Improve text output generation

* wiki2html.py (mktgt): Add 3rd argument. All callers updated.
* wiki2text.py (wiki_ns_name,mktgt): New functions
(str_link,str_tmpl): Rewrite
* wikicvt.py: New argument --input-text (not used yet)

1 files changed, 46 insertions, 34 deletions

diff --git a/wiki2text.py b/wiki2text.py
index 51a7853..564809f 100644
--- a/wiki2text.py
+++ b/wiki2text.py
@@ -17,6 +17,7 @@
 
 from wikimarkup import *
 from types import TupleType
+from wikins import wiki_ns_re, wiki_ns
 import urllib
 
 class TextWikiMarkup (WikiMarkup):
@@ -43,52 +44,60 @@ class TextWikiMarkup (WikiMarkup):
         if 'markup' in keywords:
             self.markup = keywords['markup']
 
-    def target(self, t):
-        (qual,sep,tgt) = t.partition(':')
-        r = None
-        if tgt != '':
-            if qual == "Image":
-                t = self.image_base + '/' + urllib.quote(tgt)
-            elif qual == "Media":
-                t = self.media_base + '/' + tgt
-            elif qual in self.langtab:
-                t = self.html_base % { 'lang' : qual } + '/' + urllib.quote(tgt)
-                r = self.langtab[qual]
-            else:
-                t = self.html_base % { 'lang' : self.lang } + '/' + urllib.quote(t)
-        else:
-            t = self.html_base % { 'lang' : self.lang } + '/' + urllib.quote(t)
-        return t, r
-
     def xref(self, text, target):
         if text:
             return "%s (see %s) " % (text, target)
         else:
             return "see " + target
 
-    def str_link(self, tok, env):
+    def wiki_ns_name(self, str):
+        if str in wiki_ns[self.lang]:
+            return wiki_ns[self.lang][str]
+        elif str in wiki_ns_re[self.lang]:
+            for elt in wiki_ns_re[self.lang][str]:
+                if str.beginswith(elt[0]) and str.endswith(elt[1]):
+                    return elt[2]
+        return None            
+    def mktgt(self, tgt, lang = None):
+        if not lang:
+            lang = self.lang
+        return self.html_base % { 'lang' : lang } + urllib.quote(tgt)
+    
+    def link(self, tok, env, istmpl):
         arg = self.fmtok(tok[1], env)
         text = self.fmtok(tok[2], env)
+        (qual,sep,tgt) = arg.partition(':')
+        if tgt != '':
+            ns = self.wiki_ns_name(qual)
+            if ns:
+                if ns == 'NS_IMAGE':
+                    text = "[%s: %s]" % (qual, text if text else arg)
+                    tgt = self.image_base + '/' + \
+                                 urllib.quote(tgt) + \
+                                 '/250px-' + urllib.quote(tgt)
+                elif ns == 'NS_MEDIA':
+                    text = "[%s]" % (qual)
+                else:
+                    tgt = self.mktgt(tgt)
+            elif not istmpl and qual in self.langtab:
+                text = self.langtab[qual] + ": " + tgt
+                tgt = self.mktgt(tgt, qual)
+            else:
+                tgt = self.mktgt(tgt)
+        else:
+            tgt = self.mktgt(arg)
         if self.references:
-            (target, r) = self.target(arg)
-            return self.xref(text if text else r, target)
+            return "%s (see %s) " % (text, tgt)
+        elif not text or text == '':
+            return arg
         else:
-            (qual,sep,tgt) = arg.partition(':')
-            if sep != '':
-                return ""
-            elif text:
             return text
-            return arg
+    
+    def str_link(self, tok, env):
+        return self.link(tok, env, False)
 
     def str_tmpl(self, tok, env):
-        arg = self.fmtok(tok[1], env)
-        (target, r) = self.target(arg)
-        text = self.fmtok(tok[2], env)
-        if not text and r:
-            text = r
-        if self.references:
-            return self.xref(text, target)
-        return text
+        return self.link(tok, env, True)
         
     def str_ref(self, tok, env):
         return self.xref(self.fmtok(tok[2], env), self.fmtok(tok[1], env))
@@ -165,12 +174,15 @@ class TextWiktionaryMarkup (TextWikiMarkup):
             s = ""
             self.seq_pos=0
             for t in tok[1:]:
-                s += self.fmtok(t, env)
+                x = self.fmtok(t, env)
+                if x:
+                    s += x
                 self.seq_pos += 1
             return s
     
     def str_tmpl(self, tok, env):
         arg = self.fmtok(tok[1], env)
+        if arg and arg != '':
             if self.seq_pos > 0:
                 return arg
             else: