Surprising parse results with `.` and `-` about units

[phlptp]

Yeah, it is a balancing job between a number of different unit conventions. There are a few programs which specify that . is a multiplier, other programs or conventions where a space is a multiplier. And in a linguistic sense - is a connector that bridges two parts of a word or a prefix or modifier. So I am sure there are a few cases where the complex conglomeration of those conventions make the interpretation a bit odd. I suppose some additional consideration of situations that end in a . might be warranted as that would indicate the . being used as an abbreviation rather than as a multiplier. If there are specific unit strings that get interpreted incorrectly there are a couple different ways I put in to interpret them and bypass most of the other logic that tries to figure out some sort of meaning from strings and break it into individual components.

[phlptp]

Playing around with this and how to interpret '.' This is what I have currently

// interpret . as multiply
    EXPECT_EQ(precise::m * precise::s, unit_from_string("m.s"));
    // interpret m as abbreviation so milli
    EXPECT_EQ(precise::milli * precise::s, unit_from_string("m. s"));
    // interpret as abbreviation so millisecond
    EXPECT_EQ(precise::milli * precise::s, unit_from_string("m. s."));
    // connector so space = multiply 
    EXPECT_EQ(precise::m * precise::s, unit_from_string("m- s"));
    // connector so space = multiply 
    EXPECT_EQ(precise::milli * precise::s, unit_from_string("m-s"));
    // s is abbreviation so first . is also abbreviation so millisecond
    EXPECT_EQ(precise::milli * precise::s, unit_from_string("m.s."));

a.u. in this context would be interpreted as astronomical unit instead of a*u

basic rule would be any non numerically sensible . would be interpreted as an abbreviation if at the end or followed by a space, otherwise it could be considered a multiplier. And it must be consistent for the whole unit string. I.e if one . shows as abbreviation all will.

If any show as an abbreviation they essentially get removed
[change from previous] "m.s." is equivalent to "m. s."

[SimonHeybrock]

And it must be consistent for the whole unit string. I.e if one . shows as abbreviation all will.

👍 nice approach, that makes sense!

What about the recommended (by a number of Journals) abbreviation for "arbitrary units": arb. unit? Does this parse correctly with the suggested change?

[phlptp]

I went through and looked at a few other units that end in "unit" and added some additional string associated with them and a few modifiers as well so "arb. unit" as a well as a few other descriptions of that will get interpreted as arbitrary unit