John Hewson created PDFBOX-2327:
-----------------------------------
Summary: Glyph list ligatures are decomposed too early
Key: PDFBOX-2327
URL: https://issues.apache.org/jira/browse/PDFBOX-2327
Project: PDFBox
Issue Type: Bug
Components: PDModel
Reporter: John Hewson
Priority: Minor
The following code in GlyphList was moved from the Encoding class:
{code}
DEFAULT.nameToUnicode.put("fi", "fi");
DEFAULT.nameToUnicode.put("fl", "fl");
DEFAULT.nameToUnicode.put("ffi", "ffi");
DEFAULT.nameToUnicode.put("ff", "ff");
DEFAULT.nameToUnicode.put("pi", "pi");
{code}
Which overrides the Adobe Glyph List (AGL) entries for certain ligature
characters, replacing them with their decomposition. This is used by
TextStripper so that it may output decomposed ligatures. However, it breaks the
glyph list because the entires have been overwritten with values which are not
in the AGL.
If TextStripper wants decomposed ligatures it can perform a Unicode
compatibility decomposition on the output strings, alternatively it can
implement this simple lookup itself, where it won't adversely impact the rest
of PDFBox.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
