---
title: "prickleback_swimming"
output: rmarkdown::html_vignette
vignette: >
  %\VignetteIndexEntry{prickleback_swimming}
  %\VignetteEngine{knitr::rmarkdown}
  %\VignetteEncoding{UTF-8}
---

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

```{r setup}
library(fishmechr)
library(ggplot2)
library(tidyr)
library(dplyr)
```

This is an example of processing data from a swimming rock prickleback, *Xiphister mucosus*.

The data was tracked using Sleap (https://sleap.ai/) and comes out in the following format. `frame_idx` is the frame number, and each point along the body is
identified with the point name and `.x`, `.y`, the coordinate, and `.score`, which is a measure of the estimated accuracy of the point. All of the points together are also given a score (`instance.score`).
```{r}
head(xmucosusdata)
```

# Rearrange the data

We need to rearrange the data into "long" form, where the point names are in
a separate column, and we have single columns for the `x`, `y`, and `score` values. The function `pivot_kinematics_longer` is a convenience wrapper for `tidyr::pivot_longer` that understands the format of points from Sleap or DeepLabCut.

You need to pass in the names of your points (in order from head to tail) and the name of the column you want to put the point names in.
```{r}
pointnames <- c("Snout", "BP1", "BP2", "BP3", "BP4", "BP5", "BP6", "Tail")
xmucosusdata <- xmucosusdata |> 
  pivot_kinematics_longer(pointnames = pointnames,
                          point_to = "bodypart")
```

```{r}
head(xmucosusdata)
```

# Process the data

From here on out, the steps to process the data are quite similar to what is shown in the main vignette.

```{r}
xmucosusdata |> 
  filter(frame_idx %in% c(10, 20)) |> 
  mutate(frame_idx = factor(frame_idx)) |> 
  ggplot(aes(x = x, y = y, color = frame_idx, group = frame_idx)) +
  geom_path() +
  coord_fixed()
```

## Get arc length

```{r}
xmucosusdata <-
  xmucosusdata |> 
  group_by(frame_idx) |> 
  mutate(arclen0 = arclength(x, y))
```

```{r}
xmucosusdata |> 
  ungroup() |> 
  filter(bodypart %in% c("BP2", "BP5", "Tail")) |> 
  ggplot(aes(x = frame_idx, y = arclen0, color = bodypart)) +
  geom_point()
```

## Interpolate to constant arc length

```{r}
xmucosusdata <- xmucosusdata |> 
  interpolate_points_df(arclen0, x, y, spar = 0.2,
                        tailmethod = 'extrapolate',
                        .frame = frame_idx,
                        .point = bodypart,
                        .out = c(arclen='arclen', xs='x_s', ys='y_s'))

```

```{r}
xmucosusdata |> 
  filter(frame_idx %in% c(10, 20)) |> 
  mutate(frame_idx = factor(frame_idx)) |> 
  ggplot(aes(x = x, y, color = frame_idx, group = frame_idx)) +
  geom_point(shape = 10) +
  geom_path() +
  geom_point(aes(x = x_s, y = y_s), shape = 5) +
  coord_fixed()
```

## Interpolate the width

```{r}
xmucosusdata <-
  xmucosusdata |> 
  group_by(frame_idx) |> 
  mutate(width = interpolate_width(fishwidth$s, fishwidth$eelwidth, arclen))
```

## Get the center

```{r}
xmucosusdata <-
  xmucosusdata |> 
  ungroup() |> 
  get_midline_center_df(arclen, x_s,y_s, width=width,
                        .frame = frame_idx)
```

## Get the swimming axis

```{r}
xmucosusdata <-
  xmucosusdata |> 
  mutate(x_ctr = x_s - xcom,
         y_ctr = y_s - ycom,
         t = frame_idx / 60) |> 
  get_primary_swimming_axis_df(t, x_ctr,y_ctr,
                               .frame = frame_idx,
                               .point = bodypart)
```

```{r}
xmucosusdata |> 
  filter(frame_idx %in% c(10, 20, 30)) |> 
  ggplot(aes(x = x, y = y, color = frame_idx)) +
  geom_path(aes(group = frame_idx)) +
  geom_segment(data = ~ filter(.x, bodypart == "Tail"), 
               aes(x = xcom, y = ycom, 
                   xend = xcom - 50*swimaxis_x, 
                   yend = ycom - 50*swimaxis_y)) +
  geom_point(data = ~ filter(.x, bodypart == "Tail"), 
               aes(x = xcom, y = ycom), color = 'red') +
  facet_grid(frame_idx ~ .) +
  coord_fixed()

```