RegEx Homework

textminer/extractor.py

@@ -0,0 +1,9 @@
+import re
+
+def phone_numbers(string):
+    number = re.findall(r'(\(?\d{3}\)?[\-\.]?\s*?\d{3}[\-\.]?\d{4})', string)
+    return number
+
+def emails(string):
+    emails = re.findall(r'(\w+?\.?\w+@\w+\.\w{2,3})', string)
+    return emails

textminer/separator.py

@@ -0,0 +1,50 @@
+import re
+
+def words(string):
+    result = re.findall(r'(\b[^\s]+[a-z]+\b)', string)
+    if result == []:
+        return None
+    else:
+        return result
+
+def phone_number(string):
+    number = re.match(r'\(?(\d{3})\)?[\-\.]?\s*?(\d{3})[\-\.]?(\d{4})', string)
+    if number == None:
+        return None
+    else:
+        return {"area_code": number.group(1),
+                "number": number.group(2)+"-"+number.group(3)}
+
+def money(string):
+    money = re.match(r'^(\$)(\d+(,?\d{3,})*\.?(\d{2})?)$', string)
+    if money == None:
+        return None
+    else:
+        money2 = money.group(2).replace(',','')
+        return {"currency": money.group(1),
+                "amount": float(money2)}
+
+def zipcode(string):
+    zip = re.match(r'^(\d{5})-?(\d{4})?$', string)
+    if zip == None:
+        return None
+    else:
+        return {"zip": zip.group(1),
+                "plus4": zip.group(2)}
+
+def date(string):
+    the_date = re.match(r'^(\d{1,4})[-|/](\d{1,2})[-|/](\d{1,4})$', string)
+    if the_date == None:
+        return None
+    else:
+        if int(the_date.group(1)) > 1000:
+            year = int(the_date.group(1))
+            month = int(the_date.group(2))
+            day = int(the_date.group(3))
+        elif int(the_date.group(3)) > 1000:
+            year = int(the_date.group(3))
+            month = int(the_date.group(1))
+            day = int(the_date.group(2))
+        return {"month": month,
+                "day": day,
+                "year": year}

textminer/tests/test_extractor.py

@@ -4,7 +4,7 @@
 
 import textminer.extractor as x
 
-@xfail
+#@xfail
 def test_phone_numbers():
     text = """Dear Mr. Davis,
 
@@ -24,7 +24,7 @@ def test_phone_numbers():
 
 ## HARD MODE BEGINS
 
-@xfail
+#@xfail
 def test_emails():
     text = """Veggies es bonus vobis, proinde vos postulo essum magis kohlrabi
     welsh onion daikon [email protected] tatsoi tomatillo azuki bean garlic.