src/cm/ext/html2text.py
changeset 0 40c8f766c9b8
equal deleted inserted replaced
-1:000000000000 0:40c8f766c9b8
       
     1 #!/usr/bin/env python
       
     2 """html2text: Turn HTML into equivalent Markdown-structured text."""
       
     3 __version__ = "2.35"
       
     4 __author__ = "Aaron Swartz (me@aaronsw.com)"
       
     5 __copyright__ = "(C) 2004-2008 Aaron Swartz. GNU GPL 3."
       
     6 __contributors__ = ["Martin 'Joey' Schulze", "Ricardo Reyes"]
       
     7 
       
     8 # TODO:
       
     9 #   Support decoded entities with unifiable.
       
    10 #   Relative URL resolution
       
    11 
       
    12 if not hasattr(__builtins__, 'True'): True, False = 1, 0
       
    13 import re, sys, urllib, htmlentitydefs, codecs, StringIO, types
       
    14 import sgmllib
       
    15 sgmllib.charref = re.compile('&#([xX]?[0-9a-fA-F]+)[^0-9a-fA-F]')
       
    16 
       
    17 try: from textwrap import wrap
       
    18 except: pass
       
    19 
       
    20 # Use Unicode characters instead of their ascii psuedo-replacements
       
    21 UNICODE_SNOB = 0
       
    22 
       
    23 # Put the links after each paragraph instead of at the end.
       
    24 LINKS_EACH_PARAGRAPH = 0
       
    25 
       
    26 # Wrap long lines at position. 0 for no wrapping. (Requires Python 2.3.)
       
    27 BODY_WIDTH = 78
       
    28 
       
    29 # Don't show internal links (href="#local-anchor") -- corresponding link targets
       
    30 # won't be visible in the plain text file anyway.
       
    31 SKIP_INTERNAL_LINKS = False
       
    32 
       
    33 ### Entity Nonsense ###
       
    34 
       
    35 def name2cp(k):
       
    36     if k == 'apos': return ord("'")
       
    37     if hasattr(htmlentitydefs, "name2codepoint"): # requires Python 2.3
       
    38         return htmlentitydefs.name2codepoint[k]
       
    39     else:
       
    40         k = htmlentitydefs.entitydefs[k]
       
    41         if k.startswith("&#") and k.endswith(";"): return int(k[2:-1]) # not in latin-1
       
    42         return ord(codecs.latin_1_decode(k)[0])
       
    43 
       
    44 unifiable = {'rsquo':"'", 'lsquo':"'", 'rdquo':'"', 'ldquo':'"', 
       
    45 'copy':'(C)', 'mdash':'--', 'nbsp':' ', 'rarr':'->', 'larr':'<-', 'middot':'*',
       
    46 'ndash':'-', 'oelig':'oe', 'aelig':'ae',
       
    47 'agrave':'a', 'aacute':'a', 'acirc':'a', 'atilde':'a', 'auml':'a', 'aring':'a', 
       
    48 'egrave':'e', 'eacute':'e', 'ecirc':'e', 'euml':'e', 
       
    49 'igrave':'i', 'iacute':'i', 'icirc':'i', 'iuml':'i',
       
    50 'ograve':'o', 'oacute':'o', 'ocirc':'o', 'otilde':'o', 'ouml':'o', 
       
    51 'ugrave':'u', 'uacute':'u', 'ucirc':'u', 'uuml':'u'}
       
    52 
       
    53 unifiable_n = {}
       
    54 
       
    55 for k in unifiable.keys():
       
    56     unifiable_n[name2cp(k)] = unifiable[k]
       
    57 
       
    58 def charref(name):
       
    59     if name[0] in ['x','X']:
       
    60         c = int(name[1:], 16)
       
    61     else:
       
    62         c = int(name)
       
    63     
       
    64     if not UNICODE_SNOB and c in unifiable_n.keys():
       
    65         return unifiable_n[c]
       
    66     else:
       
    67         return unichr(c)
       
    68 
       
    69 def entityref(c):
       
    70     if not UNICODE_SNOB and c in unifiable.keys():
       
    71         return unifiable[c]
       
    72     else:
       
    73         try: name2cp(c)
       
    74         except KeyError: return "&" + c
       
    75         else: return unichr(name2cp(c))
       
    76 
       
    77 def replaceEntities(s):
       
    78     s = s.group(1)
       
    79     if s[0] == "#": 
       
    80         return charref(s[1:])
       
    81     else: return entityref(s)
       
    82 
       
    83 r_unescape = re.compile(r"&(#?[xX]?(?:[0-9a-fA-F]+|\w{1,8}));")
       
    84 def unescape(s):
       
    85     return r_unescape.sub(replaceEntities, s)
       
    86     
       
    87 def fixattrs(attrs):
       
    88     # Fix bug in sgmllib.py
       
    89     if not attrs: return attrs
       
    90     newattrs = []
       
    91     for attr in attrs:
       
    92         newattrs.append((attr[0], unescape(attr[1])))
       
    93     return newattrs
       
    94 
       
    95 ### End Entity Nonsense ###
       
    96 
       
    97 def onlywhite(line):
       
    98     """Return true if the line does only consist of whitespace characters."""
       
    99     for c in line:
       
   100         if c is not ' ' and c is not '  ':
       
   101             return c is ' '
       
   102     return line
       
   103 
       
   104 def optwrap(text):
       
   105     """Wrap all paragraphs in the provided text."""
       
   106     if not BODY_WIDTH:
       
   107         return text
       
   108     
       
   109     assert wrap, "Requires Python 2.3."
       
   110     result = ''
       
   111     newlines = 0
       
   112     for para in text.split("\n"):
       
   113         if len(para) > 0:
       
   114             if para[0] is not ' ' and para[0] is not '-' and para[0] is not '*':
       
   115                 for line in wrap(para, BODY_WIDTH):
       
   116                     result += line + "\n"
       
   117                 result += "\n"
       
   118                 newlines = 2
       
   119             else:
       
   120                 if not onlywhite(para):
       
   121                     result += para + "\n"
       
   122                     newlines = 1
       
   123         else:
       
   124             if newlines < 2:
       
   125                 result += "\n"
       
   126                 newlines += 1
       
   127     return result
       
   128 
       
   129 def hn(tag):
       
   130     if tag[0] == 'h' and len(tag) == 2:
       
   131         try:
       
   132             n = int(tag[1])
       
   133             if n in range(1, 10): return n
       
   134         except ValueError: return 0
       
   135 
       
   136 class _html2text(sgmllib.SGMLParser):
       
   137     def __init__(self, out=sys.stdout.write):
       
   138         sgmllib.SGMLParser.__init__(self)
       
   139         
       
   140         if out is None: self.out = self.outtextf
       
   141         else: self.out = out
       
   142         self.outtext = u''
       
   143         self.quiet = 0
       
   144         self.p_p = 0
       
   145         self.outcount = 0
       
   146         self.start = 1
       
   147         self.space = 0
       
   148         self.a = []
       
   149         self.astack = []
       
   150         self.acount = 0
       
   151         self.list = []
       
   152         self.blockquote = 0
       
   153         self.pre = 0
       
   154         self.startpre = 0
       
   155         self.lastWasNL = 0
       
   156         self.abbr_title = None # current abbreviation definition
       
   157         self.abbr_data = None # last inner HTML (for abbr being defined)
       
   158         self.abbr_list = {} # stack of abbreviations to write later
       
   159     
       
   160     def outtextf(self, s): 
       
   161         self.outtext += s
       
   162     
       
   163     def close(self):
       
   164         sgmllib.SGMLParser.close(self)
       
   165         
       
   166         self.pbr()
       
   167         self.o('', 0, 'end')
       
   168         
       
   169         return self.outtext
       
   170         
       
   171     def handle_charref(self, c):
       
   172         self.o(charref(c))
       
   173 
       
   174     def handle_entityref(self, c):
       
   175         self.o(entityref(c))
       
   176             
       
   177     def unknown_starttag(self, tag, attrs):
       
   178         self.handle_tag(tag, attrs, 1)
       
   179     
       
   180     def unknown_endtag(self, tag):
       
   181         self.handle_tag(tag, None, 0)
       
   182         
       
   183     def previousIndex(self, attrs):
       
   184         """ returns the index of certain set of attributes (of a link) in the
       
   185             self.a list
       
   186  
       
   187             If the set of attributes is not found, returns None
       
   188         """
       
   189         if not attrs.has_key('href'): return None
       
   190         
       
   191         i = -1
       
   192         for a in self.a:
       
   193             i += 1
       
   194             match = 0
       
   195             
       
   196             if a.has_key('href') and a['href'] == attrs['href']:
       
   197                 if a.has_key('title') or attrs.has_key('title'):
       
   198                         if (a.has_key('title') and attrs.has_key('title') and
       
   199                             a['title'] == attrs['title']):
       
   200                             match = True
       
   201                 else:
       
   202                     match = True
       
   203 
       
   204             if match: return i
       
   205 
       
   206     def handle_tag(self, tag, attrs, start):
       
   207         attrs = fixattrs(attrs)
       
   208     
       
   209         if hn(tag):
       
   210             self.p()
       
   211             if start: self.o(hn(tag)*"#" + ' ')
       
   212 
       
   213         if tag in ['p', 'div']: self.p()
       
   214         
       
   215         if tag == "br" and start: self.o("  \n")
       
   216 
       
   217         if tag == "hr" and start:
       
   218             self.p()
       
   219             self.o("* * *")
       
   220             self.p()
       
   221 
       
   222         if tag in ["head", "style", 'script']: 
       
   223             if start: self.quiet += 1
       
   224             else: self.quiet -= 1
       
   225 
       
   226         if tag in ["body"]:
       
   227             self.quiet = 0 # sites like 9rules.com never close <head>
       
   228         
       
   229         if tag == "blockquote":
       
   230             if start: 
       
   231                 self.p(); self.o('> ', 0, 1); self.start = 1
       
   232                 self.blockquote += 1
       
   233             else:
       
   234                 self.blockquote -= 1
       
   235                 self.p()
       
   236         
       
   237         if tag in ['em', 'i', 'u']: self.o("_")
       
   238         if tag in ['strong', 'b']: self.o("**")
       
   239         if tag == "code" and not self.pre: self.o('`') #TODO: `` `this` ``
       
   240         if tag == "abbr":
       
   241             if start:
       
   242                 attrsD = {}
       
   243                 for (x, y) in attrs: attrsD[x] = y
       
   244                 attrs = attrsD
       
   245                 
       
   246                 self.abbr_title = None
       
   247                 self.abbr_data = ''
       
   248                 if attrs.has_key('title'):
       
   249                     self.abbr_title = attrs['title']
       
   250             else:
       
   251                 if self.abbr_title != None:
       
   252                     self.abbr_list[self.abbr_data] = self.abbr_title
       
   253                     self.abbr_title = None
       
   254                 self.abbr_data = ''
       
   255         
       
   256         if tag == "a":
       
   257             if start:
       
   258                 attrsD = {}
       
   259                 for (x, y) in attrs: attrsD[x] = y
       
   260                 attrs = attrsD
       
   261                 if attrs.has_key('href') and not (SKIP_INTERNAL_LINKS and attrs['href'].startswith('#')): 
       
   262                     self.astack.append(attrs)
       
   263                     self.o("[")
       
   264                 else:
       
   265                     self.astack.append(None)
       
   266             else:
       
   267                 if self.astack:
       
   268                     a = self.astack.pop()
       
   269                     if a:
       
   270                         i = self.previousIndex(a)
       
   271                         if i is not None:
       
   272                             a = self.a[i]
       
   273                         else:
       
   274                             self.acount += 1
       
   275                             a['count'] = self.acount
       
   276                             a['outcount'] = self.outcount
       
   277                             self.a.append(a)
       
   278                         self.o("][" + `a['count']` + "]")
       
   279         
       
   280         if tag == "img" and start:
       
   281             attrsD = {}
       
   282             for (x, y) in attrs: attrsD[x] = y
       
   283             attrs = attrsD
       
   284             if attrs.has_key('src'):
       
   285                 attrs['href'] = attrs['src']
       
   286                 alt = attrs.get('alt', '')
       
   287                 i = self.previousIndex(attrs)
       
   288                 if i is not None:
       
   289                     attrs = self.a[i]
       
   290                 else:
       
   291                     self.acount += 1
       
   292                     attrs['count'] = self.acount
       
   293                     attrs['outcount'] = self.outcount
       
   294                     self.a.append(attrs)
       
   295                 self.o("![")
       
   296                 self.o(alt)
       
   297                 self.o("]["+`attrs['count']`+"]")
       
   298         
       
   299         if tag == 'dl' and start: self.p()
       
   300         if tag == 'dt' and not start: self.pbr()
       
   301         if tag == 'dd' and start: self.o('    ')
       
   302         if tag == 'dd' and not start: self.pbr()
       
   303         
       
   304         if tag in ["ol", "ul"]:
       
   305             if start:
       
   306                 self.list.append({'name':tag, 'num':0})
       
   307             else:
       
   308                 if self.list: self.list.pop()
       
   309             
       
   310             self.p()
       
   311         
       
   312         if tag == 'li':
       
   313             if start:
       
   314                 self.pbr()
       
   315                 if self.list: li = self.list[-1]
       
   316                 else: li = {'name':'ul', 'num':0}
       
   317                 self.o("  "*len(self.list)) #TODO: line up <ol><li>s > 9 correctly.
       
   318                 if li['name'] == "ul": self.o("* ")
       
   319                 elif li['name'] == "ol":
       
   320                     li['num'] += 1
       
   321                     self.o(`li['num']`+". ")
       
   322                 self.start = 1
       
   323             else:
       
   324                 self.pbr()
       
   325         
       
   326         if tag in ["table", "tr"] and start: self.p()
       
   327         if tag == 'td': self.pbr()
       
   328         
       
   329         if tag == "pre":
       
   330             if start:
       
   331                 self.startpre = 1
       
   332                 self.pre = 1
       
   333             else:
       
   334                 self.pre = 0
       
   335             self.p()
       
   336             
       
   337     def pbr(self):
       
   338         if self.p_p == 0: self.p_p = 1
       
   339 
       
   340     def p(self): self.p_p = 2
       
   341     
       
   342     def o(self, data, puredata=0, force=0):
       
   343         if self.abbr_data is not None: self.abbr_data += data
       
   344         
       
   345         if not self.quiet: 
       
   346             if puredata and not self.pre:
       
   347                 data = re.sub('\s+', ' ', data)
       
   348                 if data and data[0] == ' ':
       
   349                     self.space = 1
       
   350                     data = data[1:]
       
   351             if not data and not force: return
       
   352             
       
   353             if self.startpre:
       
   354                 #self.out(" :") #TODO: not output when already one there
       
   355                 self.startpre = 0
       
   356             
       
   357             bq = (">" * self.blockquote)
       
   358             if not (force and data and data[0] == ">") and self.blockquote: bq += " "
       
   359             
       
   360             if self.pre:
       
   361                 bq += "    "
       
   362                 data = data.replace("\n", "\n"+bq)
       
   363             
       
   364             if self.start:
       
   365                 self.space = 0
       
   366                 self.p_p = 0
       
   367                 self.start = 0
       
   368 
       
   369             if force == 'end':
       
   370                 # It's the end.
       
   371                 self.p_p = 0
       
   372                 self.out("\n")
       
   373                 self.space = 0
       
   374 
       
   375 
       
   376             if self.p_p:
       
   377                 self.out(('\n'+bq)*self.p_p)
       
   378                 self.space = 0
       
   379                 
       
   380             if self.space:
       
   381                 if not self.lastWasNL: self.out(' ')
       
   382                 self.space = 0
       
   383 
       
   384             if self.a and ((self.p_p == 2 and LINKS_EACH_PARAGRAPH) or force == "end"):
       
   385                 if force == "end": self.out("\n")
       
   386 
       
   387                 newa = []
       
   388                 for link in self.a:
       
   389                     if self.outcount > link['outcount']:
       
   390                         self.out("   ["+`link['count']`+"]: " + link['href']) #TODO: base href
       
   391                         if link.has_key('title'): self.out(" ("+link['title']+")")
       
   392                         self.out("\n")
       
   393                     else:
       
   394                         newa.append(link)
       
   395 
       
   396                 if self.a != newa: self.out("\n") # Don't need an extra line when nothing was done.
       
   397 
       
   398                 self.a = newa
       
   399             
       
   400             if self.abbr_list and force == "end":
       
   401                 for abbr, definition in self.abbr_list.items():
       
   402                     self.out("  *[" + abbr + "]: " + definition + "\n")
       
   403 
       
   404             self.p_p = 0
       
   405             self.out(data)
       
   406             self.lastWasNL = data and data[-1] == '\n'
       
   407             self.outcount += 1
       
   408 
       
   409     def handle_data(self, data):
       
   410         if r'\/script>' in data: self.quiet -= 1
       
   411         self.o(data, 1)
       
   412     
       
   413     def unknown_decl(self, data): pass
       
   414 
       
   415 def wrapwrite(text): sys.stdout.write(text.encode('utf8'))
       
   416 
       
   417 def html2text_file(html, out=wrapwrite):
       
   418     h = _html2text(out)
       
   419     h.feed(html)
       
   420     h.feed("")
       
   421     return h.close()
       
   422 
       
   423 def html2text(html):
       
   424     return optwrap(html2text_file(html, None))
       
   425 
       
   426 if __name__ == "__main__":
       
   427     if sys.argv[1:]:
       
   428         arg = sys.argv[1]
       
   429         if arg.startswith('http://'):
       
   430             j = urllib.urlopen(arg)
       
   431             try:
       
   432                 from feedparser import _getCharacterEncoding as enc
       
   433             except ImportError:
       
   434                    enc = lambda x, y: ('utf-8', 1)
       
   435             text = j.read()
       
   436             encoding = enc(j.headers, text)[0]
       
   437             if encoding == 'us-ascii': encoding = 'utf-8'
       
   438             data = text.decode(encoding)
       
   439 
       
   440         else:
       
   441             encoding = 'utf8'
       
   442             if len(sys.argv) > 2:
       
   443                 encoding = sys.argv[2]
       
   444             data = open(arg, 'r').read().decode(encoding)
       
   445     else:
       
   446         data = sys.stdin.read().decode('utf8')
       
   447     wrapwrite(html2text(data))
       
   448