Changeset 389

Timestamp:

01/19/2010 05:03:28 PM (8 months ago)

Author:

matt

Message:

Added wikiparser.textify, wikiparser.core.Events. Renamed Replace to EventsIf?.

Location:

projects/wikiparser/trunk

Files:

• tests/test_matches.py (modified) (2 diffs)
• wikiparser/formats/textbook/matchers.py (modified) (2 diffs)
• wikiparser/match/core.py (modified) (4 diffs)
• wikiparser/textify.py (added)

projects/wikiparser/trunk/tests/test_matches.py

@@ -2 +2 @@
 import re
 
-from wikiparser.markup import BREAK, StreamTree, tostring
+from wikiparser.markup import ATTR, BREAK, StreamTree, tostring
 from wikiparser.match.blocks import LineStart, BlockStart, BreakBefore, BreakAfter, PyConsole
-from wikiparser.match.core import Suppress, OneOrMore, ZeroOrMore, StringEnd, Not, FollowedBy
-from wikiparser.match.data import Element
+from wikiparser.match.core import (Literal, Events,
+                                   Suppress, OneOrMore, ZeroOrMore, StringEnd, Not,
+                                   FollowedBy)
+from wikiparser.match.data import Element, Attr
 from wikiparser.match.text import Mixed, Regex, Until
 
@@ -135 +137 @@
                                   (BREAK, (0, 0, {})), 'Nice, huh?\n', 85
                                   ])
-
+        
+    def test_orelse(self):
+        m = Suppress(":") + Attr("type", Regex("[^:]+") | Events("item")) + Suppress(":")
+        stream = list(m.stream(":blah:"))
+        self.assertEqual(stream, [(ATTR, ('type', 'blah', True)), 6])
+        stream = list(m.stream("::"))
+        self.assertEqual(stream, [(ATTR, ('type', 'item', True)), 2])
 
 if __name__ == '__main__':

projects/wikiparser/trunk/wikiparser/formats/textbook/matchers.py

@@ -16 +16 @@
 
 from wikiparser.markup import HTML, WIKI
-from wikiparser.match.blocks import BlockStart, BreakAfter, IndentChange, LineStart, StartMarker
-from wikiparser.match.core import Choice, Escape, FollowedBy, Optional, Suppress, ZeroOrMore
-from wikiparser.match.data import AttrIf, Element, RegexAttr
+from wikiparser.match.blocks import (BlockStart, BreakAfter, IndentChange, LineStart,
+                                     StartMarker)
+from wikiparser.match.core import (Choice, Escape, Events, FollowedBy, Optional, Suppress,
+                                   ZeroOrMore)
+from wikiparser.match.data import Attr, AttrIf, Element, RegexAttr
 from wikiparser.match.text import (CharacterEntities, Mixed, Regex, SmartQuotes,
                                    Substitution, Until)
@@ -134 +136 @@
 bullet = LineStart(type="bullet") + StartMarker("[*-]+ ", first_chars="*-")
 ordinal = LineStart(type="ord") + StartMarker("#+\\.? ", first_chars="#")
-item = (LineStart(type="item")
-        + Suppress("::")
+item = (LineStart()
+        + Suppress(":") + Attr("type", Regex("[A-Za-z_][A-Za-z_0-9]*") | Events("item")) + Suppress(":")
         + Mixed(inlinecontent_lb, ":", inlineonly=True)
         + Suppress(":\n"))

projects/wikiparser/trunk/wikiparser/match/core.py

@@ -19 +19 @@
 
 from wikiparser.events import BREAK, br, isbr
+from wikiparser.filters.streams import textof
 
 
@@ -555 +556 @@
 class Apply(Wrapper):
     """Buffers events from the sub-matcher, then calls a function on the queued
-    events. The events returned by the function, if any, are then yielded.
-    """
-
-    def __init__(self, function, matcher):
+    events. Any event(s) returned by the function are then emitted.
+    """
+
+    def __init__(self, function, matcher, astext=False):
+        """
+        :param function: the function to call on the wrapped matcher's events.
+        :param matcher: the wrapped matcher.
+        :param astext: if True, the text from the wrapped matcher's events is
+            extracted and the function is called on the extracted string,
+            instead of on a list of events.
+        """
+        
         Wrapper.__init__(self, matcher)
         self.function = function
+        self.astext = astext
     
     def parse(self, input, position, ctx):
@@ -571 +581 @@
                 
         if queue:
-            for ev in self.function(queue):
-                yield ev
+            if self.astext:
+                yield self.function(textof(queue))
+            else:
+                for ev in self.function(queue):
+                    yield ev
             yield position
 
@@ -595 +608 @@
 
 
-class Replace(Wrapper):
-    """Replaces the output of the wrapped matcher with a pre-defined stream.
+class Events(Matcher):
+    """Inserts events into the stream. This is useful as a kind of "else"
+    clause when used as the last matcher in a Choice matcher, to insert
+    pre-determined events in the case where the other matchers don't match.
+    
+    For example, say you want a matcher that either matches a pattern like
+    `:typename:`, or allows the "typename" to be missing, in which case it
+    should be set to "item"::
+    
+        Suppress(":") + Attr("type", Regex("[^:]+") | Events("item")) + Suppress(":")
+    """
+    
+    def __init__(self, *events):
+        """Takes a variable number of arguments, where the arguments are used
+        as the list of events to emit.
+        """
+        self.events = events
+        
+    def parse(self, input, position, ctx):
+        for ev in self.events:
+            yield ev
+        yield position
+
+
+class EventsIf(Wrapper):
+    """If the wrapped matcher matches, replaces its events with a predefined
+    stream.
     """