asciiunits 0.1.0

data/corpora/unitsml/expected-kinds.json

@@ -0,0 +1,502 @@
+{
+  "format": "asciiunits-oracle/v1",
+  "source": "unitsml 0.6.9",
+  "cases": [
+    {
+      "input": "m",
+      "kind": "accept"
+    },
+    {
+      "input": "mm",
+      "kind": "accept"
+    },
+    {
+      "input": "um",
+      "kind": "accept"
+    },
+    {
+      "input": "degK",
+      "kind": "accept"
+    },
+    {
+      "input": "kg",
+      "kind": "accept"
+    },
+    {
+      "input": "mm*s^-2",
+      "kind": "accept"
+    },
+    {
+      "input": "sqrt(Hz)",
+      "kind": "accept"
+    },
+    {
+      "input": "K/(kg*m)",
+      "kind": "accept"
+    },
+    {
+      "input": "cal_th/cm^2",
+      "kind": "accept"
+    },
+    {
+      "input": "Hz^10*darcy^100",
+      "kind": "accept"
+    },
+    {
+      "input": "A*C^3",
+      "kind": "accept"
+    },
+    {
+      "input": "A/C^-3",
+      "kind": "accept"
+    },
+    {
+      "input": "J/kg*K",
+      "kind": "accept"
+    },
+    {
+      "input": "dim_Theta*dim_L^2",
+      "kind": "accept"
+    },
+    {
+      "input": "kg^-2",
+      "kind": "accept"
+    },
+    {
+      "input": "mW*cm^(-2)",
+      "kind": "accept"
+    },
+    {
+      "input": "p-",
+      "kind": "accept"
+    },
+    {
+      "input": "k-",
+      "kind": "accept"
+    },
+    {
+      "input": "1",
+      "kind": "accept"
+    },
+    {
+      "input": "(m)(s)",
+      "kind": "accept"
+    },
+    {
+      "input": "m(s)",
+      "kind": "accept"
+    },
+    {
+      "input": "m s",
+      "kind": "accept"
+    },
+    {
+      "input": "m *s",
+      "kind": "accept"
+    },
+    {
+      "input": "m* s",
+      "kind": "accept"
+    },
+    {
+      "input": "m  s",
+      "kind": "accept"
+    },
+    {
+      "input": "m * s",
+      "kind": "accept"
+    },
+    {
+      "input": "km//s",
+      "kind": "accept"
+    },
+    {
+      "input": "s^-2//m",
+      "kind": "accept"
+    },
+    {
+      "input": "m sqrt(Hz)",
+      "kind": "accept"
+    },
+    {
+      "input": "(m^2*s^-1)//kg",
+      "kind": "accept"
+    },
+    {
+      "input": "dim_L",
+      "kind": "accept"
+    },
+    {
+      "input": "(dim_L*dim_M)/dim_T^2",
+      "kind": "accept"
+    },
+    {
+      "input": "sqrt(dim_M)",
+      "kind": "accept"
+    },
+    {
+      "input": "sqrt(m*s)",
+      "kind": "accept"
+    },
+    {
+      "input": "sqrt((m))",
+      "kind": "accept"
+    },
+    {
+      "input": "(sqrt(m))",
+      "kind": "accept"
+    },
+    {
+      "input": "sqrt(m)s",
+      "kind": "accept"
+    },
+    {
+      "input": "m^2*s",
+      "kind": "accept"
+    },
+    {
+      "input": "m^-1/2",
+      "kind": "accept"
+    },
+    {
+      "input": "m^(-1/2)",
+      "kind": "accept"
+    },
+    {
+      "input": "m^(1/2)",
+      "kind": "accept"
+    },
+    {
+      "input": "(m)(s)(A)",
+      "kind": "accept"
+    },
+    {
+      "input": "1m",
+      "kind": "accept"
+    },
+    {
+      "input": "1*m",
+      "kind": "accept"
+    },
+    {
+      "input": "m*1",
+      "kind": "accept"
+    },
+    {
+      "input": "((m))",
+      "kind": "accept"
+    },
+    {
+      "input": "dim_L dim_M",
+      "kind": "accept"
+    },
+    {
+      "input": "(dim_L)(dim_M)",
+      "kind": "accept"
+    },
+    {
+      "input": "(dim_L)*dim_M",
+      "kind": "accept"
+    },
+    {
+      "input": "mm*",
+      "kind": "accept"
+    },
+    {
+      "input": "k-",
+      "kind": "accept"
+    },
+    {
+      "input": "da-",
+      "kind": "accept"
+    },
+    {
+      "input": "M-",
+      "kind": "accept"
+    },
+    {
+      "input": "s^2/m^2",
+      "kind": "accept"
+    },
+    {
+      "input": "m/s^2",
+      "kind": "accept"
+    },
+    {
+      "input": "K*kg^0",
+      "kind": "accept"
+    },
+    {
+      "input": "Hz^-10",
+      "kind": "accept"
+    },
+    {
+      "input": "um^3",
+      "kind": "accept"
+    },
+    {
+      "input": "mmol",
+      "kind": "accept"
+    },
+    {
+      "input": "nm",
+      "kind": "accept"
+    },
+    {
+      "input": "kat",
+      "kind": "accept"
+    },
+    {
+      "input": "Bq*Sv^2",
+      "kind": "accept"
+    },
+    {
+      "input": "ft*lb/s^2",
+      "kind": "accept"
+    },
+    {
+      "input": "mu*s",
+      "kind": "accept"
+    },
+    {
+      "input": "m*s^-3*K^4",
+      "kind": "accept"
+    },
+    {
+      "input": "sqrt(m*s^-1)",
+      "kind": "accept"
+    },
+    {
+      "input": "cm^(3)",
+      "kind": "accept"
+    },
+    {
+      "input": "cal_th/cm^2 h",
+      "kind": "accept"
+    },
+    {
+      "input": "(a_year*s^-1)",
+      "kind": "accept"
+    },
+    {
+      "input": "m*s/",
+      "kind": "accept"
+    },
+    {
+      "input": "sqrt(sqrt(m))",
+      "kind": "accept"
+    },
+    {
+      "input": "ton-force (2000 lb)",
+      "kind": "accept"
+    },
+    {
+      "input": "ton-force (2000 lb)*m",
+      "kind": "accept"
+    },
+    {
+      "input": "ton-force (2000 lb)^2",
+      "kind": "accept"
+    },
+    {
+      "input": "m//s//kg",
+      "kind": "accept"
+    },
+    {
+      "input": "m/s/kg",
+      "kind": "accept"
+    },
+    {
+      "input": "m^2^3",
+      "kind": "reject"
+    },
+    {
+      "input": "m^(2)^3",
+      "kind": "reject"
+    },
+    {
+      "input": "m^(2^3)",
+      "kind": "reject"
+    },
+    {
+      "input": "m^(-2)^2",
+      "kind": "reject"
+    },
+    {
+      "input": "m*s^(1)",
+      "kind": "accept"
+    },
+    {
+      "input": "m*s^(-1)",
+      "kind": "accept"
+    },
+    {
+      "input": "m*s^(+1)",
+      "kind": "reject"
+    },
+    {
+      "input": "m*1/2",
+      "kind": "reject"
+    },
+    {
+      "input": "m/2",
+      "kind": "reject"
+    },
+    {
+      "input": "Mt",
+      "kind": "accept"
+    },
+    {
+      "input": "EiB",
+      "kind": "reject"
+    },
+    {
+      "input": "que?",
+      "kind": "reject"
+    },
+    {
+      "input": "m2",
+      "kind": "reject"
+    },
+    {
+      "input": "m^",
+      "kind": "reject"
+    },
+    {
+      "input": "sqrt",
+      "kind": "reject"
+    },
+    {
+      "input": "dim_Ldim_M",
+      "kind": "reject"
+    },
+    {
+      "input": "mm**",
+      "kind": "reject"
+    },
+    {
+      "input": "*m",
+      "kind": "reject"
+    },
+    {
+      "input": "()",
+      "kind": "reject"
+    },
+    {
+      "input": "sqrt()",
+      "kind": "reject"
+    },
+    {
+      "input": "m**",
+      "kind": "reject"
+    },
+    {
+      "input": "s-2",
+      "kind": "reject"
+    },
+    {
+      "input": "m^-1/2 (spaced)",
+      "kind": "reject"
+    },
+    {
+      "input": "m ^2",
+      "kind": "reject"
+    },
+    {
+      "input": "s ^ -2",
+      "kind": "reject"
+    },
+    {
+      "input": "m)s",
+      "kind": "reject"
+    },
+    {
+      "input": "m(s",
+      "kind": "reject"
+    },
+    {
+      "input": "(m",
+      "kind": "reject"
+    },
+    {
+      "input": "((m)",
+      "kind": "reject"
+    },
+    {
+      "input": "(m))",
+      "kind": "reject"
+    },
+    {
+      "input": "ki-",
+      "kind": "reject"
+    },
+    {
+      "input": "degK-",
+      "kind": "reject"
+    },
+    {
+      "input": "m 2",
+      "kind": "reject"
+    },
+    {
+      "input": "m^2 3",
+      "kind": "reject"
+    },
+    {
+      "input": "sqrt(m)^2",
+      "kind": "reject"
+    },
+    {
+      "input": "dim_L*",
+      "kind": "reject"
+    },
+    {
+      "input": "micro*s",
+      "kind": "reject"
+    },
+    {
+      "input": "(cm)^3",
+      "kind": "reject"
+    },
+    {
+      "input": "m1",
+      "kind": "reject"
+    },
+    {
+      "input": "-",
+      "kind": "reject"
+    },
+    {
+      "input": "--",
+      "kind": "reject"
+    },
+    {
+      "input": "^",
+      "kind": "reject"
+    },
+    {
+      "input": "*",
+      "kind": "reject"
+    },
+    {
+      "input": "^2m",
+      "kind": "reject"
+    },
+    {
+      "input": "k",
+      "kind": "reject"
+    },
+    {
+      "input": "TON",
+      "kind": "reject"
+    },
+    {
+      "input": "SQRT(m)",
+      "kind": "reject"
+    },
+    {
+      "input": "Sqrt(m)",
+      "kind": "reject"
+    }
+  ]
+}

data/corpora/unitsml/inputs.txt

@@ -0,0 +1,129 @@
+# UnitsML expression corpus inputs.
+# One per line: ACCEPT or REJECT, TAB, then the expression (the content
+# inside unitsml(...)). Declared kinds are verified against the parslet
+# parser by scripts/gen_oracle.rb — the oracle wins, fix declarations here
+# if they disagree.
+ACCEPT	m
+ACCEPT	mm
+ACCEPT	um
+ACCEPT	degK
+ACCEPT	kg
+ACCEPT	mm*s^-2
+ACCEPT	sqrt(Hz)
+ACCEPT	K/(kg*m)
+ACCEPT	cal_th/cm^2
+ACCEPT	Hz^10*darcy^100
+ACCEPT	A*C^3
+ACCEPT	A/C^-3
+ACCEPT	J/kg*K
+ACCEPT	dim_Theta*dim_L^2
+ACCEPT	kg^-2
+ACCEPT	mW*cm^(-2)
+ACCEPT	p-
+ACCEPT	k-
+ACCEPT	1
+ACCEPT	(m)(s)
+ACCEPT	m(s)
+ACCEPT	m s
+ACCEPT	m *s
+ACCEPT	m* s
+ACCEPT	m  s
+ACCEPT	m * s
+ACCEPT	km//s
+ACCEPT	s^-2//m
+ACCEPT	m sqrt(Hz)
+ACCEPT	(m^2*s^-1)//kg
+ACCEPT	dim_L
+ACCEPT	(dim_L*dim_M)/dim_T^2
+ACCEPT	sqrt(dim_M)
+ACCEPT	sqrt(m*s)
+ACCEPT	sqrt((m))
+ACCEPT	(sqrt(m))
+ACCEPT	sqrt(m)s
+ACCEPT	m^2*s
+ACCEPT	m^-1/2
+ACCEPT	m^(-1/2)
+ACCEPT	m^(1/2)
+ACCEPT	(m)(s)(A)
+ACCEPT	1m
+ACCEPT	1*m
+ACCEPT	m*1
+ACCEPT	((m))
+ACCEPT	dim_L dim_M
+ACCEPT	(dim_L)(dim_M)
+ACCEPT	(dim_L)*dim_M
+ACCEPT	mm*
+ACCEPT	k-
+ACCEPT	da-
+ACCEPT	M-
+ACCEPT	s^2/m^2
+ACCEPT	m/s^2
+ACCEPT	K*kg^0
+ACCEPT	Hz^-10
+ACCEPT	um^3
+ACCEPT	mmol
+ACCEPT	nm
+ACCEPT	kat
+ACCEPT	Bq*Sv^2
+ACCEPT	ft*lb/s^2
+ACCEPT	mu*s
+ACCEPT	m*s^-3*K^4
+ACCEPT	sqrt(m*s^-1)
+ACCEPT	cm^(3)
+ACCEPT	cal_th/cm^2 h
+ACCEPT	(a_year*s^-1)
+ACCEPT	m*s/
+ACCEPT	sqrt(sqrt(m))
+ACCEPT	ton-force (2000 lb)
+ACCEPT	ton-force (2000 lb)*m
+ACCEPT	ton-force (2000 lb)^2
+ACCEPT	m//s//kg
+ACCEPT	m/s/kg
+REJECT	m^2^3
+REJECT	m^(2)^3
+REJECT	m^(2^3)
+REJECT	m^(-2)^2
+ACCEPT	m*s^(1)
+ACCEPT	m*s^(-1)
+REJECT	m*s^(+1)
+REJECT	m*1/2
+REJECT	m/2
+ACCEPT	Mt
+REJECT	EiB
+REJECT	que?
+REJECT	m2
+REJECT	m^
+REJECT	sqrt
+REJECT	dim_Ldim_M
+REJECT	mm**
+REJECT	*m
+REJECT	()
+REJECT	sqrt()
+REJECT	m**
+REJECT	s-2
+REJECT	m^-1/2 (spaced)
+REJECT	m ^2
+REJECT	s ^ -2
+REJECT	m)s
+REJECT	m(s
+REJECT	(m
+REJECT	((m)
+REJECT	(m))
+REJECT	ki-
+REJECT	degK-
+REJECT	m 2
+REJECT	m^2 3
+REJECT	sqrt(m)^2
+REJECT	dim_L*
+REJECT	micro*s
+REJECT	(cm)^3
+REJECT	m1
+REJECT	-
+REJECT	--
+REJECT	^
+REJECT	*
+REJECT	^2m
+REJECT	k
+REJECT	TON
+REJECT	SQRT(m)
+REJECT	Sqrt(m)

data/grammars/unitsml.parg

@@ -0,0 +1,79 @@
+# UnitsML AsciiUnits expression grammar.
+#
+# 1:1 transcription of unitsml-ruby lib/unitsml/parse.rb +
+# lib/unitsml/intermediate_exp_rules.rb (Parslet), rule for rule, in the
+# same alternative order — PEG ordered choice makes order decisive, and the
+# quirks (trailing-extender acceptance, guard fall-through to prefixed
+# units, no power after fenced/sqrt forms) are load-bearing.
+#
+# Terminal tables are generated from UnitsDB via scripts/gen_tables.rb and
+# embedded in the artifact at compile time.
+#
+# Capture names (unit, prefix, dimension, integer, power_numerator, extender,
+# sqrt, int_exp, sequence, first_set, second_set, open_paren, close_paren)
+# mirror the Parslet .as() names so consumers can map them 1:1.
+
+grammar UnitsmlAscii version "0.1.0" {
+
+  ws = %x09 / %x0A / %x0B / %x0C / %x0D / %x20
+  spaces = 1*ws
+  lparen = "("
+  rparen = ")"
+  hyphen = "-"
+  star = "*"
+  digit = %x30-39
+  forward_slashes = "//" / "/"
+  extender = (forward_slashes / star) as extender
+  number = ["-"] 1*digit
+  slashed_number = (number [forward_slashes number]) as integer
+  end_guard = !ANY / &(extender / lparen / rparen / spaces)
+
+  unit = alt from_table "units" column "name" as unit
+  single_letter_prefixes = alt from_table "prefixes_single" column "name" as prefix
+  double_letter_prefixes = alt from_table "prefixes_double" column "name" as prefix
+  dimensions = alt from_table "dimensions" column "name" as dimension
+
+  power = "^" slashed_number_int_exp as power_numerator
+  slashed_number_int_exp = slashed_number / (lparen slashed_number_named_int_exp as int_exp rparen)
+  slashed_number_named_int_exp = (lparen slashed_number_int_exp rparen) / slashed_number_int_exp
+  unit_and_power = unit [power]
+  sqrt_sequence = "sqrt" lparen prefixes_units_int_exp as sqrt rparen
+  sqrt_dimensions = "sqrt" lparen dimension_rules_int_exp as sqrt rparen
+  extended_prefixed_units = extender [spaces] prefixes_units_int_exp as sequence
+  powered_dimensions = dimensions [power]
+
+  prefixes_units = (sqrt_sequence [extended_prefixed_units]) / ("1" as unit [extended_prefixed_units]) / (unit_and_power extended_prefixed_units) / (unit_and_power end_guard) / (double_letter_prefixes unit_and_power extended_prefixed_units) / (double_letter_prefixes unit_and_power) / (single_letter_prefixes unit_and_power extended_prefixed_units) / (single_letter_prefixes unit_and_power)
+
+  wrapper_prefixes_units_value = int_exp_prefixes_units / prefixes_units_int_exp
+  int_exp_prefixes_units = lparen [spaces] named_int_exp_prefixes_units as int_exp [spaces] rparen
+  prefixes_units_named_int_exp = (lparen as open_paren [spaces] wrapper_prefixes_units_value as int_exp [spaces] rparen as close_paren)
+  named_int_exp_prefixes_units = (prefixes_units_named_int_exp as int_exp [(prefixes_units_int_exp / extended_prefixed_units)]) / prefixes_units_int_exp
+
+  implicit_extended = (prefixes_units as first_set spaces as extender prefixes_units as second_set) / (int_exp_prefixes_units as first_int_exp_set [spaces] int_exp_prefixes_units as second_int_exp_set) / (prefixes_units as first_set [spaces] int_exp_prefixes_units as second_int_exp_set) / (int_exp_prefixes_units as first_int_exp_set [spaces] prefixes_units as second_set)
+
+  prefixes_units_int_exp = (implicit_extended as implicit_extended [spaces] [prefixes_units_int_exp as sequence]) / (implicit_extended as implicit_extended [spaces] extender [spaces] [prefixes_units_int_exp as sequence]) / (int_exp_prefixes_units [spaces] extender [spaces] [prefixes_units_int_exp as sequence]) / (prefixes_units [spaces] extender [spaces] [prefixes_units_int_exp as sequence]) / (int_exp_prefixes_units [prefixes_units_int_exp as sequence]) / (prefixes_units [prefixes_units_int_exp as sequence])
+
+  dimension_rules = (sqrt_dimensions / powered_dimensions) [extended_dimension_rules]
+  extended_dimension_rules = [spaces] extender [spaces] dimension_rules_int_exp as sequence
+  int_exp_dimension_rules = lparen dimension_rules_named_exp as int_exp rparen [extended_dimension_rules]
+  dimension_rules_named_exp = ((lparen as open_paren dimension_rules_int_exp as int_exp rparen as close_paren) as int_exp [spaces] (dimension_rules_int_exp / extended_dimension_rules) as sequence) / dimension_rules_int_exp
+
+  dimension_rules_int_exp = (dimension_rules as first_set spaces as extender dimension_rules as second_set [dimension_rules_int_exp]) / (dimension_rules as first_set [spaces] int_exp_dimension_rules as second_int_exp_set [dimension_rules_int_exp]) / (int_exp_dimension_rules as first_int_exp_set [spaces] dimension_rules as second_set [dimension_rules_int_exp]) / (int_exp_dimension_rules as first_int_exp_set [spaces] int_exp_dimension_rules as second_int_exp_set [dimension_rules_int_exp]) / int_exp_dimension_rules / dimension_rules
+
+  expression = prefixes_units_int_exp / dimension_rules_int_exp / (single_letter_prefixes hyphen) / (double_letter_prefixes hyphen)
+
+  entry expression: expression
+
+  test expression {
+    accept "mm*s^-2"
+    accept "sqrt(Hz)"
+    accept "K/(kg*m)"
+    accept "p-"
+    accept "(m)(s)"
+    accept "mm*"
+    accept "sqrt(sqrt(m))"
+    reject "que?"
+    reject "m2"
+    reject "dim_Ldim_M"
+  }
+}

data/lib/asciiunits/version.rb

@@ -0,0 +1,5 @@
+# frozen_string_literal: true
+
+module Asciiunits
+  VERSION = "0.1.0"
+end