NFA macro + better NFA matcher (#58)

* NFA macro
* better NFA matcher
* benchmarks

.gitignore

@@ -1,6 +1,7 @@
 nimcache/
-src/regex
 src/regex.js
 tests/tests
 tests/tests.js
 docs/ugh
+bin/*
+bench/bench

.travis.yml

@@ -1,7 +1,6 @@
 services:
   - docker
 env:
-  - NIM=0.19.0
   - NIM=0.19.6
   - NIM=0.20.0
   - NIM=0.20.2

CHANGELOG.md

@@ -1,3 +1,8 @@
+v0.14
+==================
+
+* Drop Nim 0.19.0 support (0.19.6 is supported)
+
 v0.13.1
 ==================
 

bench/README.md

@@ -0,0 +1,16 @@
+## Benchmarks
+
+Run first:
+
+```
+nimble install nimbench
+nimble develop
+```
+
+Run benchmarks:
+
+```
+nim c -r -d:release bench/bench.nim
+```
+
+> Try -d:danger as well, but release is what most users will set

bench/bench.nim

@@ -0,0 +1,112 @@
+import nimbench
+import unicode
+from re import nil
+from regex import nil
+
+var text = ""
+for _ in 0 .. 100000:
+  text.add("a")
+text.add("sol")
+for _ in 0 .. 100000:
+  text.add("b")
+#text.add("ฅ")
+
+var pattern2 = re.re"^\w*sol\w*$"
+
+bench(re_sol, m):
+  var d: bool
+  for i in 0 ..< m:
+    d = re.match(text, pattern2)
+  doNotOptimizeAway(d)
+
+const pattern4 = regex.re(r"\w*sol\w*") #, {regex.RegexFlag.reAscii})
+
+benchRelative(regex_sol, m):
+  var m2: regex.RegexMatch
+  for i in 0 ..< m:
+    discard regex.match(text, pattern4, m2)
+  doNotOptimizeAway(m2)
+
+var dummyTextNums = """650-253-0001"""
+
+var pattern_nums = re.re"^[0-9]+-[0-9]+-[0-9]+$"
+
+bench(re_nums, m):
+  var d: bool
+  for i in 0 ..< m:
+    d = re.match(dummyTextNums, pattern_nums)
+  doNotOptimizeAway(d)
+
+const n_pattern_nums = regex.re"[0-9]+-[0-9]+-[0-9]+"
+
+benchRelative(regex_nums, m):
+  var m2: regex.RegexMatch
+  for i in 0 ..< m:
+    discard regex.match(dummyTextNums, n_pattern_nums, m2)
+  doNotOptimizeAway(m2)
+
+var pattern_nums2 = re.re"^[0-9]+..*$"
+
+bench(re_nums2, m):
+  var d: bool
+  for i in 0 ..< m:
+    d = re.match(dummyTextNums, pattern_nums2)
+  doNotOptimizeAway(d)
+
+const n_pattern_nums2 = regex.re"[0-9]+..*"
+
+benchRelative(regex_nums2, m):
+  var m3: regex.RegexMatch
+  for i in 0 ..< m:
+    discard regex.match(dummyTextNums, n_pattern_nums2, m3)
+  doNotOptimizeAway(m3)
+
+var lits_find_re = re.re"do|re|mi|fa|sol"
+
+bench(re_lits_find, m):
+  var d: int
+  for i in 0 ..< m:
+    d = re.find(text, lits_find_re)
+  doNotOptimizeAway(d)
+
+const lits_find = regex.re"do|re|mi|fa|sol"
+
+benchRelative(regex_lits_find, m):
+  var m2: regex.RegexMatch
+  for i in 0 ..< m:
+    discard regex.find(text, lits_find, m2)
+  doNotOptimizeAway(m2)
+
+const bench_text = staticRead("input-text.txt")
+
+var email_find_all_re = re.re"[\w\.+-]+@[\w\.-]+\.[\w\.-]+"
+
+bench(re_email_find_all, m):
+  var d = 0
+  for i in 0 ..< m:
+    for _ in re.findAll(bench_text, email_find_all_re):
+      d += 1
+  doAssert d == 92
+  doNotOptimizeAway(d)
+
+const email_find_all = regex.re"(?-u)[\w\.+-]+@[\w\.-]+\.[\w\.-]+"
+
+benchRelative(email_find_all, m):
+  var d = 0
+  for i in 0 ..< m:
+    for _ in regex.findAll(bench_text, email_find_all):
+      d += 1
+  doAssert d == 92
+  doNotOptimizeAway(d)
+
+when false:
+  bench(runes, m):
+    for i in text.runes:
+      memoryClobber()
+
+bench(dummy, m):
+  for i in 0 ..< m:
+    memoryClobber()
+
+when isMainModule:
+  runBenchmarks()