Add simple XML parsing wrapper

Also simplify check_region_error.

src/curl.cpp

@@ -31,9 +31,6 @@
 #include <pthread.h>
 #include <assert.h>
 #include <curl/curl.h>
-#include <libxml/xpath.h>
-#include <libxml/xpathInternals.h>
-#include <libxml/tree.h>
 #include <iostream>
 #include <fstream>
 #include <sstream>
@@ -2531,50 +2528,6 @@ string S3fsCurl::CalcSignature(const string& method, const string& canonical_uri
   return Signature;
 }
 
-// XML in BodyData has UploadId, Parse XML body for UploadId
-bool S3fsCurl::GetUploadId(string& upload_id)
-{
-  bool result = false;
-
-  if(!bodydata){
-    return result;
-  }
-  upload_id.clear();
-
-  xmlDocPtr doc;
-  if(NULL == (doc = xmlReadMemory(bodydata->str(), bodydata->size(), "", NULL, 0))){
-    return result;
-  }
-  if(NULL == doc->children){
-    S3FS_XMLFREEDOC(doc);
-    return result;
-  }
-  for(xmlNodePtr cur_node = doc->children->children; NULL != cur_node; cur_node = cur_node->next){
-    // For DEBUG
-    // string cur_node_name(reinterpret_cast<const char *>(cur_node->name));
-    // printf("cur_node_name: %s\n", cur_node_name.c_str());
-
-    if(XML_ELEMENT_NODE == cur_node->type){
-      string elementName = reinterpret_cast<const char*>(cur_node->name);
-      // For DEBUG
-      // printf("elementName: %s\n", elementName.c_str());
-
-      if(cur_node->children){
-        if(XML_TEXT_NODE == cur_node->children->type){
-          if(elementName == "UploadId") {
-            upload_id = reinterpret_cast<const char *>(cur_node->children->content);
-            result    = true;
-            break;
-          }
-        }
-      }
-    }
-  }
-  S3FS_XMLFREEDOC(doc);
-
-  return result;
-}
-
 void S3fsCurl::insertV4Headers()
 {
   string server_path = type == REQTYPE_LISTBUCKET ? "/" : path;
@@ -3449,8 +3402,7 @@ int S3fsCurl::PreMultipartPostRequest(const char* tpath, headers_t& meta, string
     return result;
   }
 
-  // Parse XML body for UploadId
+  if(!simple_parse_xml(bodydata->str(), bodydata->size(), "UploadId", upload_id)){
-  if(!S3fsCurl::GetUploadId(upload_id)){
     delete bodydata;
     bodydata = NULL;
     return -1;
@@ -3793,33 +3745,12 @@ bool S3fsCurl::CopyMultipartPostCallback(S3fsCurl* s3fscurl)
 
 bool S3fsCurl::CopyMultipartPostComplete()
 {
-  // parse ETag from response
+  std::string etag;
-  xmlDocPtr doc;
+  partdata.uploaded = simple_parse_xml(bodydata->str(), bodydata->size(), "ETag", etag);
-  if(NULL == (doc = xmlReadMemory(bodydata->str(), bodydata->size(), "", NULL, 0))){
+  if(etag.size() >= 2 && *etag.begin() == '"' && *etag.rbegin() == '"'){
-    return false;
+    etag.assign(etag.substr(1, etag.size() - 2));
   }
-  if(NULL == doc->children){
+  partdata.etaglist->at(partdata.etagpos).assign(etag);
-    S3FS_XMLFREEDOC(doc);
-    return false;
-  }
-  for(xmlNodePtr cur_node = doc->children->children; NULL != cur_node; cur_node = cur_node->next){
-    if(XML_ELEMENT_NODE == cur_node->type){
-      string elementName = reinterpret_cast<const char*>(cur_node->name);
-      if(cur_node->children){
-        if(XML_TEXT_NODE == cur_node->children->type){
-          if(elementName == "ETag") {
-            string etag = reinterpret_cast<const char *>(cur_node->children->content);
-            if(etag.size() >= 2 && *etag.begin() == '"' && *etag.rbegin() == '"'){
-              etag.assign(etag.substr(1, etag.size() - 2));
-            }
-            partdata.etaglist->at(partdata.etagpos).assign(etag);
-            partdata.uploaded = true;
-          }
-        }
-      }
-    }
-  }
-  S3FS_XMLFREEDOC(doc);
 
   delete bodydata;
   bodydata = NULL;

src/curl.h

@@ -379,7 +379,6 @@ class S3fsCurl
     void insertAuthHeaders();
     std::string CalcSignatureV2(const std::string& method, const std::string& strMD5, const std::string& content_type, const std::string& date, const std::string& resource);
     std::string CalcSignature(const std::string& method, const std::string& canonical_uri, const std::string& query_string, const std::string& strdate, const std::string& payload_hash, const std::string& date8601);
-    bool GetUploadId(std::string& upload_id);
     int GetIAMCredentials(void);
 
     int UploadMultipartPostSetup(const char* tpath, int part_num, const std::string& upload_id);

src/s3fs.cpp

@@ -3766,39 +3766,30 @@ static int s3fs_utility_processing(time_t abort_time)
 
 //
 // If calling with wrong region, s3fs gets following error body as 400 error code.
-// "<Error><Code>AuthorizationHeaderMalformed</Code><Message>The authorization header is 
+// "<Error>
-//  malformed; the region 'us-east-1' is wrong; expecting 'ap-northeast-1'</Message>
+//    <Code>AuthorizationHeaderMalformed</Code>
-//  <Region>ap-northeast-1</Region><RequestId>...</RequestId><HostId>...</HostId>
+//    <Message>The authorization header is malformed; the region 'us-east-1' is wrong; expecting 'ap-northeast-1'</Message>
+//    <Region>ap-northeast-1</Region>
+//    <RequestId>...</RequestId>
+//    <HostId>...</HostId>
 //  </Error>"
 //
-// So this is cheep codes but s3fs should get correct region automatically.
+// So this is cheap code but s3fs should get correct region automatically.
 //
-static bool check_region_error(const char* pbody, const string& currentep, string& expectregion)
+static bool check_region_error(const char* pbody, size_t len, string& expectregion)
 {
   if(!pbody){
     return false;
   }
-  const char* region;
+
-  const char* regionend;
+  std::string code;
-  if(NULL == (region = strcasestr(pbody, "<Message>The authorization header is malformed; the region "))){
+  if(!simple_parse_xml(pbody, len, "Code", code) || code != "AuthorizationHeaderMalformed"){
     return false;
   }
-  // check current endpoint region in body.
+
-  if(NULL == (region = strcasestr(region, currentep.c_str()))){
+  if(!simple_parse_xml(pbody, len, "Region", expectregion)){
     return false;
   }
-  if(NULL == (region = strcasestr(region, "expecting \'"))){
-    return false;
-  }
-  region += strlen("expecting \'");
-  if(NULL == (regionend = strchr(region, '\''))){
-    return false;
-  }
-  string strtmp(region, (regionend - region));
-  if(0 == strtmp.length()){
-    return false;
-  }
-  expectregion = strtmp;
 
   return true;
 }
@@ -3825,7 +3816,7 @@ static int s3fs_check_service()
       // check region error(for putting message or retrying)
       BodyData* body = s3fscurl.GetBodyData();
       string    expectregion;
-      if(check_region_error(body->str(), endpoint, expectregion)){
+      if(check_region_error(body->str(), body->size(), expectregion)){
         // [NOTE]
         // If endpoint is not specified(using us-east-1 region) and
         // an error is encountered accessing a different region, we

src/s3fs_util.cpp

@@ -31,6 +31,9 @@
 #include <pthread.h>
 #include <sys/types.h>
 #include <dirent.h>
+#include <libxml/xpath.h>
+#include <libxml/xpathInternals.h>
+#include <libxml/tree.h>
 
 #include <string>
 #include <sstream>
@@ -972,6 +975,50 @@ bool is_need_check_obj_detail(headers_t& meta)
   return true;
 }
 
+bool simple_parse_xml(const char* data, size_t len, const char* key, std::string& value)
+{
+  bool result = false;
+
+  if(!data || !key){
+    return result;
+  }
+  value.clear();
+
+  xmlDocPtr doc;
+  if(NULL == (doc = xmlReadMemory(data, len, "", NULL, 0))){
+    return result;
+  }
+
+  if(NULL == doc->children){
+    S3FS_XMLFREEDOC(doc);
+    return result;
+  }
+  for(xmlNodePtr cur_node = doc->children->children; NULL != cur_node; cur_node = cur_node->next){
+    // For DEBUG
+    // string cur_node_name(reinterpret_cast<const char *>(cur_node->name));
+    // printf("cur_node_name: %s\n", cur_node_name.c_str());
+
+    if(XML_ELEMENT_NODE == cur_node->type){
+      string elementName = reinterpret_cast<const char*>(cur_node->name);
+      // For DEBUG
+      // printf("elementName: %s\n", elementName.c_str());
+
+      if(cur_node->children){
+        if(XML_TEXT_NODE == cur_node->children->type){
+          if(elementName == key) {
+            value = reinterpret_cast<const char *>(cur_node->children->content);
+            result    = true;
+            break;
+          }
+        }
+      }
+    }
+  }
+  S3FS_XMLFREEDOC(doc);
+
+  return result;
+}
+
 //-------------------------------------------------------------------
 // Help
 //-------------------------------------------------------------------

src/s3fs_util.h

@@ -133,6 +133,7 @@ time_t cvtIAMExpireStringToTime(const char* s);
 time_t get_lastmodified(const char* s);
 time_t get_lastmodified(headers_t& meta);
 bool is_need_check_obj_detail(headers_t& meta);
+bool simple_parse_xml(const char* data, size_t len, const char* key, std::string& value);
 
 void show_usage(void);
 void show_help(void);