Cleaned up coerce behavior

- followed the convention for including a regex
- ignore digit sequences longer than 16 characters
- explain expected behavior more carefully in README
- update tests to reflect expected behavior

Author: joeledwards

README.md

@@ -382,3 +382,7 @@ consumes all remaining characters which satisfy at least a partial semver
 Longer versions are simply truncated (`4.6.3.9.2-alpha2` becomes `4.6.3`).
 All surrounding text is simply ignored (`v3.4 replaces v3.3.1` becomes `3.4.0`).
 Only text which lacks digits will fail coercion (`version one` is not valid).
+The maximum  length for any semver component considered for coercion is 16 characters;
+longer components will be ignored (`10000000000000000.4.7.4` becomes `4.7.4`).
+The maximum value for any semver component is `Integer.MAX_SAFE_INTEGER || (2**53 - 1)`;
+higher value components are invalid (`9999999999999999.4.7.4` is likely invalid).

semver.js

@@ -21,6 +21,9 @@ exports.SEMVER_SPEC_VERSION = '2.0.0';
 var MAX_LENGTH = 256;
 var MAX_SAFE_INTEGER = Number.MAX_SAFE_INTEGER || 9007199254740991;
 
+// Max safe segment length for coercion.
+var MAX_SAFE_COMPONENT_LENGTH = 16;
+
 // The actual regexps go on exports.re
 var re = exports.re = [];
 var src = exports.src = [];
@@ -156,6 +159,15 @@ src[XRANGE] = '^' + src[GTLT] + '\\s*' + src[XRANGEPLAIN] + '$';
 var XRANGELOOSE = R++;
 src[XRANGELOOSE] = '^' + src[GTLT] + '\\s*' + src[XRANGEPLAINLOOSE] + '$';
 
+// Coercion.
+// Extract anything that could conceivably be a part of a valid semver
+var COERCE = R++;
+src[COERCE] = '(?:^|[^\\d])' +
+              '(\\d{1,' + MAX_SAFE_COMPONENT_LENGTH + '})' +
+              '(?:\\.(\\d{1,' + MAX_SAFE_COMPONENT_LENGTH + '}))?' +
+              '(?:\\.(\\d{1,' + MAX_SAFE_COMPONENT_LENGTH + '}))?' +
+              '(?:$|[^\\d])';
+
 // Tilde ranges.
 // Meaning is "reasonably at or greater than"
 var LONETILDE = R++;
@@ -1298,20 +1310,15 @@ function intersects(r1, r2, loose) {
 exports.coerce = coerce;
 function coerce(version) {
   if (version instanceof SemVer)
-    return version
+    return version;
 
   if (typeof version !== 'string')
-    return null
-
-  if (version.length > MAX_LENGTH)
     return null;
 
-  var match = version.match(/([vV]?\d+(?:[.]\d+){0,2})/)
-  if (match == null)
-    return null
+  var match = version.match(re[COERCE]);
 
-  var parts = match[0].split(/[.]/)
-  var semver = ['0', '0', '0'].map(function (ph, idx) { return parts[idx] || ph } ).join('.')
+  if (match == null)
+    return null;
 
-  return parse(semver)
+  return parse((match[1] || '0') + '.' + (match[2] || '0') + '.' + (match[3] || '0')); 
 }

test/coerce.js

@@ -2,11 +2,15 @@ var tap = require('tap');
 var test = tap.test;
 var semver = require('../semver.js');
 var coerce = semver.coerce;
+var valid = semver.valid;
 
-test('\ncoerce tests', function(t) {
-  var tooLong = '123' + '.1'.repeat(127);
-  var justRight = '12' + '.1'.repeat(127);
+function r(text) {
+  return function (count) {
+    return text.repeat(count);
+  };
+}
 
+test('\ncoerce tests', function(t) {
   // Expected to be null (cannot be coerced).
   [
     null,
@@ -15,13 +19,22 @@ test('\ncoerce tests', function(t) {
     '',
     '.',
     'version one',
-    tooLong
+    r('9')(16),
+    r('1')(17),
+    'a' + r('9')(16),
+    'a' + r('1')(17),
+    r('9')(16) + 'a',
+    r('1')(17) + 'a',
+    r('9')(16) + '.4.7.4',
+    r('9')(16) + '.' + r('2')(16) + '.' + r('3')(16),
+    r('1')(16) + '.' + r('9')(16) + '.' + r('3')(16),
+    r('1')(16) + '.' + r('2')(16) + '.' + r('9')(16)
   ].forEach(function (input) {
     var msg = 'coerce(' + input + ') should be null'
     t.same(coerce(input), null, msg)
   });
 
-  // Expected to be the same.
+  // Expected to be the valid.
   [
     [semver.parse('1.2.3'), '1.2.3'],
     ['.1', '1.0.0'],
@@ -49,6 +62,8 @@ test('\ncoerce tests', function(t) {
     ['v1.2', '1.2.0'],
     ['v1.2.3', '1.2.3'],
     ['v1.2.3.4', '1.2.3'],
+    [' 1', '1.0.0'],
+    ['1 ', '1.0.0'],
     ['1 0', '1.0.0'],
     ['1 1', '1.0.0'],
     ['1.1 1', '1.1.0'],
@@ -65,13 +80,39 @@ test('\ncoerce tests', function(t) {
     ['v2', '2.0.0'],
     ['v3.4 replaces v3.3.1', '3.4.0'],
     ['4.6.3.9.2-alpha2', '4.6.3'],
-    [justRight, '12.1.1']
+    [r('1')(17) + '.2', '2.0.0'],
+    [r('1')(17) + '.2.3', '2.3.0'],
+    ['1.' + r('2')(17) + '.3', '1.0.0'],
+    ['1.2.' + r('3')(17), '1.2.0'],
+    [r('1')(17) + '.2.3.4', '2.3.4'],
+    ['1.' + r('2')(17) + '.3.4', '1.0.0'],
+    ['1.2.' + r('3')(17) + '.4', '1.2.0'],
+    [r('1')(17) + '.' + r('2')(16) + '.' + r('3')(16),
+     r('2')(16) + '.' + r('3')(16) + '.0'],
+    [r('1')(16) + '.' + r('2')(17) + '.' + r('3')(16),
+     r('1')(16) + '.0.0'],
+    [r('1')(16) + '.' + r('2')(16) + '.' + r('3')(17),
+     r('1')(16) + '.' + r('2')(16) + '.0'],
+    ['11' + r('.1')(126), '11.1.1'],
+    [r('1')(16), r('1')(16) + '.0.0'],
+    ['a' + r('1')(16), r('1')(16) + '.0.0'],
+    [r('1')(16) + '.2.3.4', r('1')(16) + '.2.3'],
+    ['1.' + r('2')(16) + '.3.4', '1.' + r('2')(16) + '.3'],
+    ['1.2.' + r('3')(16) + '.4', '1.2.' + r('3')(16)],
+    [r('1')(16) + '.' + r('2')(16) + '.' + r('3')(16),
+     r('1')(16) + '.' + r('2')(16) + '.' + r('3')(16)],
+    ['1.2.3.' + r('4')(252) + '.5', '1.2.3'],
+    ['1.2.3.' + r('4')(1024), '1.2.3'],
+    [r('1')(17) + '.4.7.4', '4.7.4']
   ].forEach(function (tuple) {
     var input = tuple[0]
     var expected = tuple[1]
     var msg = 'coerce(' + input + ') should become ' + expected
     t.same((coerce(input) || {}).version, expected, msg)
   });
 
+  t.same(valid(coerce('42.6.7.9.3-alpha')), '42.6.7')
+  t.same(valid(coerce('v2')), '2.0.0')
+
   t.done();
-});
+});