[Libreoffice-commits] online.git: common/Protocol.hpp wsd/LOOLWSD.cpp

[Jan Holesovsky (via logerrit)]

 common/Protocol.hpp |   13 +++++++++++++
 wsd/LOOLWSD.cpp     |   22 +++++++++++-----------
 2 files changed, 24 insertions(+), 11 deletions(-)

New commits:
commit f637f4a75af56ae16116a88730660d018f306131
Author:     Jan Holesovsky <kendy at collabora.com>
AuthorDate: Fri Nov 8 21:37:21 2019 +0100
Commit:     Jan Holesovsky <kendy at collabora.com>
CommitDate: Fri Nov 8 22:24:08 2019 +0100

    killpoco: Remove StringTokenizer from LOOLWSD.cpp.
    
    Adds possibility to tokenize using a regex easily.
    
    Change-Id: Ie327d4faabec330c76d4cadb1d14bbe1527d332f
    Reviewed-on: https://gerrit.libreoffice.org/82333
    Reviewed-by: Jan Holesovsky <kendy at collabora.com>
    Tested-by: Jan Holesovsky <kendy at collabora.com>

diff --git a/common/Protocol.hpp b/common/Protocol.hpp
index b915d1a7c..e1a5d426e 100644
--- a/common/Protocol.hpp
+++ b/common/Protocol.hpp
@@ -15,6 +15,7 @@
 #include <cstring>
 #include <iomanip>
 #include <map>
+#include <regex>
 #include <sstream>
 #include <string>
 #include <vector>
@@ -140,6 +141,18 @@ namespace LOOLProtocol
         return tokenize(s.data(), s.size(), delimiter);
     }
 
+    /// Tokenize according to the regex, potentially skip empty tokens.
+    inline
+    std::vector<std::string> tokenize(const std::string& s, const std::regex& pattern, bool skipEmpty = false)
+    {
+        std::vector<std::string> tokens;
+        if (skipEmpty)
+            std::copy_if(std::sregex_token_iterator(s.begin(), s.end(), pattern, -1), std::sregex_token_iterator(), std::back_inserter(tokens), [](std::string in) { return !in.empty(); });
+        else
+            std::copy(std::sregex_token_iterator(s.begin(), s.end(), pattern, -1), std::sregex_token_iterator(), std::back_inserter(tokens));
+        return tokens;
+    }
+
     inline
     std::vector<int> tokenizeInts(const char* data, const size_t size, const char delimiter = ',')
     {
diff --git a/wsd/LOOLWSD.cpp b/wsd/LOOLWSD.cpp
index 583455d93..4fad66287 100644
--- a/wsd/LOOLWSD.cpp
+++ b/wsd/LOOLWSD.cpp
@@ -92,7 +92,6 @@ using Poco::Net::PartHandler;
 #include <Poco/Process.h>
 #include <Poco/SAX/InputSource.h>
 #include <Poco/StreamCopier.h>
-#include <Poco/StringTokenizer.h>
 #include <Poco/TemporaryFile.h>
 #include <Poco/URI.h>
 #include <Poco/Util/AbstractConfiguration.h>
@@ -162,7 +161,6 @@ using Poco::Net::MessageHeader;
 using Poco::Net::NameValueCollection;
 using Poco::Path;
 using Poco::StreamCopier;
-using Poco::StringTokenizer;
 using Poco::TemporaryFile;
 using Poco::URI;
 using Poco::Util::Application;
@@ -2190,19 +2188,21 @@ private:
             }
             else
             {
-                StringTokenizer reqPathTokens(request.getURI(), "/?", StringTokenizer::TOK_IGNORE_EMPTY | StringTokenizer::TOK_TRIM);
-                if (reqPathTokens.count() > 1 && reqPathTokens[0] == "lool" && reqPathTokens[1] == "clipboard")
+                // split by /'s and ?'s
+                std::vector<std::string> reqPathTokens(LOOLProtocol::tokenize(request.getURI(), std::regex(R"(\s*[\/\?]\s*)"), /*skipEmpty =*/ true));
+
+                if (reqPathTokens.size() > 1 && reqPathTokens[0] == "lool" && reqPathTokens[1] == "clipboard")
                 {
 //                    Util::dumpHex(std::cerr, "clipboard:\n", "", socket->getInBuffer()); // lots of data ...
                     handleClipboardRequest(request, message, disposition);
                 }
                 else if (!(request.find("Upgrade") != request.end() && Poco::icompare(request["Upgrade"], "websocket") == 0) &&
-                    reqPathTokens.count() > 0 && reqPathTokens[0] == "lool")
+                    reqPathTokens.size() > 0 && reqPathTokens[0] == "lool")
                 {
                     // All post requests have url prefix 'lool'.
                     handlePostRequest(request, message, disposition);
                 }
-                else if (reqPathTokens.count() > 2 && reqPathTokens[0] == "lool" && reqPathTokens[2] == "ws" &&
+                else if (reqPathTokens.size() > 2 && reqPathTokens[0] == "lool" && reqPathTokens[2] == "ws" &&
                          request.find("Upgrade") != request.end() && Poco::icompare(request["Upgrade"], "websocket") == 0)
                 {
                     std::string decodedUri; // WOPISrc
@@ -2532,8 +2532,8 @@ private:
         Poco::Net::HTTPResponse response;
         std::shared_ptr<StreamSocket> socket = _socket.lock();
 
-        StringTokenizer tokens(request.getURI(), "/?");
-        if (tokens.count() > 2 && tokens[2] == "convert-to")
+        std::vector<std::string> tokens(LOOLProtocol::tokenize(request.getURI(), std::regex(R"(\s*[\/\?]\s*)")));
+        if (tokens.size() > 2 && tokens[2] == "convert-to")
         {
             // Validate sender - FIXME: should do this even earlier.
             if (!allowConvertTo(socket->clientAddress(), request, true))
@@ -2559,7 +2559,7 @@ private:
             bool bFullSheetPreview = sFullSheetPreview == "true" ? true : false;
 
             // prefer what is in the URI
-            if (tokens.count() > 3)
+            if (tokens.size() > 3)
                 format = tokens[3];
 
             bool sent = false;
@@ -2659,7 +2659,7 @@ private:
             }
             return;
         }
-        else if (tokens.count() >= 4 && tokens[3] == "insertfile")
+        else if (tokens.size() >= 4 && tokens[3] == "insertfile")
         {
             LOG_INF("Insert file request.");
 
@@ -2700,7 +2700,7 @@ private:
                 }
             }
         }
-        else if (tokens.count() >= 6)
+        else if (tokens.size() >= 6)
         {
             LOG_INF("File download request.");
             // TODO: Check that the user in question has access to this file!