mirror of
https://github.com/DSpace/DSpace.git
synced 2025-10-15 14:03:17 +00:00
2e853aee50
to do that conversion, and the resulting html and pdf documents. All of the work is done in subdirectories. This is for review, comment, and adjustment. Still to be done: - Integrate the docbook to html and pdf into the Maven build. - Merge changes from the Foundation version of docs into HTML. - Remove the legacy HTML source. - Remove the generated html and pdf once the build does it. - Various small formatting cleanups. git-svn-id: http://scm.dspace.org/svn/repo/branches/dspace-1_5_x@3042 9c30dcfa-912a-0410-8fc2-9e0234be79fd
77 lines
1.9 KiB
Python
Executable File
77 lines
1.9 KiB
Python
Executable File
#! /usr/bin/env python
|
|
|
|
""" Utility program to try to gently line wrap preformatted text inside an XML tag
|
|
|
|
"""
|
|
|
|
import sys, string
|
|
import codecs
|
|
|
|
def wrapElementLines(wrap=72,src=sys.stdin, dst=sys.stdout, srcenc='iso8859', dstenc='utf-8', elem='screen'):
|
|
|
|
# Tell python how to encode the output (generally utf-8)
|
|
(e,d,sr,sw) = codecs.lookup(dstenc)
|
|
out = sw(dst)
|
|
|
|
# Tell python how to interpret the input (generally iso8859)
|
|
(e,d,sr,sw) = codecs.lookup(srcenc)
|
|
inp = sr(src)
|
|
|
|
inElem = False
|
|
openElem = "<"+elem+">"
|
|
closeElem = "</"+elem+">"
|
|
|
|
for line in inp:
|
|
try:
|
|
index = line.index(openElem)
|
|
inElem = True
|
|
except:
|
|
pass
|
|
if inElem:
|
|
try:
|
|
line.index(closeElem)
|
|
inElem = False
|
|
except:
|
|
while len(line) > wrap:
|
|
lim = indexlimit(line,wrap)
|
|
try:
|
|
space = line[0:lim].rindex(' ')
|
|
out.write(line[0:space]+"\n\t") # " \\\n\t"
|
|
line = line[space+1:]
|
|
except:
|
|
out.write(line[0:lim]+"\n") # "\\\n"
|
|
line = line[wrap:]
|
|
out.write(line)
|
|
|
|
def indexlimit(string,limit=72):
|
|
tag=False
|
|
ent=False
|
|
ptr=0
|
|
cnt=0
|
|
for ch in string:
|
|
ptr = ptr + 1
|
|
if ch == '<' and not tag:
|
|
tag=True
|
|
elif ch == '>' and tag:
|
|
tag=False
|
|
elif ch == '&' and not ent:
|
|
ent=True
|
|
elif ch == ';' and ent:
|
|
cnt = cnt + 1
|
|
ent=False
|
|
elif not (tag or ent):
|
|
cnt = cnt + 1
|
|
if cnt >= limit:
|
|
return ptr
|
|
return ptr
|
|
|
|
src = sys.stdin
|
|
dst = sys.stdout
|
|
|
|
if len(sys.argv) > 1:
|
|
src = open(sys.argv[1])
|
|
if len(sys.argv) > 2:
|
|
dst = open(sys.argv[2],"w")
|
|
|
|
wrapElementLines(src=src,dst=dst,wrap=70,srcenc="utf-8")
|