textfile not picking up document names

[conjugateprior]

Should textfile pick up document names? In quanteda 0.9.7-17 (CRAN's current version) on R 3.3.0 it does not.

corp <- corpus(textfile("folder/*.txt"))
summary(corp)

Corpus consisting of 12 documents.

   Text Types Tokens Sentences
  text1  2112  17094       798
  text2   522   1896        89
  text3  1379   6439       290
  text4   650   2612       143
  text5   715   2300       125
  text6   400   1206        53
  text7   641   2891       162
  text8   460   1473        71
  text9   413   1256        66
 text10   455   1201        59
 text11  1502   9805       422
 text12   643   2177       106

  ...

It requires a subsequent docnames argument.

[kbenoit]

It should be automatic - the previous textfile assigned the filename to the docname. @adamobeng can you restore this behaviour?

[adamobeng]

Sorry I missed that. Thanks for noticing it, @conjugateprior!

As of 0.9.6-0 (cc31d3c), this only worked for txt files where the argument to textfile() was a glob pattern or a vector of filenames.

/tmp/tests
├── csv
│   ├── dog.csv
│   └── fox.csv
└── txt
    ├── dog.txt
    └── fox.txt

> names(textfile("/tmp/tests/txt/*")@texts)
[1] "dog.txt" "fox.txt"
> names(textfile("/tmp/tests/txt/fox.txt")@texts)
NULL
> names(textfile(c("/tmp/tests/txt/fox.txt", "/tmp/tests/txt/dog.txt"))@texts)
[1] "fox.txt" "dog.txt"
> names(textfile("/tmp/tests/csv/*", textField='text')@texts)
NULL
> names(textfile("/tmp/tests/csv/fox.csv", textField='text')@texts)
NULL

With pull request 225, the behaviour is as expected:

> names(textfile("/tmp/tests/txt/*")@texts)
[1] "dog.txt" "fox.txt"
> names(textfile("/tmp/tests/txt/fox.txt")@texts)
[1] "fox.txt"
> names(textfile(c("/tmp/tests/txt/fox.txt", "/tmp/tests/txt/dog.txt"))@texts)
[1] "fox.txt" "dog.txt"
> names(textfile("/tmp/tests/csv/*", textField='text')@texts)
[1] "dog.csv" "fox.csv"
> names(textfile("/tmp/tests/csv/fox.csv", textField='text')@texts)
[1] "fox.csv"