listed_danish_comp_analysis.Rmd

---
title: "R Notebook"
output: html_notebook
---

```{r}
library(tidyverse)
library(magrittr)
library(stringi)
library(stringr)
library(readtext)
library(dplyr)
library(curl)
library(pdftools)
library(cld2)
library(tabulizer)
library(googleLanguageR)
```

```{r}

lowerize <- function(mystring){
  rem_punct <- regex("[[:punct:]]")
  str_replace_all(
            str_squish(
              str_sub(
                str_to_lower(
                  str_remove_all(mystring, rem_punct), locale = "en"),
              start=1, end=35)
            ),
          " ", "_")
}
```

```{r}
dl_comps <- read_csv("listed_danish_companies.csv")
names(dl_comps) <- lowerize(names(dl_comps))
names(dl_comps)
```


```{r}
dl_comps %<>% mutate(company_name=iconv(company_name, from="UTF-8", to="ASCII//TRANSLIT"))

dl_comps %<>% mutate(folder=str_sub(lowerize(str_remove_all(company_name, "A/S")), start=1, end=15))

dl_comps %<>% mutate(pubyear=list(c("2021","2020","2019")))

dl_comps %<>% unnest(pubyear)

dl_comps %<>% mutate(full_path=str_c("listed_danish_companies",folder,str_c(folder,"_",pubyear,".pdf"), sep = "/"))

dl_comps %<>% data.frame()
```


```{r}
dirvec <-unique(dl_comps$folder)

create_folders(dirvec) <- function (){
  setwd("listed_danish_companies")
  for (i in 1:length(dirvec)) {
    dir.create(dirvec[i], recursive = F)  
  }
}
# create_folders(divec)
```


```{r}
setwd("listed_danish_companies")

write_csv(dl_comps, "listed_danish_companies_fullpath.csv")
```

### Text extraction

```{r}
page_vector <- c()
text_vector <- c()

dl_comps %<>% mutate(text="")
dl_comps %<>% mutate(pages="")
for (i in 1:nrow(dl_comps)) {
  page_vector <- c (1:pdf_info(dl_comps$full_path[i])$pages)
  text_vector <- list(extract_text(dl_comps$full_path[i], page_vector))
  
  dl_comps$pages[i] <- list(page_vector)
  dl_comps$text[i] <- text_vector
}


```


```{r}
dl_comps_unnested <- dl_comps %>% unnest(c(pages,text))

dl_comps_unnested %<>% mutate(lang=detect_language(text))

dl_comps_unnested %>% group_by(lang) %>% count

#dl_comps_unnested_filtered <- dl_comps_unnested %>% filter(lang=="da")
```


```{r}
# dl_comps_unnested_filtered %<>% slice(-206) # produces an error

#translated <- translate(dataset=dl_comps_unnested_filtered,
#            content.field="text",
#            source.lang="da",
#            target.lang="en",
#            google.api.key="mykey")

gl_auth("g_auth.json")

for (i in 1:nrow(dl_comps_unnested)) {
  
  if(!dl_comps_unnested$lang[i]=="en"|is.na(dl_comps_unnested$lang[i])){

    translated <- gl_translate(dl_comps_unnested$text[i], target="en")
    dl_comps_unnested$lang[i]<-"en"
    dl_comps_unnested$text[i] <- translated$translatedText
  
  }

}

dl_comps_unnested$text#[206]=="en"

#dl_comps_unnested %<>% mutate(text=case_when(lang=="da"~gl_translate(text, target="en")$translatedText))


vignette("translation", package = "googleLanguageR")
gl_translate()
# %<>% mutate(lang=detect_language(text))

```
```{r}

write.csv(dl_comps_unnested, "listed_danish_companies_text.csv")
```