We are facing a series of errors in our logs that we inferred are produced by PDFBox:
ERROR [pdmodel.font.PDCIDFont] Error: Could not parse predefined CMAP file for 'Adobe-Identity-UCS'.
These errors started when we stored PDF files with Japanese content in Alfresco, which content is extracted by using PDFBox.
We made some investigations and found out that the font we use for those samples (Sazanami Mincho) is not responsible for this, as we generated PDFs with MS Mincho, MS Gothic, and with many other Japanese fonts, and the errors were still present.
The only variation we found here is that in some cases the errors do not appear with PDFBox v. 1.7.
Also, we generated PDF sample in Japanese by using XEP formatter, in order to check if FOP is responsible for the errors, but got the errors.
We also found out that this kind of errors, for various fonts, were reportedly related to PDFBox. They were partially solved with version 1.6 and 1.7.