I have gutil's sysinfo.cc file, indentical to this one:
https://github.com/cloudera/kudu/blob/master/src/kudu/gutil/sysinfo.cc#L248
I would like to know if the InitializeSystemInfo() function works properly for Linux, given I'm not too familiar with C++.
I can't find a closing bracket for this do statement on line 248, so would it calculate numcpus properly still?
do { // we'll exit when the last read didn't read anything
// Move the next line to the beginning of the buffer
const int oldlinelen = strlen(line);
if (sizeof(line) == oldlinelen + 1) // oldlinelen took up entire line
line[0] = '\0';
else // still other lines left to save
memmove(line, line + oldlinelen+1, sizeof(line) - (oldlinelen+1));
// Terminate the new line, reading more if we can't find the newline
char* newline = strchr(line, '\n');
if (newline == NULL) {
const int linelen = strlen(line);
const int bytes_to_read = sizeof(line)-1 - linelen;
CHECK_GT(bytes_to_read, 0); // because the memmove recovered >=1 bytes
chars_read = read(fd, line + linelen, bytes_to_read);
line[linelen + chars_read] = '\0';
newline = strchr(line, '\n');
}
if (newline != NULL)
*newline = '\0';
The closing brace is in line 305.
The programming style in this file is hard to read since the function is several screens long and uses several #ifdef branches. It is necessary though. This source file hides the operating-system-specific parts from the rest of the code, therefore it may look this complicated.
The other code can just access the cpuinfo_cycles_per_second variable, which makes a really simple API.
I have a file with outputs like so,
Joel[12813/76561198008530803] was taken by Huey[3924/252820590256923]
I have set up a getline that will scan for the word taken and store it into a string, I tried using substr but I'm having an issue. The issue is that the numbers between the brackets can be different sizes so when I try something along the lines of find_first_of("/"); in combination with substr, I can't use the second parameter or it might cut off numbers. I get everything past the first forward slash.
What I want in the end is to extract just this portion 76561198008530803 of the string.
What I tried,
while (getline(read, line))
{
if (line.find("taken") != string::npos) {
size_t pos = line.find_first_of("/");
extension = line.substr(pos + 1);
output << extension << endl;
}
Thanks for taking the time!
You can use std::string::find_first_of repeatedly.
size_t pos = line.find_first_of("/");
size_t pos2 = line.find_first_of("]");
std::string extension = line.substr(pos + 1,pos2-1-pos);
I have a lack of understanding about streams. The idea is, to read a file to the ifstream and then working with it. Extract Data from the stream to a string, and discard the part which is now in a string from the stream. Is that possible? Or how to handle those problems?
The following method, is for inserting a file which is properly read by the ifstream. (its a text file, containing informations about "Lost" episodes, its an episodeguide. It works fine, for one element of the class episodes. Every time i instantiate a episode file, i want to check the stream of that file, discard the informations about one episode (its indicated by "****", then the next episode starts) and process the informations discarded in a string. If I create a new object of Episode I want to discard the next informations about the episodes after "****" to the next "****" and so on.
void Episode::read(ifstream& in) {
string contents((istreambuf_iterator<char>(in)), istreambuf_iterator<char>());
size_t episodeEndPos = contents.find("****");
if ( episodeEndPos == -1) {
in.ignore(numeric_limits<char>::max());
in.clear(), in.sync();
fullContent = contents;
}
else { // empty stream for next episode
in.ignore(episodeEndPos + 4);
fullContent = contents.substr(0, episodeEndPos);
}
// fill attributes
setNrHelper();
setTitelHelper();
setFlashbackHelper();
setDescriptionHelper();
}
I tried it with inFile >> words (to read the words, this is a way to get the words out of the stream) another way i was thinking about is, to use .ignore (to ignore an amount of characters in the stream). But that doesnt work as intended. Sorry for my bad english, hopefully its clear what i want to do.
If your goal is at each call of Read() to read the next episode and advance in the file, then the trick is to to use tellg() and seekg() to bookmark the position and update it:
void Episode::Read(ifstream& in) {
streampos pos = in.tellg(); // backup current position
string fullContent;
string contents((istreambuf_iterator<char>(in)), istreambuf_iterator<char>());
size_t episodeEndPos = contents.find("****");
if (episodeEndPos == -1) {
in.ignore(numeric_limits<char>::max());
in.clear(), in.sync();
fullContent = contents;
}
else { // empty stream for next episode
fullContent = contents.substr(0, episodeEndPos);
in.seekg(pos + streamoff(episodeEndPos + 4)); // position file at next episode
}
}
In this way, you can call several time your function, every call reading the next episode.
However, please note that your approach is not optimised. When you construct your contents string from a stream iterator, you load the full rest of the file in the memory, starting at the current position in the stream. So here you keep on reading and reading again big subparts of the file.
Edit: streamlined version adapted to your format
You just need to read the line, check if it's not a separator line and concatenate...
void Episode::Read(ifstream& in) {
string line;
string fullContent;
while (getline(in, line) && line !="****") {
fullContent += line + "\n";
}
cout << "DATENSATZ: " << fullContent << endl; // just to verify content
// fill attributes
//...
}
The code you got reads the entire stream in one go just to use some part of the read text to initialize an object. Imagining a gigantic file that is almost certainly a bad idea. The easier approach is to just read until the end marker is found. In an ideal world, the end marker is easily found. Based on comments it seems to be on a line of its own which would make it quite easy:
void Episode::read(std::istream& in) {
std::string text;
for (std::string line; in >> line && line != "****"; ) {
text += line + "\n";
}
fullContent = text;
}
If the separate isn't on a line of its own, you could use code like this instead:
void Episode::read(std::istream& in) {
std::string text;
for (std::istreambuf_iterator<char> it(in), end; it != end; ++it) {
text.push_back(*it);
if (*it == '*' && 4u <= text.size() && text.substr(text.size() - 4) == "****") {
break;
}
if (4u <= text.size() && text.substr(text.size() - 4u) == "****") {
text.resize(text.size() - 4u);
}
fullContent = text;
}
Both of these approaches would simple read the file from start to end and consume the characters to be extracted in the process, stopping as soon as reading of one record is done.
I'm trying to read a ini file in a value listbox.
Example below works, but i don't know why.
ReadSectionValues contains a string list of ini lines.
How does Rad Studio parse the lines with:
ListValues->Names[i] is first part of the line and ListValues->Values[ListValues->Names[i]] is the second part?
int i;
try
{
//ShowMessage( ListBox1->Items->Strings[ListBox1->ItemIndex] );
TStringList *ListValues = new TStringList;
TIniFile* SettingsFile = new TIniFile(ExtractFilePath(Application->ExeName) + "settings.ini");
String s;
s = ListBox1->Items->Strings[ListBox1->ItemIndex];
SettingsFile->ReadSectionValues( s , ListValues);
for (i = 0; i < (ListValues->Count); i++) {
//ShowMessage(ListValues->Names[i]);
//ShowMessage(ListValues->Values[ListValues->Names[i]]);
vList1->InsertRow(ListValues->Names[i] , ListValues->Values[ListValues->Names[i]],True);
}
delete SettingsFile;
delete ListValues;
}
catch(Exception* e)
{
ShowMessage(e->Message);
}
Please explain, Rad stuido help found no explanation.
void __fastcall ReadSectionValues(
const System::UnicodeString Section,
System::Classes::TStrings* Strings
)
is a method, which gets all lines of ini-file section with name Section and stores them in TStrings-object Strings. Note that these strings have format name=value.
TStrings class has two access properties Names and Values. Their parse algorithm is very simple. If you get stringsObject->Values[1] it takes second line from stringsObject and splits it into two strings on = (or other value of NameValueSeparator property of stringsObject). The string to the left of = (separator) is returned as name (by property Name) and the string to the right of = is returned as value (by property Value).
Closed. This question is opinion-based. It is not currently accepting answers.
Want to improve this question? Update the question so it can be answered with facts and citations by editing this post.
Closed 1 year ago.
The community reviewed whether to reopen this question 1 year ago and left it closed:
Original close reason(s) were not resolved
Improve this question
I'm trying to parse an INI file using C++. Any tips on what is the best way to achieve this? Should I use the Windows API tools for INI file processing (with which I am totally unfamiliar), an open-source solution or attempt to parse it manually?
You can use the Windows API functions, such as GetPrivateProfileString() and GetPrivateProfileInt().
If you need a cross-platform solution, try Boost's Program Options library.
I have never parsed ini files, so I can't be too specific on this issue.
But i have one advice:
Don't reinvent the wheel as long as an existing one meets your requirements
http://en.wikipedia.org/wiki/INI_file#Accessing_INI_files
http://sdl-cfg.sourceforge.net/
http://sourceforge.net/projects/libini/
http://www.codeproject.com/KB/files/config-file-parser.aspx
Good luck :)
If you are already using Qt
QSettings my_settings("filename.ini", QSettings::IniFormat);
Then read a value
my_settings.value("GroupName/ValueName", <<DEFAULT_VAL>>).toInt()
There are a bunch of other converter that convert your INI values into both standard types and Qt types. See Qt documentation on QSettings for more information.
I use SimpleIni. It's cross-platform.
this question is a bit old, but I will post my answer. I have tested various INI classes (you can see them on my website) and I also use simpleIni because I want to work with INI files on both windows and winCE.
Window's GetPrivateProfileString() works only with the registry on winCE.
It is very easy to read with simpleIni. Here is an example:
#include "SimpleIni\SimpleIni.h"
CSimpleIniA ini;
ini.SetUnicode();
ini.LoadFile(FileName);
const char * pVal = ini.GetValue(section, entry, DefaultStr);
inih is a simple ini parser written in C, it comes with a C++ wrapper too. Example usage:
#include "INIReader.h"
INIReader reader("test.ini");
std::cout << "version="
<< reader.GetInteger("protocol", "version", -1) << ", name="
<< reader.Get("user", "name", "UNKNOWN") << ", active="
<< reader.GetBoolean("user", "active", true) << "\n";
The author has also a list of existing libraries here.
Have you tried libconfig; very JSON-like syntax. I prefer it over XML configuration files.
I ended up using inipp which is not mentioned in this thread.
https://github.com/mcmtroffaes/inipp
Was a MIT licensed header only implementation which was simple enough to add to a project and 4 lines to use.
If you are interested in platform portability, you can also try Boost.PropertyTree. It supports ini as persistancy format, though the property tree my be 1 level deep only.
Unless you plan on making the app cross-platform, using the Windows API calls would be the best way to go. Just ignore the note in the API documentation about being provided only for 16-bit app compatibility.
I know this question is very old, but I came upon it because I needed something cross platform for linux, win32... I wrote the function below, it is a single function that can parse INI files, hopefully others will find it useful.
rules & caveats:
buf to parse must be a NULL terminated string. Load your ini file into a char array string and call this function to parse it.
section names must have [] brackets around them, such as this [MySection], also values and sections must begin on a line without leading spaces. It will parse files with Windows \r\n or with Linux \n line endings. Comments should use # or // and begin at the top of the file, no comments should be mixed with INI entry data. Quotes and ticks are trimmed from both ends of the return string. Spaces are only trimmed if they are outside of the quote. Strings are not required to have quotes, and whitespaces are trimmed if quotes are missing. You can also extract numbers or other data, for example if you have a float just perform a atof(ret) on the ret buffer.
// -----note: no escape is nessesary for inner quotes or ticks-----
// -----------------------------example----------------------------
// [Entry2]
// Alignment = 1
// LightLvl=128
// Library = 5555
// StrValA = Inner "quoted" or 'quoted' strings are ok to use
// StrValB = "This a "quoted" or 'quoted' String Value"
// StrValC = 'This a "tick" or 'tick' String Value'
// StrValD = "Missing quote at end will still work
// StrValE = This is another "quote" example
// StrValF = " Spaces inside the quote are preserved "
// StrValG = This works too and spaces are trimmed away
// StrValH =
// ----------------------------------------------------------------
//12oClocker super lean and mean INI file parser (with section support)
//set section to 0 to disable section support
//returns TRUE if we were able to extract a string into ret value
//NextSection is a char* pointer, will be set to zero if no next section is found
//will be set to pointer of next section if it was found.
//use it like this... char* NextSection = 0; GrabIniValue(X,X,X,X,X,&NextSection);
//buf is data to parse, ret is the user supplied return buffer
BOOL GrabIniValue(char* buf, const char* section, const char* valname, char* ret, int retbuflen, char** NextSection)
{
if(!buf){*ret=0; return FALSE;}
char* s = buf; //search starts at "s" pointer
char* e = 0; //end of section pointer
//find section
if(section)
{
int L = strlen(section);
SearchAgain1:
s = strstr(s,section); if(!s){*ret=0; return FALSE;} //find section
if(s > buf && (*(s-1))!='\n'){s+=L; goto SearchAgain1;} //section must be at begining of a line!
s+=L; //found section, skip past section name
while(*s!='\n'){s++;} s++; //spin until next line, s is now begining of section data
e = strstr(s,"\n["); //find begining of next section or end of file
if(e){*e=0;} //if we found begining of next section, null the \n so we don't search past section
if(NextSection) //user passed in a NextSection pointer
{ if(e){*NextSection=(e+1);}else{*NextSection=0;} } //set pointer to next section
}
//restore char at end of section, ret=empty_string, return FALSE
#define RESTORE_E if(e){*e='\n';}
#define SAFE_RETURN RESTORE_E; (*ret)=0; return FALSE
//find valname
int L = strlen(valname);
SearchAgain2:
s = strstr(s,valname); if(!s){SAFE_RETURN;} //find valname
if(s > buf && (*(s-1))!='\n'){s+=L; goto SearchAgain2;} //valname must be at begining of a line!
s+=L; //found valname match, skip past it
while(*s==' ' || *s == '\t'){s++;} //skip spaces and tabs
if(!(*s)){SAFE_RETURN;} //if NULL encounted do safe return
if(*s != '='){goto SearchAgain2;} //no equal sign found after valname, search again
s++; //skip past the equal sign
while(*s==' ' || *s=='\t'){s++;} //skip spaces and tabs
while(*s=='\"' || *s=='\''){s++;} //skip past quotes and ticks
if(!(*s)){SAFE_RETURN;} //if NULL encounted do safe return
char* E = s; //s is now the begining of the valname data
while(*E!='\r' && *E!='\n' && *E!=0){E++;} E--; //find end of line or end of string, then backup 1 char
while(E > s && (*E==' ' || *E=='\t')){E--;} //move backwards past spaces and tabs
while(E > s && (*E=='\"' || *E=='\'')){E--;} //move backwards past quotes and ticks
L = E-s+1; //length of string to extract NOT including NULL
if(L<1 || L+1 > retbuflen){SAFE_RETURN;} //empty string or buffer size too small
strncpy(ret,s,L); //copy the string
ret[L]=0; //null last char on return buffer
RESTORE_E;
return TRUE;
#undef RESTORE_E
#undef SAFE_RETURN
}
How to use... example....
char sFileData[] = "[MySection]\r\n"
"MyValue1 = 123\r\n"
"MyValue2 = 456\r\n"
"MyValue3 = 789\r\n"
"\r\n"
"[MySection]\r\n"
"MyValue1 = Hello1\r\n"
"MyValue2 = Hello2\r\n"
"MyValue3 = Hello3\r\n"
"\r\n";
char str[256];
char* sSec = sFileData;
char secName[] = "[MySection]"; //we support sections with same name
while(sSec)//while we have a valid sNextSec
{
//print values of the sections
char* next=0;//in case we dont have any sucessful grabs
if(GrabIniValue(sSec,secName,"MyValue1",str,sizeof(str),&next)) { printf("MyValue1 = [%s]\n",str); }
if(GrabIniValue(sSec,secName,"MyValue2",str,sizeof(str),0)) { printf("MyValue2 = [%s]\n",str); }
if(GrabIniValue(sSec,secName,"MyValue3",str,sizeof(str),0)) { printf("MyValue3 = [%s]\n",str); }
printf("\n");
sSec = next; //parse next section, next will be null if no more sections to parse
}
Maybe a late answer..But, worth knowing options..If you need a cross-platform solution , definitely you can try GLIB,, its interesting.. (https://developer.gnome.org/glib/stable/glib-Key-value-file-parser.html)