From 530c5ea14a89c9d31b2cc302edb50430328dd5a9 Mon Sep 17 00:00:00 2001
From: Daniel Sjoberg <danield.sjoberg@gmail.com>
Date: Wed, 5 Jun 2024 17:14:11 -0700
Subject: [PATCH] udpated `str_extract()` and added `str_extract_all()`

---
 R/standalone-stringr.R                   | 33 +++++++-----------------
 tests/testthat/test-standalone-stringr.R | 21 ++++++++++++++-
 2 files changed, 29 insertions(+), 25 deletions(-)

diff --git a/R/standalone-stringr.R b/R/standalone-stringr.R
index ea5cbdb..1857948 100644
--- a/R/standalone-stringr.R
+++ b/R/standalone-stringr.R
@@ -25,25 +25,6 @@ str_squish <- function(string, fixed = FALSE) {
   return(string)
 }
 
-
-str_extract <- function(string, pattern, fixed = FALSE) {
-  result <- sapply(string, function(x) {
-    # Adjust the pattern if fixed is TRUE
-    if (fixed) {
-      # Escape special characters to treat them as literal in regex
-      pattern <- gsub("([][{}()+*^$.|\\\\?])", "\\\\\\1", pattern)
-    }
-    match_pos <- regexpr(pattern = pattern, text = x, perl = TRUE)
-    if (match_pos > 0) {
-      regmatches(x, match_pos)
-    } else {
-      NA_character_
-    }
-  }, USE.NAMES = FALSE)
-
-  return(result)
-}
-
 str_remove <- function (string, pattern, fixed = FALSE) {
   sub (x = string, pattern = pattern, replacement = "", fixed = fixed)
 }
@@ -53,11 +34,15 @@ str_remove_all <- function(string, pattern, fixed = FALSE) {
 }
 
 str_extract <- function(string, pattern, fixed = FALSE) {
-  ifelse(
-    str_detect(string, pattern, fixed = fixed),
-    regmatches(x = string, m = regexpr(pattern = pattern, text = string, fixed = fixed)),
-    NA_character_
-  )
+  res <- rep(NA_character_, length.out = length(string))
+  res[str_detect(string, pattern, fixed = fixed)] <-
+    regmatches(x = string, m = regexpr(pattern = pattern, text = string, fixed = fixed))
+
+  res
+}
+
+str_extract_all <- function(string, pattern, fixed = FALSE) {
+  regmatches(x = string, m = gregexpr(pattern = pattern, text = string, fixed = fixed))
 }
 
 str_detect <- function(string, pattern, fixed = FALSE) {
diff --git a/tests/testthat/test-standalone-stringr.R b/tests/testthat/test-standalone-stringr.R
index 3de5490..8a6400c 100644
--- a/tests/testthat/test-standalone-stringr.R
+++ b/tests/testthat/test-standalone-stringr.R
@@ -59,7 +59,7 @@ test_that("str_remove_all() works", {
 })
 
 test_that("str_extract() works", {
-  shopping_list <- c("apples x43", "bag of flour", "bag of sugar", "milk x2")
+  shopping_list <- c("apples x43", "bag of flour", "bag of sugar, bag of sugar", "milk x2")
 
   s <- str_extract(shopping_list, "[a-z]+")
   expect_identical(s, c("apples", "bag", "bag", "milk"))
@@ -81,6 +81,25 @@ test_that("str_extract() works", {
   expect_identical(s, stringr::str_extract(shopping_list, "[a-z]+"))
 })
 
+test_that("str_extract_all() works", {
+  shopping_list <- c("apples x43", "bag of flour", "bag of sugar, bag of sugar", "milk x2")
+
+  s <- str_extract_all(shopping_list, "[a-z]+")
+  expect_identical(s, stringr::str_extract_all(shopping_list, "[a-z]+"))
+
+  s <- str_extract_all(shopping_list, "([a-z]+) of ([a-z]+)")
+  expect_identical(s, stringr::str_extract_all(shopping_list, "([a-z]+) of ([a-z]+)"))
+
+  s_notfixed <- str_extract_all(shopping_list, "\\d")
+  expect_identical(s_notfixed, stringr::str_extract_all(shopping_list, "\\d"))
+
+  s_fixed <- str_extract_all(shopping_list, "\\d", fixed = TRUE)
+  expect_identical(s_fixed, stringr::str_extract_all(shopping_list, stringr::fixed("\\d")))
+
+  s <- str_extract_all(shopping_list, "[a-z]+")
+  expect_identical(s, stringr::str_extract_all(shopping_list, "[a-z]+"))
+})
+
 test_that("str_detect() works", {
   fruits <- c("apple", "banana", "pear", "pineapple")