fix: improve postgres-style interval parsing (#177)

Inspired by some examples in #176
electric-sql · Jul 24, 2024 · bbb377e · bbb377e
1 parent 04f305f
commit bbb377e
Show file tree

Hide file tree

Showing 2 changed files with 70 additions and 19 deletions.
diff --git a/.changeset/old-lamps-hear.md b/.changeset/old-lamps-hear.md
@@ -0,0 +1,5 @@
+---
+"@core/sync-service": patch
+---
+
+fix: correctly parse larger set of Postgres intervals with signs
diff --git a/packages/sync-service/lib/pg_interop/interval/postgres_and_sql_parser.ex b/packages/sync-service/lib/pg_interop/interval/postgres_and_sql_parser.ex
@@ -5,24 +5,30 @@ defmodule PgInterop.Interval.PostgresAndSQLParser do
   alias PgInterop.Interval
 
   @parse_part_regexes [
-    unmarked_end: ~r/(?<=\s|^)(?<second>\d+(?:\.\d+)?)(?=\s*$)/,
-    microsecond: ~r/(?<=\s|^)(?<microsecond>\d+(?:\.\d+)?)\s*(?:us|usecs?|microseconds?)(?=\s|$)/,
-    millisecond: ~r/(?<=\s|^)(?<millisecond>\d+(?:\.\d+)?)\s*(?:ms|msecs?|milliseconds?)(?=\s|$)/,
-    second: ~r/(?<=\s|^)(?<second>\d+(?:\.\d+)?)\s*(?:s|secs?|seconds?)(?=\s|$)/,
-    minute: ~r/(?<=\s|^)(?<minute>\d+(?:\.\d+)?)\s*(?:m|mins?|minutes?)(?=\s|$)/,
-    hour: ~r/(?<=\s|^)(?<hour>\d+(?:\.\d+)?)\s*(?:h|hours?)(?=\s|$)/,
-    day: ~r/(?<=\s|^)(?<day>\d+(?:\.\d+)?)\s*(?:d|days?)(?=\s|$)/,
-    week: ~r/(?<=\s|^)(?<week>\d+(?:\.\d+)?)\s*(?:w|weeks?)(?=\s|$)/,
-    month: ~r/(?<=\s|^)(?<month>\d+(?:\.\d+)?)\s*(?:m|mons?|months?)(?=\s|$)/,
-    year: ~r/(?<=\s|^)(?<year>\d+(?:\.\d+)?)\s*(?:y|years?)(?=\s|$)/,
-    decade: ~r/(?<=\s|^)(?<decade>\d+(?:\.\d+)?)\s*(?:decs?|decades?)(?=\s|$)/,
-    century: ~r/(?<=\s|^)(?<century>\d+(?:\.\d+)?)\s*(?:c|cent|century|centuries)(?=\s|$)/,
-    millennium: ~r/(?<=\s|^)(?<millennium>\d+(?:\.\d+)?)\s*(?:mils|millenniums)(?=\s|$)/,
-    sql_ym: ~r/(?<=\s|^)(?<year>\d+)-(?<month>\d+)(?=\s|$)/,
-    sql_dhm: ~r/(?<=\s|^)(?<day>\d+(?:\.\d+)?)?\s+(?<hour>\d+):(?<minute>\d+)(?=\s|$)/,
+    unmarked_end: ~r/(?<=\s|^)(?<second>(?:\+|-)?\s*\d+(?:\.\d+)?)(?=\s*$)/,
+    microsecond:
+      ~r/(?<=\s|^)(?<microsecond>(?:\+|-)?\s*\d+(?:\.\d+)?)\s*(?:us|usecs?|microseconds?)(?=\s|$)/,
+    millisecond:
+      ~r/(?<=\s|^)(?<millisecond>(?:\+|-)?\s*\d+(?:\.\d+)?)\s*(?:ms|msecs?|milliseconds?)(?=\s|$)/,
+    second: ~r/(?<=\s|^)(?<second>(?:\+|-)?\s*\d+(?:\.\d+)?)\s*(?:s|secs?|seconds?)(?=\s|$)/,
+    minute: ~r/(?<=\s|^)(?<minute>(?:\+|-)?\s*\d+(?:\.\d+)?)\s*(?:m|mins?|minutes?)(?=\s|$)/,
+    hour: ~r/(?<=\s|^)(?<hour>(?:\+|-)?\s*\d+(?:\.\d+)?)\s*(?:h|hours?)(?=\s|$)/,
+    day: ~r/(?<=\s|^)(?<day>(?:\+|-)?\s*\d+(?:\.\d+)?)\s*(?:d|days?)(?=\s|$)/,
+    week: ~r/(?<=\s|^)(?<week>(?:\+|-)?\s*\d+(?:\.\d+)?)\s*(?:w|weeks?)(?=\s|$)/,
+    month: ~r/(?<=\s|^)(?<month>(?:\+|-)?\s*\d+(?:\.\d+)?)\s*(?:m|mons?|months?)(?=\s|$)/,
+    year: ~r/(?<=\s|^)(?<year>(?:\+|-)?\s*\d+(?:\.\d+)?)\s*(?:y|years?)(?=\s|$)/,
+    decade: ~r/(?<=\s|^)(?<decade>(?:\+|-)?\s*\d+(?:\.\d+)?)\s*(?:decs?|decades?)(?=\s|$)/,
+    century:
+      ~r/(?<=\s|^)(?<century>(?:\+|-)?\s*\d+(?:\.\d+)?)\s*(?:c|cent|century|centuries)(?=\s|$)/,
+    millennium:
+      ~r/(?<=\s|^)(?<millennium>(?:\+|-)?\s*\d+(?:\.\d+)?)\s*(?:mils|millenniums)(?=\s|$)/,
+    sql_ym: ~r/(?<=\s|^)(?<sign>\+|-)?\s*(?<year>\d+)-(?<month>\-?\d+)(?=\s|$)/,
+    sql_dhm:
+      ~r/(?<=\s|^)(?<day>(?:\+|-)?\s*\d+(?:\.\d+)?)?\s+(?<sign>\+|-)?\s*(?<hour>\d+):(?<minute>\d+)(?=\s|$)/,
     sql_dhms:
-      ~r/(?<=\s|^)(?<day>\d+(?:\.\d+)?)?\s+(?<hour>\d+):(?<minute>\d+):(?<second>\d+(?:\.\d+)?)(?=\s|$)/,
-    sql_dms: ~r/(?<=\s|^)(?<day>\d+(?:\.\d+)?)?\s+(?<minute>\d+):(?<second>\d+\.\d+)(?=\s|$)/
+      ~r/(?<=\s|^)(?<day>(?:\+|-)?\s*\d+(?:\.\d+)?)?\s+(?<sign>\+|-)?\s*(?<hour>\d+):(?<minute>\d+):(?<second>\d+(?:\.\d+)?)(?=\s|$)/,
+    sql_dms:
+      ~r/(?<=\s|^)(?<day>(?:\+|-)?\s*\d+(?:\.\d+)?)?\s+(?<sign>\+|-)?\s*(?<minute>\d+):(?<second>\d+\.\d+)(?=\s|$)/
   ]
 
   @doc """
@@ -37,24 +43,39 @@ defmodule PgInterop.Interval.PostgresAndSQLParser do
       iex> parse("@ 1-2")
       {:ok, Interval.parse!("P1Y2M")}
 
+      iex> parse("-1-2 +3 -4:05:06")
+      {:ok, Interval.parse!("P-1Y-2M3DT-4H-5M-6S")}
+
+      iex> parse("-1-2 -5:10.1")
+      {:ok, Interval.parse!("P-1Y-2MT-5M-10.1S")}
+
       iex> parse("3 4:05:06")
       {:ok, Interval.parse!("P3DT4H5M6S")}
 
       iex> parse("1 year 2 months 3 days 4 hours 5 minutes 6 seconds")
       {:ok, Interval.parse!("P1Y2M3DT4H5M6S")}
 
-      iex> parse("1 year 2-1 3 days 2")
+      iex> parse("1 year 2-1 3 days +2")
       {:ok, Interval.parse!("P3Y1M3DT2S")}
 
+      iex> parse("1 year 2-1 3 days -2")
+      {:ok, Interval.parse!("P3Y1M3DT-2S")}
+
       iex> parse("1 year 2-1 3 days 2:2")
       {:ok, Interval.parse!("P3Y1M3DT2H2M")}
 
       iex> parse("1 year 2-1 3 days 2.2")
       {:ok, Interval.parse!("P3Y1M3DT2.2S")}
 
+      iex> parse("-1-2 -5:10.1")
+      {:ok, Interval.parse!("P-1Y-2MT-5M-10.1S")}
+
       iex> parse("1.3 cent 100-11 10.3")
       {:ok, Interval.parse!("P230Y11MT10.3S")}
 
+      iex> parse("-   1 year -2 mons +3 days -  04:05:06")
+      {:ok, Interval.parse!("P-1Y-2M3DT-4H-5M-6S")}
+
       iex> parse("0.1 mils 1 cent 1 decade 1 year 1 month 1 week 1 day 1 hour 1 minute 1 second 1000 ms 1000000 us")
       {:ok, Interval.parse!("P211Y1M8DT1H1M3S")}
 
@@ -92,6 +113,7 @@ defmodule PgInterop.Interval.PostgresAndSQLParser do
       with :ok <- validate_parts(Map.new(parsed_parts)) do
         {:ok,
          parsed_parts
+         |> Enum.map(&apply_sign/1)
          |> Keyword.values()
          |> Enum.reject(&is_nil/1)
          |> Enum.flat_map(&Enum.map(&1, fn {k, v} -> build_duration(k, parse_float!(v)) end))
@@ -102,10 +124,34 @@ defmodule PgInterop.Interval.PostgresAndSQLParser do
     end
   end
 
+  defp apply_sign({k, %{"sign" => sign} = v}) when sign == "+" or sign == "",
+    do: {k, Map.delete(v, "sign")}
+
+  defp apply_sign({k, %{"sign" => "-"} = v}) do
+    v =
+      v
+      |> Map.delete("sign")
+      |> Map.new(fn
+        {"day", _} = day -> day
+        {part, val} -> {part, "-" <> val}
+      end)
+
+    {k, v}
+  end
+
+  defp apply_sign(no_sign), do: no_sign
+
   defp parse_float!(""), do: 0
 
   defp parse_float!(v) do
-    {float, ""} = Float.parse(v)
+    prepared =
+      case v do
+        "+" <> number -> String.trim(number)
+        "-" <> number -> "-" <> String.trim(number)
+        number -> String.trim(number)
+      end
+
+    {float, ""} = Float.parse(prepared)
     float
   end