On 03/07/2011 10:52 AM, Ramon F Herrera wrote:
Boost::regex [...] My only problem is spaces. I design typical mini computer language syntaxes, for example:
Table[1, 4] = Value AND Array[8, 4]
My initial approach was to get rid of all spaces before parsing (I recall being taught ages ago that the FORTRAN compiler does just that).
Most of the world seems to prefer languages with C-like syntax.
Such strategy works in some cases, but the above line would become:
Table[1,4]=ValueANDArray[8,4]
Which is obviously bad news. I solved this particular case with this syntax:
Table[1, 4] = Value && Array[8, 4]
My latest "language" looks like this and has the same problem described above:
Table rows [1758, 1904, 2053, 2201, 2345, 2497] Table cols [372, 880, 1336, 1756, 2083, 2439]
I guess I could insert a colon or something between the first two words, but I am sure there has to be a better way.
I usually just use \w+ in the pattern where there needs to be a space and \w* in the pattern where space is optional. E.g. "\\w*Table\\w+rows\\w*" If you still want to continue throwing away spaces you could match for the merged identifier possibilities: "Table(rows|cols)" or "Table([a-z]+)" - Marsh