Merge pull request #114 from MineralsCloud:TimedItem

Deprecate `parse_clock`, use `parse(TimedItem)`
MineralsCloud · Oct 26, 2023 · 6bf3998 · 6bf3998
2 parents 417e75c + 4e91ca8
commit 6bf3998
Show file tree

Hide file tree

Showing 3 changed files with 48 additions and 58 deletions.
diff --git a/Project.toml b/Project.toml
@@ -7,6 +7,7 @@ version = "0.4.0"
 AbInitioSoftwareBase = "df5135bc-470e-46c6-b451-292e27ca5b84"
 CrystallographyBase = "93b1d1cd-a8ea-4aa5-adb1-b2407ea0ba8d"
 DataFrames = "a93c6f00-e57d-5684-b7b6-d8193f3e46c0"
+Dates = "ade2ca70-3891-5945-98fb-dc099432e06a"
 PyFortran90Namelists = "e44308e6-bd5b-11e9-2850-49daf8f1ec40"
 QuantumESPRESSOBase = "51b62caa-b28f-11e9-38c2-1f67cb498e05"
 ReadableRegex = "cbbcb084-453d-4c4c-b292-e315607ba6a4"

diff --git a/src/PWscf/output.jl b/src/PWscf/output.jl
@@ -1,14 +1,20 @@
-# using Dates: DateTime, DateFormat
+using Dates: Hour, Minute, Millisecond
 using DataFrames: AbstractDataFrame, DataFrame, groupby
 using QuantumESPRESSOBase.PWscf
 using VersionParsing: vparse
 
+export TimedItem
+
 struct SubroutineError
     name::String
     cerr::String
     msg::String
 end
 
+struct ParseError <: Exception
+    msg::String
+end
+
 export Diagonalization,
     Preamble,
     Davidson,
@@ -29,7 +35,6 @@ export Diagonalization,
     parse_fft_dimensions,
     parse_iteration_head,
     parse_electrons_energies,
-    parse_clock,
     parse_input_name,
     isoptimized,
     isjobdone,
@@ -391,39 +396,39 @@ function parse_fft_dimensions(str::AbstractString)::Maybe{NamedTuple}
     return (; zip((:ng, :nr1, :nr2, :nr3), parsed)...)
 end # function parse_fft_dimensions
 
-function parse_clock(str::AbstractString)::Maybe{AbstractDataFrame}
-    m = match(TIME_BLOCK, str)
-    m === nothing && return nothing
-    content = only(m.captures)
+struct TimedItem
+    name::String
+    cpu::Millisecond
+    wall::Millisecond
+    calls::Maybe{UInt64}
+end
 
-    info = DataFrame(;
-        subroutine=String[], item=String[], CPU=Float64[], wall=Float64[], calls=Int[]
-    )
-    for regex in [
-        SUMMARY_TIME_BLOCK
-        INIT_RUN_TIME_BLOCK
-        ELECTRONS_TIME_BLOCK
-        C_BANDS_TIME_BLOCK
-        SUM_BAND_TIME_BLOCK
-        EGTERG_TIME_BLOCK
-        H_PSI_TIME_BLOCK
-        GENERAL_ROUTINES_TIME_BLOCK
-        PARALLEL_ROUTINES_TIME_BLOCK
-    ]
-        block = match(regex, content)
-        if block !== nothing
-            for m in eachmatch(TIME_ITEM, block[:body])
-                push!(
-                    info,
-                    [block[:head] m[1] map(x -> parse(Float64, x), m.captures[2:4])...],
-                )
-            end
-        end
+function Base.parse(::Type{TimedItem}, str::AbstractString)
+    matched = match(TIMED_ITEM, str)
+    if isnothing(matched)
+        nothing
+    else
+        name, cpu, wall = matched[1], parsetime(matched[2]), parsetime(matched[3])
+        return TimedItem(
+            name, cpu, wall, isnothing(matched[4]) ? nothing : parse(UInt64, matched[5])
+        )
     end
-    # m = match(TERMINATED_DATE, content)
-    # info["terminated date"] = parse(DateTime, m.captures[1], DateFormat("H:M:S"))
-    return info
-end # function parse_clock
+end
+
+function parsetime(str::AbstractString)
+    compound = match(r"(\d+)h\s*(\d+)m", str)
+    seconds = match(r"(\d+\.\d{2})s", str)
+    if isnothing(seconds) && !isnothing(compound)
+        hours = parse(UInt64, compound[1])
+        minutes = parse(UInt64, compound[2])
+        return convert(Millisecond, Hour(hours) + Minute(minutes))
+    elseif isnothing(compound) && !isnothing(seconds)
+        seconds = parse(Float64, seconds[1])
+        return Millisecond(1000seconds)
+    else
+        throw(ParseError("unrecognized time format!"))
+    end
+end
 
 function parse_input_name(str::AbstractString)
     m = match(READING_INPUT_FROM, str)

diff --git a/src/PWscf/regexes.jl b/src/PWscf/regexes.jl
@@ -200,31 +200,15 @@ const CONVERGED_ELECTRONS_ENERGY = Regex(
     "m",
 )
 const TIME_BLOCK = r"(init_run\X+?This run was terminated on:.*)"
-# This format is from https://github.com/QEF/q-e/blob/4132a64/PW/src/print_clock_pw.f90#L29-L33.
-const SUMMARY_TIME_BLOCK = r"""
-(?<head>)
-(?<body>
-init_run\s+:.*
-\s*electrons\s+:.*
-\s*(?:update_pot\s+.*)?  # This does not always exist.
-\s*(?:forces\s+:.*)?     # This does not always exist.
-\s*(?:stress\s+:.*)?     # This does not always exist.
-)
-"""mx
-const TIME_ITEM = Regex(
-    "\\s*([\\w0-9:]+)\\s+:\\s*$(FIXED_POINT_REAL)s\\sCPU\\s*$(FIXED_POINT_REAL)s\\sWALL\\s\\(\\s*$INTEGER\\scalls\\)",
-)
-# This format is from https://github.com/QEF/q-e/blob/4132a64/PW/src/print_clock_pw.f90#L35-L36.
-const INIT_RUN_TIME_BLOCK = r"Called by (?<head>init_run):(?<body>\X+?)^\s*$"m
-# This format is from https://github.com/QEF/q-e/blob/4132a64/PW/src/print_clock_pw.f90#L53-L54.
-const ELECTRONS_TIME_BLOCK = r"Called by (?<head>electrons):(?<body>\X+?)^\s*$"m
-# This format is from https://github.com/QEF/q-e/blob/4132a64/PW/src/print_clock_pw.f90#L78-L79.
-const C_BANDS_TIME_BLOCK = r"Called by (?<head>c_bands):(?<body>\X+?)^\s*$"m
-const SUM_BAND_TIME_BLOCK = r"Called by (?<head>sum_band):(?<body>\X+?)^\s*$"m
-const EGTERG_TIME_BLOCK = r"Called by (?<head>\*egterg):(?<body>\X+?)^\s*$"m
-const H_PSI_TIME_BLOCK = r"Called by (?<head>h_psi):(?<body>\X+?)^\s*$"m
-const GENERAL_ROUTINES_TIME_BLOCK = r"(?<head>General routines)(?<body>\X+?)^\s*$"m
-const PARALLEL_ROUTINES_TIME_BLOCK = r"(?<head>Parallel routines)(?<body>\X+?)^\s*$"m
+const TIME_FORMAT = r"(\d+h\s*\d+m|\d+\.\d{2}s)"
+const TIMED_ITEM =
+    r"([\w:]+)\s*:\s*" *
+    TIME_FORMAT *
+    r"\s*CPU\s*" *
+    TIME_FORMAT *
+    r"\s*WALL" *
+    r"(\s*\(\s*(\d+)\s*calls\))?" *
+    r"$"  # Match the last row
 const TERMINATED_DATE = r"This run was terminated on:(.+)"  # TODO: Date
 const JOB_DONE = r"JOB DONE\."
 # These formats are from https://github.com/QEF/q-e/blob/4132a64/UtilXlib/error_handler.f90#L48-L68.