Add integration tests for symbols/unicode

Signed-off-by: Elijah Zupancic <[email protected]>

Author: dekobon

test/data/bucket-1/a/%@!*()=$#^&|.txt

@@ -0,0 +1 @@
+We are but selling water next to a river.

test/data/bucket-1/a/abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ0123456789.txt

@@ -0,0 +1,4 @@
+"Where the is not one thing, what then?"
+"Throw it away!"
+"With not one thing, what there is to throw away?"
+"Then carry it off!"

test/data/bucket-1/a/plus+plus.txt

@@ -0,0 +1,27 @@
+代悲白頭翁　　　Lament for the White-Haired Old Man
+洛陽城東桃李花　In the east of Luoyang City, Peach blossoms abound
+飛來飛去落誰家　Their petals float around, coming and going, to whose house will they fall?
+洛陽女児惜顔色　Girls in Luoyang cherish their complexion
+行逢落花長歎息　They breathe a deep sigh upon seeing the petals fall
+今年花落顔色改　This year the petals fall and their complexion changes
+明年花開復誰在　Who will be there when the flowers bloom next year?
+已見松柏摧為薪　I've seen the pines and cypresses destroyed and turned into firewood
+更聞桑田変成海　I hear that the mulberry fields have fallen into the sea
+古人無復洛城東　The people of old never came back to the east of Luoyang City
+今人還對落花風　The people of today likewise face the falling flowers in the wind
+年年歳歳花相似　Year after year, flowers look alike
+歳歳年年人不同　Year after year, the people are not the same
+寄言全盛紅顔子　I want you to get this message, my child, you are in your prime, with a rosy complexion
+應憐半死白頭翁　Take pity on the half-dead white-haired old man
+此翁白頭真可憐　You really must take pity on this white-haired old man
+伊昔紅顔美少年　For once upon a time, I used to be a red-faced handsome young man
+公子王孫芳樹下　A child of noble birth under a fragrant tree
+清歌妙舞落花前　Singing and dancing in front of the falling petals
+光禄池臺開錦繍　At the platform before the mirror pond, beautiful autumn leaves opening all around
+将軍楼閣畫神仙　The general’s pavilion is painted with gods and goddesses
+一朝臥病無相識　Once I was sick and no one knew me
+三春行楽在誰邉　Who will be at the shore for the spring outing?
+宛轉蛾眉能幾時　For how long will the moths gracefully turn about?
+須臾鶴髪亂如絲　The crane’s feathers are like tangled threads for just a moment
+但看古来歌舞地　Yet, look at the ancient places of song and dance
+惟有黄昏鳥雀悲　Only in twilight, do the birds lament

test/data/bucket-1/a/これは This is ASCII системы חן .txt

@@ -0,0 +1 @@
+What name do you use to call out to the emptiness?

test/integration/test_api.sh

@@ -154,17 +154,24 @@ assertHttpRequestEquals "HEAD" "b/c/d.txt" "200"
 assertHttpRequestEquals "HEAD" "b/c/../e.txt" "200"
 assertHttpRequestEquals "HEAD" "b/e.txt" "200"
 assertHttpRequestEquals "HEAD" "b//e.txt" "200"
+assertHttpRequestEquals "HEAD" "a/abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ0123456789.txt" "200"
 
 # Weird filenames
 assertHttpRequestEquals "HEAD" "b/c/=" "200"
 assertHttpRequestEquals "HEAD" "b/c/@" "200"
 assertHttpRequestEquals "HEAD" "b/c/'(1).txt" "200"
 assertHttpRequestEquals "HEAD" "%D1%81%D0%B8%D1%81%D1%82%D0%B5%D0%BC%D1%8B/%25bad%25file%25name%25" "200"
+assertHttpRequestEquals "HEAD" 'a/plus+plus.txt' "200"
 if [ ${is_windows} == "0" ]; then
   assertHttpRequestEquals "HEAD" "a/c/あ" "200"
   assertHttpRequestEquals "HEAD" "b/クズ箱/ゴミ.txt" "200"
   assertHttpRequestEquals "HEAD" "системы/system.txt" "200"
   assertHttpRequestEquals "HEAD" "b/ブツブツ.txt" "200"
+  # The following two objects do not get encoded correctly by curl when requested using their
+  # unicode names. The are provided as URL encoded as below. This is the same type of encoding
+  # expected by S3 and divergence from it will not work with nginx either.
+  assertHttpRequestEquals "HEAD" 'a/%25%40%21%2A%28%29%3D%24%23%5E%26%7C.txt' "200"
+  assertHttpRequestEquals "HEAD" 'a/%E3%81%93%E3%82%8C%E3%81%AF%E3%80%80This%20is%20ASCII%20%D1%81%D0%B8%D1%81%D1%82%D0%B5%D0%BC%D1%8B%20%20%D7%97%D7%9F%20.txt' "200"
 fi
 
 # Expected 400s
@@ -219,18 +226,25 @@ fi
 
 # Verify GET is working
 assertHttpRequestEquals "GET" "a.txt" "data/bucket-1/a.txt"
+assertHttpRequestEquals "GET" "a/abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ0123456789.txt" "data/bucket-1/a/abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ0123456789.txt"
 assertHttpRequestEquals "GET" "a.txt?some=param&that=should&be=stripped#aaah" "data/bucket-1/a.txt"
 assertHttpRequestEquals "GET" "b/c/d.txt" "data/bucket-1/b/c/d.txt"
 assertHttpRequestEquals "GET" "b/c/=" "data/bucket-1/b/c/="
 assertHttpRequestEquals "GET" "b/c/'(1).txt" "data/bucket-1/b/c/'(1).txt"
 assertHttpRequestEquals "GET" "b/e.txt" "data/bucket-1/b/e.txt"
 assertHttpRequestEquals "GET" "%D1%81%D0%B8%D1%81%D1%82%D0%B5%D0%BC%D1%8B/%25bad%25file%25name%25" "data/bucket-1/системы/%bad%file%name%"
+assertHttpRequestEquals "GET" 'a/plus+plus.txt' "data/bucket-1/a/plus+plus.txt"
 
 if [ ${is_windows} == "0" ]; then
   assertHttpRequestEquals "GET" "a/c/あ" "data/bucket-1/a/c/あ"
   assertHttpRequestEquals "GET" "b/ブツブツ.txt" "data/bucket-1/b/ブツブツ.txt"
   assertHttpRequestEquals "GET" "b/クズ箱/ゴミ.txt" "data/bucket-1/b/クズ箱/ゴミ.txt"
   assertHttpRequestEquals "GET" "системы/system.txt" "data/bucket-1/системы/system.txt"
+  # The following two objects do not get encoded correctly by curl when requested using their
+  # unicode names. The are provided as URL encoded as below. This is the same type of encoding
+  # expected by S3 and divergence from it will not work with nginx either.
+  assertHttpRequestEquals "GET" 'a/%25%40%21%2A%28%29%3D%24%23%5E%26%7C.txt' 'data/bucket-1/a/%@!*()=$#^&|.txt'
+  assertHttpRequestEquals "GET" 'a/%E3%81%93%E3%82%8C%E3%81%AF%E3%80%80This%20is%20ASCII%20%D1%81%D0%B8%D1%81%D1%82%D0%B5%D0%BC%D1%8B%20%20%D7%97%D7%9F%20.txt' "data/bucket-1/a/これは　This is ASCII системы  חן .txt"
 fi
 
 if [ "${index_page}" == "1" ]; then