From fc054f6f41bc2dd7c1bc76cbaf591771563287ab Mon Sep 17 00:00:00 2001
From: Zach Borboa <zachborboa@gmail.com>
Date: Sat, 28 Apr 2018 02:09:14 -0700
Subject: [PATCH] Add script to generate compressed urls test file.

Omit the timestamp and the file name from the file header to make the
file deterministic.
Refresh the urls test file.
---
 tests/PHPCurlClass/urls.csv.gz | Bin 74129 -> 74120 bytes
 tests/generate_urls.py         |  61 +++++++++++++++++++++++++++++++++
 tests/generate_urls.sh         |  10 ++++++
 3 files changed, 71 insertions(+)
 create mode 100755 tests/generate_urls.sh

diff --git a/tests/PHPCurlClass/urls.csv.gz b/tests/PHPCurlClass/urls.csv.gz
index 055d81d46ac7a94aa08da7883f8b915168c3be21..cc1972bae5fca90da7c9c4872c03fef1224a0ae5 100644
GIT binary patch
delta 24
dcmbPun5E+|3zvL12LlK&F*gct6=tkc1OQJ!1!VvL

delta 33
ocmeA;%rfyXi?Dn*2ghT^Cs9nyrA0Z#ddbCQ42@h{xftsd0mKFh#sB~S

diff --git a/tests/generate_urls.py b/tests/generate_urls.py
index fc28d16..ab468db 100644
--- a/tests/generate_urls.py
+++ b/tests/generate_urls.py
@@ -54,9 +54,70 @@ first_queries = ['', '?a=1', '?a=647&b=s564']
 second_queries = ['', '?a=sdf', '?a=cvb&b=987']
 fragments = ['', '#foo', '#bar']
 
+additional_tests = [
+    {
+        'args': [
+            'http://www.example.com/',
+            '',
+        ],
+        'expected': 'http://www.example.com/',
+    },
+    {
+        'args': [
+            'http://www.example.com/',
+            'foo',
+        ],
+        'expected': 'http://www.example.com/foo',
+    },
+    {
+        'args': [
+            'http://www.example.com/',
+            '/foo',
+        ],
+        'expected': 'http://www.example.com/foo',
+    },
+    {
+        'args': [
+            'http://www.example.com/',
+            '/foo/',
+        ],
+        'expected': 'http://www.example.com/foo/',
+    },
+    {
+        'args': [
+            'http://www.example.com/',
+            '/dir/page.html',
+        ],
+        'expected': 'http://www.example.com/dir/page.html',
+    },
+    {
+        'args': [
+            'http://www.example.com/dir1/page2.html',
+            '/dir/page.html',
+        ],
+        'expected': 'http://www.example.com/dir/page.html',
+    },
+    {
+        'args': [
+            'http://www.example.com/dir1/page2.html',
+            'dir/page.html',
+        ],
+        'expected': 'http://www.example.com/dir1/dir/page.html',
+    },
+    {
+        'args': [
+            'http://www.example.com/dir1/dir3/page.html',
+            '../dir/page.html',
+        ],
+        'expected': 'http://www.example.com/dir1/dir/page.html',
+    },
+]
+
 with open('urls.csv', 'wt') as f:
     csvwriter = csv.writer(f, quotechar='"', quoting=csv.QUOTE_ALL)
     csvwriter.writerow(['first_url', 'second_url', 'expected'])
+    for test in additional_tests:
+        csvwriter.writerow([test['args'][0], test['args'][1], test['expected']])
     for first_domain, second_domain in product(first_authorities, second_authorities):
         for first_path, second_path in product(first_paths, second_paths):
             for first_query, second_query in product(first_queries, second_queries):
diff --git a/tests/generate_urls.sh b/tests/generate_urls.sh
new file mode 100755
index 0000000..175f148
--- /dev/null
+++ b/tests/generate_urls.sh
@@ -0,0 +1,10 @@
+SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
+cd "${SCRIPT_DIR}"
+
+set -x
+
+python "generate_urls.py" &&
+    ([[ -f "urls.csv.gz" ]] && rm --verbose "urls.csv.gz" || exit 0) &&
+    gzip --verbose --best --no-name "urls.csv" &&
+    gzip --verbose --test "urls.csv.gz" &&
+    mv --verbose "urls.csv.gz" "PHPCurlClass/urls.csv.gz"