Merge pull request #41 from louib/better_replace_function

😹😹 Improve tokenization

README.md

@@ -72,6 +72,13 @@ Note: if you want colors you need to specify the `--color` option.
 
 To invoke help run `emojify -h`.
 
+Testing
+-------
+
+The project can be tested using [bats](https://github.com/sstephenson/bats#installing-bats-from-source):
+```bash
+$ bats emojify_tests.bats
+```
 
 Related
 -------

emojify

@@ -2791,21 +2791,55 @@ declare -A emojis=(
 )
 
 # Gets emoji from the associative array
-# Returns raw emoji character or word
-to_emoji () {
+# Returns raw emoji character or original token
+emojify_token () {
     local value=${emojis[$1]}
-    [[ -n $value ]] && echo -e "$value " || echo "$1"
+    [[ -n $value ]] && echo -e "$value" || echo "$1"
 }
 
-# Function to parse a line, split it into an array of words and then emojify
-# each word.
-parse_line () {
-    IFS=' '; read -ra words_arr <<< "$*"
-    out=()
-    for word in "${words_arr[@]}"; do
-        out+=("$(to_emoji "$word")")
+# Function to emojify a line.
+emojify_line () {
+    line=$*
+    current_token=""
+    emojified_line=""
+    for (( i=0; i<${#line}; i++ )); do
+        current_char="${line:$i:1}"
+
+        if [[ -z "$current_token" ]]; then
+            # starting a new token.
+            if [[ $current_char =~ ^:$ ]]; then
+                current_token+=$current_char
+            # not starting a new token.
+            else
+                emojified_line+=$current_char
+            fi
+        else
+            # finishing the current token.
+            if [[ $current_char =~ ^:$ ]]; then
+                emojified_token=$(emojify_token "$current_token$current_char")
+
+                # nothing changed. we still need the ending ':' as it might
+                # serve for an upcoming emoji 😁😁🎉🎉
+                if [[ "$emojified_token" == "$current_token$current_char" ]]; then
+                    emojified_line+=$current_token
+                    current_token=$current_char
+                else
+                    emojified_line+=$emojified_token
+                    current_token=""
+                fi
+            # continuing the current token.
+            elif [[ $current_char =~ ^[a-zA-Z0-9_+-]$ ]]; then
+                current_token+=$current_char
+            # dropping the current token.
+            else
+                current_token+=$current_char
+                emojified_line+=$current_token
+                current_token=""
+            fi
+        fi
     done
-    echo "${out[*]}"
+    emojified_line+=$current_token
+    echo "$emojified_line"
 }
 
 # Function to print help info.
@@ -2850,12 +2884,12 @@ if [[ -n $1 ]]; then
 
     * )
         check_version
-        parse_line "$*"
+        emojify_line "$*"
         ;;
     esac
 else
     check_version
     while IFS=''; read -r line || [ -n "$line" ]; do
-        parse_line "$line"
+        emojify_line "$line"
     done
 fi

emojify_tests.bats

@@ -0,0 +1,54 @@
+#!/usr/bin/env bats
+
+@test "handles an input without a single emoji 😿" {
+  result=$(./emojify "no emoji :(")
+  [ "$result" = "no emoji :(" ]
+}
+
+@test "handles an input with a single emoji 😹" {
+  result=$(./emojify "an emoji :grin:")
+  [ "$result" = "an emoji 😁" ]
+}
+
+@test "handles an input with a lot of emojis 😻" {
+  result=$(./emojify "emojis :grin::grin: :tada:yay:champagne:")
+  [ "$result" = "emojis 😁😁 🎉yay🍾" ]
+}
+
+@test "handles emojis with underscores and numbers 💯" {
+  result=$(./emojify "this is perfect :100: :1st_place_medal:")
+  [ "$result" = "this is perfect 💯 🥇" ]
+}
+
+@test "handles emojis with + and - 👍" {
+  result=$(./emojify "great :+1::+1::-1:")
+  [ "$result" = "great 👍👍👎" ]
+}
+
+@test "handles right-hand side emojis 👉" {
+  result=$(./emojify ":not_an_emoji:point_right:")
+  [ "$result" = ":not_an_emoji👉" ]
+  result=$(./emojify "::::point_right:")
+  [ "$result" = ":::👉" ]
+}
+
+@test "handles punctuations just after aliases" {
+  result=$(./emojify "Enter the :airplane:!")
+  [ "$result" = "Enter the ✈️!" ]
+}
+
+@test "handles multiple spaces after an emoji" {
+  result=$(./emojify ":sparkles:   Three spaces")
+  [ "$result" = "✨   Three spaces" ]
+  result=$(./emojify ":sparkles:     Five spaces")
+  [ "$result" = "✨     Five spaces" ]
+  result=$(./emojify ":sparkles: One space")
+  [ "$result" = "✨ One space" ]
+}
+
+@test "handles the examples from the readme 😉" {
+  result=$(./emojify "Hey, I just :raising_hand: you, and this is :scream: , but here's my :calling: , so :telephone_receiver: me, maybe?")
+  [ "$result" = "Hey, I just 🙋 you, and this is 😱 , but here's my 📲 , so 📞 me, maybe?" ]
+  result=$(./emojify "To :bee: , or not to :bee: : that is the question... To take :muscle: against a :ocean: of troubles, and by opposing, end them?")
+  [ "$result" = "To 🐝 , or not to 🐝 : that is the question... To take 💪 against a 🌊 of troubles, and by opposing, end them?" ]
+}