sql fixes

--HG--
branch : pmacs2

code_examples/MakeEfileServer.sql

@@ -2,6 +2,15 @@
 -- these tables are distinct from the esub tables
 -- they are used to proxy communications between the IRS and the BDS
 
+create function foo() returns integer as '
+declare
+    x integer := 30;
+begin
+    raise notice ''blarg''; --urk
+    RETURN x;
+end;
+' language plgpsql;
+
 -- this table stores all the various username/password pairs we use
 -- the one with the most recent date is assumed to be current
 CREATE TABLE e2_password (

lex2.py

@@ -24,7 +24,8 @@ class Token(object):
             names = self.parent.domain()
         else:
             names = []
-        names.append(self.rule.name)
+        if self.name != 'middle':
+            names.append(self.rule.name)
         return names
     def fqlist(self):
         if self.parent is not None:
@@ -301,8 +302,17 @@ class DualRegionRule(Rule):
         self.middle   = middle
         self.grammar2 = grammar2
         self.end      = end
-        self.start_re = re.compile(start)
+        #self.start_re = re.compile(start)
+        self.start_re = self._compile_start()
         self._set_group(group)
+
+    def _compile_start(self):
+        return re.compile(self.start)
+    def _compile_middle(self, d):
+        return re.compile(self.middle % d)
+    def _compile_end(self, d):
+        return re.compile(self.end % d)
+
     def _add_from_regex(self, name, lexer, parent, m, matchd={}):
         s = m.group(0)
         token = self.make_token(lexer, s, name, parent, matchd)
@@ -343,7 +353,8 @@ class DualRegionRule(Rule):
         d1 = parent.matchd
         assert parent.name == 'start'
         null_t = None
-        middle_re = re.compile(self.middle % d1)
+        #middle_re = re.compile(self.middle % d1)
+        middle_re = self._compile_middle(d1)
         d2 = {}
 
         # ok, so as long as we aren't done (we haven't found an end token),
@@ -428,7 +439,8 @@ class DualRegionRule(Rule):
         #assert parent.name == 'middle'
         d3 = parent.matchd
         null_t = None
-        end_re = re.compile(self.end % d3)
+        #end_re = re.compile(self.end % d3)
+        end_re = self._compile_end(d3)
 
         # ok, so as long as we aren't done (we haven't found an end token),
         # keep reading input
@@ -509,6 +521,10 @@ class Grammar:
         for rule in self.rules:
             if hasattr(rule, 'grammar') and rule.grammar is None:
                 rule.grammar = self
+            if hasattr(rule, 'grammar1') and rule.grammar is None:
+                rule.grammar = self
+            if hasattr(rule, 'grammar2') and rule.grammar is None:
+                rule.grammar = self
 
 grammars = {}
 grammars['null'] = Grammar()
@@ -600,3 +616,10 @@ class Lexer:
         else:
             raise StopIteration
 
+class NocaseDualRegionRule(DualRegionRule):
+    def _compile_start(self):
+        return re.compile(self.start, re.IGNORECASE)
+    def _compile_middle(self, d):
+        return re.compile(self.middle % d, re.IGNORECASE)
+    def _compile_end(self, d):
+        return re.compile(self.end % d, re.IGNORECASE)

mode_sql.py

@@ -1,18 +1,60 @@
 import color, mode2, tab2
-from lex2 import Grammar, PatternRule, NocasePatternRule, RegionRule
+from lex2 import Grammar, PatternRule, NocasePatternRule, RegionRule, NocaseRegionRule, DualRegionRule, NocaseDualRegionRule
 from mode_python import StringGrammar
 
+class PlPgSqlGrammar(Grammar):
+    rules = [
+        PatternRule(r'comment', r'--.*\n$'),
+        RegionRule(r'comment', '/\*', Grammar, '\*/'),
+        PatternRule(r'delimiter', r':=|[():;,\.\$\[\]]'),
+    
+        NocasePatternRule(r'attribute', r'(?:check|exists|unique|not null|default|primary key|minvalue|foreign key|references)(?![A-Za-z0-9_])'),
+        NocasePatternRule(r'keyword', r'(?:declare|begin|end|raise notice|return)'),
+        NocasePatternRule(r'operator', r'(?:case|when|then|else|end|not|and|or|is not|is|in|not in)(?![A-Za-z0-9_])'),
+        NocasePatternRule(r'keyword', r'(?:create database|create index|create sequence|create table|create trigger|create view|select|insert|update|delete|drop database|drop index|drop sequence|drop table|drop trigger|drop view|create user|alter user|drop user|drop function|grant|revoke|create function|create or replace function|create or replace view|create language|create operator|create type)(?![A-Za-z0-9_])'),
+        NocasePatternRule(r'pseudokeyword', r'(?:returns|language|right join|left join|inner join|outer join|join|where|null|true|false|into|values|as|from|order by|asc|desc|limit|distinct|cascade|using|on)(?![A-Za-z0-9_])'),
+        NocasePatternRule(r'type', r'(?:void|row|serial|varchar|float|integer|int|text|timestamptz|timestamp|datetz|date|timetz|time|boolean|bool)(?![A-Za-z0-9_])'),
+        PatternRule(r'builtin', r'(?:nextval|current_timestamp|current_time|current_date)(?![A-Za-z0-9_])'),
+        RegionRule(r'string', "''", StringGrammar, "''"),
+        RegionRule(r'quoted', '"', StringGrammar, '"'),
+        PatternRule(r'bareword', r'[A-Za-z0-9_]+'),
+        PatternRule(r'empty', r'^ *\n$'),
+        PatternRule(r'eol', r'\n'),
+    ]
+
+class FunctionGrammar(Grammar):
+    rules = [
+        PatternRule(r'comment', r'--.*\n$'),
+        RegionRule(r'comment', '/\*', Grammar, '\*/'),
+        PatternRule(r'delimiter', r':=|[():;,\.\$\[\]]'),
+        
+        PatternRule(r'name', r'[a-zA-Z_][a-zA-Z0-9_]*(?=\()'),
+        NocasePatternRule(r'keyword', r'(?:as|returns|language)'),
+        NocasePatternRule(r'type', r'(?:void|row|serial|varchar|float|integer|int|text|timestamptz|timestamp|datetz|date|timetz|time|boolean|bool)(?![A-Za-z0-9_])'),
+    
+        NocasePatternRule(r'language', r'(?<=language ) *[a-zA-Z_][a-zA-Z0-9_]+'),
+    
+        RegionRule(r'definition', "'", PlPgSqlGrammar, "'(?!')"),
+        PatternRule(r'bareword', r'[A-Za-z0-9_]+'),
+        PatternRule(r'empty', r'^ *\n$'),
+        PatternRule(r'eol', r'\n'),
+    ]
+
 class SqlGrammar(Grammar):
     rules = [
         PatternRule(r'comment', r'--.*\n$'),
         RegionRule(r'comment', '/\*', Grammar, '\*/'),
-        PatternRule(r'delimiter', r'[();,\.:\$\[\]]'),
+        PatternRule(r'delimiter', r':=|[():;,\.\$\[\]]'),
+    
+        NocaseRegionRule(r'function', r'create function', FunctionGrammar, r';'),
+        NocaseRegionRule(r'function', r'create or replace function', FunctionGrammar, r';'),
+    
         NocasePatternRule(r'attribute', r'(?:check|exists|unique|not null|default|primary key|minvalue|foreign key|references)(?![A-Za-z0-9_])'),
         NocasePatternRule(r'operator', r'(?:case|when|then|else|end|not|and|or|is not|is|in|not in)(?![A-Za-z0-9_])'),
         NocasePatternRule(r'keyword', r'(?:create database|create index|create sequence|create table|create trigger|create view|select|insert|update|delete|drop database|drop index|drop sequence|drop table|drop trigger|drop view|create user|alter user|drop user|drop function|grant|revoke|create function|create or replace function|create or replace view|create language|create operator|create type)(?![A-Za-z0-9_])'),
         NocasePatternRule(r'pseudokeyword', r'(?:returns|language|right join|left join|inner join|outer join|join|where|null|true|false|into|values|as|from|order by|asc|desc|limit|distinct|cascade|using|on)(?![A-Za-z0-9_])'),
         NocasePatternRule(r'type', r'(?:void|row|serial|varchar|float|integer|int|text|timestamptz|timestamp|datetz|date|timetz|time|boolean|bool)(?![A-Za-z0-9_])'),
-        PatternRule(r'function', r'(?:nextval|current_timestamp|current_time|current_date)(?![A-Za-z0-9_])'),
+        PatternRule(r'builtin', r'(?:nextval|current_timestamp|current_time|current_date)(?![A-Za-z0-9_])'),
         RegionRule(r'string', "'", StringGrammar, "'"),
         RegionRule(r'quoted', '"', StringGrammar, '"'),
         PatternRule(r'bareword', r'[A-Za-z0-9_]+'),
@@ -34,6 +76,14 @@ class SqlTabber(tab2.StackTabber):
         token  = self.get_token(y, i)
         if token.name == 'delimiter' and token.string == ';':
             self._opt_pop('cont')
+        elif token.name == 'keyword':
+            if token.string == 'declare':
+                self._opt_append('declare', currlvl + 4)
+            elif token.string == 'begin':
+                currlvl -= 4
+            elif token.string == 'end':
+                self._opt_pop('declare')
+                currlvl = self.get_curr_level()
 
         if self.is_rightmost_token(y, i):
             if not self._empty() and token.name == 'continuation':
@@ -61,7 +111,7 @@ class Sql(mode2.Fundamental):
             'keyword':        color.build('cyan', 'default'),
             'pseudokeyword':  color.build('cyan', 'default'),
             'type':           color.build('green', 'default'),
-            'function':       color.build('yellow', 'default'),
+            'builtin':        color.build('yellow', 'default'),
             'quoted':         color.build('yellow', 'default'),
             'string.start':   color.build('green', 'default'),
             'string.null':    color.build('green', 'default'),
@@ -69,6 +119,17 @@ class Sql(mode2.Fundamental):
             'string.octal':   color.build('magenta', 'default'),
             'string.end':     color.build('green', 'default'),
             'bareword':       color.build('default', 'default'),
+
+            'function.start':    color.build('cyan', 'default'),
+            'function.null':     color.build('default', 'default'),
+            'function.name':     color.build('magenta', 'default'),
+            'function.language': color.build('magenta', 'default'),
+            'function.end':      color.build('default', 'default'),
+
+            'function.definition.start':    color.build('magenta', 'default'),
+            'function.definition.bareword': color.build('magenta', 'default'),
+            'function.definition.null':     color.build('magenta', 'default'),
+            'function.definition.end':      color.build('magenta', 'default'),
         }
     def name(self):
         return "Sql"