Implement GNU awk strftime

CHANGELOG.md

@@ -51,6 +51,7 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ### Added
 
+* `awk` implements GNU-compatible `strftime([format [, timestamp [, utc-flag]]])` arguments and defaults
 * `mix bash_fixtures.gen printf` enumerates a printf conversion × flag/width/precision matrix (`printf_matrix`) — #70 item 2
 * `mix bash_fixtures.gen test` enumerates a `test`/`[` operator × revealing-shape matrix (`test_matrix`) — #70 item 2
 * `mix bash_fixtures.gen varop` enumerates a `${var op word}` parameter-expansion matrix (`varop_matrix`) — #70 item 2

lib/just_bash/commands/awk/evaluator.ex

@@ -7,9 +7,12 @@ defmodule JustBash.Commands.Awk.Evaluator do
   """
 
   alias JustBash.Commands.Awk.{AST, Formatter}
+  alias JustBash.Commands.Date, as: DateCommand
   alias JustBash.FS
   alias JustBash.Limit
 
+  @default_strftime_format "%a %b %e %H:%M:%S %Z %Y"
+
   @type state :: %{
           nr: non_neg_integer(),
           nf: non_neg_integer(),
@@ -1607,6 +1610,27 @@ defmodule JustBash.Commands.Awk.Evaluator do
     Formatter.format_printf(format, args)
   end
 
+  defp evaluate_function("strftime", [], state) do
+    format =
+      state.arrays
+      |> Map.get("PROCINFO", %{})
+      |> Map.get("strftime", @default_strftime_format)
+
+    format_strftime(format, DateTime.utc_now(), false)
+  end
+
+  defp evaluate_function("strftime", [format], _state) do
+    format_strftime(format, DateTime.utc_now(), false)
+  end
+
+  defp evaluate_function("strftime", [format, timestamp], _state) do
+    format_strftime(format, timestamp, false)
+  end
+
+  defp evaluate_function("strftime", [format, timestamp, utc_flag], _state) do
+    format_strftime(format, timestamp, truthy?(utc_flag))
+  end
+
   # Math functions
   defp evaluate_function("int", [arg], _state) do
     parse_number(arg) |> trunc()
@@ -1664,6 +1688,29 @@ defmodule JustBash.Commands.Awk.Evaluator do
 
   defp evaluate_function(_name, _args, _state), do: ""
 
+  defp format_strftime(format, %DateTime{} = datetime, utc?) do
+    datetime
+    |> strftime_datetime(utc?)
+    |> DateCommand.format_datetime(to_string(format))
+  end
+
+  defp format_strftime(format, timestamp, utc?) do
+    case timestamp |> parse_number() |> trunc() |> DateTime.from_unix() do
+      {:ok, datetime} ->
+        datetime
+        |> strftime_datetime(utc?)
+        |> DateCommand.format_datetime(to_string(format))
+
+      {:error, :invalid_unix_time} ->
+        ""
+    end
+  end
+
+  # GNU awk names the forced UTC zone "GMT". Without the flag, JustBash uses
+  # its UTC sandbox clock and keeps the normal "UTC" zone name.
+  defp strftime_datetime(datetime, true), do: %{datetime | zone_abbr: "GMT"}
+  defp strftime_datetime(datetime, false), do: datetime
+
   # Format a value for output - integers print without .0
   # Format array keys: 0.0 -> "0", 1.0 -> "1", "1.0" -> "1", etc.
   defp format_array_key(value) when is_float(value) do

lib/just_bash/commands/date.ex

@@ -528,7 +528,9 @@ defmodule JustBash.Commands.Date do
   #
   # `yr_spec` is the padding flag a compound conversion forwards to the year
   # fields of its sub-format, and nil at the top level. See `yearish/3`.
-  defp format_datetime(datetime, format), do: scan(format, datetime, nil, [])
+  @doc false
+  @spec format_datetime(DateTime.t(), binary()) :: binary()
+  def format_datetime(datetime, format), do: scan(format, datetime, nil, [])
 
   defp scan(<<>>, _datetime, _yr_spec, acc), do: acc |> Enum.reverse() |> IO.iodata_to_binary()
 

test/commands/awk_test.exs

@@ -483,6 +483,75 @@ defmodule JustBash.Commands.AwkTest do
     end
   end
 
+  describe "strftime() function" do
+    test "formats explicit timestamps in local UTC and forced UTC" do
+      bash = JustBash.new()
+
+      {result, _} =
+        JustBash.exec(
+          bash,
+          ~s|TZ=UTC awk 'BEGIN { print strftime("%F %T %Z %z", 0); print strftime("%F %T %Z %z", 1718458200, 1) }'|
+        )
+
+      assert result.stdout ==
+               "1970-01-01 00:00:00 UTC +0000\n2024-06-15 13:30:00 GMT +0000\n"
+
+      assert result.exit_code == 0
+    end
+
+    test "uses the current time and documented format defaults" do
+      before = DateTime.utc_now() |> DateTime.to_unix()
+
+      {result, _} =
+        JustBash.exec(
+          JustBash.new(),
+          ~s|TZ=UTC awk 'BEGIN { print strftime(); print strftime("%s") }'|
+        )
+
+      after_time = DateTime.utc_now() |> DateTime.to_unix()
+      [default_time, timestamp, ""] = String.split(result.stdout, "\n")
+
+      assert default_time =~
+               ~r/^(Mon|Tue|Wed|Thu|Fri|Sat|Sun) (Jan|Feb|Mar|Apr|May|Jun|Jul|Aug|Sep|Oct|Nov|Dec) [ 0-9][0-9] [0-9]{2}:[0-9]{2}:[0-9]{2} UTC [0-9]{4}$/
+
+      assert String.to_integer(timestamp) in before..after_time
+      assert result.exit_code == 0
+    end
+
+    test "uses PROCINFO strftime as the default format" do
+      {result, _} =
+        JustBash.exec(
+          JustBash.new(),
+          ~s|TZ=UTC awk 'BEGIN { PROCINFO["strftime"] = "%F"; print length(strftime()) }'|
+        )
+
+      assert result.stdout == "10\n"
+      assert result.exit_code == 0
+    end
+
+    test "coerces invalid timestamp text to the epoch" do
+      {result, _} =
+        JustBash.exec(
+          JustBash.new(),
+          ~s|awk 'BEGIN { print strftime("%F", "not-a-timestamp", 1) }'|
+        )
+
+      assert result.stdout == "1970-01-01\n"
+      assert result.exit_code == 0
+    end
+
+    test "returns an empty string for an out-of-range timestamp or empty format" do
+      {result, _} =
+        JustBash.exec(
+          JustBash.new(),
+          ~s|awk 'BEGIN { print strftime("%Y", 1e100, 1); print strftime("", 0, 1) }'|
+        )
+
+      assert result.stdout == "\n\n"
+      assert result.exit_code == 0
+    end
+  end
+
   describe "gsub and sub" do
     test "gsub replaces all occurrences" do
       bash = JustBash.new(files: %{"/data.txt" => "hello world\n"})

test/fixtures/bash_cases/awk.json

@@ -530,6 +530,21 @@
       "name": "awk edge cases comparison: awk string coercion in arithmetic",
       "script": "echo '10abc 5' | awk '{print $1 + $2}'",
       "content_hash": "587a86f56c3a1f53"
+    },
+    {
+      "name": "awk strftime comparison: explicit timestamp and UTC flag",
+      "script": "TZ=UTC LC_ALL=C awk 'BEGIN { print strftime(\"%F %T %Z %z\", 0); print strftime(\"%F %T %Z %z\", 1718458200, 1) }'",
+      "content_hash": "4b852164672823b4"
+    },
+    {
+      "name": "awk strftime comparison: invalid timestamp inputs and empty format",
+      "script": "TZ=UTC LC_ALL=C awk 'BEGIN { print strftime(\"%F\", \"not-a-timestamp\", 1); print strftime(\"%Y\", 1e100, 1); print strftime(\"\", 0, 1) }'",
+      "content_hash": "4ae0bc0e390c8664"
+    },
+    {
+      "name": "awk strftime comparison: omitted arguments use documented defaults",
+      "script": "TZ=UTC LC_ALL=C awk 'BEGIN { value = strftime(); print length(value), (index(value, \" UTC \") > 0); print length(strftime(\"%F\")); PROCINFO[\"strftime\"] = \"%Y\"; print length(strftime()) }'",
+      "content_hash": "f8b630f4b23a831d"
     }
   ]
 }

test/fixtures/bash_expected/awk.json

@@ -741,6 +741,27 @@
       "name": "awk edge cases comparison: awk string coercion in arithmetic",
       "stderr": "",
       "stdout": "15\n"
+    },
+    {
+      "content_hash": "4b852164672823b4",
+      "exit_code": 0,
+      "name": "awk strftime comparison: explicit timestamp and UTC flag",
+      "stderr": "",
+      "stdout": "1970-01-01 00:00:00 UTC +0000\n2024-06-15 13:30:00 GMT +0000\n"
+    },
+    {
+      "content_hash": "4ae0bc0e390c8664",
+      "exit_code": 0,
+      "name": "awk strftime comparison: invalid timestamp inputs and empty format",
+      "stderr": "",
+      "stdout": "1970-01-01\n\n\n"
+    },
+    {
+      "content_hash": "f8b630f4b23a831d",
+      "exit_code": 0,
+      "name": "awk strftime comparison: omitted arguments use documented defaults",
+      "stderr": "",
+      "stdout": "28 1\n10\n4\n"
     }
   ],
   "suite": "awk"