Unicode character mappings don't seem to actually map to the appropriate characters in the PDF output. I'd really like ligatures and circled alphanums, but can't get 'em 'cause FOP apparently mistranslates some of the codes. The summary is one example.
The mappings depend on the font. If the font does not contain the character then it will output a #.
The font, Times New Roman, definitely contains the fi ligature, as I have other documents using the same font, but produced via other means (pdftex, for example). All docs use the base fonts from Adobe - no embedded fonts.
The problem here appears to be that FOP sets up a Times-Roman font with a platform specific encoding. In the most likely case, which is Win-Ansi, the fi-ligature Unicode character can not be mapped. A work around is to use a Unicode user font with a glyph for the ligature.
This works for me in latest Trunk (rev 584703), and probably in FOP 0.94
batch transition pre-FOP1.0 resolved+fixed bugs to closed+fixed