From 61f9e3534fdaaf9c7d76e04d161ddaed96c672c3 Mon Sep 17 00:00:00 2001
From: Dan Wheeler <dan@dropbox.com>
Date: Sun, 13 Sep 2015 23:55:39 -0700
Subject: [PATCH] add regex precedence to cut down on redundant regex matches +
 regex matching tests

---
 src/matching.coffee       | 36 ++++++++++++++++++++++++++++++++----
 test/test-matching.coffee | 23 +++++++++++++++++++++++
 2 files changed, 55 insertions(+), 4 deletions(-)

diff --git a/src/matching.coffee b/src/matching.coffee
index 171784db..2e0eab4b 100644
--- a/src/matching.coffee
+++ b/src/matching.coffee
@@ -43,14 +43,23 @@ L33T_TABLE =
   z: ['2']
 
 REGEXEN =
-  alpha_lower:  /[A-Z]{2,}/g
-  alpha_upper:  /[a-z]{2,}/g
-  alpha:        /[a-zA-Z]{2,}/g
   alphanumeric: /[a-zA-Z0-9]{2,}/g
+  alpha:        /[a-zA-Z]{2,}/g
+  alpha_lower:  /[a-z]{2,}/g
+  alpha_upper:  /[A-Z]{2,}/g
   digits:       /\d{2,}/g
   symbols:      /[\W_]{2,}/g # includes non-latin unicode chars
   recent_year:  /19\d\d|200\d|201\d/g
 
+REGEX_PRECEDENCE =
+  alphanumeric: 0
+  alpha:        1
+  alpha_lower:  2
+  alpha_upper:  2
+  digits:       2
+  symbols:      2
+  recent_year:  3
+
 DATE_MAX_YEAR = 2050
 DATE_MIN_YEAR = 1000
 DATE_SPLITS =
@@ -402,7 +411,26 @@ matching =
           j: rx_match.index + rx_match[0].length - 1
           regex_name: name
           regex_match: rx_match
-    @sorted matches
+    # currently, match list includes a bunch of redundancies:
+    # ex for every alpha_lower match, also an alpha and alphanumeric match of the same [i,j].
+    # ex for every recent_year match, also an alphanumeric match and digits match.
+    # use precedence to filter these redundancies out.
+    precedence_map = {} # maps from 'i-j' to current highest precedence
+    get_key = (match) -> "#{match.i}-#{match.j}"
+    for match in matches
+      key = get_key match
+      precedence = REGEX_PRECEDENCE[match.regex_name]
+      if key of precedence_map
+        highest_precedence = precedence_map[key]
+        continue if highest_precedence >= precedence
+      precedence_map[key] = precedence
+    filtered_matches = []
+    for match in matches
+      key = get_key match
+      precedence = REGEX_PRECEDENCE[match.regex_name]
+      if precedence_map[key] == precedence
+        filtered_matches.push match
+    @sorted filtered_matches
 
   #-------------------------------------------------------------------------------
   # date matching ----------------------------------------------------------------
diff --git a/test/test-matching.coffee b/test/test-matching.coffee
index f05c6b93..3a75e4c6 100644
--- a/test/test-matching.coffee
+++ b/test/test-matching.coffee
@@ -427,6 +427,29 @@ test 'repeat matching', (t) ->
   t.end()
 
 
+test 'regex matching', (t) ->
+  for [pattern, name] in [
+    ['aaa', 'alpha_lower']
+    ['a7c8D9', 'alphanumeric']
+    ['aAaA', 'alpha']
+    ['1922', 'recent_year']
+    ['&@*#', 'symbols']
+    ['94113', 'digits']
+    ]
+    matches = matching.regex_match pattern
+    msg = "matches #{pattern} as a #{name} pattern"
+    check_matches msg, t, matches, 'regex', [pattern], [[0, pattern.length - 1]],
+      regex_name: [name]
+
+  password = 'a7c8D9vvv2015'
+  matches = matching.regex_match password
+  ijs = [[0, 12], [6, 8], [9, 12]]
+  msg = "matches multiple overlapping regex patterns"
+  check_matches msg, t, matches, 'regex', ['a7c8D9vvv2015', 'vvv', '2015'], ijs,
+    regex_name: ['alphanumeric', 'alpha_lower', 'recent_year']
+  t.end()
+
+
 test 'date matching', (t) ->
   for sep in ['', ' ', '-', '/', '\\', '_', '.']
     password = "13#{sep}2#{sep}1921"