A drop-in replacement for base::strsplit
.
strsplit_ctl(
x,
split,
fixed = FALSE,
perl = FALSE,
useBytes = FALSE,
warn = getOption("fansi.warn", TRUE),
term.cap = getOption("fansi.term.cap", dflt_term_cap()),
ctl = "all",
normalize = getOption("fansi.normalize", FALSE),
carry = getOption("fansi.carry", FALSE),
terminate = getOption("fansi.terminate", TRUE)
)
Like base::strsplit
, with Control Sequences excluded.
a character vector, or, unlike base::strsplit
an object that can
be coerced to character.
character vector (or object which can be coerced to such)
containing regular expression(s) (unless fixed = TRUE
)
to use for splitting. If empty matches occur, in particular if
split
has length 0, x
is split into single characters.
If split
has length greater than 1, it is re-cycled along
x
.
logical. If TRUE
match split
exactly, otherwise
use regular expressions. Has priority over perl
.
logical. Should Perl-compatible regexps be used?
logical. If TRUE
the matching is done
byte-by-byte rather than character-by-character, and inputs with
marked encodings are not converted. This is forced (with a warning)
if any input is found which is marked as "bytes"
(see Encoding
).
TRUE (default) or FALSE, whether to warn when potentially
problematic Control Sequences are encountered. These could cause the
assumptions fansi
makes about how strings are rendered on your display
to be incorrect, for example by moving the cursor (see ?fansi
).
At most one warning will be issued per element in each input vector. Will
also warn about some badly encoded UTF-8 strings, but a lack of UTF-8
warnings is not a guarantee of correct encoding (use validUTF8
for
that).
character a vector of the capabilities of the terminal, can
be any combination of "bright" (SGR codes 90-97, 100-107), "256" (SGR codes
starting with "38;5" or "48;5"), "truecolor" (SGR codes starting with
"38;2" or "48;2"), and "all". "all" behaves as it does for the ctl
parameter: "all" combined with any other value means all terminal
capabilities except that one. fansi
will warn if it encounters SGR codes
that exceed the terminal capabilities specified (see term_cap_test
for details). In versions prior to 1.0, fansi
would also skip exceeding
SGRs entirely instead of interpreting them. You may add the string "old"
to any otherwise valid term.cap
spec to restore the pre 1.0 behavior.
"old" will not interact with "all" the way other valid values for this
parameter do.
character, which Control Sequences should be treated
specially. Special treatment is context dependent, and may include
detecting them and/or computing their display/character width as zero. For
the SGR subset of the ANSI CSI sequences, and OSC hyperlinks, fansi
will also parse, interpret, and reapply the sequences as needed. You can
modify whether a Control Sequence is treated specially with the ctl
parameter.
"nl": newlines.
"c0": all other "C0" control characters (i.e. 0x01-0x1f, 0x7F), except for newlines and the actual ESC (0x1B) character.
"sgr": ANSI CSI SGR sequences.
"csi": all non-SGR ANSI CSI sequences.
"url": OSC hyperlinks
"osc": all non-OSC-hyperlink OSC sequences.
"esc": all other escape sequences.
"all": all of the above, except when used in combination with any of the above, in which case it means "all but".
TRUE or FALSE (default) whether SGR sequence should be
normalized out such that there is one distinct sequence for each SGR code.
normalized strings will occupy more space (e.g. "\033[31;42m" becomes
"\033[31m\033[42m"), but will work better with code that assumes each SGR
code will be in its own escape as crayon
does.
TRUE, FALSE (default), or a scalar string, controls whether to
interpret the character vector as a "single document" (TRUE or string) or
as independent elements (FALSE). In "single document" mode, active state
at the end of an input element is considered active at the beginning of the
next vector element, simulating what happens with a document with active
state at the end of a line. If FALSE each vector element is interpreted as
if there were no active state when it begins. If character, then the
active state at the end of the carry
string is carried into the first
element of x
(see "Replacement Functions" for differences there). The
carried state is injected in the interstice between an imaginary zeroeth
character and the first character of a vector element. See the "Position
Semantics" section of substr_ctl
and the "State Interactions" section
of ?fansi
for details. Except for strwrap_ctl
where NA
is
treated as the string "NA"
, carry
will cause NA
s in inputs to
propagate through the remaining vector elements.
TRUE (default) or FALSE whether substrings should have
active state closed to avoid it bleeding into other strings they may be
prepended onto. This does not stop state from carrying if carry = TRUE
.
See the "State Interactions" section of ?fansi
for details.
Control Sequences are non-printing characters or sequences of characters.
Special Sequences are a subset of the Control Sequences, and include CSI
SGR sequences which can be used to change rendered appearance of text, and
OSC hyperlinks. See fansi
for details.
Several factors could affect the exact output produced by fansi
functions across versions of fansi
, R
, and/or across systems.
In general it is best not to rely on exact fansi
output, e.g. by
embedding it in tests.
Width and grapheme calculations depend on locale, Unicode database
version, and grapheme processing logic (which is still in development), among
other things. For the most part fansi
(currently) uses the internals of
base::nchar(type='width')
, but there are exceptions and this may change in
the future.
How a particular display format is encoded in Control Sequences is
not guaranteed to be stable across fansi
versions. Additionally, which
Special Sequences are re-encoded vs transcribed untouched may change.
In general we will strive to keep the rendered appearance stable.
To maximize the odds of getting stable output set normalize_state
to
TRUE
and type
to "chars"
in functions that allow it, and
set term.cap
to a specific set of capabilities.
fansi
is unaware of text directionality and operates as if all strings are
left to right (LTR). Using fansi
function with strings that contain mixed
direction scripts (i.e. both LTR and RTL) may produce undesirable results.
This function works by computing the position of the split points after
removing Control Sequences, and uses those positions in conjunction with
substr_ctl
to extract the pieces. This concept is borrowed from
crayon::col_strsplit
. An important implication of this is that you cannot
split by Control Sequences that are being treated as Control Sequences.
You can however limit which control sequences are treated specially via the
ctl
parameters (see examples).
?fansi
for details on how Control Sequences are
interpreted, particularly if you are getting unexpected results,
normalize_state
for more details on what the normalize
parameter does,
state_at_end
to compute active state at the end of strings,
close_state
to compute the sequence required to close active state.
strsplit_ctl("\033[31mhello\033[42m world!", " ")
## Splitting by newlines does not work as they are _Control
## Sequences_, but we can use `ctl` to treat them as ordinary
strsplit_ctl("\033[31mhello\033[42m\nworld!", "\n")
strsplit_ctl("\033[31mhello\033[42m\nworld!", "\n", ctl=c("all", "nl"))
Run the code above in your browser using DataLab