Changeset 487

Timestamp:

09/09/08 16:17:26 (18 months ago)

Author:

tom

Message:

reverted html=False to full_strip=True

Files:

• libraries/web/wikipedia.py (modified) (10 diffs)

libraries/web/wikipedia.py

@@ -427 +427 @@
 class WikipediaPage:
     
-    def __init__(self, title, markup, light=False, html=False):
+    def __init__(self, title, markup, light=False, full_strip=True):
         
         """ Wikipedia page parser.
@@ -433 +433 @@
         The expected markup is the stuff in Wikipedia's edit textarea.
         With light=True, it will onlt parse links to other articles (which is faster).
-        With html=True, it will preserve some HTML markup (links, bold, italic).
+        With full_strip=False, it will preserve some HTML markup (links, bold, italic).
         
         """
@@ -439 +439 @@
         self.title = title
         self.markup = markup
-        self.html = html
+        self.full_strip = full_strip
         
         self.disambiguation = []
@@ -546 +546 @@
         Does some commonsense maintenance as well,
         like collapsing multiple spaces.
-        If you specified html=True for WikipediaPage instance,
+        If you specified full_strip=False for WikipediaPage instance,
         some markup is preserved as HTML (links, bold, italic).
         
@@ -552 +552 @@
         
         # Strip bold and italic.
-        if not self.html:
+        if self.full_strip:
             markup = markup.replace("'''", "")
             markup = markup.replace("''", "")
@@ -569 +569 @@
         # Strip links, keeping the display alias.
         # We'll strip the ending ]] later.
-        if not self.html:
+        if self.full_strip:
             markup = re.sub(r"\[\[[^\]]*?\|", "", markup)
         else:
@@ -626 +626 @@
         
         # Strip all HTML except <math> tags.
-        if not self.html:
+        if self.full_strip:
             markup = strip_tags(markup, exclude=["math"], linebreaks=True)
         
@@ -1547 +1547 @@
     
     def __init__(self, q, language="en", light=False, wait=10, asynchronous=False, cached=True,
-                 case_sensitive=False, html=False):
+                 case_sensitive=False, full_strip=True):
         
         """ A download manager for Wikipedia pages.
@@ -1561 +1561 @@
         
         self._light = light
-        self._html = html
+        self._full_strip = full_strip
         
         if cached: 
@@ -1587 +1587 @@
             data = ""
 
-        WikipediaPage.__init__(self,  title, data, light=self._light, html=self._html)
+        WikipediaPage.__init__(self,  title, data, light=self._light, full_strip=self._full_strip)
 
 def search(q, language="en", light=False, wait=10, asynchronous=False, cached=True, 
-           case_sensitive=False, html=False):
-    return WikipediaSearch(q, language, light, wait, asynchronous, cached, case_sensitive, html)
+           case_sensitive=False, full_strip=True):
+    return WikipediaSearch(q, language, light, wait, asynchronous, cached, case_sensitive, full_strip)
 
 ######################################################################################################