I have built PDF2SVG (https://bitbucket.org/petermr/pdf2svg/wiki/Home) on
top of PDFBox, and I use
org.apache.pdfbox.pdmodel.font.PDFont.getFontWidth( byte[] c, int offset,
int length )
to get the width of characters, which I then use to calculate spacing and
thereby words. I have processed many documents successfully, but have found
one that causes problems:
apps.who.int/iris/bitstream/10665/143216/1/roadmapsitrep_14Nov2014_eng.pdf
Paragraphs 1 ("A total ...") and 2 ("Following...") have a font
"ABCDEE+Calibri" where nearly all character widths are 1000 (which is
clearly wrong). Para 3 ("In Mali"...) apparently has the same font name
("ABCDEE+Calibri") but has different spacings for each character, which
then give a proper layout.
I don't know whether the document or my code is wrong and I'd be grateful
for a very quick reality-check. I've run it through PDFTextStripper and it
behaves properly. [I don't use PDFTextStripper because I want to preserve
individual characters, including styles, weights and sub/superscripts.]
(a) Do paragraphs 1 and 2 use the same font as 3? What are the font names?
(b) Is it allowed to have the same name for 2 different fonts (it would be
very bad...)
(c) How does PDFTextStripper calculate spaces? From the Font, or by some
other heuristics?
(d) is there a debugging tool on PDFBox I could be using for this sort of
problem?
Many thanks.
P.
--
Peter Murray-Rust
Reader in Molecular Informatics
Unilever Centre, Dep. Of Chemistry
University of Cambridge
CB2 1EW, UK
+44-1223-763069
