[Pywikipedia-l] SVN: [3923] trunk/pywikiparser - pywikibot

30 Jul 2007

Revision: 3923
Author:   valhallasw
Date:     2007-07-29 21:09:30 +0000 (Sun, 29 Jul 2007)
Log Message:
-----------
Added lexer symbold STAR COLON SEMICOLON HASH
Parser now accepts all input (but does not parse everything)
Modified Paths:
--------------
    trunk/pywikiparser/Lexer.py
    trunk/pywikiparser/Parser.py
Modified: trunk/pywikiparser/Lexer.py
===================================================================

--- trunk/pywikiparser/Lexer.py	2007-07-29 20:29:29 UTC (rev 3922)
+++ trunk/pywikiparser/Lexer.py	2007-07-29 21:09:30 UTC (rev 3923)
@@ -25,6 +25,10 @@
                ('PIPE',        '|     Pipe symbol'),
                ('EQUAL_SIGN',  '=     Equal sign'),
                ('APOSTROPHE',  ''     Apostrophe'),
+               ('STAR',        '*     Star sign'),
+               ('COLON',       ':     Colon'),
+               ('SEMICOLON',   ';     Semicolon'),
+               ('HASH',        '#     Hash symbol'),
                ('CURL_OPEN',   '{     Curly bracket open'),
                ('CURL_CLOSE',  '}     Curly bracket close'),
                ('ANGL_OPEN',   '<     Angular bracket open'),
@@ -56,7 +60,7 @@
         try:
             c = self.getchar()
             while True:
-                if (c in ('[', ']', '{', '}', '<', '>', '=', ''')):
+                if (c in ('[', ']', '{', '}', '<', '>', '=', ''', '*', ':', ';', '#')):
                     if text:
                         yield (Tokens.TEXT, text)
                         text = ''
@@ -76,6 +80,10 @@
                         elif (c == '>'): yield (Tokens.ANGL_CLOSE, num)
                         elif (c == '='): yield (Tokens.EQUAL_SIGN, num)
                         elif (c == '''): yield(Tokens.APOSTROPHE, num)
+                        elif (c == '*'): yield (Tokens.STAR,       num)
+                        elif (c == ':'): yield (Tokens.COLON,      num)
+                        elif (c == ';'): yield (Tokens.SEMICOLON,  num)
+                        elif (c == '#'): yield (Tokens.HASH,       num)
                     c = t
                 elif (c == '|'):
                     if text:
Modified: trunk/pywikiparser/Parser.py
===================================================================
--- trunk/pywikiparser/Parser.py	2007-07-29 20:29:29 UTC (rev 3922)
+++ trunk/pywikiparser/Parser.py	2007-07-29 21:09:30 UTC (rev 3923)
@@ -58,6 +58,8 @@
     def parseEOF(self):
         token = self.expect(Tokens.EOF)
         raise StopIteration
+    
+    # Special functions that directly access the storage tree
def parseNEWPAR(self):
         token = self.expect(Tokens.NEWPAR)
@@ -113,6 +115,8 @@
             self.bold = True
         return []
+    # Functions that return the input directly
+    
     def parseSQRE_CLOSE(self):
         token = self.expect(Tokens.SQRE_CLOSE)
         return [']'*token[1]]
@@ -133,6 +137,22 @@
         token = self.expect(Tokens.ANGL_CLOSE)
         return ['>'*token[1]]
+    def parseSTAR(self):
+        token = self.expect(Tokens.STAR)
+        return ['*'*token[1]]
+        
+    def parseCOLON(self):
+        token = self.expect(Tokens.COLON)
+        return [':'*token[1]]
+        
+    def parseSEMICOLON(self):
+        token = self.expect(Tokens.SEMICOLON)
+        return [';'*token[1]]
+        
+    def parseHASH(self):
+        token = self.expect(Tokens.HASH)
+        return ['#'*token[1]]
+
     def parseTAB_NEWLINE(self):
         token = self.expect(Tokens.TAB_NEWLINE)
         return ['|-']
@@ -140,7 +160,11 @@
     def parseTAB_CLOSE(self):
         token = self.expect(Tokens.TAB_CLOSE)
         return ['|}']
+
+
+    # True parser callers
+
     def parseWHITESPACE(self):
         return self.parseTEXT()
@@ -156,15 +180,66 @@
             return []
def parseSQRE_OPEN(self):
+        try:
+            return self.parseWikilink()
+        except ParseError: pass
+            
+        self.lex.undo()
+        try:
+            return self.parseExternallink()
+        except ParseError: pass
+        
+        self.lex.undo()
         token = self.expect(Tokens.SQRE_OPEN)
+        return ['['*token[1]]
+        
     def parseCURL_OPEN(self):
+        try:
+            return self.parseTemplateparam()
+        except ParseError: pass
+        
+        self.lex.undo()
+        try:
+            return self.parseTemplate()
+        except ParseError: pass
+
+        self.lex.undo()
         token = self.expect(Tokens.CURL_OPEN)
+        return ['{'*token[1]]
+                
     def parseANGL_OPEN(self):
+        try:
+            return self.parseHTML()
+        except ParseError: pass
+        
+        self.lex.undo()
         token = self.expect(Tokens.ANGL_OPEN)
+        return ['<'*token[1]]
+
     def parseTAB_OPEN(self):
+        try:
+            return self.parseWikitable()
+        except ParseError: pass
+        
+        self.lex.undo()
         token = self.expect(Tokens.TAB_OPEN)
-
-
-
+        return ['{|']
-             
\ No newline at end of file
+    def parseWikilink(self):
+        raise ParseError("Needs implementation")
+        
+    def parseExternallink(self):
+        raise ParseError("Needs implementation")
+    
+    def parseTemplateparam(self):
+        raise ParseError("Needs implementation")
+        
+    def parseTemplate(self):
+        raise ParseError("Needs implementation")
+        
+    def parseHTML(self):
+        raise ParseError("Needs implementation")
+        
+    def parseWikitable(self):
+    	raise ParseError("Needs implementation")
+    	
\ No newline at end of file

    