# HG changeset patch
# User jbe
# Date 1389495467 -3600
# Node ID 0c4841af07a528d6e1ccc559782e9ed709942c22
# Parent  db7ad8e4f78bf3bfe0eaf62611a85a352d10ce2b
String truncating by counting Unicode codepoints in format.string(...)
(grapheme cluster boundary detection not implemented)

diff -r db7ad8e4f78b -r 0c4841af07a5 framework/env/format/string.lua
--- a/framework/env/format/string.lua	Sun Nov 04 18:51:32 2012 +0100
+++ b/framework/env/format/string.lua	Sun Jan 12 03:57:47 2014 +0100
@@ -1,20 +1,76 @@
 --[[--
-text =                 -- a string
+text =                                          -- a string
 format.string(
-  value,               -- any value where tostring(value) gives a reasonable result
+  value,                               -- any value where tostring(value) gives a reasonable result
   {
-    nil_as = nil_text  -- text to be returned for a nil value
+    nil_as          = nil_text,        -- text to be returned for a nil value
+    truncate_mode   = "codepoints",    -- performe truncating by counting UTF-8 codepoints ("codepoints") or Unicode grapheme clusters ("graphmeclusters")
+                                       -- (currently only "codepoints" are supported and this option may be omitted)
+    truncate_at     = truncate_at,     -- truncate string after the given number of UTF-8 codepoints (or Unicode grapheme clusterst)
+    truncate_if     = truncate_if,     -- truncate only if length exceeds the given number of UTF-8 codepoints (or Unicode grapheme clusters)
+    truncate_suffix = truncate_suffix  -- string to append, if string was truncated
   }
 )
 
-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.
+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.
 
 --]]--
 
+function truncate_codepoints(str, truncate_at_codepoint, truncate_if_codepoint, suffix)
+  local byte_pos = 0
+  local codepoint_pos = 0
+  local truncate_at_byte
+  truncate_at_codepoint = truncate_at_codepoint or truncate_if_codepoint
+  truncate_if_codepoint = truncate_if_codepoint or truncate_at_codepoint
+  while true do
+    b1, b2, b3, b4 = string.byte(str, byte_pos, byte_pos+3)
+    if b1 then
+      if codepoint_pos > truncate_if_codepoint then
+        return string.sub(str, 1, truncate_at_byte or byte_pos) .. (suffix or "")
+      end
+      if codepoint_pos == truncate_at_codepoint then
+        truncate_at_byte = byte_pos
+      end
+      if b1 < 128 then
+        byte_pos = byte_pos + 1
+      elseif b1 >= 192 and b1 < 248 then
+        if b2 and b2 >= 128 and b2 < 192 then
+          if b1 < 240 and b3 and b3 >= 128 and b3 < 192 then
+            if b1 < 224 and b4 and b4 >= 128 and b4 < 192 then
+              byte_pos = byte_pos + 4
+            else
+              byte_pos = byte_pos + 3
+            end
+          else
+            byte_pos = byte_pos + 2
+          end
+        else
+          byte_pos = byte_pos + 1
+        end
+      else
+        byte_pos = byte_pos + 1
+      end
+      codepoint_pos = codepoint_pos + 1
+    else
+      break
+    end
+  end
+  return str
+end
+
 function format.string(str, options)
   local options = options or {}
   if str == nil then
     return options.nil_as or ""
+  elseif options.truncate_at or options.truncate_if then
+    -- TODO: Unicode grapheme cluster boundary detetion is not implemented
+    -- (Unicode codepoints are used instead)
+    return truncate_codepoints(
+      tostring(str),
+      options.truncate_at,
+      options.truncate_if,
+      options.truncate_suffix
+    )
   else
     return tostring(str)
   end