monkey patch for over, #27

pkg/tests/window-functions.R

@@ -1,4 +1,37 @@
 
+over  =
+  function (expr, partition = NULL, order = NULL, frame = NULL)
+  {
+    args <- (!is.null(partition)) + (!is.null(order)) + (!is.null(frame))
+    if (args == 0) {
+      stop("Must supply at least one of partition, order, frame",
+           call. = FALSE)
+    }
+    if (!is.null(partition)) {
+      partition <- build_sql("PARTITION BY ",sql_vector(partition,
+                                                         collapse = ", ",  parens = FALSE))
+    }
+    if (!is.null(order)) {
+      order <- build_sql("ORDER BY ", sql_vector(order, collapse = ", ", parens = FALSE))
+    }
+    if (!is.null(frame)) {
+      if (is.numeric(frame))
+        frame <- rows(frame[1], frame[2])
+      frame <- build_sql("ROWS ", frame)
+    }
+    over <- sql_vector(compact(list(partition, order, frame)),
+                       parens = TRUE)
+    build_sql(expr, " OVER ", over)
+  }
+
+environment (over) = environment(select_)
+
+assignInNamespace(
+  x = "over",
+  ns = "dplyr",
+  value = over)
+
+
 library(Lahman)
 batting = copy_to(my_db, Batting)
 batting <- select(batting, playerid, yearid, teamid, g, ab:h)
@@ -11,8 +44,10 @@ filter(players, min_rank(desc(h)) <= 2 & h > 0)
 mutate(players, g_rank = min_rank(g))
 
 # For each player, find every year that was better than the previous year
+#broken
 filter(players, g > lag(g))
 # For each player, compute avg change in games played per year
+#broken
 mutate(players, g_change = (g - lag(g)) / (yearid - lag(yearid)))
 
 # For each player, find all where they played more games than average