™N
cK•;c       s¬     d  Z    k Z k Z k Z k Z k Z  k Z  d Z  d d „ Z b d d d „ Z	 z d „  Z
 ~ d „  Z — e d j o  ˜ e e e i ƒ e i ƒ n d S(	   s0   
  html2text.py

  convert an html doc to text

iP   c    sã    t  i d d |  ƒ }   t  i d d |  ƒ }   t i |  d d ƒ }   g  }  t  i d |  ƒ }  xÈ | d  r» }  t i	 | d  ƒ d	 j oˆ  t  i
 d
 | ƒ }  | o  t  i
 d | ƒ } n   | o1 ! | i d | | i d ƒ | i d ƒ !ƒ n # | i d ƒ n % | i | ƒ qu W& t i | d ƒ }  ' t  i d d |  ƒ }  ( t  i d d d d d |  ƒ }  ) t  i d d |  ƒ }  * | o + t  i d d |  ƒ }  n - t  i d d |  ƒ }  . | o / t  i d d |  ƒ }  n 1 t  i d d |  ƒ }  2 t  i d d |  ƒ }  3 t  i d d |  ƒ }  4 t  i d d |  ƒ }  5 t  i d d  |  ƒ }  6 | o” 7 t  i d! d" |  ƒ }  8 t  i d# d$ |  ƒ }  9 t  i d% d& |  ƒ }  : t  i d' d$ |  ƒ }  ; t  i d( d) |  ƒ }  < t  i d* d+ |  ƒ }  n1 > t  i d( d |  ƒ }  ? t  i d* d  |  ƒ }  @ t  i d, d |  ƒ }  A t  i d- d |  ƒ }  B t  i d. d |  ƒ }  C t  i d/ d0 |  ƒ }  D t  i d1 d0 |  ƒ }  E t  i d2 d |  ƒ }  F t i |  d ƒ } G g  } H xt| d H rg}	 I |	 d3 j o J | d } nAK |	 d4 j o. L | d } M | d j  o M d } n nN |	 d5 j o O | d } nâ P |	 d6 j o. Q | d } R | d j  o R d } n n¤ T | o U d7 | t i |	 ƒ }	 V t  i d8 d |	 ƒ }
 W t  i d9 d |
 ƒ }
 X t d: t |
 ƒ d: } Y | i d | |	 ƒ n [ | i d7 | |	 ƒ qW\ t i | d ƒ }  ] t  i d2 d |  ƒ }  ^ t  i d; d< |  ƒ }  _ t  i d9 d |  ƒ }  ` |  | | f Sd  S(=   Ns   \s+s    s   (?s)<!--.*?-->s    s   
s   (?si)(<img.*?>)i    i   s   <imgs   (?si)alt\s*=\s*"([^"]*)"s   (?si)alt\s*=([^\s]*)s   %si   s   [img]s   (?i)<br>s   (?i)<hr[^>]*>s   -i2   s   (?i)<li>s   
* s   (?i)<h[0-9]>s   
[32ms   (?i)</h[0-9]>s   [0m
s   (?i)<ul>s   
<UL>
s	   (?i)</ul>s   
</UL>
s   (?i)<center>s
   
<CENTER>
s   (?i)</center>s   
</CENTER>
s   (?i)<b>s   [1ms   (?i)</b>s   [0ms   (?i)<i>s   [2ms   (?i)</i>s   (?i)<title>s   
<CENTER>
[31ms   (?i)</title>s   [0m
</CENTER>
s   (?i)<p>s   (?i)<tr[^>]*>s   (?i)</table>s   (?i)<td[^>]*>s   	s   (?i)<th[^>]*>s    *
s   <UL>s   </UL>s   <CENTER>s	   </CENTER>s     s   \[.*?ms   <[^>]*>i   s   


*s   

(   s   res   subs   datas   strings   replaces   outputs   splits   imgss   imgs   lowers   searchs   alts   appends   starts   ends   joins   use_ansis   liness   lines   indents   centers   strips   nlines   WIDTHs   lens   c(   s   datas   centers   indents   use_ansis   outputs   imgss   imgs   alts   liness   lines   nlines   cs   ./html2text.pys   tag_replace s   	 	
1$


	 	  
i    c 	   sx  b c t  i d |  ƒ } d g  } e d } f d } g x„ | d g rw } h | d  d j o5 i t | | | | ƒ \ } } } j | i | ƒ n) l t  i d d | ƒ } m | i | ƒ q? Wn t i | ƒ }  o t  i d d |  ƒ }  p t  i d	 d
 |  ƒ }  q t  i d d |  ƒ }  r | oL s t  i d d |  ƒ }  t t  i d d |  ƒ }  u t  i d d |  ƒ }  n w |  Sd  S(   Ns   (?s)(<pre>[^<]*</pre>)i    i   s   <pre>s   (?i)</*pre>s    s   &gt;s   >s   &lt;s   <s   &nbsp;s    s   &copy;s   ©s   &eacute;s   és   &egrave;s   è(   s   res   splits   datas   pres   outs   indents   centers   parts   tag_replaces   use_ansis   ress   appends   subs   strings   joins	   is_latin1(	   s   datas   use_ansis	   is_latin1s   pres   outs   indents   centers   parts   ress   ./html2text.pys	   html2textb s(   			 	!
c    s   z { d |  GH| t GHd  S(   Ns   usage: %s --help <htmlfile>(   s   prognames   __doc__(   s   prognames   ./html2text.pys   usagez s   c 
   sý   ~  | d } € t i | d d d g ƒ \ } } ‚ xA | d ‚ r4 \ } } ƒ | d j o „ t | ƒ … d  Sn qA W‡ t | ƒ o ˆ | d } n Š | d } Œ t
 | ƒ }  | i ƒ  }	 Ž | i ƒ   |	 o ‘ t |	 ƒ GHn	 “ d GHd  S(   Ni    i   s    s   helps   --helps   Document contained no data(   s   argvs   prognames   getopts   alists   argss   fields   vals   usages   lens   files   opens   fps   reads   datas   closes	   html2text(
   s   argcs   argvs   prognames   alists   argss   fields   vals   files   fps   datas   ./html2text.pys   main~ s    % 
s   __main__N(   s   __doc__s   oss   syss   strings   times   getopts   res   WIDTHs   tag_replaces	   html2texts   usages   mains   __name__s   lens   argv(    s   ./html2text.pys   ? s   !		Q