Add metweight tool

[WolfgangDrescher]

I added a metweight tool that analyzes kern files with Tool_meter and creates a new spine for each selected voice (-k or -s) that outputs the meter weight of every token.

Generated data tokens are as follows:

Weight class	Abbreviation	Full word	Number
strong	"s"	"strong"	"1"
half-strong	"hs"	"half-strong"	"2"
weak	"w"	"weak"	"3"
unclassified	"u"	"unclassified"	"4"

These options are available:

define("f|full=b",         "print full text labels (strong/half-strong/weak) instead of abbreviations");
define("i|integer=b",      "print integer rank labels (1/2/3) instead of abbreviations");
define("x|cdata=b",        "label the spine **cdata-metweight instead of **metweight");
define("k|kern-tracks=s",  "process only the specified kern spines");
define("s|spine|spines=s", "process only the specified spines");
define("n|null=b",         "always use the null token . for unclassified positions");

Example usage:

Command:

metweight -nx -k1,$ ./bach-370-chorales/kern/chor029.krn | myank -m 1-4 | ridxx -LG

Output:

**kern	**cdata-metweight	**kern	**kern	**kern	**cdata-metweight
*clefF4	*	*clefGv2	*clefG2	*clefG2	*
*k[f#]	*	*k[f#]	*k[f#]	*k[f#]	*
*G:	*	*G:	*G:	*G:	*
*M4/4	*	*M4/4	*M4/4	*M4/4	*
*met(c)	*	*met(c)	*met(c)	*met(c)	*
*MM100	*	*MM100	*MM100	*MM100	*
=1-	=1-	=1-	=1-	=1-	=1-
4G	s	4B	4d	4g	s
4F#	w	4A	4d	4a	w
4G	hs	[4G	4d	4b	hs
4D	w	8GL]	4d	4a	w
.	.	8F#J	.	.	.
=2	=2	=2	=2	=2	=2
4E	s	4G	8dL	4g	s
.	.	.	8c#J	.	.
8F#L	w	4A	4d	4f#	w
8GJ	.	.	.	.	.
4A	hs	8AL	4c#	4e	hs
.	.	8GJ	.	.	.
4D;	w	4F#;	4A;	4d;	w
=3	=3	=3	=3	=3	=3
4E	s	4B	4e	4g	s
4F#	w	4A	4d	4a	w
4G	hs	8GL	4d	4b	hs
.	.	8F#J	.	.	.
4AA	w	8EL	4a	4cc	w
.	.	8F#J	.	.	.
=4	=4	=4	=4	=4	=4
8BBL	s	4.G	4d	4b	s
8CJ	.	.	.	.	.
4D	w	.	4d	4a	w
.	.	8F#	.	.	.
2GG;	hs	2B;	2d;	2g;	hs
=	=	=	=	=	=
*-	*-	*-	*-	*-	*-

[craigsapp]

Metweight is a bit long, I have usd metpos (Metrical position) in the paste (humdrum extras).

Why are there two analysis spines? Ideally the analytic output:

Better is one, in the first spine. (displayed below the system).

Also it would be useful add a -c or --color option to color code the analyisis as demonstrated above.

Also it would be useful to add a -n or --numeric option to convert the categories to numbers: 0 would be the beat level, -1 is the half note -2 is the whole note, +1 is the eighth note.and so on (it is log_2(rhythm / 3), such as eigthth note: log_2(8/4) = log_2(2) = +1. whole note is (log_2(1/4) = -2.

for ternerary probably take the floor log_2. 12 → 8 = log_2(8/2) = +1. or log_2(12/4) = log_2(3) = +1.58 = floor(1.58) = 1.

Also it would be good to add a -b option to highlight only the beat level and higher of the music. (0 numeric and negative values).

Here is adding composite rhythm analysis:

Composite only:

[WolfgangDrescher]

Metweight is a bit long

Personally I don't mind the long name. But is this a good terminology? I am not a native speaker. metlev does something a bit different than my implementation (plus, its currently not running see #117)

I have usd metpos (Metrical position) in the paste (humdrum extras).

I could not get `metpos` running (I think it's AWK and not humextra). Click here to see bash stdout.

Command:

metpos ./bach-370-chorales/kern/chor029.krn

stdout:

!!!!SEGMENT: chor029.krn
!!!COM: Bach, Johann Sebastian
!!!CDT: 1685/02/21/-1750/07/28/
!!!OTL@@DE: Freu dich sehr, o meine Seele
!!!SCT: BWV 32/6
!!!PC#: 29
!!!AGN: chorale
**kern	**kern	**kern	**kern	**metpos
*ICvox	*ICvox	*ICvox	*ICvox	*
*Ibass	*Itenor	*Ialto	*Isoprn	*
*I"Bass	*I"Tenor	*I"Alto	*I"Soprano	*
*>[A,A,B]	*>[A,A,B]	*>[A,A,B]	*>[A,A,B]	*
*>norep[A,B]	*>norep[A,B]	*>norep[A,B]	*>norep[A,B]	*
*>A	*>A	*>A	*>A	*
*clefF4	*clefGv2	*clefG2	*clefG2	*
*k[f#]	*k[f#]	*k[f#]	*k[f#]	*
*G:	*G:	*G:	*G:	*
*M4/4	*M4/4	*M4/4	*M4/4	*M4/4
*M4/4	*M4/4	*M4/4	*M4/4	*M4/4
*M4/4	*M4/4	*M4/4	*M4/4	*M4/4
*M4/4	*M4/4	*M4/4	*M4/4	*M4/4
*met(c)	*met(c)	*met(c)	*met(c)	*
!!xywh-#24:169,0,2736,477
=1-	=1-	=1-	=1-	=1-
metpos: ERROR: Data record encountered prior to timebase definition at line 23.

Why are there two analysis spines?

This comes from the -k1,$ arg. I just used a random "complicated" metweight command to show some of the options available. The command produces a **metweight spine for each **kern spine. In my use case I need to be able to scan each voice for strong and weak beats to provide detailed search options for the Bach chorales corpus.

The output I actually need and would prefer looks like this:

metweight ./bach-370-chorales/kern/chor029.krn | myank -m 1-4 | ridxx -LG

**kern	**metweight	**kern	**metweight	**kern	**metweight	**kern	**metweight
*clefF4	*	*clefGv2	*	*clefG2	*	*clefG2	*
*k[f#]	*	*k[f#]	*	*k[f#]	*	*k[f#]	*
*G:	*	*G:	*	*G:	*	*G:	*
*M4/4	*	*M4/4	*	*M4/4	*	*M4/4	*
*met(c)	*	*met(c)	*	*met(c)	*	*met(c)	*
*MM100	*	*MM100	*	*MM100	*	*MM100	*
=1-	=1-	=1-	=1-	=1-	=1-	=1-	=1-
4G	s	4B	s	4d	s	4g	s
4F#	w	4A	w	4d	w	4a	w
4G	hs	[4G	hs	4d	hs	4b	hs
4D	w	8GL]	.	4d	w	4a	w
.	.	8F#J	u	.	.	.	.
=2	=2	=2	=2	=2	=2	=2	=2
4E	s	4G	s	8dL	s	4g	s
.	.	.	.	8c#J	u	.	.
8F#L	w	4A	w	4d	w	4f#	w
8GJ	u	.	.	.	.	.	.
4A	hs	8AL	hs	4c#	hs	4e	hs
.	.	8GJ	u	.	.	.	.
4D;	w	4F#;	w	4A;	w	4d;	w
=3	=3	=3	=3	=3	=3	=3	=3
4E	s	4B	s	4e	s	4g	s
4F#	w	4A	w	4d	w	4a	w
4G	hs	8GL	hs	4d	hs	4b	hs
.	.	8F#J	u	.	.	.	.
4AA	w	8EL	w	4a	w	4cc	w
.	.	8F#J	u	.	.	.	.
=4	=4	=4	=4	=4	=4	=4	=4
8BBL	s	4.G	s	4d	s	4b	s
8CJ	u	.	.	.	.	.	.
4D	w	.	.	4d	w	4a	w
.	.	8F#	u	.	.	.	.
2GG;	hs	2B;	hs	2d;	hs	2g;	hs
=:|!	=:|!	=:|!	=:|!	=:|!	=:|!	=:|!	=:|!
*-	*-	*-	*-	*-	*-	*-	*-

Better is one, in the first spine. (displayed below the system).

You can already do this with -k1/-s1.

Also it would be useful add a -c or --color option to color code the analyisis as demonstrated above.

This is a nice idea I have implemented it in 48a9216.

I have also implemented --composite in 8db83b4. Maybe I should remove it again and allow usage of metweight in combination with composite without having Tool_composite inside of Tool_metweight.

Also it would be useful to add a -n or --numeric option to convert the categories to numbers: 0 would be the beat level, -1 is the half note -2 is the whole note, +1 is the eighth note.and so on (it is log_2(rhythm / 3), such as eigthth note: log_2(8/4) = log_2(2) = +1. whole note is (log_2(1/4) = -2.

for ternerary probably take the floor log_2. 12 → 8 = log_2(8/2) = +1. or log_2(12/4) = log_2(3) = +1.58 = floor(1.58) = 1.

Also it would be good to add a -b option to highlight only the beat level and higher of the music. (0 numeric and negative values).

I will have to think about this in a bit more detail. For now I think the logic I need is slightly different, but I need to check out this approach.

Here is my kern file that I am using for debugging:

**kern
*M4/2
=
8gL
8g
8g
8gJ
8gL
8g
8g
8gJ
8gL
8g
8g
8gJ
8gL
8g
8g
8gJ
=||
*M3/2
8gL
8g
8g
8gJ
8gL
8g
8g
8gJ
8gL
8g
8g
8gJ
=||
*M2/2
8gL
8g
8g
8gJ
8gL
8g
8g
8gJ
=||
*M6/4
16gL
16g
16g
16gJ
16gL
16g
16g
16gJ
16gL
16g
16g
16gJ
16gL
16g
16g
16gJ
16gL
16g
16g
16gJ
16gL
16g
16g
16gJ
=||
*M4/4
16gL
16g
16g
16gJ
16gL
16g
16g
16gJ
16gL
16g
16g
16gJ
16gL
16g
16g
16gJ
=||
*M3/4
16gL
16g
16g
16gJ
16gL
16g
16g
16gJ
16gL
16g
16g
16gJ
=||
*M2/4
16gL
16g
16g
16gJ
16gL
16g
16g
16gJ
=||
*M5/4
16gL
16g
16g
16gJ
16gL
16g
16g
16gJ
16gL
16g
16g
16gJ
16gL
16g
16g
16gJ
16gL
16g
16g
16gJ
=||
*M12/8
16gLL
16gJ
16gL
16gJ
16gL
16gJJ
16gLL
16gJ
16gL
16gJ
16gL
16gJJ
16gLL
16gJ
16gL
16gJ
16gL
16gJJ
16gLL
16gJ
16gL
16gJ
16gL
16gJJ
=||
*M9/8
16gLL
16gJ
16gL
16gJ
16gL
16gJJ
16gLL
16gJ
16gL
16gJ
16gL
16gJJ
16gLL
16gJ
16gL
16gJ
16gL
16gJJ
=||
*M6/8
16gLL
16gJ
16gL
16gJ
16gL
16gJJ
16gLL
16gJ
16gL
16gJ
16gL
16gJJ
=||
*M3/8
32gL
32g
32g
32gJ
32gL
32g
32g
32gJ
32gL
32g
32g
32gJ
==
*-