//parse apache access.log entry into components
bool ApacheCombinedLog::parseCommit(RCommit& commit) {
std::string line;
std::vector<std::string> matches;
if(!logf->getNextLine(line)) return false;
apache_entry_start.match(line, &matches);
if(matches.size()!=4) {
return 0;
}
//get details
commit.username = matches[0];
//std::string user = matches[1];
//parse timestamp
struct tm time_str;
std::string request_str = matches[3];
std::string datestr = matches[2];
apache_entry_date.match(datestr, &matches);
if(matches.size()!=8) {
return 0;
}
int day = atoi(matches[0].c_str());
int year = atoi(matches[2].c_str());
int hour = atoi(matches[3].c_str());
int minute = atoi(matches[4].c_str());
int second = atoi(matches[5].c_str());
// int zone = atoi(matches[7].c_str());
//negative timezone
// if(strcmp(matches[6].c_str(), "-")==0) {
// zone = -zone;
// }
int month=0;
for(int i=0;i<12;i++) {
if(matches[1] == months[i]) {
month=i;
break;
}
}
time_str.tm_year = year - 1900;
time_str.tm_mon = month;
time_str.tm_mday = day;
time_str.tm_hour = hour;
time_str.tm_min = minute;
time_str.tm_sec = second;
time_str.tm_isdst = -1;
commit.timestamp = mktime(&time_str);
matches.clear();
apache_entry_request.match(request_str, &matches);
if(matches.size() < 5) {
return false;
}
std::string rtype = matches[0];
std::string file = matches[1];
std::string proto = matches[2];
int code = atoi(matches[3].c_str());
int bytes = atol(matches[4].c_str());
//remove args from url
int argpos = file.rfind("?");
if(argpos != std::string::npos) {
file = file.substr(0,argpos);
}
if(file.size()==0) file = "/";
//name index pages
if(file[file.size()-1] == '/') {
file += "index.html";
}
std::string action = "A";
commit.addFile(file, action);
std::string refer;
std::string agent;
if(matches.size() > 5) {
std::string agentstr = matches[5];
matches.clear();
apache_entry_agent.match(agentstr, &matches);
if(matches.size()>1) {
refer = matches[0];
agent = matches[1];
}
}
return true;
}
bool SVNCommitLog::parseCommit(RCommit& commit) {
//fprintf(stderr,"parsing svn log\n");
std::string line;
if(!getNextLine(line)) return false;
//start of log entry
if(!svn_logentry_start.match(line)) {
//is this the start of the document
if(!svn_xml_tag.match(line)) return false;
//fprintf(stderr,"found xml tag\n");
//if so find the first logentry tag
bool found_logentry = false;
while(getNextLine(line)) {
if(svn_logentry_start.match(line)) {
found_logentry = true;
break;
}
}
if(!found_logentry) return false;
}
//fprintf(stderr,"found logentry\n");
logentry.clear();
logentry.append(line);
logentry.append("\n");
//fprintf(stderr,"found opening tag\n");
bool endfound = false;
while(getNextLine(line)) {
logentry.append(line);
logentry.append("\n");
if(svn_logentry_end.match(line)) {
//fprintf(stderr,"found closing tag\n");
endfound=true;
break;
}
}
//incomplete commit
if(!endfound) return false;
//fprintf(stderr,"read logentry\n");
TiXmlDocument doc;
if(!doc.Parse(logentry.c_str())) return false;
//fprintf(stderr,"try to parse logentry: %s\n", logentry.c_str());
TiXmlElement* leE = doc.FirstChildElement( "logentry" );
std::vector<std::string> entries;
if(!leE) return false;
//parse date
TiXmlElement* dateE = leE->FirstChildElement( "date" );
if(!dateE) return false;
std::string timestamp_str(dateE->GetText());
if(!svn_logentry_timestamp.match(timestamp_str, &entries))
return false;
struct tm time_str;
time_str.tm_year = atoi(entries[0].c_str()) - 1900;
time_str.tm_mon = atoi(entries[1].c_str()) - 1;
time_str.tm_mday = atoi(entries[2].c_str());
time_str.tm_hour = atoi(entries[3].c_str());
time_str.tm_min = atoi(entries[4].c_str());
time_str.tm_sec = atoi(entries[5].c_str());
time_str.tm_isdst = -1;
commit.timestamp = timegm(&time_str);
//parse author
TiXmlElement* authorE = leE->FirstChildElement("author");
if(authorE != 0) {
std::string author(authorE->GetText());
if(author.empty()) author = "Unknown";
commit.username = author;
}
TiXmlElement* pathsE = leE->FirstChildElement( "paths" );
//log entries sometimes dont have any paths
if(!pathsE) return true;
//parse changes
for(TiXmlElement* pathE = pathsE->FirstChildElement("path"); pathE != 0; pathE = pathE->NextSiblingElement()) {
//parse path
const char* kind = pathE->Attribute("kind");
const char* action = pathE->Attribute("action");
//check for action
if(action == 0) continue;
bool is_dir = false;
//if has the 'kind' attribute (old versions of svn dont have this), check if it is a dir
if(kind != 0 && strcmp(kind,"dir") == 0) {
//accept only deletes for directories
if(strcmp(action, "D") != 0) continue;
is_dir = true;
}
std::string file(pathE->GetText());
std::string status(action);
if(file.empty()) continue;
if(status.empty()) continue;
//append trailing slash if is directory
if(is_dir && file[file.size()-1] != '/') {
file = file + std::string("/");
}
commit.addFile(file, status);
}
//fprintf(stderr,"parsed logentry\n");
//read files
return true;
}
bool CVS2CLCommitLog::parseCommit(RCommit& commit) {
//fprintf(stderr,"parsing cvs2cl log\n");
std::string line;
if(!getNextLine(line)) return false;
//start of log entry
if(!cvs2cl_logentry_start.match(line)) {
//is this the start of the document
if(!cvs2cl_xml_tag.match(line)) return false;
//fprintf(stderr,"found xml tag\n");
//if so find the first logentry tag
bool found_logentry = false;
while(getNextLine(line)) {
if(cvs2cl_logentry_start.match(line)) {
found_logentry = true;
break;
}
}
if(!found_logentry) return false;
}
//fprintf(stderr,"found logentry\n");
logentry.clear();
logentry.append(line);
logentry.append("\n");
//fprintf(stderr,"found opening tag\n");
bool endfound = false;
while(getNextLine(line)) {
logentry.append(line);
logentry.append("\n");
if(cvs2cl_logentry_end.match(line)) {
//fprintf(stderr,"found closing tag\n");
endfound=true;
break;
}
}
//incomplete commit
if(!endfound) return false;
//fprintf(stderr,"read logentry\n");
TiXmlDocument doc;
if(!doc.Parse(logentry.c_str())) return false;
//fprintf(stderr,"try to parse logentry: %s\n", logentry.c_str());
TiXmlElement* leE = doc.FirstChildElement( "entry" );
std::vector<std::string> entries;
if(!leE) return false;
//parse date
TiXmlElement* dateE = leE->FirstChildElement( "isoDate" );
if(!dateE) return false;
std::string timestamp_str(dateE->GetText());
if(!cvs2cl_logentry_timestamp.match(timestamp_str, &entries))
return false;
struct tm time_str;
time_str.tm_year = atoi(entries[0].c_str()) - 1900;
time_str.tm_mon = atoi(entries[1].c_str()) - 1;
time_str.tm_mday = atoi(entries[2].c_str());
time_str.tm_hour = atoi(entries[3].c_str());
time_str.tm_min = atoi(entries[4].c_str());
time_str.tm_sec = atoi(entries[5].c_str());
time_str.tm_isdst = -1;
commit.timestamp = mktime(&time_str);
//parse author
TiXmlElement* authorE = leE->FirstChildElement("author");
if(authorE != 0) {
std::string author(authorE->GetText());
if(author.empty()) author = "Unknown";
commit.username = author;
}
//parse changes
for(TiXmlElement* fileE = leE->FirstChildElement("file"); fileE != 0; fileE = fileE->NextSiblingElement()) {
TiXmlElement* state = fileE->FirstChildElement("cvsstate");
TiXmlElement* name = fileE->FirstChildElement("name");
//check for state
if(name == 0 || state == 0) continue;
std::string status = strcmp(state->GetText(), "dead") == 0 ? "D" : "M";
std::string file(name->GetText());
if(file.empty()) continue;
commit.addFile(file, status);
}
//fprintf(stderr,"parsed logentry\n");
//read files
return true;
}
bool GitCommitLog::parseCommit(RCommit& commit) {
std::string line;
char filesizeBuf[256];
commit.username = "";
while(logf->getNextLine(line) && line.size()) {
if(line.find("commit_id:") == 0) {
commit.commit_id = line.substr(10);
if(!logf->getNextLine(line)) return false;
//username follows user prefix
commit.username = line.substr(5);
if(!logf->getNextLine(line)) return false;
commit.timestamp = atol(line.c_str());
//this isnt a commit we are parsing, abort
if(commit.timestamp == 0) return false;
continue;
}
//should see username before files
if(commit.username.empty()) return false;
size_t tab = line.find('\t');
//incorrect log format
if(tab == std::string::npos || tab == 0 || tab == line.size()-1) continue;
std::string status = line.substr(tab - 1, 1);
std::string file = line.substr(tab + 1);
if(file.empty()) continue;
//check for and remove double quotes
if(file.find('"') == 0 && file.rfind('"') == file.size()-1) {
if(file.size()<=2) continue;
file = file.substr(1,file.size()-2);
}
std::string size_temp_file = temp_file + "_filesize";
memset(filesizeBuf, 0, sizeof(char) * 256);
snprintf(filesizeBuf, sizeof(char) * 256, GitGetFileSizeCommand.c_str(), commit.commit_id.c_str(), file.c_str(), size_temp_file.c_str());
systemCommand(filesizeBuf);
std::ifstream ifs ( size_temp_file );
std::string filesize_str;
ifs >> filesize_str;
size_t filesize = 0;
if(!filesize_str.empty()) {
filesize = std::stoi(filesize_str);
}
commit.addFile(file, status, filesize);
}
//check we at least got a username
if(commit.username.empty()) return false;
return true;
}
