changes to adapt to compressed line table format (#526)

Co-authored-by: Shuhei Kadowaki <aviatesk@gmail.com>

Author: vtjnash

TypedSyntax/src/node.jl

@@ -76,7 +76,8 @@ function TypedSyntaxNode(rootnode::SyntaxNode, src::CodeInfo, mappings, symtyps)
     # There may be ambiguous assignments back to the source; preserve just the unambiguous ones
     node2ssa = IdDict{SyntaxNode,Int}(only(list) => i for (i, list) in pairs(mappings) if length(list) == 1)
     # Copy `rootnode`, adding type annotations
-    trootnode = TypedSyntaxNode(nothing, nothing, TypedSyntaxData(rootnode.data::SyntaxData, src, gettyp(node2ssa, rootnode, src)))
+    typ = gettyp(node2ssa, rootnode, src)
+    trootnode = TypedSyntaxNode(nothing, nothing, TypedSyntaxData(rootnode.data::SyntaxData, src, typ))
     addchildren!(trootnode, rootnode, src, node2ssa, symtyps, mappings)
     # Add argtyps to signature
     fnode = get_function_def(trootnode)
@@ -439,6 +440,76 @@ function collect_symbol_nodes!(symlocs::AbstractDict, node)
     return symlocs
 end
 
+## utility function to extract the line number at a particular program counter (ignoring inlining).
+## return <= 0 if there is no line number change caused by this statement
+@static if VERSION ≥ v"1.12.0-DEV.173"
+function getline(lt::Core.DebugInfo, i::Int)
+    while true
+        codeloc = Base.IRShow.getdebugidx(lt, i)
+        line::Int = codeloc[1]
+        line < 0 && return 0 # broken or disabled debug info?
+        line == 0 && return 0 # no line number update (though maybe inlining changed)
+        ltnext = lt.linetable
+        if ltnext === nothing
+            return line
+        end
+        i = line
+        lt = ltnext
+    end
+end
+function getnextline(lt::Core.DebugInfo, i::Int, Δline)
+    while true
+        codeloc = Base.IRShow.getdebugidx(lt, i)
+        line::Int = codeloc[1]
+        line < 0 && return typemax(Int) # broken or disabled debug info?
+        if line == 0
+            i += 1
+            continue
+        end
+        ltnext = lt.linetable
+        if ltnext === nothing
+            break
+        end
+        i = line
+        lt = ltnext
+    end
+    # now that we have line i and a list of all lines with code on them lt
+    # find the next largest line number in this list greater than i, or return typemax(Int)
+    j = i+1
+    currline = Base.IRShow.getdebugidx(lt, i)[1]
+    while j ≤ typemax(Int)
+        codeloc = Base.IRShow.getdebugidx(lt, j)
+        line::Int = codeloc[1]
+        line < 0 && break
+        if line == 0 || currline == line
+            j += 1
+        else
+            return line + Δline
+        end
+    end
+    return typemax(Int)
+end
+
+else # VERSION < v"1.12.0-DEV.173"
+function getline(lt, j)
+    linfo = (j == 0 ? first(lt) : lt[j])::Core.LineInfoNode
+    linfo.inlined_at == 0 && return linfo.line
+    @assert linfo.method === Symbol("macro expansion")
+    linfo = lt[linfo.inlined_at]::Core.LineInfoNode
+    return linfo.line
+end
+function getnextline(lt, j, Δline)
+    j == 0 && return typemax(Int)
+    j += 1
+    while j <= length(lt)
+        linfo = lt[j]::Core.LineInfoNode
+        linfo.inlined_at == 0 && return linfo.line + Δline
+        j += 1
+    end
+    return typemax(Int)
+end
+end # @static if
+
 # Main logic for mapping `src.code[i]` to node(s) in the SyntaxNode tree
 # Success: when we map it to a unique node
 # Δline is the (Revise) offset of the line number
@@ -471,8 +542,13 @@ function map_ssas_to_source(src::CodeInfo, mi::MethodInstance, rootnode::SyntaxN
     # Append (to `mapped`) all nodes in `targets` that are consistent with the line number of the `i`th stmt
     # (Essentially `copy!(mapped, filter(predicate, targets))`)
     function append_targets_for_line!(mapped#=::Vector{nodes}=#, i::Int, targets#=::Vector{nodes}=#)
-        j = src.codelocs[i]
-        lt = src.linetable::Vector
+        @static if VERSION ≥ v"1.12.0-DEV.173"
+            j = i
+            lt = src.debuginfo
+        else
+            j = src.codelocs[i]
+            lt = src.linetable::Vector
+        end
         start = getline(lt, j) + Δline
         stop = getnextline(lt, j, Δline) - 1
         linerange = start : stop
@@ -853,25 +929,6 @@ function symloc_key(sym::Symbol)
     return sym
 end
 
-function getline(lt, j)
-    linfo = (j == 0 ? first(lt) : lt[j])::Core.LineInfoNode
-    linfo.inlined_at == 0 && return linfo.line
-    @assert linfo.method === Symbol("macro expansion")
-    linfo = lt[linfo.inlined_at]::Core.LineInfoNode
-    return linfo.line
-end
-
-function getnextline(lt, j, Δline)
-    j == 0 && return typemax(Int)
-    j += 1
-    while j <= length(lt)
-        linfo = lt[j]::Core.LineInfoNode
-        linfo.inlined_at == 0 && return linfo.line + Δline
-        j += 1
-    end
-    return typemax(Int)
-end
-
 function find_identifier_or_tuplechild(node::AbstractSyntaxNode, sym)
     kind(node) == K"Identifier" && node.val === sym && return node, true
     if kind(node) == K"tuple"   # tuple destructuring

src/codeview.jl

@@ -127,9 +127,6 @@ function cthulhu_typed(io::IO, debuginfo::Symbol,
             # We empty the body when filling kwargs
             istruncated = isempty(children(body))
             idxend = istruncated ? JuliaSyntax.last_byte(sig) : lastindex(tsn.source)
-            if any(iszero, src.codelocs)
-                @warn "Some line information is missing, type-assignment may be incomplete"
-            end
             if src.slottypes === nothing
                 @warn "Inference terminated in an incomplete state due to argument-type changes during recursion"
             end

src/interpreter.jl

@@ -153,7 +153,8 @@ function CC.src_inlining_policy(interp::CthulhuInterpreter,
             src::Any, info::CCCallInfo, stmt_flag::UInt32)
     end
 end
-CC.retrieve_ir_for_inlining(cached_result::CodeInstance, src::OptimizedSource) = CC.copy(src.ir)
+CC.retrieve_ir_for_inlining(cached_result::CodeInstance, src::OptimizedSource) =
+    CC.retrieve_ir_for_inlining(cached_result.def, src.ir::IRCode, true)
 CC.retrieve_ir_for_inlining(mi::MethodInstance, src::OptimizedSource, preserve_local_sources::Bool) =
     CC.retrieve_ir_for_inlining(mi, src.ir, preserve_local_sources)
 elseif VERSION ≥ v"1.11.0-DEV.879"

src/reflection.jl

@@ -309,7 +309,7 @@ function find_caller_of(interp::AbstractInterpreter, callee::MethodInstance, cal
         callsites, _ = find_callsites(interp′, src, infos, caller, slottypes, optimize)
         callsites = allow_unspecialized ? filter(cs->maybe_callsite(cs, callee), callsites) :
                                           filter(cs->is_callsite(cs, callee), callsites)
-        foreach(cs -> add_sourceline!(locs, src, cs.id), callsites)
+        foreach(cs -> add_sourceline!(locs, src, cs.id, caller), callsites)
     end
     # Consolidate by method, but preserve the order
     prlookup = Dict{Tuple{Symbol,Symbol},Int}()
@@ -335,17 +335,25 @@ function find_caller_of(interp::AbstractInterpreter, callee::MethodInstance, cal
     return ulocs
 end
 
-function add_sourceline!(locs, CI, stmtidx::Int)
-    if isa(CI, IRCode)
-        stack = Base.IRShow.compute_loc_stack(CI.linetable, CI.stmts.line[stmtidx])
+function add_sourceline!(locs::Vector{Tuple{Core.LineInfoNode,Int}}, src::Union{CodeInfo,IRCode}, stmtidx::Int, caller::MethodInstance)
+    @static if VERSION ≥ v"1.12.0-DEV.173"
+    stack = Base.IRShow.buildLineInfoNode(src.debuginfo, caller, stmtidx)
+    for (i, di) in enumerate(stack)
+        loc = Core.LineInfoNode(Main, di.method, di.file, di.line, zero(Int32))
+        push!(locs, (loc, i-1))
+    end
+    else # VERSION < v"1.12.0-DEV.173"
+    if isa(src, IRCode)
+        stack = Base.IRShow.compute_loc_stack(src.linetable, src.stmts.line[stmtidx])
         for (i, idx) in enumerate(stack)
-            line = CI.linetable[idx]
+            line = src.linetable[idx]
             line.line == 0 && continue
-            push!(locs, (CI.linetable[idx], i-1))
+            push!(locs, (src.linetable[idx], i-1))
         end
     else
-        push!(locs, (CI.linetable[CI.codelocs[stmtidx]], 0))
+        push!(locs, (src.linetable[src.codelocs[stmtidx]], 0))
     end
+    end # @static if
     return locs
 end
 

test/test_Cthulhu.jl

@@ -814,7 +814,6 @@ end
     @test lines == [line2]
 end
 
-
 @testset "ascend interface" begin
     m = Module()
     @eval m begin