alexandergunnarson / lsm

lsm

We tried a few approaches here.

Favored approach — pdfseparate + Java Aspose PDF + HTML index page (current approach)

Code contained elsewhere.

Discarded approaches

pdf2htmlEX

This results in great performance and formatting, but the HTML comes out garbled so it's hard for Google Translate to make sense of it.

echo '
#!/bin/bash
docker run -ti --rm -v `pwd`:/pdf bwits/pdf2htmlex pdf2htmlEX "$@"
' > /usr/local/bin/pdf2htmlEX && chmod +x /usr/local/bin/pdf2htmlEX

# All-in-one — High quality visual, and medium-low translatability
pdf2htmlEX --embed cfijo --split-pages 1 --dest-dir lsm --page-filename page-%d.html lsm.pdf

# Individual page — High quality visual, but low translatability
pdf2htmlEX page-1.pdf page-1.html
pdf2htmlEX --correct-text-visibility 1 page-1.pdf page-1.html

# Individual page — High quality visual, and medium-low translatability
pdf2htmlEX --embed cfijo --dest-dir page-1 page-1.pdf page-1.html
pdf2htmlEX --embed cfijo --dest-dir page-9 page-9.pdf page-9.html
pdf2htmlEX --correct-text-visibility 1 --embed cfijo --dest-dir page-9 page-9.pdf page-9.html

mupdf

Images aren't preserved.

brew install mupdf
mutool convert -o page-1.html page-1.pdf

pdftohtml

Low quality visual, but good translatability.

pdftohtml page-1.pdf page-1.html

Unilex transcript

This is a very immature tool that doesn't do a good job.

git clone https://github.com/fmalina/unilex-transcript
cd unilex-transcript
pip3 install -r requirements.txt

echo
<<EOF
DATA_DIR = '/Users/alexandergunnarson/Creating/Code/alexandergunnarson/lsm/pdfs/test/'
PDF_DIR  = DATA_DIR+'PDF'
HTML_DIR = DATA_DIR+'HTML'
# used by ttf.py to access full original fonts to compare with the broken ones
FULL_FONTS_PATH  = '/Users/f/SITES/etc/ttf'
BULLETS =  ('', '')
REMOVE_BEFORE = ()
REPLACE_AFTER = ()
EOF | tee config.py

# Adapted from https://github.com/fmalina/unilex-transcript/blob/51a14361a16b4d9c6e565e1ed5f742ac2be93dba/pdf2html.py
docker run -ti --rm -v /Users/alexandergunnarson/Creating/Code/alexandergunnarson/lsm/pdfs/test:/pdf bwits/pdf2htmlex pdf2htmlEX \
  --embed-external-font 0 \
  --external-hint-tool ttfautohint \
  --process-nontext 0 \
  --embed cfijo \
  --dest-dir ./HTML/page-9 \
  PDF/page-9.pdf \
   page-9.html

./transcript.py