From acee32bd1c66a43fd9719476a3ca82a13c24e88e Mon Sep 17 00:00:00 2001
From: Mathieu Westphal <mathieu.westphal@kitware.com>
Date: Thu, 13 Feb 2020 16:38:12 +0000
Subject: [PATCH] Adding DecodeUrl and GetFilePathFromURI methods
---
SystemTools.cxx | 38 ++++++++++++++++++++++++++++++++++----
SystemTools.hxx.in | 14 ++++++++++++--
testSystemTools.cxx | 30 ++++++++++++++++++++++++++++++
3 files changed, 76 insertions(+), 6 deletions(-)
diff --git a/SystemTools.cxx b/SystemTools.cxx
index 39873e63..c68487a3 100644
--- a/SystemTools.cxx
+++ b/SystemTools.cxx
@@ -123,9 +123,9 @@ extern char** environ;
#define VTK_URL_PROTOCOL_REGEX "([a-zA-Z0-9]*)://(.*)"
#define VTK_URL_REGEX \
- "([a-zA-Z0-9]*)://(([A-Za-z0-9]+)(:([^:@]+))?@)?([^:@/]+)(:([0-9]+))?/" \
+ "([a-zA-Z0-9]*)://(([A-Za-z0-9]+)(:([^:@]+))?@)?([^:@/]*)(:([0-9]+))?/" \
"(.+)?"
-
+#define VTK_URL_BYTE_REGEX "%[0-9a-fA-F][0-9a-fA-F]"
#ifdef _MSC_VER
# include <sys/utime.h>
#else
@@ -4513,7 +4513,7 @@ std::string SystemTools::GetOperatingSystemNameAndVersion()
bool SystemTools::ParseURLProtocol(const std::string& URL,
std::string& protocol,
- std::string& dataglom)
+ std::string& dataglom, bool decode)
{
// match 0 entire url
// match 1 protocol
@@ -4526,13 +4526,17 @@ bool SystemTools::ParseURLProtocol(const std::string& URL,
protocol = urlRe.match(1);
dataglom = urlRe.match(2);
+ if (decode) {
+ dataglom = DecodeURL(dataglom);
+ }
+
return true;
}
bool SystemTools::ParseURL(const std::string& URL, std::string& protocol,
std::string& username, std::string& password,
std::string& hostname, std::string& dataport,
- std::string& database)
+ std::string& database, bool decode)
{
kwsys::RegularExpression urlRe(VTK_URL_REGEX);
if (!urlRe.find(URL))
@@ -4556,9 +4560,35 @@ bool SystemTools::ParseURL(const std::string& URL, std::string& protocol,
dataport = urlRe.match(8);
database = urlRe.match(9);
+ if (decode) {
+ username = DecodeURL(username);
+ password = DecodeURL(password);
+ hostname = DecodeURL(hostname);
+ dataport = DecodeURL(dataport);
+ database = DecodeURL(database);
+ }
+
return true;
}
+// ----------------------------------------------------------------------
+std::string SystemTools::DecodeURL(const std::string& url)
+{
+ kwsys::RegularExpression urlByteRe(VTK_URL_BYTE_REGEX);
+ std::string ret;
+ for (size_t i = 0; i < url.length(); i++) {
+ if (urlByteRe.find(url.substr(i, 3))) {
+ ret +=
+ static_cast<char>(strtoul(url.substr(i + 1, 2).c_str(), nullptr, 16));
+ i += 2;
+ } else {
+ ret += url[i];
+ }
+ }
+ return ret;
+}
+
+// ----------------------------------------------------------------------
// These must NOT be initialized. Default initialization to zero is
// necessary.
static unsigned int SystemToolsManagerCount;
diff --git a/SystemTools.hxx.in b/SystemTools.hxx.in
index c4ab9d4f..d4a93faf 100644
--- a/SystemTools.hxx.in
+++ b/SystemTools.hxx.in
@@ -935,22 +935,32 @@ public:
* Parse a character string :
* protocol://dataglom
* and fill protocol as appropriate.
+ * decode the dataglom using DecodeURL if set to true.
* Return false if the URL does not have the required form, true otherwise.
*/
static bool ParseURLProtocol(const std::string& URL, std::string& protocol,
- std::string& dataglom);
+ std::string& dataglom, bool decode = false);
/**
* Parse a string (a URL without protocol prefix) with the form:
* protocol://[[username[':'password]'@']hostname[':'dataport]]'/'[datapath]
* and fill protocol, username, password, hostname, dataport, and datapath
* when values are found.
+ * decode all string except the protocol using DecodeUrl if set to true.
* Return true if the string matches the format; false otherwise.
*/
static bool ParseURL(const std::string& URL, std::string& protocol,
std::string& username, std::string& password,
std::string& hostname, std::string& dataport,
- std::string& datapath);
+ std::string& datapath, bool decode = false);
+
+ /**
+ * Decode the percent-encoded string from an URL or an URI
+ * into their correct char values.
+ * Does not perform any other sort of validation.
+ * Return the decoded string
+ */
+ static std::string DecodeURL(const std::string& url);
private:
/**
diff --git a/testSystemTools.cxx b/testSystemTools.cxx
index 1f3a15b5..7032a100 100644
--- a/testSystemTools.cxx
+++ b/testSystemTools.cxx
@@ -1077,6 +1077,34 @@ static bool CheckCopyFileIfDifferent()
return ret;
}
+static bool CheckURLParsing()
+{
+ bool ret = true;
+ std::string url = "http://user:pw@hostname:42/full/url.com";
+
+ std::string protocol, username, password, hostname, dataport, database;
+ kwsys::SystemTools::ParseURL(url, protocol, username, password, hostname,
+ dataport, database);
+ if (protocol != "http" || username != "user" || password != "pw" ||
+ hostname != "hostname" || dataport != "42" ||
+ database != "full/url.com") {
+ std::cerr << "Incorrect URL parsing" << std::endl;
+ ret = false;
+ }
+
+ std::string uri =
+ "file://hostname/path/to/"
+ "a%20file%20with%20str%C3%A0ng%C3%A8%20ch%40r%20and%20s%C2%B5aces";
+ kwsys::SystemTools::ParseURL(uri, protocol, username, password, hostname,
+ dataport, database, true);
+ if (protocol != "file" || hostname != "hostname" ||
+ database != "path/to/a file with strà ngè ch@r and sµaces") {
+ std::cerr << "Incorrect URL parsing or decoding" << std::endl;
+ ret = false;
+ }
+ return ret;
+}
+
int testSystemTools(int, char* [])
{
bool res = true;
@@ -1124,5 +1152,7 @@ int testSystemTools(int, char* [])
res &= CheckCopyFileIfDifferent();
+ res &= CheckURLParsing();
+
return res ? 0 : 1;
}
--
GitLab