tr2tex

#!/usr/bin/env python3

import os
import re
import sys

def end(l):
   # look for " .", etc.
   p = re.compile(r'\s+([\.,:;])')
   l = p.sub(r'\1', l)
   # look for ) (
   p = re.compile(r'\s+\)([\.,]*) \(\n')
   m = p.search(l)
   if m != None:
      l = p.sub(r'', l)
      if len(m.groups()) > 0:
         l = "(" + l + ")%s\n" % m.groups()[0]
      else:
         l = "(" + l + ")\n"
   # look for ''  ``
   p = re.compile(r"\s+''([\.,]*) ``\n")
   m = p.search(l)
   if m != None:
      l = p.sub(r'', l)
      if len(m.groups()) > 0:
         l = "``" + l + "''%s\n" % m.groups()[0]
      else:
         l = "``" + l + "''\n"
   # check for " )"
   p = re.compile(r'\s+([\)])')
   l = p.sub(r'\1', l)
   return l

def ig2dotdot():
   line = f.readline()
   while line:
      if line.startswith('..'):
         return
      print("%", line, end="")
      line = f.readline()

def chapter(l):
   p = re.compile(r'.chapter ([:\w]+) "(.+)"')
   l = p.sub(r'\\chapter{\2}\n\\label{\1}', l)
   print(l, end="")

def chapter1(l):
   p = re.compile(r'.chapterlike "(.+)"')
   l = p.sub(r'\\chapter*{\1}\n', l)
   print(l, end="")

def comment(l):
   p = re.compile(r'\.\\"(.*)')
   l = p.sub(r'%% \1', l)
   print(l, end="")
   
def index(l):
   p = re.compile(r'.index ([-\w]+)')
   l = p.sub(r'\\index{\1}', l)
   p = re.compile(r'.index "(.*)"')
   l = p.sub(r'\\index{\1}', l)
   l = end(l)
   print(l, end="")

def defindex(l):
   p = re.compile(r'.italic-index ([-\w]+)')
   l = p.sub(r'\\textit{\1}\\index{\1}', l)
   p = re.compile(r'.italic-index "(.*)"')
   l = p.sub(r'\\textit{\1}\\index{\1}', l)
   l = end(l)
   print(l, end="")

codep = r'([\/\.\->_\w\&<\(\)\#\[\]\|=\*%\!]+)'

def code(l):
   p = re.compile(r'.code "(.*)"')
   l = p.sub(r'\\lstinline{\1}', l)
   p = re.compile(r'.code ' + codep)
   l = p.sub(r'\\lstinline{\1}', l)
   l = end(l)
   print(l, end="")

def codeindex(l):
   p = re.compile(r'.code-index ' + codep)
   l = p.sub(r'\\lstinline{\1}\\index{\1@\\lstinline{\1}}', l)
   p = re.compile(r'.code-index "(.*)"')
   l = p.sub(r'\\lstinline{\1}\\index{\1@\\lstinline{\1}}', l)
   l = end(l)
   print(l, end="")

def section(l):
   p = re.compile(r'.section "(.+)"')
   l = p.sub(r'\\section{\1}', l)
   print(l, end="")

def figref(l):
   p = re.compile(r'.figref ([:\w]+)')
   m = p.search(l)
   if p != None:
      n = m.groups()[0].split(':')
      if len(n) > 1:
         n = n[1]
      else:
         n = n[0]
      l = p.sub(r'Figure~\\ref{fig:%s}' % n, l)
      l = end(l)
   print(l, end="")

def caption(fname):
   with open(fname) as f:  
      line = f.readline()
      while line:
         if not line.startswith("."):
            return line.rstrip('\n')
         line = f.readline()

def istable(fname):
   with open(fname) as f:  
      line = f.readline()
      while line:
         if line.startswith(".TS"):
            return True
         line = f.readline()
      return False

def table(fname):
   try: 
      with open(fname) as f:  
         print(r'\begin{tabular}{ll}')
         line = f.readline()  # F1
         line = f.readline()  # .TS
         line = f.readline()  # center
         line = f.readline()  # lb lb
         line = f.readline()  # l l.
         line = f.readline()  # heading
         p = re.compile(r'([\w\ ]+)\t+([\w]+)\n')
         line = p.sub(r'{\\bf \1} & {\\bf \2} \\\\', line)
         print(line)
         print("\midrule")
         # table
         line = f.readline()
         while line:
            if line.startswith(".TE"):
               break
            p = re.compile(r'([\(\)\w]+)\t+(.+)\n')
            line = p.sub(r'\1 & \2 \\\\', line)
            print(line)
            line = f.readline()
         line = f.readline() # F2
         line = f.readline() # caption
         print("\end{tabular}")
         print("\caption{" + line.strip("\n") + "}")
   except IOError:
      print("error: couldn't open %s" % fname, file=sys.stderr)

def insert(l):
   l = l.rstrip('\n')
   fname = l.split(' ')[1]
   try: 
      with open(fname) as f:
         line = f.readline()
         while line:
            print(line, end="")
            line = f.readline()
   except IOError:
      print("error: couldn't open %s" % fname, file=sys.stderr)

def listing(l):
   print(r'\begin{lstlisting}[]')
   line = f.readline()
   while line:
      if line.startswith(".so"):
         insert(line)
      elif line.startswith(".P2"):
         break
      else:
         print(line, end="")
      line = f.readline()
   print("\end{lstlisting}");
    
def figure(l):
   p = re.compile(r'.figure (.+)\n')
   l = p.sub(r'\1', l)
   name = "fig/%s.t" % l
   print("")
   print(r'\begin{figure}[t]')
   print("\center")
   if istable(name):
      table(name)
   else:
      print("\includegraphics[scale=0.5]{fig/%s.eps}" % l)
      print("\caption{%s}" % caption(name))
   print("\label{fig:%s}" % l)
   print("\end{figure}")

def italic(l):
   p = re.compile(r'.italic ([\w]+)')
   l = p.sub(r'\\textit{\1}', l)
   p = re.compile(r'.italic "(.*)"')
   l = p.sub(r'\\textit{\1}', l)

   l = end(l)
   print(l, end="")

def address(l):
   p = re.compile(r'.address (\w+)')
   l = p.sub(r'\\texttt{\1}', l)
   l = end(l)
   print(l, end="")

def register(l):
   p = re.compile(r'.register (\w+)')
   l = p.sub(r'\\texttt{\%\1}', l)
   l = end(l)
   print(l, end="")

def lineref(l):
   l = l.replace('!', '\\')
   p = re.compile(r'.line (.*):\/(.*)\/([+-]?\d+)?')
   l = p.sub(r'\\lineref{\1:/\2/\3}', l)
   p = re.compile(r'.line (.*):(\d+)')
   l = p.sub(r'\\lineref{\1:\2}', l)
   l = end(l)
   print(l, end="")

def linerefs(l):
   l = l.replace('!', '\\')
   p = re.compile(r'.lines (.*):\/(.*)\/([+-]?\d+)?,\/(.*)\/([+-]?\d+)?')
   l = p.sub(r'\\linerefs{\1:/\2/\3,/\4/\5}', l)
   l = end(l)
   print(l, end="")

def indent(l):
   p = re.compile(r'.IP (.*)\n')
   l = p.sub(r'\\paragraph{\\textbullet}', l)
   l = end(l)
   print(l, end="")

def file(l):
   p = re.compile(r'.file ([-\w\/\.]+)')
   l = p.sub(r'\\fileref{\1}', l)
   p = re.compile(r'.file "(.*)"')
   l = p.sub(r'\\fileref{\1}', l)
   l = end(l)
   print(l, end="")
   
def doref(l):
   # replace references in a line
   p = re.compile(r' \\\*\[([\w:]*)\]')
   l = p.sub(r'~\\ref{\1}', l)
   return l

def doquote(l):
   # replace references in a line
   p = re.compile(r'"([\w]+)"')
   l = p.sub(r"``\1''", l)
   return l

def docarrot(l):
   p = re.compile(r'(\d+)\^(\d+)(-\d+)*')
   l = p.sub(r'$\1^{\2}\3$', l)
   return l

def dourl(l):
   p = re.compile(r'https://([\w\.\/]+)')
   l = p.sub(r'\\url{https://\1}', l)
   return l

def tr2tex(l):
   if l.startswith(". "):
      l = "." + l[2:]
      
   if l.startswith(".ig"):
      ig2dotdot()
   elif l.startswith(".chapter "):
      chapter(l)
   elif l.startswith(".chapterlike"):
      chapter1(l)
   elif l.startswith(".PP"):
      print("")
   elif l.startswith(".code "):
      code(l)
   elif l.startswith(".italic-index"):
      defindex(l)
   elif l.startswith(".index"):
      index(l)
   elif l.startswith(".code-index"):
      codeindex(l)
   elif l.startswith(".section"):
      section(l)
   elif l.startswith(r'.\"'):
      comment(l)
   elif l.startswith(r'.figref'):
      figref(l)
   elif l.startswith(r'.figure '):
      figure(l)
   elif l.startswith(r'.italic'):
      italic(l)
   elif l.startswith(r'.address '):
      address(l)
   elif l.startswith(r'.register '):
      register(l)
   elif l.startswith(r'.line '):
      lineref(l)
   elif l.startswith(r'.lines '):
      linerefs(l)
   elif l.startswith(r'.P1'):
      listing(l)
   elif l.startswith(r'.IP'):
      indent(l)
   elif l.startswith(r'.file'):
      file(l)
   elif l.startswith(r'.sheet'):
      return
   elif l.startswith(r"."):
      print("TODO", l)
   else:
      l = docarrot(l)
      l = doref(l)
      l = doquote(l)
      l = dourl(l)
      print(l, end="")
   
with open(sys.argv[1]) as f:  
   line = f.readline()
   while line:
       tr2tex(line)
       line = f.readline()