jbe/bsw@0: --[[--
jbe@471: text =                                             -- a string
jbe/bsw@0: format.string(
jbe@111:   value,                                           -- any value where tostring(value) gives a reasonable result
jbe/bsw@0:   {
jbe@111:     nil_as                = nil_text,              -- text to be returned for a nil value
jbe@111:     truncate_mode         = "codepoints",          -- performe truncating by counting UTF-8 codepoints ("codepoints") or Unicode grapheme clusters ("graphmeclusters")
jbe@111:                                                    -- (currently only "codepoints" are supported and this option may be omitted)
jbe@111:     truncate_at           = truncate_at,           -- truncate string after the given number of UTF-8 codepoints (or Unicode grapheme clusters)
jbe@111:     truncate_suffix       = truncate_suffix,       -- string to append, if string was truncated (use boolean true for Unicode ellipsis)
jbe@112:     truncate_count_suffix = truncate_count_suffix  -- if true, then the total length (including suffix) may not exceed the given length
jbe/bsw@0:   }
jbe/bsw@0: )
jbe/bsw@0: 
jbe@110: Formats a value as a text by calling tostring(...), unless the value is nil, in which case the text returned is chosen by the 'nil_as' option. Using the 'truncate_*' parameters, it is possible to show only the beginning of a string.
jbe/bsw@0: 
jbe/bsw@0: --]]--
jbe/bsw@0: 
jbe@111: local function codepoint_count(str)
jbe@111:   return #string.gsub(str, '[\128-\255][\128-\191]?[\128-\191]?[\128-\191]?', 'x')
jbe@111: end
jbe@111: 
jbe@111: local function codepoint_truncate(str, length)
jbe@111:   local byte_pos = 1
jbe@111:   local count = 0
jbe@111:   while count < length do
jbe@110:     b1, b2, b3, b4 = string.byte(str, byte_pos, byte_pos+3)
jbe@111:     if not b2 then
jbe@111:       break
jbe@111:     end
jbe@111:     b3 = b3 or 0
jbe@111:     b4 = b4 or 0
jbe@111:     if b1 >= 128 and b2 >= 128 and b2 <= 191 then
jbe@111:       if b3 >= 128 and b3 <= 191 then
jbe@111:         if b4 >= 128 and b4 <= 191 then
jbe@111:           byte_pos = byte_pos + 4
jbe@111:           count = count + 1
jbe@111:         elseif count + 1 < length and b4 < 128 then
jbe@111:           byte_pos = byte_pos + 4
jbe@111:           count = count + 2
jbe@110:         else
jbe@111:           byte_pos = byte_pos + 3
jbe@111:           count = count + 1
jbe@111:         end
jbe@111:       elseif count + 1 < length and b3 < 128 then
jbe@111:         if count + 2 < length and b4 < 128 then
jbe@111:           byte_pos = byte_pos + 4
jbe@111:           count = count + 3
jbe@111:         else
jbe@111:           byte_pos = byte_pos + 3
jbe@111:           count = count + 2
jbe@110:         end
jbe@110:       else
jbe@111:         byte_pos = byte_pos + 2
jbe@111:         count = count + 1
jbe@110:       end
jbe@111:     elseif count + 1 < length and b2 < 128 then
jbe@111:       if count + 2 < length and b3 < 128 then
jbe@111:         if count + 3 < length and b4 < 128 then
jbe@111:           byte_pos = byte_pos + 4
jbe@111:           count = count + 4
jbe@111:         else
jbe@111:           byte_pos = byte_pos + 3
jbe@111:           count = count + 3
jbe@111:         end
jbe@111:       else
jbe@111:         byte_pos = byte_pos + 2
jbe@111:         count = count + 2
jbe@111:       end
jbe@110:     else
jbe@111:       byte_pos = byte_pos + 1
jbe@111:       count = count + 1
jbe@110:     end
jbe@110:   end
jbe@111:   return string.sub(str, 1, byte_pos-1)
jbe@110: end
jbe@110: 
jbe/bsw@0: function format.string(str, options)
jbe/bsw@0:   local options = options or {}
jbe/bsw@0:   if str == nil then
jbe/bsw@0:     return options.nil_as or ""
jbe@111:   elseif options.truncate_at then
jbe@111:     str = tostring(str)
jbe@110:     -- TODO: Unicode grapheme cluster boundary detetion is not implemented
jbe@110:     -- (Unicode codepoints are used instead)
jbe@111:     local truncate_suffix = options.truncate_suffix
jbe@111:     if truncate_suffix == true then
jbe@111:       truncate_suffix = '\226\128\166'
jbe@111:     elseif not truncate_suffix then
jbe@111:       truncate_suffix = ''
jbe@111:     end
jbe@112:     if options.truncate_count_suffix and truncate_suffix then
jbe@111:       local suffix_length = codepoint_count(truncate_suffix)
jbe@111:       if codepoint_count(str) > options.truncate_at then
jbe@111:         return (
jbe@111:           codepoint_truncate(str, options.truncate_at - suffix_length) ..
jbe@111:           truncate_suffix
jbe@111:         )
jbe@111:       else
jbe@111:         return str
jbe@111:       end
jbe@111:     else
jbe@111:       if codepoint_count(str) > options.truncate_at then
jbe@111:         return codepoint_truncate(str, options.truncate_at) .. truncate_suffix
jbe@111:       else
jbe@111:         return str
jbe@111:       end
jbe@111:     end
jbe/bsw@0:   else
jbe/bsw@0:     return tostring(str)
jbe/bsw@0:   end
jbe/bsw@0: end