Merge pull request jump-dev#1646 from JuliaOpt/bl/print

Implement model printing

docs/src/quickstart.md

@@ -28,6 +28,11 @@ used to specify the optimizer to be used:
 ```julia
 julia> model = Model(with_optimizer(GLPK.Optimizer))
 A JuMP Model
+Feasibility problem with:
+Variables: 0
+Model mode: Automatic
+CachingOptimizer state: NoOptimizer
+Solver name: No optimizer attached.
 ```
 
 ```@meta

docs/src/variables.md

@@ -28,6 +28,7 @@ To illustrate these three types of variables, consider the following JuMP code
 ```jldoctest variables
 julia> model = Model()
 A JuMP Model
+Feasibility problem with:
 Variables: 0
 Model mode: Automatic
 CachingOptimizer state: NoOptimizer

src/macros.jl

@@ -884,6 +884,7 @@ To minimize the value of the variable `x`, do as follows:
 ```jldoctest @objective; setup = :(using JuMP)
 julia> model = Model()
 A JuMP Model
+Feasibility problem with:
 Variables: 0
 Model mode: Automatic
 CachingOptimizer state: NoOptimizer

src/objective.jl

@@ -90,6 +90,7 @@ Error if the objective is not convertible to type `T`.
 ```jldoctest objective_function; setup = :(using JuMP)
 julia> model = Model()
 A JuMP Model
+Feasibility problem with:
 Variables: 0
 Model mode: Automatic
 CachingOptimizer state: NoOptimizer

src/print.jl

@@ -147,33 +147,91 @@ wrap_in_inline_math_mode(str) = "\$ $str \$"
 ## Model
 #------------------------------------------------------------------------
 function Base.show(io::IO, model::Model)
+    plural(n) = (n==1 ? "" : "s")
     println(io, "A JuMP Model")
+    sense = objective_sense(model)
+    if sense == MOI.MaxSense
+        print(io, "Maximization")
+    elseif sense == MOI.MinSense
+        print(io, "Minimization")
+    else
+        print(io, "Feasibility")
+    end
+    println(io, " problem with:")
     # TODO: Consider allowing a JuMP model to have a string name.
-    println(io, "Variables: ", num_variables(model))
+    println(io, "Variable", plural(num_variables(model)), ": ",
+            num_variables(model))
     # https://github.com/JuliaOpt/JuMP.jl/issues/1556
     # TODO: This doesn't account for nonlinear objectives
     # println(io, "\tObjective function type:",
     #            MOI.get(model, MOI.ObjectiveFunctionType()))
     for (F, S) in MOI.get(model, MOI.ListOfConstraints())
         num_constraints = MOI.get(model, MOI.NumberOfConstraints{F, S}())
-        println(io, "`$F`-in-`$S`: $num_constraints constraints")
+        println(io, "`$F`-in-`$S`: $num_constraints constraint",
+                plural(num_constraints))
     end
     if !iszero(num_nl_constraints(model))
-        println(io, "Nonlinear: ", num_nl_constraints(model), " constraints")
+        println(io, "Nonlinear: ", num_nl_constraints(model), " constraint",
+                plural(num_nl_constraints(model)))
     end
     model_mode = mode(model)
     println(io, "Model mode: ", model_mode)
     if model_mode == Manual || model_mode == Automatic
         println(io, "CachingOptimizer state: ",
                 MOIU.state(backend(model)))
     end
-    println(io, "Solver name: ", solver_name(model))
-    names_in_scope = collect(keys(object_dictionary(model)))
+    # The last print shouldn't have a new line
+    print(io, "Solver name: ", solver_name(model))
+    names_in_scope = sort(collect(keys(object_dictionary(model))))
     if !isempty(names_in_scope)
-        println(io, "Names registered in the model: ",
-                    join(string.(names_in_scope), ", "))
+        println(io)
+        print(io, "Names registered in the model: ",
+              join(string.(names_in_scope), ", "))
+    end
+end
+
+function Base.print(io::IO, model::Model)
+    print(io, model_string(REPLMode, model))
+end
+function Base.show(io::IO, ::MIME"text/latex", model::Model)
+    print(io, wrap_in_math_mode(model_string(IJuliaMode, model)))
+end
+function model_string(print_mode, model::Model)
+    ijl = print_mode == IJuliaMode
+    sep = ijl ? " & " : " "
+    eol = ijl ? "\\\\\n" : "\n"
+    sense = objective_sense(model)
+    str = ""
+    if sense == MOI.MaxSense
+        str *= ijl ? "\\max" : "Max"
+    elseif sense == MOI.MinSense
+        str *= ijl ? "\\min" : "Min"
+    else
+        str *= ijl ? "\\text{feasibility}" : "Feasibility"
     end
-    # TODO: The last print shouldn't have a new line
+    if sense != MOI.FeasibilitySense
+        if ijl
+            str *= "\\quad"
+        end
+        str *= sep
+        str *= function_string(print_mode,
+                               objective_function(model, QuadExpr))
+    end
+    str *= eol
+    str *= ijl ? "\\text{Subject to} \\quad" : "Subject to" * eol
+    for (F, S) in MOI.get(model, MOI.ListOfConstraints())
+        for idx in MOI.get(model, MOI.ListOfConstraintIndices{F, S}())
+            # FIXME the shape may be incorrect here
+            shape = S <: MOI.AbstractScalarSet ? ScalarShape() : VectorShape()
+            cref = ConstraintRef(model, idx, shape)
+            con = constraint_object(cref)
+            str *= sep * constraint_string(print_mode, con) * eol
+        end
+    end
+    if ijl
+        str = "\\begin{alignat*}{1}" * str * "\\end{alignat*}\n"
+    end
+    return str
 end
 
 #------------------------------------------------------------------------
@@ -307,10 +365,10 @@ end
 # `JuMP.jump_function` or `JuMP.function_string` and either `JuMP.moi_set` or
 # `JuMP.in_set_string` should be implemented.
 function Base.show(io::IO, ref::ConstraintRef)
-    print(io, constraint_string(REPLMode, name(ref), constraint_object(ref)))
+    print(io, constraint_string(REPLMode, ref))
 end
 function Base.show(io::IO, ::MIME"text/latex", ref::ConstraintRef)
-    print(io, constraint_string(IJuliaMode, name(ref), constraint_object(ref)))
+    print(io, constraint_string(IJuliaMode, ref))
 end
 
 """
@@ -403,12 +461,14 @@ function in_set_string(print_mode, constraint::AbstractConstraint)
     return in_set_string(print_mode, moi_set(constraint))
 end
 
-# constraint_object is a JuMP constraint object like AffExprConstraint.
-# Assumes a .func and .set member.
-function constraint_string(print_mode, constraint_name, constraint_object)
+function constraint_string(print_mode, constraint_object::AbstractConstraint)
     func_str = function_string(print_mode, constraint_object)
     in_set_str = in_set_string(print_mode, constraint_object)
-    constraint_without_name = func_str * " " * in_set_str
+    return func_str * " " * in_set_str
+end
+function constraint_string(print_mode, constraint_name,
+                           constraint_object::AbstractConstraint)
+    constraint_without_name = constraint_string(print_mode, constraint_object)
     if print_mode == IJuliaMode
         constraint_without_name = wrap_in_inline_math_mode(constraint_without_name)
     end
@@ -418,6 +478,9 @@ function constraint_string(print_mode, constraint_name, constraint_object)
         return constraint_name * " : " * constraint_without_name
     end
 end
+function constraint_string(print_mode, ref::ConstraintRef)
+    return constraint_string(print_mode, name(ref), constraint_object(ref))
+end
 
 #------------------------------------------------------------------------
 ## NonlinearExprData

test/old/print.jl

@@ -336,113 +336,6 @@ end
     end
 
 
-
-    @testset "Model" begin
-        le, ge, eq, fa = repl[:leq], repl[:geq], repl[:eq], repl[:for_all]
-        inset, dots = repl[:in], repl[:dots]
-        infty, union = repl[:infty], repl[:union]
-        Vert, sub2 = repl[:Vert], repl[:sub2]
-
-        #------------------------------------------------------------------
-
-        mod_1 = Model()
-        @variable(mod_1, a>=1)
-        @variable(mod_1, b<=1)
-        @variable(mod_1, -1<=c<=1)
-        @variable(mod_1, a1>=1,Int)
-        @variable(mod_1, b1<=1,Int)
-        @variable(mod_1, -1<=c1<=1,Int)
-        @variable(mod_1, x, Bin)
-        @variable(mod_1, y)
-        @variable(mod_1, z, Int)
-        @variable(mod_1, sos[1:3], Bin)
-        @variable(mod_1, 2 <= si <= 3, SemiInt)
-        @variable(mod_1, 2 <= sc <= 3, SemiCont)
-        @variable(mod_1, fi == 9)
-        @objective(mod_1, Max, a - b + 2a1 - 10x)
-        @constraint(mod_1, a + b - 10c - 2x + c1 <= 1)
-        @constraint(mod_1, a*b <= 2)
-        addSOS1(mod_1, [i*sos[i] for i in 1:3])
-        @constraint(mod_1, norm(sos) + a <= 1)
-
-        io_test(REPLMode, mod_1, """
-    Max a - b + 2 a1 - 10 x
-    Subject to
-     a + b - 10 c - 2 x + c1 $le 1
-     a*b - 2 $le 0
-     SOS1: {1 sos[1], 2 sos[2], 3 sos[3]}
-     $(Vert)[sos[1],sos[2],sos[3]]$(Vert)$(sub2) $le -a + 1
-     sos[i] $inset {0,1} $fa i $inset {1,2,3}
-     a $ge 1
-     b $le 1
-     -1 $le c $le 1
-     a1 $ge 1, integer
-     b1 $le 1, integer
-     -1 $le c1 $le 1, integer
-     x $inset {0,1}
-     y
-     z, integer
-     si $inset {2,$dots,3} $union {0}
-     sc $inset [2,3] $union {0}
-     fi = 9
-    """, repl=:print)
-
-        io_test(REPLMode, mod_1, """
-    Maximization problem with:
-     * 1 linear constraint
-     * 1 quadratic constraint
-     * 1 SOS constraint
-     * 1 SOC constraint
-     * 15 variables: 4 binary, 4 integer, 1 semicontinuous, 1 semi-integer
-    Solver is default solver""", repl=:show)
-
-        io_test(IJuliaMode, mod_1, """
-    \\begin{alignat*}{1}\\max\\quad & a - b + 2 a1 - 10 x\\\\
-    \\text{Subject to} \\quad & a + b - 10 c - 2 x + c1 \\leq 1\\\\
-     & a\\times b - 2 \\leq 0\\\\
-     & SOS1: \\{1 sos[1], 2 sos[2], 3 sos[3]\\}\\\\
-     & \\Vert[sos_{1},sos_{2},sos_{3}]\\Vert_2 $le -a + 1\\\\
-     & sos_{i} \\in \\{0,1\\} \\quad\\forall i \\in \\{1,2,3\\}\\\\
-     & a \\geq 1\\\\
-     & b \\leq 1\\\\
-     & -1 \\leq c \\leq 1\\\\
-     & a1 \\geq 1, \\in \\mathbb{Z}\\\\
-     & b1 \\leq 1, \\in \\mathbb{Z}\\\\
-     & -1 \\leq c1 \\leq 1, \\in \\mathbb{Z}\\\\
-     & x \\in \\{0,1\\}\\\\
-     & y\\\\
-     & z, \\in \\mathbb{Z}\\\\
-     & si \\in \\{2,\\dots,3\\} \\cup \\{0\\}\\\\
-     & sc \\in \\[2,3\\] \\cup \\{0\\}\\\\
-     & fi = 9\\\\
-    \\end{alignat*}
-    """)
-
-        #------------------------------------------------------------------
-
-        mod_2 = Model()
-        @variable(mod_2, x, Bin)
-        @variable(mod_2, y, Int)
-        @constraint(mod_2, x*y <= 1)
-
-        io_test(REPLMode, mod_2, """
-    Feasibility problem with:
-     * 0 linear constraints
-     * 1 quadratic constraint
-     * 2 variables: 1 binary, 1 integer
-    Solver is default solver""", repl=:show)
-
-        mod_2 = Model()
-        @variable(mod_2, x)
-        @constraint(mod_2, x <= 3)
-
-        io_test(REPLMode, mod_2, """
-    Feasibility problem with:
-     * 1 linear constraint
-     * 1 variable
-    Solver is default solver""", repl=:show)
-    end
-
     @testset "changing variable categories" begin
         le, ge, fa = repl[:leq], repl[:geq], repl[:for_all]
         inset, dots = repl[:in], repl[:dots]