lecture-5.Rmd

---
output: 
  revealjs::revealjs_presentation:
    theme: white
    transition: none
    incremental: false
    css: custom.css
    self_contained: true
    center: true
    md_extensions: +fenced_code_attributes
editor_options: 
  chunk_output_type: console
---


```{r setup, include=FALSE}
knitr::opts_chunk$set(warning=FALSE, message=FALSE, error=FALSE, dpi = 400,fig.cap = "")
```

# File manipulation

## Typical tasks

- Updating reference tiers
- Updating tokenization
- Adding new annotations
- Removing unnecessary tiers
- Changing linguistic types

## Lots of this can be done anywhere

- This is just rather simple XML manipulation
- Let's say we rename a tier:

```{r}
library(tidyverse)
library(xml2)

rename_tier <- function(filename, current_name, new_name, suffix){
    read_xml(filename) %>%
    xml_find_all(glue("//TIER[@TIER_ID='{current_name}']")) %>%
    walk(~ xml_set_attr(.x, 'LINGUISTIC_TYPE_ID', new_name)) %>%
    xml_find_first('/') %>%
    write_xml(str_replace(filename, '.eaf', glue('{suffix}.eaf')))
}
```

##

- Or we rename a type:

```{r}
rename_type <- function(filename, current_name, new_name, suffix){
    read_xml(filename) %>%
    xml_find_all(glue("//TIER[@LINGUISTIC_TYPE_REF='{current_name}']")) %>%
    walk(~ xml_set_attr(.x, 'LINGUISTIC_TYPE_REF', new_name)) %>%
    xml_find_all("../LINGUISTIC_TYPE[@LINGUISTIC_TYPE_ID='{current_name}']") %>% 
    walk(~ xml_set_attr(.x, 'LINGUISTIC_TYPE_ID', new_name)) %>%
    xml_find_first('/') %>%
    write_xml(str_replace(filename, '.eaf', glue('{suffix}.eaf')))
}

```

- Already one extra thing to remember!
- However, things can be way more complicated!

## pympi

- Python package maintained by Mart Lubbers
- Has quite thorough [documentation](http://dopefishh.github.io/pympi/Elan.html)
- We will explore some of the basic functionalities next

## {data-background-image="https://imgur.com/mbVZFh1.png"}

## To sum up

- Very useful, well done package
- I would have bit different philosophy to deal with time codes
- But there are probably reasons I don't know!
- I have opened GitHub issue for the most urgent question (which I think relates to others)

## reticulate

```{r}
library(reticulate)

pympi <- import('pympi')

elan_file <- pympi$Eaf(file_path = '/Users/niko/github/testcorpus/kpv_udo20120330SazinaJS-encounter.eaf')

elan_file$get_tier_names()

elan_file$get_annotation_data_for_tier(id_tier = 'orth@NTP-M-1986')


```

## Up next

### Discussion?
### Working with your files?