Merge pull request #104 from etiennebacher/more-speedup

Import `collapse` to reduce the time taken by `%in%`

DESCRIPTION

@@ -1,6 +1,6 @@
 Package: zoomerjoin
 Title: Superlatively Fast Fuzzy Joins
-Version: 0.1.4
+Version: 0.1.2.9000
 Authors@R: c(
     person("Beniamino", "Green", , "[email protected]", role = c("aut", "cre", "cph")),
     person("Etienne", "Bacher", email = "[email protected]", role = "ctb"),
@@ -15,6 +15,7 @@ Roxygen: list(markdown = TRUE)
 RoxygenNote: 7.3.1
 SystemRequirements: Cargo (>= 1.56) (Rust's package manager), rustc
 Imports:
+    collapse,
     dplyr,
     tibble,
     tidyr

NEWS.md

@@ -1,3 +1,7 @@
+# zoomerjoin (development version)
+
+* Several performance improvements (#101, #104).
+
 # zoomerjoin 0.1.2
 
 * Submitted Package to CRAN

R/jaccard_join_core.R

@@ -152,20 +152,25 @@ jaccard_join <- function (a, b, mode, by, salt_by, n_gram_width, n_bands,
         return(matches)
     }
 
-    not_matched_a <- ! seq(nrow(a)) %in% match_table[,1]
-    not_matched_b <- ! seq(nrow(b)) %in% match_table[,2]
-
-    if (mode == "left") {
-        matches <- dplyr::bind_rows(matches,a[not_matched_a,])
-    } else if (mode == "right") {
-        matches <- dplyr::bind_rows(matches,b[not_matched_b,])
-    } else if (mode == "full") {
-        matches <- dplyr::bind_rows(matches,a[not_matched_a,],b[not_matched_b,])
-    } else if (mode == "anti") {
-        matches <- dplyr::bind_rows(a[not_matched_a,], b[not_matched_b,])
-    } else {
-        stop("Invalid Mode Selected!")
-    }
-
-    return(matches)
+    switch(
+        mode,
+        "left" = {
+            not_matched_a <- collapse::`%!iin%`(seq(nrow(a)), match_table[, 1])
+            matches <- dplyr::bind_rows(matches, a[not_matched_a,])
+        },
+        "right" = {
+            not_matched_b <- collapse::`%!iin%`(seq(nrow(b)), match_table[, 2])
+            matches <- dplyr::bind_rows(matches, b[not_matched_b,])
+        },
+        "full" = {
+            not_matched_a <- collapse::`%!iin%`(seq(nrow(a)), match_table[, 1])
+            not_matched_b <- collapse::`%!iin%`(seq(nrow(b)), match_table[, 2])
+            matches <- dplyr::bind_rows(matches, a[not_matched_a,], b[not_matched_b,])
+        },
+        "anti" = {
+            not_matched_a <- collapse::`%!iin%`(seq(nrow(a)), match_table[, 1])
+            not_matched_b <- collapse::`%!iin%`(seq(nrow(b)), match_table[, 2])
+            matches <- dplyr::bind_rows(a[not_matched_a,], b[not_matched_b,])
+        }
+    )
 }