Issue 45: Implement the regular expression library

Add initial support for the regular expression library. This implements findall(), search(), and match() The most complete implementation is findall() which is only limited by the incompatibiliies between the JS regexp and Python regexp. The bggest problem is the different interpretation of () in the JS regexp. Patterns like: "From .*@(\\S*)" Return the whole match instead of the piece inside the parenthesis. search() and match() are barely there. They properly return None if there is no match and return a MatchObject when there is a hit. but the returned MatchObject is useless (i.e. no data and no methods).
Duncanr-glitch · Jan 27, 2013 · 9aa25b7 · 9aa25b7
1 parent 45d2ad8
commit 9aa25b7
Show file tree

Hide file tree

Showing 5 changed files with 442 additions and 0 deletions.
diff --git a/src/lib/re/__init__.js b/src/lib/re/__init__.js
@@ -0,0 +1,57 @@
+var $builtinmodule = function(name)
+{
+    var mod = {};
+
+    var matchobj = function($gbl, $loc) {
+        $loc.__init__ = new Sk.builtin.func(function(self,thematch) {
+            self.thematch = thematch;
+        });
+    }
+
+    mod.MatchObject = Sk.misceval.buildClass(mod, matchobj, 'MatchObject', []);
+
+    // Internal function to return a Python list of strings 
+    // From a JS regular expression string
+    mod._findre = function(res, string) {
+        var re = eval(res);
+        var matches = string.v.match(re);
+        retval = new Sk.builtin.list();
+        if ( matches == null ) return retval;
+        for (var i = 0; i < matches.length; ++i) {
+            var sitem = new Sk.builtin.str(matches[i]);
+            retval.v.push(sitem);
+        }
+        return retval;
+    }
+
+    mod.findall = new Sk.builtin.func(function(pattern, string, flags) {
+        var res = "/"+pattern.v.replace("/","\\/")+"/g";
+        var re = eval(res);
+        var matches = string.v.match(re);
+        retval = new Sk.builtin.list();
+        if ( matches == null ) return retval;
+        for (var i = 0; i < matches.length; ++i) {
+            var sitem = new Sk.builtin.str(matches[i]);
+            retval.v.push(sitem);
+        }
+        return retval;
+    });
+
+    mod.search = new Sk.builtin.func(function(pattern, string, flags) {
+        var res = "/"+pattern.v.replace("/","\\/")+"/";
+        lst = mod._findre(res,string);
+        if ( lst.v.length < 1 ) return null;
+        var mob = Sk.misceval.callsim(mod.MatchObject, lst);
+        return mob;
+    });
+
+    mod.match = new Sk.builtin.func(function(pattern, string, flags) {
+        var res = "/^"+pattern.v.replace("/","\\/")+"/";
+        lst = mod._findre(res,string);
+        if ( lst.v.length < 1 ) return null;
+        var mob = Sk.misceval.callsim(mod.MatchObject, lst);
+        return mob;
+    });
+
+    return mod;
+}
diff --git a/test/run/t339.py b/test/run/t339.py
@@ -0,0 +1,51 @@
+import re
+
+val = re.findall("From","dlkjdsljkdlkdsjlk")
+print val
+if len(val) == 0 : print "Correct 1"
+else : print "InCorrect 1"
+
+val = re.findall("From","dlkjd From kdsjlk")
+print val
+if len(val) == 1 : print "Correct 2"
+else : print "InCorrect 2"
+
+val = re.findall("From","From dlkjd From kdsjlk")
+print val
+if len(val) == 2 : print "Correct 3"
+else : print "InCorrect 3"
+
+val = re.findall("[0-9]+/[0-9]+","1/2 1/3 3/4 1/8 fred 10/0")
+print val
+if len(val) == 5 : print "Correct 4"
+else : print "InCorrect 4"
+
+# Won't work because JS match does not deal with ()
+# print re.findall("From .*@(\\S*)","From csev@umich.edu Sat 09:25:14")
+
+# These return either None or a trivial MatchObject with no methods
+
+val = re.search("From","dlkjdsljkdlkdsjlk")
+if val is None: print "Correct 5"
+else : print "InCorrect 5",val
+
+val = re.search("From","dlkjd From kdsjlk")
+if val is not None: print "Correct 6"
+else : print "InCorrect 6",val
+
+val =  re.search("From","From dlkjd From kdsjlk")
+if val is not None: print "Correct 7"
+else : print "InCorrect 7",val
+
+val = re.match("From","dlkjdsljkdlkdsjlk")
+if val is None: print "Correct 8"
+else : print "InCorrect 8",val
+
+val = re.match("From","dlkjd From kdsjlk")
+if val is None: print "Correct 9"
+else : print "InCorrect 9",val
+
+val = re.match("From","From dlkjd From kdsjlk")
+if val is not None: print "Correct 10"
+else : print "InCorrect 10",val
+
diff --git a/test/run/t339.py.real b/test/run/t339.py.real
@@ -0,0 +1,14 @@
+[]
+Correct 1
+['From']
+Correct 2
+['From', 'From']
+Correct 3
+['1/2', '1/3', '3/4', '1/8', '10/0']
+Correct 4
+Correct 5
+Correct 6
+Correct 7
+Correct 8
+Correct 9
+Correct 10
diff --git a/test/run/t339.py.symtab b/test/run/t339.py.symtab
@@ -0,0 +1,54 @@
+Sym_type: module
+Sym_name: top
+Sym_lineno: 0
+Sym_nested: False
+Sym_haschildren: False
+-- Identifiers --
+name: None
+  is_referenced: True
+  is_imported: False
+  is_parameter: False
+  is_global: True
+  is_declared_global: False
+  is_local: False
+  is_free: False
+  is_assigned: False
+  is_namespace: False
+  namespaces: [
+  ]
+name: len
+  is_referenced: True
+  is_imported: False
+  is_parameter: False
+  is_global: True
+  is_declared_global: False
+  is_local: False
+  is_free: False
+  is_assigned: False
+  is_namespace: False
+  namespaces: [
+  ]
+name: re
+  is_referenced: True
+  is_imported: True
+  is_parameter: False
+  is_global: False
+  is_declared_global: False
+  is_local: True
+  is_free: False
+  is_assigned: False
+  is_namespace: False
+  namespaces: [
+  ]
+name: val
+  is_referenced: True
+  is_imported: False
+  is_parameter: False
+  is_global: False
+  is_declared_global: False
+  is_local: True
+  is_free: False
+  is_assigned: True
+  is_namespace: False
+  namespaces: [
+  ]