forked from GitHub/gf-core
(un)lexmixed: added the other math environments than $ used in latex
This commit is contained in:
@@ -1367,13 +1367,13 @@ stringOpOptions = sort $ [
|
||||
("from_utf8","decode from utf8 (default)"),
|
||||
("lextext","text-like lexer"),
|
||||
("lexcode","code-like lexer"),
|
||||
("lexmixed","mixture of text and code (code between $...$)"),
|
||||
("lexmixed","mixture of text and code, as in LaTeX (code between $...$, \\(...)\\, \\[...\\])"),
|
||||
("to_cp1251","encode to cp1251 (Cyrillic used in Bulgarian resource)"),
|
||||
("to_html","wrap in a html file with linebreaks"),
|
||||
("to_utf8","encode to utf8 (default)"),
|
||||
("unlextext","text-like unlexer"),
|
||||
("unlexcode","code-like unlexer"),
|
||||
("unlexmixed","mixture of text and code (code between $...$)"),
|
||||
("unlexmixed","mixture of text and code (code between $...$, \\(...)\\, \\[...\\])"),
|
||||
("unchars","unlexer that puts no spaces between tokens"),
|
||||
("unwords","unlexer that puts a single space between tokens (default)"),
|
||||
("words","lexer that assumes tokens separated by spaces (default)")
|
||||
|
||||
@@ -67,18 +67,18 @@ lexLatexCode = restoreBackslash . lexCode where --- quick hack: postprocess Hask
|
||||
|
||||
-- | LaTeX style lexer, with "math" environment using Code between $...$
|
||||
lexMixed :: String -> [String]
|
||||
lexMixed = concat . alternate False where
|
||||
alternate env s = case s of
|
||||
_:_ -> case break (=='$') s of
|
||||
(t,[]) -> lex env t : []
|
||||
(t,c:m) -> lex env t : [[c]] : alternate (not env) m
|
||||
_ -> []
|
||||
lexMixed = concat . alternate False [] where
|
||||
alternate env t s = case s of
|
||||
'$':cs -> lex env (reverse t) : ["$"] : alternate (not env) [] cs
|
||||
'\\':c:cs | elem c "()[]" -> lex env (reverse t) : [['\\',c]] : alternate (not env) [] cs
|
||||
c:cs -> alternate env (c:t) cs
|
||||
_ -> [lex env (reverse t)]
|
||||
lex env = if env then lexLatexCode else lexText
|
||||
|
||||
unlexMixed :: [String] -> String
|
||||
unlexMixed = capitInit . concat . alternate False where
|
||||
alternate env s = case s of
|
||||
_:_ -> case break (=="$") s of
|
||||
_:_ -> case break (flip elem ["$","\\[","\\]","\\(","\\)"]) s of
|
||||
(t,[]) -> unlex env t : []
|
||||
(t,c:m) -> unlex env t : sep env c m : alternate (not env) m
|
||||
_ -> []
|
||||
|
||||
Reference in New Issue
Block a user