check_text: Check Text For Potential Problems

In trinker/textmod: Text Cleaning Tools

Description

check_text - Uncleaned text may result in errors, warnings, and incorrect results in subsequent analysis. check_text checks text for potential problems and suggests possible fixes. Potential text anomalies that are detected include: factors, missing ending punctuation, empty cells, double punctuation, non-space after comma, no alphabetic characters, non-ASCII, missing value, and potentially misspelled words.

available_check - Provide a data.frame view of all the available checks in the check_text function.

Usage

check_text(x, file = NULL, checks = NULL, n = 10, ...)

available_checks()

Arguments

x	The text variable.
file	A connection, or a character string naming the file to print to. If NULL prints to the console. Note that this is assigned as an attribute and passed to print.
checks	A vector of checks to include from which_are. If checks = NULL, all checks from which_are which be used. Note that all meta checks will be conducted (see which_are for details on meta checks).
n	The number of affected elements to print out (the rest are truncated).
...	ignored.

Value

Returns a list with the following potential text faults report:

• contraction- Text elements that contain contractions

• date- Text elements that contain dates

• digit- Text elements that contain digits/numbers

• email- Text elements that contain email addresses

• emoticon- Text elements that contain emoticons

• empty- Text elements that contain empty text cells (all white space)

• escaped- Text elements that contain escaped back spaced characters

• hash- Text elements that contain Twitter style hash tags (e.g., #rstats)

• html- Text elements that contain HTML markup

• incomplete- Text elements that contain incomplete sentences (e.g., uses ending punctuation like ...)

• kern- Text elements that contain kerning (e.g., 'The B O M B!')

• list_column- Text variable that is a list column

• missing_value- Text elements that contain missing values

• misspelled- Text elements that contain potentially misspelled words

• no_alpha- Text elements that contain elements with no alphabetic (a-z) letters

• no_endmark- Text elements that contain elements with missing ending punctuation

• no_space_after_comma- Text elements that contain commas with no space afterwards

• non_ascii- Text elements that contain non-ASCII text

• non_character- Text variable that is not a character column (likely factor)

• non_split_sentence- Text elements that contain unsplit sentences (more than one sentence per element)

• tag- Text elements that contain Twitter style handle tags (e.g., @trinker)

• time- Text elements that contain timestamps

• url- Text elements that contain URLs

Note

The output is a list containing meta checks and elemental checks but prints as a pretty formatted output with potential problem elements, the accompanying text, and possible suggestions to fix the text.

Examples

## Not run: 
v <- list(c('foo', 'bar'), NA, c('hello', 'world'))
check_text(v)

w <- factor(unlist(v))
check_text(w)

x <- c("i like", "<p>i want. </p>thet them ther .", "I am ! that|", "", NA, 
    "&quot;they&quot;,were there", ".", "   ", "?", "3;", "I like goud eggs!", 
    "i 4like...", "\\tgreat",  'She said "yes"')
check_text(x)
print(check_text(x), include.text=FALSE)
check_text(x, checks = c('non_split_sentence', 'no_endmark'))
elementals <- available_checks()[is_meta != TRUE,][['fun']]
check_text(
    x, 
    checks = elementals[
        !elementals %in% c('non_split_sentence', 'no_endmark')
    ]
)

y <- c("A valid sentence.", "yet another!")
check_text(y)

z <- rep("dfsdsd'nt", 120)
check_text(z)

## End(Not run)

trinker/textmod documentation built on Nov. 3, 2021, 7:20 p.m.