LaTeX reader: fix siunitx unit commands...

...they should only be recognized in siunitx contexts.

For example, `\l` outside of an siunitx context should be l-slash,
not l (for liter)!

Closes #4842.

1 files changed, 182 insertions, 174 deletions

diff --git a/src/Text/Pandoc/Readers/LaTeX.hs b/src/Text/Pandoc/Readers/LaTeX.hs
index afd0f34b9..ebcb9fecf 100644
--- a/src/Text/Pandoc/Readers/LaTeX.hs
+++ b/src/Text/Pandoc/Readers/LaTeX.hs
@@ -735,7 +735,7 @@ dosiunitx = do
   skipopts
   value <- tok
   valueprefix <- option "" $ bracketed tok
-  unit <- inlineCommand' <|> tok
+  unit <- grouped (mconcat <$> many1 siUnit) <|> siUnit <|> tok
   let emptyOr160 "" = ""
       emptyOr160 _  = "\160"
   return . mconcat $ [valueprefix,
@@ -744,11 +744,187 @@ dosiunitx = do
                       emptyOr160 unit,
                       unit]
 
--- siunitx's \square command
-dosquare :: PandocMonad m => LP m Inlines
-dosquare = do
-  unit <- inlineCommand' <|> tok
-  return . mconcat $ [unit, "\178"]
+siUnit :: PandocMonad m => LP m Inlines
+siUnit = do
+  Tok _ (CtrlSeq name) cmd <- anyControlSeq
+  if name == "square"
+     then do
+       unit <- grouped (mconcat <$> many1 siUnit) <|> siUnit <|> tok
+       return . mconcat $ [unit, "\178"]
+     else
+       case M.lookup name siUnitMap of
+            Just il -> return il
+            Nothing -> mzero
+
+siUnitMap :: M.Map Text Inlines
+siUnitMap = M.fromList
+  [ ("fg", str "fg")
+  , ("pg", str "pg")
+  , ("ng", str "ng")
+  , ("ug", str "μg")
+  , ("mg", str "mg")
+  , ("g", str "g")
+  , ("kg", str "kg")
+  , ("amu", str "u")
+  , ("pm", str "pm")
+  , ("nm", str "nm")
+  , ("um", str "μm")
+  , ("mm", str "mm")
+  , ("cm", str "cm")
+  , ("dm", str "dm")
+  , ("m", str "m")
+  , ("km", str "km")
+  , ("as", str "as")
+  , ("fs", str "fs")
+  , ("ps", str "ps")
+  , ("ns", str "ns")
+  , ("us", str "μs")
+  , ("ms", str "ms")
+  , ("s", str "s")
+  , ("fmol", str "fmol")
+  , ("pmol", str "pmol")
+  , ("nmol", str "nmol")
+  , ("umol", str "μmol")
+  , ("mmol", str "mmol")
+  , ("mol", str "mol")
+  , ("kmol", str "kmol")
+  , ("pA", str "pA")
+  , ("nA", str "nA")
+  , ("uA", str "μA")
+  , ("mA", str "mA")
+  , ("A", str "A")
+  , ("kA", str "kA")
+  , ("ul", str "μl")
+  , ("ml", str "ml")
+  , ("l", str "l")
+  , ("hl", str "hl")
+  , ("uL", str "μL")
+  , ("mL", str "mL")
+  , ("L", str "L")
+  , ("hL", str "hL")
+  , ("mHz", str "mHz")
+  , ("Hz", str "Hz")
+  , ("kHz", str "kHz")
+  , ("MHz", str "MHz")
+  , ("GHz", str "GHz")
+  , ("THz", str "THz")
+  , ("mN", str "mN")
+  , ("N", str "N")
+  , ("kN", str "kN")
+  , ("MN", str "MN")
+  , ("Pa", str "Pa")
+  , ("kPa", str "kPa")
+  , ("MPa", str "MPa")
+  , ("GPa", str "GPa")
+  , ("mohm", str "mΩ")
+  , ("kohm", str "kΩ")
+  , ("Mohm", str "MΩ")
+  , ("pV", str "pV")
+  , ("nV", str "nV")
+  , ("uV", str "μV")
+  , ("mV", str "mV")
+  , ("V", str "V")
+  , ("kV", str "kV")
+  , ("W", str "W")
+  , ("uW", str "μW")
+  , ("mW", str "mW")
+  , ("kW", str "kW")
+  , ("MW", str "MW")
+  , ("GW", str "GW")
+  , ("J", str "J")
+  , ("uJ", str "μJ")
+  , ("mJ", str "mJ")
+  , ("kJ", str "kJ")
+  , ("eV", str "eV")
+  , ("meV", str "meV")
+  , ("keV", str "keV")
+  , ("MeV", str "MeV")
+  , ("GeV", str "GeV")
+  , ("TeV", str "TeV")
+  , ("kWh", str "kWh")
+  , ("F", str "F")
+  , ("fF", str "fF")
+  , ("pF", str "pF")
+  , ("K", str "K")
+  , ("dB", str "dB")
+  , ("angstrom", str "Å")
+  , ("arcmin", str "′")
+  , ("arcminute", str "′")
+  , ("arcsecond", str "″")
+  , ("astronomicalunit", str "ua")
+  , ("atomicmassunit", str "u")
+  , ("atto", str "a")
+  , ("bar", str "bar")
+  , ("barn", str "b")
+  , ("becquerel", str "Bq")
+  , ("bel", str "B")
+  , ("candela", str "cd")
+  , ("celsius", str "°C")
+  , ("centi", str "c")
+  , ("coulomb", str "C")
+  , ("dalton", str "Da")
+  , ("day", str "d")
+  , ("deca", str "d")
+  , ("deci", str "d")
+  , ("decibel", str "db")
+  , ("degreeCelsius",str "°C")
+  , ("degree", str "°")
+  , ("deka", str "d")
+  , ("electronvolt", str "eV")
+  , ("exa", str "E")
+  , ("farad", str "F")
+  , ("femto", str "f")
+  , ("giga", str "G")
+  , ("gram", str "g")
+  , ("hectare", str "ha")
+  , ("hecto", str "h")
+  , ("henry", str "H")
+  , ("hertz", str "Hz")
+  , ("hour", str "h")
+  , ("joule", str "J")
+  , ("katal", str "kat")
+  , ("kelvin", str "K")
+  , ("kilo", str "k")
+  , ("kilogram", str "kg")
+  , ("knot", str "kn")
+  , ("liter", str "L")
+  , ("litre", str "l")
+  , ("lumen", str "lm")
+  , ("lux", str "lx")
+  , ("mega", str "M")
+  , ("meter", str "m")
+  , ("metre", str "m")
+  , ("milli", str "m")
+  , ("minute", str "min")
+  , ("mmHg", str "mmHg")
+  , ("mole", str "mol")
+  , ("nano", str "n")
+  , ("nauticalmile", str "M")
+  , ("neper", str "Np")
+  , ("newton", str "N")
+  , ("ohm", str "Ω")
+  , ("Pa", str "Pa")
+  , ("pascal", str "Pa")
+  , ("percent", str "%")
+  , ("per", str "/")
+  , ("peta", str "P")
+  , ("pico", str "p")
+  , ("radian", str "rad")
+  , ("second", str "s")
+  , ("siemens", str "S")
+  , ("sievert", str "Sv")
+  , ("steradian", str "sr")
+  , ("tera", str "T")
+  , ("tesla", str "T")
+  , ("tonne", str "t")
+  , ("volt", str "V")
+  , ("watt", str "W")
+  , ("weber", str "Wb")
+  , ("yocto", str "y")
+  , ("yotta", str "Y")
+  , ("zepto", str "z")
+  , ("zetta", str "Z")
+  ]
 
 lit :: String -> LP m Inlines
 lit = pure . str
@@ -1594,174 +1770,6 @@ inlineCommands = M.union inlineLanguageCommands $ M.fromList
   , ("acsp", doAcronymPlural "abbrv")
   -- siuntix
   , ("SI", dosiunitx)
-  -- units of siuntix
-  , ("fg", lit "fg")
-  , ("pg", lit "pg")
-  , ("ng", lit "ng")
-  , ("ug", lit "μg")
-  , ("mg", lit "mg")
-  , ("g", lit "g")
-  , ("kg", lit "kg")
-  , ("amu", lit "u")
-  , ("pm", lit "pm")
-  , ("nm", lit "nm")
-  , ("um", lit "μm")
-  , ("mm", lit "mm")
-  , ("cm", lit "cm")
-  , ("dm", lit "dm")
-  , ("m", lit "m")
-  , ("km", lit "km")
-  , ("as", lit "as")
-  , ("fs", lit "fs")
-  , ("ps", lit "ps")
-  , ("ns", lit "ns")
-  , ("us", lit "μs")
-  , ("ms", lit "ms")
-  , ("s", lit "s")
-  , ("fmol", lit "fmol")
-  , ("pmol", lit "pmol")
-  , ("nmol", lit "nmol")
-  , ("umol", lit "μmol")
-  , ("mmol", lit "mmol")
-  , ("mol", lit "mol")
-  , ("kmol", lit "kmol")
-  , ("pA", lit "pA")
-  , ("nA", lit "nA")
-  , ("uA", lit "μA")
-  , ("mA", lit "mA")
-  , ("A", lit "A")
-  , ("kA", lit "kA")
-  , ("ul", lit "μl")
-  , ("ml", lit "ml")
-  , ("l", lit "l")
-  , ("hl", lit "hl")
-  , ("uL", lit "μL")
-  , ("mL", lit "mL")
-  , ("L", lit "L")
-  , ("hL", lit "hL")
-  , ("mHz", lit "mHz")
-  , ("Hz", lit "Hz")
-  , ("kHz", lit "kHz")
-  , ("MHz", lit "MHz")
-  , ("GHz", lit "GHz")
-  , ("THz", lit "THz")
-  , ("mN", lit "mN")
-  , ("N", lit "N")
-  , ("kN", lit "kN")
-  , ("MN", lit "MN")
-  , ("Pa", lit "Pa")
-  , ("kPa", lit "kPa")
-  , ("MPa", lit "MPa")
-  , ("GPa", lit "GPa")
-  , ("mohm", lit "mΩ")
-  , ("kohm", lit "kΩ")
-  , ("Mohm", lit "MΩ")
-  , ("pV", lit "pV")
-  , ("nV", lit "nV")
-  , ("uV", lit "μV")
-  , ("mV", lit "mV")
-  , ("V", lit "V")
-  , ("kV", lit "kV")
-  , ("W", lit "W")
-  , ("uW", lit "μW")
-  , ("mW", lit "mW")
-  , ("kW", lit "kW")
-  , ("MW", lit "MW")
-  , ("GW", lit "GW")
-  , ("J", lit "J")
-  , ("uJ", lit "μJ")
-  , ("mJ", lit "mJ")
-  , ("kJ", lit "kJ")
-  , ("eV", lit "eV")
-  , ("meV", lit "meV")
-  , ("keV", lit "keV")
-  , ("MeV", lit "MeV")
-  , ("GeV", lit "GeV")
-  , ("TeV", lit "TeV")
-  , ("kWh", lit "kWh")
-  , ("F", lit "F")
-  , ("fF", lit "fF")
-  , ("pF", lit "pF")
-  , ("K", lit "K")
-  , ("dB", lit "dB")
-  , ("angstrom", lit "Å")
-  , ("arcmin", lit "′")
-  , ("arcminute", lit "′")
-  , ("arcsecond", lit "″")
-  , ("astronomicalunit", lit "ua")
-  , ("atomicmassunit", lit "u")
-  , ("atto", lit "a")
-  , ("bar", lit "bar")
-  , ("barn", lit "b")
-  , ("becquerel", lit "Bq")
-  , ("bel", lit "B")
-  , ("candela", lit "cd")
-  , ("celsius", lit "°C")
-  , ("centi", lit "c")
-  , ("coulomb", lit "C")
-  , ("dalton", lit "Da")
-  , ("day", lit "d")
-  , ("deca", lit "d")
-  , ("deci", lit "d")
-  , ("decibel", lit "db")
-  , ("degreeCelsius",lit "°C")
-  , ("degree", lit "°")
-  , ("deka", lit "d")
-  , ("electronvolt", lit "eV")
-  , ("exa", lit "E")
-  , ("farad", lit "F")
-  , ("femto", lit "f")
-  , ("giga", lit "G")
-  , ("gram", lit "g")
-  , ("hectare", lit "ha")
-  , ("hecto", lit "h")
-  , ("henry", lit "H")
-  , ("hertz", lit "Hz")
-  , ("hour", lit "h")
-  , ("joule", lit "J")
-  , ("katal", lit "kat")
-  , ("kelvin", lit "K")
-  , ("kilo", lit "k")
-  , ("kilogram", lit "kg")
-  , ("knot", lit "kn")
-  , ("liter", lit "L")
-  , ("litre", lit "l")
-  , ("lumen", lit "lm")
-  , ("lux", lit "lx")
-  , ("mega", lit "M")
-  , ("meter", lit "m")
-  , ("metre", lit "m")
-  , ("milli", lit "m")
-  , ("minute", lit "min")
-  , ("mmHg", lit "mmHg")
-  , ("mole", lit "mol")
-  , ("nano", lit "n")
-  , ("nauticalmile", lit "M")
-  , ("neper", lit "Np")
-  , ("newton", lit "N")
-  , ("ohm", lit "Ω")
-  , ("Pa", lit "Pa")
-  , ("pascal", lit "Pa")
-  , ("percent", lit "%")
-  , ("per", lit "/")
-  , ("peta", lit "P")
-  , ("pico", lit "p")
-  , ("radian", lit "rad")
-  , ("second", lit "s")
-  , ("siemens", lit "S")
-  , ("sievert", lit "Sv")
-  , ("square", dosquare)
-  , ("steradian", lit "sr")
-  , ("tera", lit "T")
-  , ("tesla", lit "T")
-  , ("tonne", lit "t")
-  , ("volt", lit "V")
-  , ("watt", lit "W")
-  , ("weber", lit "Wb")
-  , ("yocto", lit "y")
-  , ("yotta", lit "Y")
-  , ("zepto", lit "z")
-  , ("zetta", lit "Z")
   -- hyphenat
   , ("bshyp", lit "\\\173")
   , ("fshyp", lit "/\173")