I'm just getting into sparklr (and R actually) and slowly getting the hang of figuring out what works and what doesn't. Working on converting a straight R script to work with sparklyr. I've managed to replace grepl with regexp_replace and work out the differences in the regex formats..
I'm a bit stuck on this one tho...I am loading files (spark_read_json) that can contain non UTF-8 chars, and want to remove them.
The code that works to remove these chars in straight R is
fileline <- fileline %>% mutate(text = iconv(text, "", "UTF-8", sub=" "))
.. and this doesn't work with sparlyr. Looks like iconv isn't available.
I'm not sure what alternative to use for this. Hive doesn't appear to have an equivalent. And spark_read_json doesn't have an option like read_csv has...
There's a possible regex approach here:
Remove non-utf8 characters from string
But I was just wondering if there is something a little less involved already available...
Thanks
Related
I'm trying to take an input path and convert the slashes from backwards to forwards as part of an R package. There are some suggestions to do this here and elsewhere, but all of these solutions require some amount of interaction by the user: either copying a path, or selecting a path from a prompt. I would be happy if I could find a solution that works on strings, but when I try to use strings with slashes in R, I run into problems.
# example path
path <- "C:\aaa\bbb\fff\"\n" # I have to add this \n linebreak for R to save the object without an error
# replace slashes in string
gsub(pattern="\"", replacement="/", x=path) # I have to put an extra quote in the pattern to avoid error, but this makes it not work.
Note that to avoid errors in this example, I had to save the path with a linebreak and there is an extra quote in the pattern for gsub. This is only a problem on Windows computers, but I need my package to work on Windows. Thanks in advance for suggestions.
I've got a strange text file with a bunch of NUL characters in it (actually about 10 such files), and I'd like to programmatically replace them from within R. Here is a link to one of the files.
With the aid of this question I've finally figured out a better-than-ad-hoc way of going into each file and find-and-replacing the nuisance characters. It turns out that each pair of them should correspond to one space ([NUL][NUL]->) to maintain the intended line width of the file (which is crucial for reading these as fixed-width further down the road).
However, for robustness' sake, I prefer a more automable approach to the solution, ideally (for organization's sake) something I could add at the beginning of an R script I'm writing to clean up the files. This question looked promising but the accepted answer is insufficient - readLines throws an error whenever I try to use it on these files (unless I activate skipNul).
Is there any way to get the lines of this file into R so I could use gsub or whatever else to fix this issue without resorting to external programs?
You want to read the file as binary then you can substitute the NULs, e.g. to replace them by spaces:
r = readBin("00staff.dat", raw(), file.info("00staff.dat")$size)
r[r==as.raw(0)] = as.raw(0x20) ## replace with 0x20 = <space>
writeBin(r, "00staff.txt")
str(readLines("00staff.txt"))
# chr [1:155432] "000540952Anderson Shelley J FW1949 2000R000000000000119460007620 3 0007000704002097907KGKG1616"| __truncated__ ...
You could also substitute the NULs with a really rare character (such as "\01") and work on the string in place, e.g., let's say if you want to replace two NULs ("\00\00") with one space:
r = readBin("00staff.dat", raw(), file.info("00staff.dat")$size)
r[r==as.raw(0)] = as.raw(1)
a = gsub("\01\01", " ", rawToChar(r), fixed=TRUE)
s = strsplit(a, "\n", TRUE)[[1]]
str(s)
# chr [1:155432] "000540952Anderson Shelley J FW1949 2000R000000000000119460007620 3 0007000704002097907KGKG1616"| __truncated__
Am using the RODBC library to bring data into R. I have a long query that I want to pass a variable to, much like this SO user.
Problem is that R interprets the whitespace/carriage returns in my query as a newline '\n'.
The accepted solution for this question suggests to simply break up the text into chunks and then paste() together - which works, but ideally I'd like to keep the whitespace intact - makes it easier to test/verify the behavior of the query over in the database before pasting into R.
In other languages I'm familiar with there's a simple line continuation character - indeed, several of the comments on the accepted answer are looking for an approach similar to python's \.
I found an aside to a workaround using strwrap deep in the bowels of an R discussion lists, so in the interest of making the internet better I will post it here. However, if someone can point the direction toward a more elegant/straightforward solution, I will happily accept your answer.
I don't know if you will find this helpful or not, but I have eventually gravitated towards keeping my SQL separate from my R scripts. Keeping the query in my R script, except for very very short ones, I find gets unreadable very quickly.
These days, I tend to keep queries that are more than a single line in their own separate .sql file. Then I can keep them nice and formatted and readable in a nice text editor, and read them into R as needed via something like this:
read_sql <- function(path){
stopifnot(file.exists(path))
sql <- readChar(path,nchar = file.info(path)$size)
sql
}
For binding parameters into the queries, I just keep a %s where the parameter will go in the .sql file, and then add in the parameters in R using sprintf.
I've been much happier this way, as I was finding that cluttering up my R scripts with really long paste statements and multi-line character objects was making my code really hard to read.
R's strwrap will destroy whitespace, including newline characters, per the documentation.
Essentially, you can get the desired behavior by initially letting R introduce line breaks/newline \ns, and then immediately stripping them out.
#make query using PASTE
query_1 <- paste("SELECT map.ps_studentid
,students.first_name || ' ' || students.last_name AS full_name
,map.testritscore
,map.termname
,map.measurementscale
FROM map$comprehensive_with_growth map
JOIN students
ON map.ps_studentid = students.id
WHERE map.termname = '",map_term,"'", sep='')
#remove newline characters introduced above.
#width is an arbitrary big number-
#it just needs to be longer than your string.
query_1 <- strwrap(query_1, width=10000, simplify=TRUE)
#execute the query
map_njask <- sqlQuery(XE, query_1)
query <- gsub(pattern='\\s',replacement="",x=query)
Try using sprintf to get variable substitution, and then replacing all newlines and whitespace.
See my answer to a similar question for details.
I am trying to read this file (3.8mb) using its fixed-width structure as described in the following link.
This command:
a <- read.fwf('~/ccsl.txt',c(2,30,6,2,30,8,10,11,6,8))
Produces an error:
line 37 did not have 10 elements
After replicating the issue with different values of the skip option, I figured that the lines causing the problem all contain the "#" symbol.
Is there any way to get around it?
As #jverzani already commented, this problem is probably the fact that the # sign often used as a character to signal a comment. Setting the comment.char input argument of read.fwf to something other than # could fix the problem. I'll leave my answer below as a more general case that you can use on any character that causes problems (e.g. the 's in the Dutch city name 's Gravenhage).
I've had this problem occur with other symbols. The approach I took was to simply replace the # by either nothing, or by a character which does not generate the error. In my case it was no problem to simply replace the character, but this might not be possible in your case.
So my approach would be to delete the symbol that generates the error, or replace by another character. This can be done using a text editor (find and replace), in an R script, or using some linux tools called grep and sed. If you want to do this in an R script, use scan or readLines to read the lines. Once the text is in memory, you can use sub to replace the character.
If you cannot replace the character, I would try the following approach: replace the character by a character that does not generate an error, read it into R using read.fwf, and finally replace the character by the # character.
Following up on the answer above: to get all characters to be read as literals, use both comment.char="" and quote="" (the latter takes care of #PaulHiemstra's problem with single-quotes in Dutch proper nouns) in the call to read.fwf (this is documented in ?read.table).
I've been using gsub("toreplace","replacement", myvector) to clean out data in R. While this works for commas and the like, removing "$" has no effect. So if I do gsub("$","",myvector) all the dollar signs remain in place.
I think this is because $ is a special character in R. I tried escaping it "\$" but that yields the same result (no effect). And I couldn't find a resource on escaping special characters in R.
Obviously I should do this in preprocessing. But I was wondering if anyone out there knew how to either a) escape special characters in R b) get rid of pesky $ in R directly. For science.
You have to escape it twice, first for R, second for the regex.
gsub('\\$', '', c("a$a", "bb$"))
[1] "aa" "bb"
See ?Quotes for details on quoting and escaping.
Use fixed = TRUE:
gsub('$', '', c("a$a", "bb$"), fixed = TRUE)
Then you don't need to worry about any special characters. In stringr, this is implemented a little differently:
library(stringr)
str_replace_all(c("$100","ta$ty"), fixed("$"), "")
Thanks to DiggyF and James for the examples!
Escaping characters can be a pain some times, but just putting it in square brackets (make it a character class) helps with this:
> gsub("[$]","",c("$100","ta$ty"))
[1] "100" "taty"
if you have $ followed by number in set of data columns (e.g. $400,000) there is an easier way that worked like charm for me.
data%>%
mutate_at(5:6, parse_number)
where 5:6 are the data column numbers.