This is due to the fact that FT_Set_Char_Size is called even for Type 3 fonts in pdf_unicode.c's extracttext. Also: Font Differences in the /a65 form aren't recognized (/a65 standing for the ASCII 65 character 'A'). See http://code.google.com/p/sumatrapdf/source/detail?r=1311
Created attachment 5214 [details] sample PDF See http://drop.io/issue_589 for the source.
After we added a heuristic to pdf_lookupagl() to also look for "a%d" patterns the Type3 fonts generated by TeX & co become searchable.