,
  `~1` intercepts, `:=` defined parameters)

Pre-mark focal paths so they surface in the report zone:

```{r, eval=FALSE}
mellio_open(fit, focal = c("visual =~ x1", "textual =~ x4"))
```

### Descriptives

For variable summaries:

```{r, eval=FALSE}
mellio_open(mtcars$mpg, name = "Fuel efficiency (mpg)")
# Or from summary() output
mellio_open(summary(mtcars$mpg))
```

The card shows *M*, *SD*, *N*, and range.

## What's on the card

Each Result Card carries:

- **Type + call** — what test you ran, deparsed exactly
- **Extracted fields** — statistic, p, df, estimate, etc.
- **Formatted preview** — APA 7 inline string, copyable
- **Provenance** — R and package versions, plus data fingerprints when
  available
- **Raw R output** — verbatim `print()` for audit

Local provenance fields such as user name, host name, working directory,
git state, and script path are opt-in:

```{r, eval=FALSE}
options(mellio.provenance = "full")
```

Set `options(mellio.provenance = FALSE)` to omit provenance metadata.

## Saving to Library

The **Save to Library** button on the card persists it. When you're
signed into Mellio, cards sync to your account and follow you across
devices. When you're not signed in, cards live in browser
localStorage.

The Library lists everything you've saved with search and type
filters. Click any card to reopen it.

## Methods paragraph and reproducibility export

From the Library view:

- **Methods** generates an APA-style Methods paragraph from your
  saved cards, citing the R packages you actually used (lavaan,
  lme4, etc.) with versions.
- **Export** downloads either the full JSON (lossless) or a
  flattened CSV for inclusion in a reproducibility statement.

## RStudio addin

Once `mellio` is loaded, RStudio's **Addins** menu gets a
**Send to Mellio** entry. Highlight an R expression (`t.test(...)`)
and run the addin — same effect as typing `mellio_open()`.

## Non-interactive use

`mellio_open()` opens a browser, which is interactive. For scripts and
knitr documents, use `mellio_payload()` to get the JSON payload directly
without launching a browser:

```{r, eval=FALSE}
p <- mellio_payload(t.test(extra ~ group, data = sleep))
mellio_to_json(p, pretty = TRUE)
```

Use `citation("mellio")` for the package citation.