Dear LyX developers,
some LICR (LaTeX internal character representation) macros require
termination by a {}, so that the macro name can be separated from the
following text.
Example: "\\OE", e.g. used in \OE{}vre
However, "lib/unicodesymbols" has also replacement code that must not be
terminated with {}.
Examples: "ff" (ff ligature)
"\\-" (no-break-hyphen)
"\\mkern3mu\\mathchar'26\\mkern-12mu d" (dj character in math)
To distinguish cases that require/don't require termination, the
"notermination" flag (with values "text", "math", "both", "none") was
introduced (set by default if a command ends with "}").
I propose to replace this flag with a detection algorithm to make the
"unicodesymbols" file easier to read/edit:
TeX's macro name parsing can be described as:
After a backslash, take every alphabetical character (a-z,A-Z) until
reaching a non-alphabetical character (space, digits, punctuation, ...)
If the first character after the backslash is no letter, the first
non-letter character is used as macro name. Otherwise the macro name is
the sequence of letters up to the first non-letter.
We could reverse this parsing to determine whether a command in
"unicodesymbols" requires termination:
Test the characters in the command string in reverse order:
* Only if the last characters are a backslash followed by letters, the
command needs termination
* if there is no backslash, no termination is required (like with "ff")
* a non-letter character (everthing exept a-zA-Z) abort the loop.
The regular expression would be "\\[a-zA-Z]+$" (literal backslash,
followed by one or more characters in the range a-zA-Z, anchored at the
end of the string).
A possible Python implementation would be
def requires_termination(command):
"""check whether `command` must be terminated with '{}'"""
for c in list(command).__reversed__():
if c == "\\":
return True
if not c.isalpha():
return False
return False
>>> requires_termination(r'\foo')
True
>>> requires_termination(r'was ist \foo')
True
>>> requires_termination(r'\foo bar')
False
>>> requires_termination(r'\foo{a}')
False
>>> requires_termination(r'\~')
False
>>> requires_termination(r'ffl')
False
Would it be difficult, to implement something in this line in C++?
Günter
