[GR-19220] Support Time.new with string timestamp argument and error …

…when invalid (#3702)

PullRequest: truffleruby/4389

CHANGELOG.md

@@ -17,6 +17,7 @@ Compatibility:
 * Set `RbConfig::CONFIG['archincludedir']` (#3396, @andrykonchin).
 * Support the index/length arguments for the string argument to `String#bytesplice` added in 3.3 (#3656, @rwstauner).
 * Implement `rb_str_strlen()` (#3697, @Th3-M4jor).
+* Support `Time.new` with String argument and error when invalid (#3693, @rwstauner).
 
 Performance:
 

spec/ruby/core/time/new_spec.rb

@@ -485,6 +485,8 @@ def zone.local_to_utc(t)
         Time.new("2020-12-25 00:56:17 +0900").should == t
         Time.new("2020-12-25 00:57:47 +090130").should == t
         Time.new("2020-12-25T00:56:17+09:00").should == t
+
+        Time.new("2020-12-25T00:56:17.123456+09:00").should == Time.utc(2020, 12, 24, 15, 56, 17, 123456)
       end
 
       it "accepts precision keyword argument and truncates specified digits of sub-second part" do
@@ -511,6 +513,16 @@ def zone.local_to_utc(t)
         Time.new("2021-12-25 00:00:00", in: "-01:00").to_s.should == "2021-12-25 00:00:00 -0100"
       end
 
+      it "returns Time of Jan 1 for string with just year" do
+        Time.new("2021").should == Time.new(2021, 1, 1)
+        Time.new("2021").zone.should == Time.new(2021, 1, 1).zone
+        Time.new("2021").utc_offset.should == Time.new(2021, 1, 1).utc_offset
+      end
+
+      it "returns Time of Jan 1 for string with just year in timezone specified with in keyword argument" do
+        Time.new("2021", in: "+17:00").to_s.should == "2021-01-01 00:00:00 +1700"
+      end
+
       it "converts precision keyword argument into Integer if is not nil" do
         obj = Object.new
         def obj.to_int; 3; end
@@ -539,108 +551,128 @@ def obj.to_int; 3; end
       it "raises ArgumentError if part of time string is missing" do
         -> {
           Time.new("2020-12-25 00:56 +09:00")
-        }.should raise_error(ArgumentError, "missing sec part: 00:56 ")
+        }.should raise_error(ArgumentError, /missing sec part: 00:56 |can't parse:/)
 
         -> {
           Time.new("2020-12-25 00 +09:00")
-        }.should raise_error(ArgumentError, "missing min part: 00 ")
+        }.should raise_error(ArgumentError, /missing min part: 00 |can't parse:/)
+      end
+
+      ruby_version_is "3.2.3" do
+        it "raises ArgumentError if the time part is missing" do
+          -> {
+            Time.new("2020-12-25")
+          }.should raise_error(ArgumentError, /no time information|can't parse:/)
+        end
       end
 
       it "raises ArgumentError if subsecond is missing after dot" do
         -> {
           Time.new("2020-12-25 00:56:17. +0900")
-        }.should raise_error(ArgumentError, "subsecond expected after dot: 00:56:17. ")
+        }.should raise_error(ArgumentError, /subsecond expected after dot: 00:56:17. |can't parse:/)
       end
 
       it "raises ArgumentError if String argument is not in the supported format" do
         -> {
           Time.new("021-12-25 00:00:00.123456 +09:00")
-        }.should raise_error(ArgumentError, "year must be 4 or more digits: 021")
+        }.should raise_error(ArgumentError, /year must be 4 or more digits: 021|can't parse:/)
 
         -> {
           Time.new("2020-012-25 00:56:17 +0900")
-        }.should raise_error(ArgumentError, /\Atwo digits mon is expected after [`']-': -012-25 00:\z/)
+        }.should raise_error(ArgumentError, /\Atwo digits mon is expected after [`']-': -012-25 00:\z|can't parse:/)
 
         -> {
           Time.new("2020-2-25 00:56:17 +0900")
-        }.should raise_error(ArgumentError, /\Atwo digits mon is expected after [`']-': -2-25 00:56\z/)
+        }.should raise_error(ArgumentError, /\Atwo digits mon is expected after [`']-': -2-25 00:56\z|can't parse:/)
 
         -> {
           Time.new("2020-12-215 00:56:17 +0900")
-        }.should raise_error(ArgumentError, /\Atwo digits mday is expected after [`']-': -215 00:56:\z/)
+        }.should raise_error(ArgumentError, /\Atwo digits mday is expected after [`']-': -215 00:56:\z|can't parse:/)
 
         -> {
           Time.new("2020-12-25 000:56:17 +0900")
-        }.should raise_error(ArgumentError, "two digits hour is expected:  000:56:17 ")
+        }.should raise_error(ArgumentError, /two digits hour is expected:  000:56:17 |can't parse:/)
 
         -> {
           Time.new("2020-12-25 0:56:17 +0900")
-        }.should raise_error(ArgumentError, "two digits hour is expected:  0:56:17 +0")
+        }.should raise_error(ArgumentError, /two digits hour is expected:  0:56:17 \+0|can't parse:/)
 
         -> {
           Time.new("2020-12-25 00:516:17 +0900")
-        }.should raise_error(ArgumentError, /\Atwo digits min is expected after [`']:': :516:17 \+09\z/)
+        }.should raise_error(ArgumentError, /\Atwo digits min is expected after [`']:': :516:17 \+09\z|can't parse:/)
 
         -> {
           Time.new("2020-12-25 00:6:17 +0900")
-        }.should raise_error(ArgumentError, /\Atwo digits min is expected after [`']:': :6:17 \+0900\z/)
+        }.should raise_error(ArgumentError, /\Atwo digits min is expected after [`']:': :6:17 \+0900\z|can't parse:/)
 
         -> {
           Time.new("2020-12-25 00:56:137 +0900")
-        }.should raise_error(ArgumentError, /\Atwo digits sec is expected after [`']:': :137 \+0900\z/)
+        }.should raise_error(ArgumentError, /\Atwo digits sec is expected after [`']:': :137 \+0900\z|can't parse:/)
 
         -> {
           Time.new("2020-12-25 00:56:7 +0900")
-        }.should raise_error(ArgumentError, /\Atwo digits sec is expected after [`']:': :7 \+0900\z/)
+        }.should raise_error(ArgumentError, /\Atwo digits sec is expected after [`']:': :7 \+0900\z|can't parse:/)
 
         -> {
           Time.new("2020-12-25 00:56. +0900")
-        }.should raise_error(ArgumentError, "fraction min is not supported: 00:56.")
+        }.should raise_error(ArgumentError, /fraction min is not supported: 00:56\.|can't parse:/)
 
         -> {
           Time.new("2020-12-25 00. +0900")
-        }.should raise_error(ArgumentError, "fraction hour is not supported: 00.")
+        }.should raise_error(ArgumentError, /fraction hour is not supported: 00\.|can't parse:/)
       end
 
       it "raises ArgumentError if date/time parts values are not valid" do
         -> {
           Time.new("2020-13-25 00:56:17 +09:00")
-        }.should raise_error(ArgumentError, "mon out of range")
+        }.should raise_error(ArgumentError, /(mon|argument) out of range/)
 
         -> {
           Time.new("2020-12-32 00:56:17 +09:00")
-        }.should raise_error(ArgumentError, "mday out of range")
+        }.should raise_error(ArgumentError, /(mday|argument) out of range/)
 
         -> {
           Time.new("2020-12-25 25:56:17 +09:00")
-        }.should raise_error(ArgumentError, "hour out of range")
+        }.should raise_error(ArgumentError, /(hour|argument) out of range/)
 
         -> {
           Time.new("2020-12-25 00:61:17 +09:00")
-        }.should raise_error(ArgumentError, "min out of range")
+        }.should raise_error(ArgumentError, /(min|argument) out of range/)
 
         -> {
           Time.new("2020-12-25 00:56:61 +09:00")
-        }.should raise_error(ArgumentError, "sec out of range")
+        }.should raise_error(ArgumentError, /(sec|argument) out of range/)
 
         -> {
           Time.new("2020-12-25 00:56:17 +23:59:60")
-        }.should raise_error(ArgumentError, "utc_offset out of range")
+        }.should raise_error(ArgumentError, /(utc_offset|argument) out of range/)
 
         -> {
           Time.new("2020-12-25 00:56:17 +24:00")
-        }.should raise_error(ArgumentError, "utc_offset out of range")
+        }.should raise_error(ArgumentError, /(utc_offset|argument) out of range/)
 
         -> {
           Time.new("2020-12-25 00:56:17 +23:61")
-        }.should raise_error(ArgumentError, '"+HH:MM", "-HH:MM", "UTC" or "A".."I","K".."Z" expected for utc_offset: +23:61')
+        }.should raise_error(ArgumentError, /#{Regexp.escape('"+HH:MM", "-HH:MM", "UTC" or "A".."I","K".."Z" expected for utc_offset: +23:61')}|can't parse:/)
       end
 
       it "raises ArgumentError if string has not ascii-compatible encoding" do
         -> {
           Time.new("2021-11-31 00:00:60 +09:00".encode("utf-32le"))
         }.should raise_error(ArgumentError, "time string should have ASCII compatible encoding")
       end
+
+      it "raises ArgumentError if string doesn't start with year" do
+        -> {
+          Time.new("a\nb")
+        }.should raise_error(ArgumentError, "can't parse: \"a\\nb\"")
+      end
+
+      it "raises ArgumentError if string has extra characters after offset" do
+        -> {
+          Time.new("2021-11-31 00:00:59 +09:00 abc")
+        }.should raise_error(ArgumentError, /can't parse.+ abc/)
+      end
     end
   end
 end

spec/tags/core/time/new_tags.txt

@@ -20,15 +20,6 @@ fails:Time.new with a timezone argument subject's class implements .find_timezon
 fails:Time.new with a timezone argument subject's class implements .find_timezone method calls .find_timezone to build a time object if passed zone name as a timezone argument
 fails:Time.new with a timezone argument subject's class implements .find_timezone method does not call .find_timezone if passed any not string/numeric/timezone timezone argument
 fails:Time.new with a timezone argument :in keyword argument could be a timezone object
-fails:Time.new with a timezone argument Time.new with a String argument parses an ISO-8601 like format
 fails:Time.new with a timezone argument Time.new with a String argument accepts precision keyword argument and truncates specified digits of sub-second part
-fails:Time.new with a timezone argument Time.new with a String argument returns Time in timezone specified in the String argument
-fails:Time.new with a timezone argument Time.new with a String argument returns Time in timezone specified in the String argument even if the in keyword argument provided
-fails:Time.new with a timezone argument Time.new with a String argument returns Time in timezone specified with in keyword argument if timezone isn't provided in the String argument
 fails:Time.new with a timezone argument Time.new with a String argument converts precision keyword argument into Integer if is not nil
 fails:Time.new with a timezone argument Time.new with a String argument raise TypeError is can't convert precision keyword argument into Integer
-fails:Time.new with a timezone argument Time.new with a String argument raises ArgumentError if part of time string is missing
-fails:Time.new with a timezone argument Time.new with a String argument raises ArgumentError if subsecond is missing after dot
-fails:Time.new with a timezone argument Time.new with a String argument raises ArgumentError if String argument is not in the supported format
-fails:Time.new with a timezone argument Time.new with a String argument raises ArgumentError if date/time parts values are not valid
-fails:Time.new with a timezone argument Time.new with a String argument raises ArgumentError if string has not ascii-compatible encoding

src/main/ruby/truffleruby/core/time.rb

@@ -403,15 +403,19 @@ def at(sec, sub_sec = undefined, unit = undefined, **kwargs)
       time
     end
 
-    def new(year = undefined, month = nil, day = nil, hour = nil, minute = nil, second = nil, utc_offset = nil, **options)
+    def new(year = undefined, month = undefined, day = nil, hour = nil, minute = nil, second = nil, utc_offset = nil, **options)
       if utc_offset && options[:in]
         raise ArgumentError, 'timezone argument given as positional and keyword arguments'
       end
 
       utc_offset ||= options[:in]
+      month_undefined = Primitive.undefined?(month)
+      month = nil if month_undefined
 
       if Primitive.undefined?(year)
         utc_offset ? self.now.getlocal(utc_offset) : self.now
+      elsif Primitive.is_a?(year, String) && month_undefined
+        Truffle::TimeOperations.new_from_string(self, year, **options)
       elsif Primitive.nil? utc_offset
         Truffle::TimeOperations.compose(self, :local, year, month, day, hour, minute, second)
       elsif utc_offset == :std

src/main/ruby/truffleruby/core/truffle/time_operations.rb

@@ -91,5 +91,35 @@ def self.compose(time_class, utc_offset, p1, p2 = nil, p3 = nil, p4 = nil, p5 =
 
       Primitive.time_s_from_array(time_class, sec, min, hour, mday, month, year, nsec, is_dst, is_utc, utc_offset)
     end
+
+    def self.new_from_string(time_class, str, **options)
+      raise ArgumentError, 'time string should have ASCII compatible encoding' unless str.encoding.ascii_compatible?
+
+      # Fast path for well-formed strings.
+      if /\A (?<year>\d{4,5})
+             (?:
+               - (?<month>\d{2})
+               - (?<mday> \d{2})
+               [ T] (?<hour> \d{2})
+                  : (?<min>  \d{2})
+                  : (?<sec>  \d{2})
+                  (?:\. (?<usec> \d+) )?
+              \s* (?<offset>\S+)?
+             )?\z/x =~ str
+        return self.compose(time_class, self.utc_offset_for_compose(offset || options[:in]), year, month, mday, hour, min, sec, usec)
+      end
+
+      raise ArgumentError, "can't parse: #{str.inspect}"
+    end
+
+    def self.utc_offset_for_compose(utc_offset)
+      if Primitive.nil?(utc_offset)
+        :local
+      elsif Time.send(:utc_offset_in_utc?, utc_offset)
+        :utc
+      else
+        Truffle::Type.coerce_to_utc_offset(utc_offset)
+      end
+    end
   end
 end