process_trilinears.Rmd

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

# Load Libraries

```{r,message=F}
library(tidyverse)
fs::dir_ls("math",regexp = "\\.txt")
```

# Process first 100

{% raw %}
```{r}
fname <- "math/high_prio_tri_points.txt"
df <- read_file(fname) %>%
  read_delim(delim="=",col_names = c("kimberling","center_name","trilinears")) %>%
  mutate_all(str_trim) %>%
  mutate_at(vars(trilinears),~str_replace_all(.," : ",",")) %>%
  # curly braces: github pages jekyll pukes
  mutate(math=sprintf('{"%s",{%s},"%s"}',kimberling,trilinears,center_name)) %>%
  mutate(n=str_sub(kimberling,start=3,end=-2)%>%as.integer) %>%
  arrange(n)
df
```
{% endraw %}

```{r}
1:100 %>% setdiff(df$n)
```

```{r}
df$math %>% 
  str_c(collapse=",\r\n") %>%
  write_file("math/trilinear_output.txt")
```

Will try to purify by hand

```{r}
fs::dir_ls("math")
```

Which have the wrong # of fields

```{r}
fname_low <- "math/low_prio_tri_points.txt"
tibble(fields=count_fields(fname_low,tokenizer_delim(delim="=")))%>%
  mutate(line=row_number())%>% 
  filter(fields!=3)
```

```{r}
fix_etc <- function(s) s %>%
  str_replace_all("π/(\\d)","Pi\\1") %>%
  str_replace_all("sec\\((.) - (.)\\)", " 1/cos(\\1 - \\2)") %>%
  str_replace_all("csc\\((.) - (.)\\)", " 1/sin(\\1 - \\2)") %>%
  str_replace_all("cos\\((.) \\+ ([ABCPi\\d]{1,3})\\)", "(cos \\1 cos \\2 - sin \\1 sin \\2)") %>%
  str_replace_all("cos\\((.) - ([ABCPi\\d]{1,3})\\)", "(cos \\1 cos \\2 + sin \\1 sin \\2)") %>%
  str_replace_all("sin\\((.) \\+ ([ABCPi\\d]{1,3})\\)", "(sin \\1 cos \\2 + sin \\2 cos \\1)") %>%
  str_replace_all("sin\\((.) - ([ABCPi\\d]{1,3})\\)", "(sin \\1 cos \\2 - sin \\2 cos \\1)") %>%
  str_replace_all("(cos|sin|sec|csc|tan|cot) ","\\1")%>%
  str_replace_all("(a|b|c)(\\d)(?![ABC])","\\1^\\2 ") %>%
  str_replace_all("(abc|bca|cab)","a b c") %>%
  str_replace_all("(cba|bac|acb)","a b c") %>%
  str_replace_all("(ab|ba)","a b") %>%
  str_replace_all("(ac|ca)","a c") %>%
  str_replace_all("(bc|cb)","b c") %>%
  str_replace_all("  +"," ") %>%
  str_replace_all(" +\\)",")") %>%
  str_replace_all("cosPi","cPi") %>%
  str_replace_all("sinPi","sPi") %>%
  str_squish()
```

```{r}
rotate_vars <- function(s) s%>%
  str_replace_all("A","B~") %>%
  str_replace_all("B(?!~)","C~") %>%  
  str_replace_all("C(?!~)","A~") %>%
  str_replace_all("a(?![:alpha:])","b~") %>%
  str_replace_all("b(?![[:alpha:]~])","c~") %>%
  str_replace_all("c(?![[:alpha:]~])","a~") %>%
  str_remove_all(fixed("~"))
```

```{r}
fix_colon_colon <- function(s) {
  s1 <- s %>% str_remove_all(": *:")
  s2 <- rotate_vars(s1)
  s3 <- rotate_vars(s2)
  cat(c(s1,s2,s3))
  str_c(s1,s2,s3,sep=" : ")
}
```

{% raw %}
```{r}
df_low <- read_delim(fname_low,
                     delim="=",
                     col_names = c("kimberling","center_name","trilinears")) %>%
  mutate_at(vars(trilinears),fix_etc) %>%
  mutate(needs_rot=str_detect(trilinears,": *:")) %>%
  mutate_at(vars(trilinears),~if_else(needs_rot,map_chr(.,fix_colon_colon),.)) %>%
  mutate_all(str_trim) %>%
  mutate_at(vars(trilinears),~str_replace_all(.," : ",",")) %>%
  # do not use glue's double curly braces: github pages jekyll pukes
  mutate(math=sprintf('{"%s",{%s},"%s"}',kimberling,trilinears,center_name)) %>%
  mutate(n=str_sub(kimberling,start=3,end=-2)%>%as.integer) %>%
  arrange(n)
```

{% endraw %}

```{r}
df_low$math %>% 
  str_c(collapse=",\r\n") %>%
  write_file("math/trilinear_low_output.txt")
```

# Can I process all of ETC?

```{r}
etc <- read_file("https://faculty.evansville.edu/ck6/encyclopedia/ETC.html") %>%
  str_replace_all(fixed("&nbsp;")," ") %>%
  str_split(fixed("\r\n")) %>%
  first %>%
  str_squish() %>%
  keep(~str_length(.)>0)

etc %>% write_lines("math/etc.txt")
```

```{r}
etc_headers <- read_file("math/etc.txt") %>%
  str_locate_all(regex('(?<=\n<h3 id="X\\d{1,4}">)[^<]+?(?=<)',dotall=T)) %>%
  first

etc_headers_str <- read_file("math/etc.txt") %>%
  str_sub(etc_headers) %>%
  str_replace_all(fixed("\n")," ")

etc_headers_str %>% write_lines("math/etc_headers.txt")

etc_headers_str %>% length
```

Which ones contain & weirdness

```{r}
etc_headers_str %>%
  keep(str_detect,fixed("&"))
```

```{r}
replace_ampersand <- function(s) s%>%
  str_replace_all(c("&Psi;"="Ψ",
                    "&Lambda;"="Λ",
                    "&cap;"="∩"))
etc_headers_str %>%
  keep(str_detect,fixed("&")) %>%
  replace_ampersand
```

Write without Ampersand

```{r}
etc_headers_str_noamp <- etc_headers_str %>%
  replace_ampersand
etc_headers_str_noamp %>%
  write_lines("math/etc_headers_noamp.txt")
```

## Get Trilienars from each header

<h3 id="X1">X(1) = INCENTER<a id="Incenter" name="Incenter"></a></h3>
Trilinears 1 : 1 : 1<br>

```{r}
etc_headers_x <- etc_headers_str%>%str_extract("X\\(\\d+?\\)")
```


```{r}
etc_999 <- read_file("math/etc.txt") %>%
  str_sub(etc_headers%>%as.data.frame%>%mutate(end=end+1000)%>%as.matrix) %>%
 # str_extract("([tT]rilinears.*\\n)+") %>%
  str_extract(regex("[tT]rilinears.*?(?=([bB]arycentrics|X\\())",dotall = T)) %>%
  str_replace_all(c("<br>"="",
                    "&nbsp ; "="",
                    "&pi;"="π",
                    "&omega;"="ω",
                    "<sup>([^<]+)</sup>"="^\\1 ",
                    "<sub>([^<]+)</sub>"="_\\1",
                    "\\["="(",
                    "\\]"=")",
                    "(?<=[^abc])([abc])([abc])(?=[^abc])"="\\1 \\2",
                    "(?<=[^abc])([abc])([abc])([abc])(?=[^abc])"="\\1 \\2 \\3")) %>%
  str_replace_all(c("([:alpha:]) +\\^"="\\1^",
                    " +\\)"=")",
                    ":\\s+:"="::",
                    "<[^<>]{1,2}>"="")) %>%
  str_remove("[tT]rilinears") %>% # first one for split
  str_remove(regex("\\(Peter.*?Moses.*?\\)",ignore_case = T)) %>%
  str_split("([tT]rilinears|=)") %>%
  map(str_squish) %>%
  map(str_remove,"<p>.*") %>%
  map(head,2)
write_lines(etc_999%>%map2_chr(etc_headers_str_noamp,
                               ~str_c(c(.y,.x),collapse="|")),'math/etc_999.txt')
length(etc_999)
```



```{r}
etc_headers <- read_file("math/etc.txt") %>%
  head(10) %>%
  str_extract_all(regex('(?<=\n<h3 id="X\\d{1,4}">)(.*?)(?=</h3>).*?([Tt]rilnears][^<]+)',dotall=T)) %>%
  first
```