|
|
|
context [-b regexp] [-d string] [-e regexp] [-i regexp] [-n n] [-o regexp] [-p n] [inputfile ...]
In its simplest mode of operation
context
will join a specified number of successive data records together
to form a single output record.
By way of example, consider a file (named input) consisting of a single
spine whose data records contain the numbers 1 through 6 on separate lines:
The command
**numbers 1 2 3 4 5 6 *-
context -n 3 input
will produce the following output:
Notice that the output file has been padded with null tokens so that
the number of output records matches the number of input records.
By invoking the
-p
option, the padded null tokens can be placed either at the beginning or the
end of the file, or split between beginning and end.
For example, the command
**numbers 1 2 3 2 3 4 3 4 5 4 5 6 . . *-
context -n 3 -p 3 input
will place the trailing null tokens in the above example at the beginning of the output. The -p option defaults to the value 0.
If null tokens are present in the input, they remain in place, yet do not affect the congealed data records. For example, if a null token was present between the numbers 1 and 2 in the above input, the command
context -n 4 -p 1 input
would produce an output beginning with a single padded null token:
Rather than specifying a fixed number of congealed data records,
input records can alternatively be amalgamated according to the
signifiers present in the input data itself.
The
-e
option allows the user to specify an "end" signifier.
When this signifier is encountered in the input,
the input record is appended to the current congealed record
-- which is then output -- and a new congealed output record begins.
End signifiers are defined as string patterns using the
regular expression
syntax (see regexp (5)).
For example, given an input of six successive numbers, the command
**numbers.1 2 3 4.2 3 4 53 4 5 6..*-
context -e [0246] input
would produce the following output:
**numbers1 2.3 4.5 6.*-
Options are specified in the command line. Note that the -b and -e options are mutually exclusive with the -n, -p, and -i options.
-b regexp begin a new output record starting with token matching regexp -d string use string as output delimiter for input records rather than the space character -e regexp begin a new output record starting after token matching regexp -h displays a help screen summarizing the command syntax -i regexp ignore any records matching regexp when counting -n n amalgamate n input data records for each output record -o regexp omit any records matching regexp from amalgamated output; do not count -p n pad n (normally trailing) null tokens at the beginning of the output spine
In the default operation, context separates amalgamated tokens by inserting a space character. (Thus the input tokens are treated as subtokens in a Humdrum multiple-stop.) The -d option allows the user to define an alternative string as the subtoken delimiter.
The -n option allows the user to specify the maximum number of data records assembled into a single output record.
The -b option allows the user to specify a "begin" marker. When this marker (regexp) is matched in the input, any current congealed record is output, and a new congealed record begins. Begin signifiers are defined as string patterns using the regular expression syntax.
With the -e option, if context encounters a data record matching regexp then it appends the current input record to the current assembled output record and begins assembling a new record with the following input record.
The -i option is used only with -n; it causes any data records matching regexp to be ignored in the counting of amalgamated tokens. Such "uncounted" records are nevertheless output.
The -o option causes data records matching regexp to be omitted from the output.
The -p option may be used in conjunction with -n. Normally, the output from context -n is padded with trailing null tokens -- one fewer in number than the value specified with -n. The -p n option causes n null tokens to be padded at the beginning of the output spine, rather than trailing at the end.
Note that tandem interpretations and comments are processed like null tokens; they are merely echoed in the output in their appropriate position. Note also that context automatically breaks a congealed output record whenever it encounters a spine-path terminator or exclusive interpretation in the input.
!! Edgar Varèse, Density 21.5 (1936)!! excerpt: mm.41-45**kern*MM72=41(16f#16e#[8gn2.g_=424g])4r4r8r(16f#16e#=436gn)(6e#6f#8g)(8f#12e#12g12dn=442a-)(4an8een[8bb-=454bb-]2.ccc#)=458r*-
A simple transformation would be to amalgamate successive data
records in overlapping groups of 3.
The following command:
context -n 3 density
would produce the following output:
Notice once again that the input and output have the same number of records.
Preserving the structure in this way allows the user to
coordinate the contextual output with the original input using the
assemble
command.
!! Edgar Varèse, Density 21.5 (1936)!! excerpt: mm.41-45**kern*MM72=41 (16f# 16e#(16f# 16e# [8gn16e# [8gn 2.g_[8gn 2.g_ =422.g_ =42 4g])=42 4g]) 4r4g]) 4r 4r4r 4r 8r4r 8r (16f#8r (16f# 16e#(16f# 16e# =4316e# =43 6gn)=43 6gn) (6e#6gn) (6e# 6f#(6e# 6f# 8g)6f# 8g) (8f#8g) (8f# 12e#(8f# 12e# 12g12e# 12g 12dn12g 12dn =4412dn =44 2a-)=44 2a-) (4an2a-) (4an 8een(4an 8een [8bb-8een [8bb- =45[8bb- =45 4bb-]=45 4bb-] 2.ccc#)4bb-] 2.ccc#) =452.ccc#) =45 8r..*-
A more useful transformation might amalgamate successive data records in overlapping groups of 3 notes; that is where rests and barlines are not counted. The following command causes input records containing either an equals-sign or the letter `r' to be ignored when counting the number of amalgamated data records:
context -n 3 -i [=r] density
The input and corresponding output are given in the left
and right spines below:
Notice that as the end of the file is approached,
context
will continue amalgamating data records until it
is no longer able to satisfy the amalgamating criteria.
If unable to complete an output record,
context
will output a null token.
!! Edgar Varèse, Density 21.5 (1936)!! excerpt: mm.41-45
**kern**kern*MM72*MM72=41=41 (16f# 16e# [8gn(16f#(16f# 16e# [8gn16e#16e# [8gn 2.g_[8gn[8gn 2.g_ =42 4g])2.g_2.g_ =42 4g]) 4r 4r 8r (16f#=42=42 4g]) 4r 4r 8r (16f# 16e#4g])4g]) 4r 4r 8r (16f# 16e#4r4r 4r 8r (16f# 16e# =43 6gn)4r4r 8r (16f# 16e# =43 6gn)8r8r (16f# 16e# =43 6gn)(16f#(16f# 16e# =43 6gn)16e#16e# =43 6gn) (6e#=43=43 6gn) (6e# 6f#6gn)6gn) (6e# 6f#(6e#(6e# 6f# 8g)6f#6f# 8g) (8f#8g)8g) (8f# 12e#(8f#(8f# 12e# 12g12e#12e# 12g 12dn12g12g 12dn =44 2a-)12dn12dn =44 2a-) (4an=44=44 2a-) (4an 8een2a-)2a-) (4an 8een(4an(4an 8een [8bb-8een8een [8bb- =45 4bb-][8bb-[8bb- =45 4bb-] 2.ccc#)=45. 4bb-]. 2.ccc#). =45. 8r. *-*-
If the above command had used the -o rather than the -i option, all of the rests and barlines would have been omitted from the output. Otherwise, the output would be the same as given above.
A more musically useful partitioning of Varèse's work might be based on slur markings. The following command uses open- and closed-slur markings to demarcate the contextual outputs; (note the need to escape the parentheses since they are regular expression metacharacters).
context -b '\(' -e '\)' -o '[=r]' density
Notice that the ensuing output (shown below) pads the output with
null tokens in order to maintain the same
number of data records as the original input.
(An output such as the following might be used as input to
a command sequence such as pc -x | pcset).
If there were any notes not embraced within a slur,
the above command would have caused them to be output on
their own line.
!! Edgar Varèse, Density 21.5 (1936)!! excerpt: mm.41-45**kern*MM72.(16f# 16e# [8gn 2.g_ 4g]).. . . . . . . (16f# 16e# 6gn). . . (6e# 6f# 8g). . (8f# 12e# 12g 12dn 2a-). . . . . (4an 8een [8bb- 4bb-] 2.ccc#). . . . . . . *-