Variable not becoming a series of values. Why? - global-variables

I created this script to detect Swing Highs and Lows and then to determine the High and Low values of these Swings. The code works for the last bar, but I would like these values to be a series I can refer back to, but it's not. It's just retaining the last value. Not sure why. I included a picture of the indicator that includes this codeand a table in the corner with the last 5 swing high values....as you can see there are different swing highs but the same value in the table. Here is the code:
//Swing Highs and Lows
swing_detection(index)=>
swing_high = false
swing_low = false
start = (index*2) - 1 // -1 so we have an even number of
swing_point_high = high[index]
swing_point_low = low[index]
//Swing Highs
for i = 0 to start
swing_high := true
if i < index
if high[i] > swing_point_high
swing_high := false
break
// Have to do checks before pivot and after seperately because we can get
// two highs of the same value in a row. Notice the > and >= difference
if i > index
if high[i] >= swing_point_high
swing_high := false
break
//Swing lows
for i = 0 to start
swing_low := true
if i < index
if low[i] < swing_point_low
swing_low := false
break
// Have to do checks before pivot and after seperately because we can get
// two lows of the same value in a row. Notice the > and >= difference
if i > index
if low[i] <= swing_point_low
swing_low := false
break
[swing_high, swing_low]
// Check for a swing
[swing_high, swing_low] = swing_detection(barsback)
//Determine Swing High and Low Values
var float swing_high_price = na
var float swing_low_price = na
if swing_high
swing_high_price := high[barsback]
if swing_low
swing_low_price := low[barsback]
Example of indicator with this code in it

I don't see any code that explicitly builds a series/array so I suppose you expect swing_high_price to be the series as you add value bar-by-bar (built by the Pine engine automatically).
However Variables defined with var needs to be handled with care or you will put gaps into their series, on which Pine usually barfs.
Instead of the IF structure at the end, try this:
swing_high_price := swing_high ? high[barsback] : swing_high_price
That is, swing_high_price will be always set on every bar. This will result in a continuous series, which keeps Pine happy.

Thank you. I got it working creating a function using valuewhen().
//Swing Highs and Lows
swing_detection(index)=>
swing_high = false
swing_low = false
start = (index*2) - 1 // -1 so we have an even number of
swing_point_high = high[index]
swing_point_low = low[index]
//Swing Highs
for i = 0 to start
swing_high := true
if i < index
if high[i] > swing_point_high
swing_high := false
break
// Have to do checks before pivot and after seperately because we can get
// two highs of the same value in a row. Notice the > and >= difference
if i > index
if high[i] >= swing_point_high
swing_high := false
break
//Swing lows
for i = 0 to start
swing_low := true
if i < index
if low[i] < swing_point_low
swing_low := false
break
// Have to do checks before pivot and after seperately because we can get
// two lows of the same value in a row. Notice the > and >= difference
if i > index
if low[i] <= swing_point_low
swing_low := false
break
[swing_high, swing_low]
// Check for a swing
[swing_high, swing_low] = swing_detection(barsback)
////////////////////////////////////////////////////////////////////////////////
swing_high_price(_vw) =>
_return=valuewhen(conf_swing_high, high[barsback],_vw)
swing_low_price(_vw) =>
_return=valuewhen(conf_swing_low, low[barsback],_vw)

Related

How to plot line only on first value of crossover condition met in Pine Script

I need help in Pine script coding, i want to plot the first crossover value alone not entire series of values for whole day. so if crossover condition is true for 1st or 2nd or 3rd or n times, i need to plot the line from the first condition met to end of day and ignore 2nd or 3rd or 4th or nth series values to plot. how to do it? and i don't want to use line.new() since i want to backtest in tradingview.com, for historical bars i want to use Plot(). so kindly help me with the code.
Strategy im trying: Added 30pts above from 5mins High-Low, if 5 mins candle crossover the 30pts at first time from High need to plot line, ignore if condition met again in a day.
`
//#version=5
indicator(title="Crossover", shorttitle="Crossover", overlay=true)
//*5Mins High Low calculation*//
inputMax = input(5, title= "ORB total time (minutes)")
sess = input("0915-0920", title="Session Time")
t = time(timeframe.period, sess + ":1234567")
hide = timeframe.isintraday and timeframe.multiplier <= inputMax
is_newbar(res) => ta.change(time(res)) != 0
in_session = not na(t)
is_first = in_session and not in_session[1]
orb_high = float(na)
orb_low = float(na)
if is_first
orb_high := high
orb_low := low
else
orb_high := orb_high[1]
orb_low := orb_low[1]
if high > orb_high and in_session
orb_high := high
if low < orb_low and in_session
orb_low := low
plot(hide ? orb_high : na , style=plot.style_line, color=orb_high[1] != orb_high ? na : color.green , title="ORB High", linewidth=3)
plot(hide ? orb_low : na , style=plot.style_line, color=orb_low[1] != orb_low ? na : color.red, title="ORB Low", linewidth=3)
//*Crossover condition*//
var Up = 0.0
Up := orb_high + 30
var b_i = 0.0
cross_over_happened = if(ta.crossover(close, Up)) and barstate.isconfirmed
1
else
0
b_i := ta.valuewhen(cross_over_happened, close, 0)
plot(b_i, color = color.black, linewidth = 2)
`
the above code will plot whenever condition met, but i need to plot only the first value of crossover condition, not for entire series. Kindly help with the code.
I would go about this in a slightly different way (which is very close to what you have done). I would add a bool that will check if a cross already happened today, and will change the value of b_i only when the cross and the new bool are true.
First set the variables that we're going to need:
var float orb_high = na
var float orb_low = na
var bool already_had_crossover_today = false
var float b_i = na
Second, reset those variables on each new day:
if ta.change(time("D"))
orb_high := high
orb_low := low
already_had_crossover_today := false
b_i := na
Third, we'll check for the crossover, regardless if it's the first one on the day or not:
Up = orb_high + 30
cross_over_happened = ta.crossover(close, Up)
And lastly, we'll check if we had a crossover and during the day a crossover hasn't happened yet:
if cross_over_happened and not already_had_crossover_today
b_i := close
already_had_crossover_today := true
Than we can just use plot, and for better visualization use the style parameter:
plot(b_i, style = plot.style_linebr)

I am not able to understand what this piece of code is doing. can anyone please explain?

excelTbl[, .(sum(`Current Period Amount`),
sum(`Comparison Period Amount`),
sum(`% Change`),
type = if (sum(`Current Period Amount`) - sum(`Comparison Period Amount`)< 0) "N" else "P"),
by = .(Class, AccountSubType)][round(V1) != 0 &
round(V2) != 0 &
(if (logicalOp == 1) {
((abs(V2 - V1) * 100 / abs(V2)) >= variancePer) &
#(abs(V3)*100 >= variancePer) &
(abs(V2 - V1) >= varianceAmt * 1000000)
} else {
((abs(V2 - V1) * 100 / abs(V2)) >= variancePer) |
#(abs(V3)*100 >= variancePer) |
(abs(V2 - V1) >= varianceAmt * 1000000)
}), 1:2]
round(V1) != 0 returns a logical vector as long as there are rows in the table. For example,
1:5 < 3
# [1] TRUE TRUE FALSE FALSE FALSE
... & round(V2) != 0 is a vectorized "AND", where it is still as long as there are rows in the table. For example,
(1:5 < 3) & (6:10 < 7)
# [1] TRUE FALSE FALSE FALSE FALSE
if (logicalOp == 1) ... else ... will calculate one of two logical vectors based on the value of logicalOp. If it is 1, then it logically "AND"s the logical vector returned from
((abs(V2 - V1) * 100 / abs(V2)) >= variancePer) &
(abs(V2 - V1) >= varianceAmt * 1000000)
otherwise the other vector.
This is vector-wise "AND"ed with the results of the two round logical vectors.
And because all of that is before the first comma (it is in the i= component of the bracket indexing), it defined which rows to return. If you have a logical vector as long as there are rows, then the rows with TRUE will be preserved, FALSE will be discarded.
While many things in data.table operate in-place (referential semantics, as opposed to R's regular copy-on-write semantics), anything that subsets the rows like this will return another data.table object and not modify the original in-place. Because of this, the value of this code chunk should be captured into a new variable.
After the comma, 1:2 selects the first two columns.
Interestingly, the first bracket block creates a data.table with six columns, named Class and AccountSubType (the groups defined with by=), V1, V2, V3, and type. However, the second bracket block only references V1 and V2 in the filtering and then discards all except Class and AccountSubType, so ... it seems there is unnecessary calculation.

Julia 1.0.2.1: Why is a variable changing value, without being assigned?

I am building a metaheuristic in Julia for study purpose.
The purpose is to find the best order of boxes.
1) I start with an initial order (random order) defined as. Order = InitOrder before my while loop.
2) For each iteration in the while loop I set CurrentOrder = Order
3) When the CurrentOrder is changed, Order changes too. Why does Order change value without being assigned? And how do I avoid it?
Version:
JuliaPro 1.0.2.1
Editor: Atom
while ( (time_ns()-timestart)/1.0e9 < RunLength && done == false ) #Stopping Criteria
done = true #Starting point
IterationCount = IterationCount + 1
BestCurrentValue = sum(H) #Worst case solutio
CurrentOrder = Order #(From,To)
for n1=1:N
for n2=1:N
if n1 != n2
(CurrentOrder,CopyTo) = SwapBox(CurrentOrder,n1,n2) #Swap boxes
(CurrentLayout,L) = DeltaCopy(CurrentLayout,CopyTo,CurrentOrder) #Delta Copy to minimise calculations
(TempLayout,L) = BLV(BinW,CurrentLayout,CopyTo,CurrentOrder,W,H,L) #Evalueate by BLV
if L < BestCurrentValue #check if TempLayout is better than Best Current
BestCurrentValue = L
BestCurrentOrder = CurrentOrder
BestCurrentLayout = CurrentLayout
end #if L<...
end #if n1 != n2
##############################################################################
CurrentOrder = Order
##############################################################################
end #n2 in N
end #n1 in N
if BestCurrentValue < BestValue
done = false #Look further
BestValue = BestCurrentValue
BestOrder = BestCurrentOrder
BestLayout = BestCurrentLayout
Order = BestOrder
end #if BestCurrentValue...
end #while
Your assignment NewOrder=Order does not copy any information in memory, it just says that the variable NewOrder should point to the same memory location as Order. Changing one of these variables will thus also change the other. If you want to copy a variable you could use NewOrder=deepcopy(Order)

How to get a slice of tuples (hour, minute)

I've got a problem that I couldn't resolve. I'm using https://github.com/kmanley/golang-tuple to create tuples.
I've got a list of minutes:
minutes := int{0, 30} // Minutes are 0 and 30
And a four parameters: start, startBreak, stop, stopBreak:
start := tuple.NewTupleFromItems(9, 30) // It represents "9:30"
startBreak := tuple.NewTupleFromItems(12, 0) // It represents "12:00"
stop := tuple.NewTupleFromItems(21, 0) // It represents "21:00"
stopBreak := tuple.NewTupleFromItems(14, 30) // It represents "14:30"
I want to get a slice of tuples (hour, minutes) using all the minutes in the minutes slice and they must not be included in the range startBreak-stopBreak (it can be equal to startBreak or stopBreak, so the range will become 12:30, 13:00, 13:30, 14:00) and stop-start (it can be equal to start and stop, so the range will become 21:30, 22:00, 22:30, ..., 8:30, 9:00).
For example, using those four parameters, the final result will be:
9:30, 10:00, 10:30, 11:00, 11:30, 12:00, 14:30, 15:00, 15:30, 16:00, 16:30, 17:00, 17:30, 18:00, 18:30, 19:00, 19:30, 20:00, 20:30, 21:00
Here is a minimal code that demonstrates this, I did not put any data validation.
func periods(minutes, start, startBreak, stopBreak, stop *tuple.Tuple) (out []tuple.Tuple) {
// next() moves current to the next minute interval
i := 0
curr := tuple.NewTupleFromItems(start.Get(0), minutes.Get(0))
next := func() {
i = (i + 1) % minutes.Len()
curr.Set(1, minutes.Get(i))
if i == 0 {
curr.Set(0, curr.Get(0).(int)+1)
}
}
for ; curr.Le(stop); next() {
if (curr.Ge(start) && curr.Le(startBreak)) || (curr.Ge(stopBreak) && curr.Le(stop)) {
out = append(out, *curr.Copy())
}
}
return out
}
Playground

Determine Whether Two Date Ranges Overlap

Given two date ranges, what is the simplest or most efficient way to determine whether the two date ranges overlap?
As an example, suppose we have ranges denoted by DateTime variables StartDate1 to EndDate1 and StartDate2 to EndDate2.
(StartA <= EndB) and (EndA >= StartB)
Proof:
Let ConditionA Mean that DateRange A Completely After DateRange B
_ |---- DateRange A ------|
|---Date Range B -----| _
(True if StartA > EndB)
Let ConditionB Mean that DateRange A is Completely Before DateRange B
|---- DateRange A -----| _
_ |---Date Range B ----|
(True if EndA < StartB)
Then Overlap exists if Neither A Nor B is true -
(If one range is neither completely after the other,
nor completely before the other,
then they must overlap.)
Now one of De Morgan's laws says that:
Not (A Or B) <=> Not A And Not B
Which translates to: (StartA <= EndB) and (EndA >= StartB)
NOTE: This includes conditions where the edges overlap exactly. If you wish to exclude that,
change the >= operators to >, and <= to <
NOTE2. Thanks to #Baodad, see this blog, the actual overlap is least of:
{ endA-startA, endA - startB, endB-startA, endB - startB }
(StartA <= EndB) and (EndA >= StartB)
(StartA <= EndB) and (StartB <= EndA)
NOTE3. Thanks to #tomosius, a shorter version reads:
DateRangesOverlap = max(start1, start2) < min(end1, end2)
This is actually a syntactical shortcut for what is a longer implementation, which includes extra checks to verify that the start dates are on or before the endDates. Deriving this from above:
If start and end dates can be out of order, i.e., if it is possible that startA > endA or startB > endB, then you also have to check that they are in order, so that means you have to add two additional validity rules:
(StartA <= EndB) and (StartB <= EndA) and (StartA <= EndA) and (StartB <= EndB)
or:
(StartA <= EndB) and (StartA <= EndA) and (StartB <= EndA) and (StartB <= EndB)
or,
(StartA <= Min(EndA, EndB) and (StartB <= Min(EndA, EndB))
or:
(Max(StartA, StartB) <= Min(EndA, EndB)
But to implement Min() and Max(), you have to code, (using C ternary for terseness),:
((StartA > StartB) ? StartA : StartB) <= ((EndA < EndB) ? EndA : EndB)
I believe that it is sufficient to say that the two ranges overlap if:
(StartDate1 <= EndDate2) and (StartDate2 <= EndDate1)
This article Time Period Library for .NET describes the relation of two time periods by the enumeration PeriodRelation:
// ------------------------------------------------------------------------
public enum PeriodRelation
{
After,
StartTouching,
StartInside,
InsideStartTouching,
EnclosingStartTouching,
Enclosing,
EnclosingEndTouching,
ExactMatch,
Inside,
InsideEndTouching,
EndInside,
EndTouching,
Before,
} // enum PeriodRelation
For reasoning about temporal relations (or any other interval relations, come to that), consider Allen's Interval Algebra. It describes the 13 possible relations that two intervals can have with respect to each other. You can find other references — "Allen Interval" seems to be an operative search term. You can also find information about these operations in Snodgrass's Developing Time-Oriented Applications in SQL (PDF available online at URL), and in Date, Darwen and Lorentzos Temporal Data and the Relational Model (2002) or Time and Relational Theory: Temporal Databases in the Relational Model and SQL (2014; effectively the second edition of TD&RM).
The short(ish) answer is: given two date intervals A and B with components .start and .end and the constraint .start <= .end, then two intervals overlap if:
A.end >= B.start AND A.start <= B.end
You can tune the use of >= vs > and <= vs < to meet your requirements for degree of overlap.
ErikE comments:
You can only get 13 if you count things funny... I can get "15 possible relations that two intervals can have" when I go crazy with it. By sensible counting, I get only six, and if you throw out caring whether A or B comes first, I get only three (no intersect, partially intersect, one wholly within other). 15 goes like this: [before:before, start, within, end, after], [start:start, within, end, after], [within:within, end, after], [end:end, after], [after:after].
I think that you cannot count the two entries 'before:before' and 'after:after'. I could see 7 entries if you equate some relations with their inverses (see the diagram in the referenced Wikipedia URL; it has 7 entries, 6 of which have a different inverse, with equals not having a distinct inverse). And whether three is sensible depends on your requirements.
----------------------|-------A-------|----------------------
|----B1----|
|----B2----|
|----B3----|
|----------B4----------|
|----------------B5----------------|
|----B6----|
----------------------|-------A-------|----------------------
|------B7-------|
|----------B8-----------|
|----B9----|
|----B10-----|
|--------B11--------|
|----B12----|
|----B13----|
----------------------|-------A-------|----------------------
If the overlap itself should be calculated as well, you can use the following formula:
overlap = max(0, min(EndDate1, EndDate2) - max(StartDate1, StartDate2))
if (overlap > 0) {
...
}
All the solutions that check a multitude of conditions based on where the ranges are in relation to one another can be greatly simplified by simply ensuring that one range starts before or at the same time as the other. You can do this by swapping the ranges if necessary up front.
Then, you can detect overlap if the second range start is:
less than or equal to the first range end (if ranges are inclusive, containing both the start and end times); or
less than (if ranges are inclusive of start and exclusive of end).
For example (assuming inclusive at both ends), there's only four possibilities for range 2, of which one is a non-overlap (the > at the end of the range means it doesn't matter where the range ends):
|-----| range 1, lines below are all range 2.
|--> : overlap.
|--> : overlap.
|---> overlap (no overlap in exclusive-of-end case).
|---> no overlap.
The endpoint of the second range doesn't affect the result at all. So, in pseudo-code, you can do something like (assuming s <= e holds for all ranges - if not, you may have top swap them as well):
def overlaps(r1, r2):
if r1.s > r2.s:
swap r1, r2
return r2.s <= r1.e
Or, the one-level-limit recursive option:
def overlaps(r1, r2):
if r1.s <= r2.s:
return r2.s <= r1.e
return overlaps(r2, r1)
If the ranges are exclusive at the end, you just have to replace <= with < in the expression you return (in both code snippets).
This greatly limits the number of checks you have to make because you remove half of the problem space early by ensuring the first range never starts after the second.
And, since "code talks", here is some Python code that shows this in action, with quite a few test cases. First, the InclusiveRange class:
class InclusiveRange:
"""InclusiveRange class to represent a lower and upper bound."""
def __init__(self, start, end):
"""Initialisation, ensures start <= end.
Args:
start: The start of the range.
end: The end of the range.
"""
self.start = min(start, end)
self.end = max(start, end)
def __repr__(self):
"""Return representation for f-string."""
return f"({self.start}, {self.end})"
def overlaps(self, other):
"""True if range overlaps with another.
Args:
other: The other InclusiveRange to check against.
"""
# Very limited recursion to ensure start of first range
# isn't after start of second.
if self.start > other.start:
return other.overlaps(self)
# Greatly simplified check for overlap.
return other.start <= self.end
Then a test case handler to allow us to nicely present the result of a single test case:
def test_case(range1, range2):
"""Single test case checker."""
# Get low and high value for "graphic" output.
low = min(range1.start, range2.start)
high = max(range1.end, range2.end)
# Output ranges and graphic.
print(f"r1={range1} r2={range2}: ", end="")
for val in range(low, high + 1):
is_in_first = range1.start <= val <= range1.end
is_in_second = range2.start <= val <= range2.end
if is_in_first and is_in_second:
print("|", end="")
elif is_in_first:
print("'", end="")
elif is_in_second:
print(",", end="")
else:
print(" ", end="")
# Finally, output result of overlap check.
print(f" - {range1.overlaps(range2)}\n")
Then finally, a decent chunk of test cases to which you can add your own if need be:
# Various test cases, add others if you doubt the correctness.
test_case(InclusiveRange(0, 1), InclusiveRange(8, 9))
test_case(InclusiveRange(0, 4), InclusiveRange(5, 9))
test_case(InclusiveRange(0, 4), InclusiveRange(4, 9))
test_case(InclusiveRange(0, 7), InclusiveRange(2, 9))
test_case(InclusiveRange(0, 4), InclusiveRange(0, 9))
test_case(InclusiveRange(0, 9), InclusiveRange(0, 9))
test_case(InclusiveRange(0, 9), InclusiveRange(4, 5))
test_case(InclusiveRange(8, 9), InclusiveRange(0, 1))
test_case(InclusiveRange(5, 9), InclusiveRange(0, 4))
test_case(InclusiveRange(4, 9), InclusiveRange(0, 4))
test_case(InclusiveRange(2, 9), InclusiveRange(0, 7))
test_case(InclusiveRange(0, 9), InclusiveRange(0, 4))
test_case(InclusiveRange(0, 9), InclusiveRange(0, 9))
test_case(InclusiveRange(4, 5), InclusiveRange(0, 9))
Running that produces the output:
r1=(0, 1) r2=(8, 9): '' ,, - False
r1=(0, 4) r2=(5, 9): ''''',,,,, - False
r1=(0, 4) r2=(4, 9): ''''|,,,,, - True
r1=(0, 7) r2=(2, 9): ''||||||,, - True
r1=(0, 4) r2=(0, 9): |||||,,,,, - True
r1=(0, 9) r2=(0, 9): |||||||||| - True
r1=(0, 9) r2=(4, 5): ''''||'''' - True
r1=(8, 9) r2=(0, 1): ,, '' - False
r1=(5, 9) r2=(0, 4): ,,,,,''''' - False
r1=(4, 9) r2=(0, 4): ,,,,|''''' - True
r1=(2, 9) r2=(0, 7): ,,||||||'' - True
r1=(0, 9) r2=(0, 4): |||||''''' - True
r1=(0, 9) r2=(0, 9): |||||||||| - True
r1=(4, 5) r2=(0, 9): ,,,,||,,,, - True
where each line has:
the two ranges being evaluated;
a graphical representation of the "range space" (from lowest start to highest end) where each character is a value in that "range space":
' indicates a value in the first range only;
, indicates a value in the second range only;
| indicates a value in both ranges; and
indicates a value in neither range.
the result of the overlap check.
You can see quite clearly that you only get true in the overlap check when there is at least one value in both ranges (i.e., a | character). Every other case gives false.
Feel free to use any other values if you want to add more test cases.
Here is yet another solution using JavaScript. Specialities of my solution:
Handles null values as infinity
Assumes that the lower bound is inclusive and the upper bound exclusive.
Comes with a bunch of tests
The tests are based on integers but since date objects in JavaScript are comparable you can just throw in two date objects as well. Or you could throw in the millisecond timestamp.
Code:
/**
* Compares to comparable objects to find out whether they overlap.
* It is assumed that the interval is in the format [from,to) (read: from is inclusive, to is exclusive).
* A null value is interpreted as infinity
*/
function intervalsOverlap(from1, to1, from2, to2) {
return (to2 === null || from1 < to2) && (to1 === null || to1 > from2);
}
Tests:
describe('', function() {
function generateTest(firstRange, secondRange, expected) {
it(JSON.stringify(firstRange) + ' and ' + JSON.stringify(secondRange), function() {
expect(intervalsOverlap(firstRange[0], firstRange[1], secondRange[0], secondRange[1])).toBe(expected);
});
}
describe('no overlap (touching ends)', function() {
generateTest([10,20], [20,30], false);
generateTest([20,30], [10,20], false);
generateTest([10,20], [20,null], false);
generateTest([20,null], [10,20], false);
generateTest([null,20], [20,30], false);
generateTest([20,30], [null,20], false);
});
describe('do overlap (one end overlaps)', function() {
generateTest([10,20], [19,30], true);
generateTest([19,30], [10,20], true);
generateTest([10,20], [null,30], true);
generateTest([10,20], [19,null], true);
generateTest([null,30], [10,20], true);
generateTest([19,null], [10,20], true);
});
describe('do overlap (one range included in other range)', function() {
generateTest([10,40], [20,30], true);
generateTest([20,30], [10,40], true);
generateTest([10,40], [null,null], true);
generateTest([null,null], [10,40], true);
});
describe('do overlap (both ranges equal)', function() {
generateTest([10,20], [10,20], true);
generateTest([null,20], [null,20], true);
generateTest([10,null], [10,null], true);
generateTest([null,null], [null,null], true);
});
});
Result when run with karma&jasmine&PhantomJS:
PhantomJS 1.9.8 (Linux): Executed 20 of 20 SUCCESS (0.003 secs / 0.004 secs)
Here is the code that does the magic:
var isOverlapping = ((A == null || D == null || A <= D)
&& (C == null || B == null || C <= B)
&& (A == null || B == null || A <= B)
&& (C == null || D == null || C <= D));
Where..
A -> 1Start
B -> 1End
C -> 2Start
D -> 2End
Proof? Check out this test console code gist.
An easy way to remember the solution would be
min(ends)>max(starts)
I would do
StartDate1.IsBetween(StartDate2, EndDate2) || EndDate1.IsBetween(StartDate2, EndDate2)
Where IsBetween is something like
public static bool IsBetween(this DateTime value, DateTime left, DateTime right) {
return (value > left && value < right) || (value < left && value > right);
}
Here's my solution in Java, which works on unbounded intervals too
private Boolean overlap (Timestamp startA, Timestamp endA,
Timestamp startB, Timestamp endB)
{
return (endB == null || startA == null || !startA.after(endB))
&& (endA == null || startB == null || !endA.before(startB));
}
The solution posted here did not work for all overlapping ranges...
----------------------|-------A-------|----------------------
|----B1----|
|----B2----|
|----B3----|
|----------B4----------|
|----------------B5----------------|
|----B6----|
----------------------|-------A-------|----------------------
|------B7-------|
|----------B8-----------|
|----B9----|
|----B10-----|
|--------B11--------|
|----B12----|
|----B13----|
----------------------|-------A-------|----------------------
my working solution was:
AND (
('start_date' BETWEEN STARTDATE AND ENDDATE) -- caters for inner and end date outer
OR
('end_date' BETWEEN STARTDATE AND ENDDATE) -- caters for inner and start date outer
OR
(STARTDATE BETWEEN 'start_date' AND 'end_date') -- only one needed for outer range where dates are inside.
)
As there have been several answers for different languages and environments, here is one for standard ANSI SQL.
In standard SQL it is as as simple as
(StartDate1, EndDate1) overlaps (StartDate2, EndDate2)
assuming all four columns are DATE or TIMESTAMP columns. It returns true if both ranges have at least one day in common (assuming DATE values)
(However not all DBMS products support that)
In PostgreSQL it's also easy to test for inclusion by using date ranges
daterange(StartDate1, EndDate1) #> daterange(StartDate2, EndDate2)
the above returns true if the second range is completely included in the first (which is different to "overlaps")
This was my javascript solution with moment.js:
// Current row dates
var dateStart = moment("2014-08-01", "YYYY-MM-DD");
var dateEnd = moment("2014-08-30", "YYYY-MM-DD");
// Check with dates above
var rangeUsedStart = moment("2014-08-02", "YYYY-MM-DD");
var rangeUsedEnd = moment("2014-08-015", "YYYY-MM-DD");
// Range covers other ?
if((dateStart <= rangeUsedStart) && (rangeUsedEnd <= dateEnd)) {
return false;
}
// Range intersects with other start ?
if((dateStart <= rangeUsedStart) && (rangeUsedStart <= dateEnd)) {
return false;
}
// Range intersects with other end ?
if((dateStart <= rangeUsedEnd) && (rangeUsedEnd <= dateEnd)) {
return false;
}
// All good
return true;
Short answer using momentjs:
function isOverlapping(startDate1, endDate1, startDate2, endDate2){
return moment(startDate1).isSameOrBefore(endDate2) &&
moment(startDate2).isSameOrBefore(endDate1);
}
the answer is based on above answers, but its shortened.
In Microsoft SQL SERVER - SQL Function
CREATE FUNCTION IsOverlapDates
(
#startDate1 as datetime,
#endDate1 as datetime,
#startDate2 as datetime,
#endDate2 as datetime
)
RETURNS int
AS
BEGIN
DECLARE #Overlap as int
SET #Overlap = (SELECT CASE WHEN (
(#startDate1 BETWEEN #startDate2 AND #endDate2) -- caters for inner and end date outer
OR
(#endDate1 BETWEEN #startDate2 AND #endDate2) -- caters for inner and start date outer
OR
(#startDate2 BETWEEN #startDate1 AND #endDate1) -- only one needed for outer range where dates are inside.
) THEN 1 ELSE 0 END
)
RETURN #Overlap
END
GO
--Execution of the above code
DECLARE #startDate1 as datetime
DECLARE #endDate1 as datetime
DECLARE #startDate2 as datetime
DECLARE #endDate2 as datetime
DECLARE #Overlap as int
SET #startDate1 = '2014-06-01 01:00:00'
SET #endDate1 = '2014-06-01 02:00:00'
SET #startDate2 = '2014-06-01 01:00:00'
SET #endDate2 = '2014-06-01 01:30:00'
SET #Overlap = [dbo].[IsOverlapDates] (#startDate1, #endDate1, #startDate2, #endDate2)
SELECT Overlap = #Overlap
the simplest
The simplest way is to use a well-engineered dedicated library for date-time work.
someInterval.overlaps( anotherInterval )
java.time & ThreeTen-Extra
The best in the business is the java.time framework built into Java 8 and later. Add to that the ThreeTen-Extra project that supplements java.time with additional classes, specifically the Interval class we need here.
As for the language-agnostic tag on this Question, the source code for both projects is available for use in other languages (mind their licenses).
Interval
The org.threeten.extra.Interval class is handy, but requires date-time moments (java.time.Instant objects) rather than date-only values. So we proceed by using the first moment of the day in UTC to represent the date.
Instant start = Instant.parse( "2016-01-01T00:00:00Z" );
Instant stop = Instant.parse( "2016-02-01T00:00:00Z" );
Create an Interval to represent that span of time.
Interval interval_A = Interval.of( start , stop );
We can also define an Interval with a starting moment plus a Duration.
Instant start_B = Instant.parse( "2016-01-03T00:00:00Z" );
Interval interval_B = Interval.of( start_B , Duration.of( 3 , ChronoUnit.DAYS ) );
Comparing to test for overlaps is easy.
Boolean overlaps = interval_A.overlaps( interval_B );
You can compare an Interval against another Interval or Instant:
abuts
contains
encloses
equals
isAfter
isBefore
overlaps
All of these use the Half-Open approach to defining a span of time where the beginning is inclusive and the ending is exclusive.
I had a situation where we had dates instead of datetimes, and the dates could overlap only on start/end. Example below:
(Green is the current interval, blue blocks are valid intervals, red ones are overlapping intervals).
I adapted Ian Nelson's answer to the following solution:
(startB <= startA && endB > startA)
|| (startB >= startA && startB < endA)
This matches all overlap cases but ignores the allowed overlap ones.
The mathematical solution given by #Bretana is good but neglects two specific details:
aspect of closed or half-open intervals
empty intervals
About the closed or open state of interval boundaries, the solution of #Bretana valid for closed intervals
(StartA <= EndB) and (EndA >= StartB)
can be rewritten for half-open intervals to:
(StartA < EndB) and (EndA > StartB)
This correction is necessary because an open interval boundary does not belong to the value range of an interval by definition.
And about empty intervals, well, here the relationship shown above does NOT hold. Empty intervals which do not contain any valid value by definition must be handled as special case. I demonstrate it by my Java time library Time4J via this example:
MomentInterval a = MomentInterval.between(Instant.now(), Instant.now().plusSeconds(2));
MomentInterval b = a.collapse(); // make b an empty interval out of a
System.out.println(a); // [2017-04-10T05:28:11,909000000Z/2017-04-10T05:28:13,909000000Z)
System.out.println(b); // [2017-04-10T05:28:11,909000000Z/2017-04-10T05:28:11,909000000Z)
The leading square bracket "[" indicates a closed start while the last bracket ")" indicates an open end.
System.out.println(
"startA < endB: " + a.getStartAsInstant().isBefore(b.getEndAsInstant())); // false
System.out.println(
"endA > startB: " + a.getEndAsInstant().isAfter(b.getStartAsInstant())); // true
System.out.println("a overlaps b: " + a.intersects(b)); // a overlaps b: false
As shown above, empty intervals violate the overlap condition above (especially startA < endB), so Time4J (and other libraries, too) has to handle it as special edge case in order to guarantee that the overlap of any arbitrary interval with an empty interval does not exist. Of course, date intervals (which are closed by default in Time4J but can be half-open, too, like empty date intervals) are handled in a similar way.
This is an extension to the excellent answer by #charles-bretana.
The answer however does not make a distinction among open, closed, and half-open (or half-closed) intervals.
Case 1: A, B are closed intervals
A = [StartA, EndA]
B = [StartB, EndB]
[---- DateRange A ------] (True if StartA > EndB)
[--- Date Range B -----]
[---- DateRange A -----] (True if EndA < StartB)
[--- Date Range B ----]
Overlap iff: (StartA <= EndB) and (EndA >= StartB)
Case 2: A, B are open intervals
A = (StartA, EndA)
B = (StartB, EndB)
(---- DateRange A ------) (True if StartA >= EndB)
(--- Date Range B -----)
(---- DateRange A -----) (True if EndA <= StartB)
(--- Date Range B ----)
Overlap iff: (StartA < EndB) and (EndA > StartB)
Case 3: A, B right open
A = [StartA, EndA)
B = [StartB, EndB)
[---- DateRange A ------) (True if StartA >= EndB)
[--- Date Range B -----)
[---- DateRange A -----) (True if EndA <= StartB)
[--- Date Range B ----)
Overlap condition: (StartA < EndB) and (EndA > StartB)
Case 4: A, B left open
A = (StartA, EndA]
B = (StartB, EndB]
(---- DateRange A ------] (True if StartA >= EndB)
(--- Date Range B -----]
(---- DateRange A -----] (True if EndA <= StartB)
(--- Date Range B ----]
Overlap condition: (StartA < EndB) and (EndA > StartB)
Case 5: A right open, B closed
A = [StartA, EndA)
B = [StartB, EndB]
[---- DateRange A ------) (True if StartA > EndB)
[--- Date Range B -----]
[---- DateRange A -----) (True if EndA <= StartB)
[--- Date Range B ----]
Overlap condition: (StartA <= EndB) and (EndA > StartB)
etc...
Finally, the general condition for two intervals to overlap is
(StartA <🞐 EndB) and (EndA >🞐 StartB)
where 🞐 turns a strict inequality into a non-strict one whenever the comparison is made between two included endpoint.
In case you're using a date range that has not ended yet (still on going) e.g. not set
endDate = '0000-00-00' you can not use BETWEEN because 0000-00-00 is not a valid date!
I used this solution:
(Startdate BETWEEN '".$startdate2."' AND '".$enddate2."') //overlap: starts between start2/end2
OR (Startdate < '".$startdate2."'
AND (enddate = '0000-00-00' OR enddate >= '".$startdate2."')
) //overlap: starts before start2 and enddate not set 0000-00-00 (still on going) or if enddate is set but higher then startdate2
If startdate2 is higher then enddate there is no overlap!
The answer is too simple for me so I have created a more generic dynamic SQL statement which checks to see if a person has any overlapping dates.
SELECT DISTINCT T1.EmpID
FROM Table1 T1
INNER JOIN Table2 T2 ON T1.EmpID = T2.EmpID
AND T1.JobID <> T2.JobID
AND (
(T1.DateFrom >= T2.DateFrom AND T1.dateFrom <= T2.DateTo)
OR (T1.DateTo >= T2.DateFrom AND T1.DateTo <= T2.DateTo)
OR (T1.DateFrom < T2.DateFrom AND T1.DateTo IS NULL)
)
AND NOT (T1.DateFrom = T2.DateFrom)
Using Java util.Date, here what I did.
public static boolean checkTimeOverlaps(Date startDate1, Date endDate1, Date startDate2, Date endDate2)
{
if (startDate1 == null || endDate1 == null || startDate2 == null || endDate2 == null)
return false;
if ((startDate1.getTime() <= endDate2.getTime()) && (startDate2.getTime() <= endDate1.getTime()))
return true;
return false;
}
For ruby I also found this:
class Interval < ActiveRecord::Base
validates_presence_of :start_date, :end_date
# Check if a given interval overlaps this interval
def overlaps?(other)
(start_date - other.end_date) * (other.start_date - end_date) >= 0
end
# Return a scope for all interval overlapping the given interval, including the given interval itself
named_scope :overlapping, lambda { |interval| {
:conditions => ["id <> ? AND (DATEDIFF(start_date, ?) * DATEDIFF(?, end_date)) >= 0", interval.id, interval.end_date, interval.start_date]
}}
end
Found it here with nice explaination ->
http://makandracards.com/makandra/984-test-if-two-date-ranges-overlap-in-ruby-or-rails
The easiest way to do it in my opinion would be to compare if either EndDate1 is before StartDate2 and EndDate2 is before StartDate1.
That of course if you are considering intervals where StartDate is always before EndDate.
If you provide a date range as input and want to find out if it overlaps with the existing date range in database, the following conditions can successfully meet your demand
Assume you provide a #StartDate and #EndDate from your form input.
conditions are :
If #StartDate is ahead of existingStartDate and behind existingEndDate then we can say #StartDate is in the middle of a existing date range, thus we can conclude it will overlap
#StartDate >=existing.StartDate And #StartDate <= existing.EndDate)
If #StartDate is behind existingStartDate but #EndDate is ahead of existingStartDate we can say that it will overlap
(#StartDate <= existing.StartDate And #EndDate >= existing.StartDate)
If #StartDate is behind existingStartDate And #EndDate is ahead of existingEndDate we can conclude that the provided date range devours a existing date range , thus overlaps
(#StartDate <= existing.StartDate And #EndDate >= existing.EndDate))
If any of the condition stands true, your provided date range overlaps with existing ones in the database.
Split the problem into cases then handle each case.
The situation 'two date ranges intersect' is covered by two cases - the first date range starts within the second, or the second date range starts within the first.
You can try this:
//custom date for example
$d1 = new DateTime("2012-07-08");
$d2 = new DateTime("2012-07-11");
$d3 = new DateTime("2012-07-08");
$d4 = new DateTime("2012-07-15");
//create a date period object
$interval = new DateInterval('P1D');
$daterange = iterator_to_array(new DatePeriod($d1, $interval, $d2));
$daterange1 = iterator_to_array(new DatePeriod($d3, $interval, $d4));
array_map(function($v) use ($daterange1) { if(in_array($v, $daterange1)) print "Bingo!";}, $daterange);
public static class NumberExtensionMethods
{
public static Boolean IsBetween(this Int64 value, Int64 Min, Int64 Max)
{
if (value >= Min && value <= Max) return true;
else return false;
}
public static Boolean IsBetween(this DateTime value, DateTime Min, DateTime Max)
{
Int64 numricValue = value.Ticks;
Int64 numericStartDate = Min.Ticks;
Int64 numericEndDate = Max.Ticks;
if (numricValue.IsBetween(numericStartDate, numericEndDate) )
{
return true;
}
return false;
}
}
public static Boolean IsOverlap(DateTime startDate1, DateTime endDate1, DateTime startDate2, DateTime endDate2)
{
Int64 numericStartDate1 = startDate1.Ticks;
Int64 numericEndDate1 = endDate1.Ticks;
Int64 numericStartDate2 = startDate2.Ticks;
Int64 numericEndDate2 = endDate2.Ticks;
if (numericStartDate2.IsBetween(numericStartDate1, numericEndDate1) ||
numericEndDate2.IsBetween(numericStartDate1, numericEndDate1) ||
numericStartDate1.IsBetween(numericStartDate2, numericEndDate2) ||
numericEndDate1.IsBetween(numericStartDate2, numericEndDate2))
{
return true;
}
return false;
}
if (IsOverlap(startdate1, enddate1, startdate2, enddate2))
{
Console.WriteLine("IsOverlap");
}
This was my solution, it returns true when the values don't overlap:
X START 1
Y END 1
A START 2
B END 2
TEST1: (X <= A || X >= B)
&&
TEST2: (Y >= B || Y <= A)
&&
TEST3: (X >= B || Y <= A)
X-------------Y
A-----B
TEST1: TRUE
TEST2: TRUE
TEST3: FALSE
RESULT: FALSE
---------------------------------------
X---Y
A---B
TEST1: TRUE
TEST2: TRUE
TEST3: TRUE
RESULT: TRUE
---------------------------------------
X---Y
A---B
TEST1: TRUE
TEST2: TRUE
TEST3: TRUE
RESULT: TRUE
---------------------------------------
X----Y
A---------------B
TEST1: FALSE
TEST2: FALSE
TEST3: FALSE
RESULT: FALSE

Resources