Combined inch-foot mentions should be combined into one conversion of distance.

[Wendelstein7]

Currently, message I am 1 inch 3 feet. corrects to I am 2.54 cm 0.914 m..

Rather, it would be favourable if it'd result in one combined SI unit: I am 0.93 m.

[Randall-Scharpf]

To add to your proposition, I know that pounds and ounces (mass units) are used in a similar way, and since I don't hear a number of the units like stones and bushels in colloquial use, I can't say that they aren't used similarly. I think it is reasonable to assume in general that if a unit x and a unit y are of the same type, and y is more precise than x, the phrase A x's and B y's, where A is an integer and B is a non-negative real value, the same treatment should be given as in the case you showed.

Here's some test cases that should work when we're done, whether we generalize and implement without directly naming inches, feet, pounds, or ounces, or if we implement specifically:

• Feet and inches:
  
  
• Pounds and ounces:
  
  

A potential related issue is sub-sub-units. Although the only case I can think of is degrees-minutes-seconds, and these are not supported units from the bot, the bot would not support converting 19 degrees, 34 minutes, and 12 seconds into 0.342 radians. However, as no units that are (to my knowledge) regularly used as sub-sub-units are implemented in the bot, I do not propose solving this issue at the present. Instead, I mention it only because if it were to be supported, its implementation would, in all likelihood, directly involve the code used to solve this issue.

Proposed Fixes

• Create a regex that searches for something imperial followed by an optional and, any whitespace, and another something imperial and processes these before the individual units are processed. If the units are of a different type from each other (e.g, I am 6 feet and 220 pounds), the additional check must fallback to the main check.
• Invoke the above additional regex check into the main regex checker by enclosing the segment after the first something imperial in an optional (occurs 0 or 1 times) flag.
• Invoke the above additional regex check into the main regex checker by enclosing the segment after the first something imperial in a non-negative loop (occurs 0 or more times) flag.
• Record the replacements of individual imperial numbers that have been made, and then if more than one of the same type has been made, check that they are consecutive and separated by a trivial separator string (just whitespace and and). When these checks are satisfied, replace the replacements with the sum of their values.

Cheers!

[jamesgeddes]

I would also like to add that stone and pounds suffers from the same issue.

User:

12st 11.2lbs.

UnitCorrector:

I think user meant to say:

76.2kg 5.08kg

In Britain, Imperial weight is usually given in stone and pounds, rather than pounds alone. Because Britain.