1
0
mirror of https://github.com/VDR4Arch/vdr.git synced 2023-10-10 13:36:52 +02:00
vdr/tools.c
2015-05-22 13:44:43 +02:00

2239 lines
53 KiB
C

/*
* tools.c: Various tools
*
* See the main source file 'vdr.c' for copyright information and
* how to reach the author.
*
* $Id: tools.c 4.1 2015/05/11 14:15:15 kls Exp $
*/
#include "tools.h"
#include <ctype.h>
#include <dirent.h>
#include <errno.h>
extern "C" {
#ifdef boolean
#define HAVE_BOOLEAN
#endif
#include <jpeglib.h>
#undef boolean
}
#include <locale.h>
#include <stdlib.h>
#include <sys/time.h>
#include <sys/vfs.h>
#include <time.h>
#include <unistd.h>
#include <utime.h>
#include "i18n.h"
#include "thread.h"
int SysLogLevel = 3;
#define MAXSYSLOGBUF 256
void syslog_with_tid(int priority, const char *format, ...)
{
va_list ap;
char fmt[MAXSYSLOGBUF];
snprintf(fmt, sizeof(fmt), "[%d] %s", cThread::ThreadId(), format);
va_start(ap, format);
vsyslog(priority, fmt, ap);
va_end(ap);
}
int BCD2INT(int x)
{
return ((1000000 * BCDCHARTOINT((x >> 24) & 0xFF)) +
(10000 * BCDCHARTOINT((x >> 16) & 0xFF)) +
(100 * BCDCHARTOINT((x >> 8) & 0xFF)) +
BCDCHARTOINT( x & 0xFF));
}
ssize_t safe_read(int filedes, void *buffer, size_t size)
{
for (;;) {
ssize_t p = read(filedes, buffer, size);
if (p < 0 && errno == EINTR) {
dsyslog("EINTR while reading from file handle %d - retrying", filedes);
continue;
}
return p;
}
}
ssize_t safe_write(int filedes, const void *buffer, size_t size)
{
ssize_t p = 0;
ssize_t written = size;
const unsigned char *ptr = (const unsigned char *)buffer;
while (size > 0) {
p = write(filedes, ptr, size);
if (p < 0) {
if (errno == EINTR) {
dsyslog("EINTR while writing to file handle %d - retrying", filedes);
continue;
}
break;
}
ptr += p;
size -= p;
}
return p < 0 ? p : written;
}
void writechar(int filedes, char c)
{
safe_write(filedes, &c, sizeof(c));
}
int WriteAllOrNothing(int fd, const uchar *Data, int Length, int TimeoutMs, int RetryMs)
{
int written = 0;
while (Length > 0) {
int w = write(fd, Data + written, Length);
if (w > 0) {
Length -= w;
written += w;
}
else if (written > 0 && !FATALERRNO) {
// we've started writing, so we must finish it!
cTimeMs t;
cPoller Poller(fd, true);
Poller.Poll(RetryMs);
if (TimeoutMs > 0 && (TimeoutMs -= t.Elapsed()) <= 0)
break;
}
else
// nothing written yet (or fatal error), so we can just return the error code:
return w;
}
return written;
}
char *strcpyrealloc(char *dest, const char *src)
{
if (src) {
int l = max(dest ? strlen(dest) : 0, strlen(src)) + 1; // don't let the block get smaller!
dest = (char *)realloc(dest, l);
if (dest)
strcpy(dest, src);
else
esyslog("ERROR: out of memory");
}
else {
free(dest);
dest = NULL;
}
return dest;
}
char *strn0cpy(char *dest, const char *src, size_t n)
{
char *s = dest;
for ( ; --n && (*dest = *src) != 0; dest++, src++) ;
*dest = 0;
return s;
}
char *strreplace(char *s, char c1, char c2)
{
if (s) {
char *p = s;
while (*p) {
if (*p == c1)
*p = c2;
p++;
}
}
return s;
}
char *strreplace(char *s, const char *s1, const char *s2)
{
char *p = strstr(s, s1);
if (p) {
int of = p - s;
int l = strlen(s);
int l1 = strlen(s1);
int l2 = strlen(s2);
if (l2 > l1) {
if (char *NewBuffer = (char *)realloc(s, l + l2 - l1 + 1))
s = NewBuffer;
else {
esyslog("ERROR: out of memory");
return s;
}
}
char *sof = s + of;
if (l2 != l1)
memmove(sof + l2, sof + l1, l - of - l1 + 1);
strncpy(sof, s2, l2);
}
return s;
}
const char *strchrn(const char *s, char c, size_t n)
{
if (n == 0)
return s;
if (s) {
for ( ; *s; s++) {
if (*s == c && --n == 0)
return s;
}
}
return NULL;
}
int strcountchr(const char *s, char c)
{
int n = 0;
if (s && c) {
for ( ; *s; s++) {
if (*s == c)
n++;
}
}
return n;
}
char *stripspace(char *s)
{
if (s && *s) {
for (char *p = s + strlen(s) - 1; p >= s; p--) {
if (!isspace(*p))
break;
*p = 0;
}
}
return s;
}
char *compactspace(char *s)
{
if (s && *s) {
char *t = stripspace(skipspace(s));
char *p = t;
while (p && *p) {
char *q = skipspace(p);
if (q - p > 1)
memmove(p + 1, q, strlen(q) + 1);
p++;
}
if (t != s)
memmove(s, t, strlen(t) + 1);
}
return s;
}
char *compactchars(char *s, char c)
{
if (s && *s && c) {
char *t = s;
char *p = s;
int n = 0;
while (*p) {
if (*p != c) {
*t++ = *p;
n = 0;
}
else if (t != s && n == 0) {
*t++ = *p;
n++;
}
p++;
}
if (n)
t--; // the last character was c
*t = 0;
}
return s;
}
cString strescape(const char *s, const char *chars)
{
char *buffer;
const char *p = s;
char *t = NULL;
while (*p) {
if (strchr(chars, *p)) {
if (!t) {
buffer = MALLOC(char, 2 * strlen(s) + 1);
t = buffer + (p - s);
s = strcpy(buffer, s);
}
*t++ = '\\';
}
if (t)
*t++ = *p;
p++;
}
if (t)
*t = 0;
return cString(s, t != NULL);
}
cString strgetval(const char *s, const char *name, char d)
{
if (s && name) {
int l = strlen(name);
const char *t = s;
while (const char *p = strstr(t, name)) {
t = skipspace(p + l);
if (p == s || *(p - 1) <= ' ') {
if (*t == d) {
t = skipspace(t + 1);
const char *v = t;
while (*t > ' ')
t++;
return cString(v, t);
break;
}
}
}
}
return NULL;
}
bool startswith(const char *s, const char *p)
{
while (*p) {
if (*p++ != *s++)
return false;
}
return true;
}
bool endswith(const char *s, const char *p)
{
const char *se = s + strlen(s) - 1;
const char *pe = p + strlen(p) - 1;
while (pe >= p) {
if (*pe-- != *se-- || (se < s && pe >= p))
return false;
}
return true;
}
bool isempty(const char *s)
{
return !(s && *skipspace(s));
}
int numdigits(int n)
{
int res = 1;
while (n >= 10) {
n /= 10;
res++;
}
return res;
}
bool isnumber(const char *s)
{
if (!s || !*s)
return false;
do {
if (!isdigit(*s))
return false;
} while (*++s);
return true;
}
int64_t StrToNum(const char *s)
{
char *t = NULL;
int64_t n = strtoll(s, &t, 10);
if (t) {
switch (*t) {
case 'T': n *= 1024;
case 'G': n *= 1024;
case 'M': n *= 1024;
case 'K': n *= 1024;
}
}
return n;
}
bool StrInArray(const char *a[], const char *s)
{
if (a) {
while (*a) {
if (strcmp(*a, s) == 0)
return true;
a++;
}
}
return false;
}
cString AddDirectory(const char *DirName, const char *FileName)
{
return cString::sprintf("%s/%s", DirName && *DirName ? DirName : ".", FileName);
}
#define DECIMAL_POINT_C '.'
double atod(const char *s)
{
static lconv *loc = localeconv();
if (*loc->decimal_point != DECIMAL_POINT_C) {
char buf[strlen(s) + 1];
char *p = buf;
while (*s) {
if (*s == DECIMAL_POINT_C)
*p = *loc->decimal_point;
else
*p = *s;
p++;
s++;
}
*p = 0;
return atof(buf);
}
else
return atof(s);
}
cString dtoa(double d, const char *Format)
{
static lconv *loc = localeconv();
char buf[16];
snprintf(buf, sizeof(buf), Format, d);
if (*loc->decimal_point != DECIMAL_POINT_C)
strreplace(buf, *loc->decimal_point, DECIMAL_POINT_C);
return buf;
}
cString itoa(int n)
{
char buf[16];
snprintf(buf, sizeof(buf), "%d", n);
return buf;
}
bool EntriesOnSameFileSystem(const char *File1, const char *File2)
{
struct stat st;
if (stat(File1, &st) == 0) {
dev_t dev1 = st.st_dev;
if (stat(File2, &st) == 0)
return st.st_dev == dev1;
else
LOG_ERROR_STR(File2);
}
else
LOG_ERROR_STR(File1);
return true; // we only return false if both files actually exist and are in different file systems!
}
int FreeDiskSpaceMB(const char *Directory, int *UsedMB)
{
if (UsedMB)
*UsedMB = 0;
int Free = 0;
struct statfs statFs;
if (statfs(Directory, &statFs) == 0) {
double blocksPerMeg = 1024.0 * 1024.0 / statFs.f_bsize;
if (UsedMB)
*UsedMB = int((statFs.f_blocks - statFs.f_bfree) / blocksPerMeg);
Free = int(statFs.f_bavail / blocksPerMeg);
}
else
LOG_ERROR_STR(Directory);
return Free;
}
bool DirectoryOk(const char *DirName, bool LogErrors)
{
struct stat ds;
if (stat(DirName, &ds) == 0) {
if (S_ISDIR(ds.st_mode)) {
if (access(DirName, R_OK | W_OK | X_OK) == 0)
return true;
else if (LogErrors)
esyslog("ERROR: can't access %s", DirName);
}
else if (LogErrors)
esyslog("ERROR: %s is not a directory", DirName);
}
else if (LogErrors)
LOG_ERROR_STR(DirName);
return false;
}
bool MakeDirs(const char *FileName, bool IsDirectory)
{
bool result = true;
char *s = strdup(FileName);
char *p = s;
if (*p == '/')
p++;
while ((p = strchr(p, '/')) != NULL || IsDirectory) {
if (p)
*p = 0;
struct stat fs;
if (stat(s, &fs) != 0 || !S_ISDIR(fs.st_mode)) {
dsyslog("creating directory %s", s);
if (mkdir(s, ACCESSPERMS) == -1) {
LOG_ERROR_STR(s);
result = false;
break;
}
}
if (p)
*p++ = '/';
else
break;
}
free(s);
return result;
}
bool RemoveFileOrDir(const char *FileName, bool FollowSymlinks)
{
struct stat st;
if (stat(FileName, &st) == 0) {
if (S_ISDIR(st.st_mode)) {
cReadDir d(FileName);
if (d.Ok()) {
struct dirent *e;
while ((e = d.Next()) != NULL) {
cString buffer = AddDirectory(FileName, e->d_name);
if (FollowSymlinks) {
struct stat st2;
if (lstat(buffer, &st2) == 0) {
if (S_ISLNK(st2.st_mode)) {
int size = st2.st_size + 1;
char *l = MALLOC(char, size);
int n = readlink(buffer, l, size - 1);
if (n < 0) {
if (errno != EINVAL)
LOG_ERROR_STR(*buffer);
}
else {
l[n] = 0;
dsyslog("removing %s", l);
if (remove(l) < 0)
LOG_ERROR_STR(l);
}
free(l);
}
}
else if (errno != ENOENT) {
LOG_ERROR_STR(FileName);
return false;
}
}
dsyslog("removing %s", *buffer);
if (remove(buffer) < 0)
LOG_ERROR_STR(*buffer);
}
}
else {
LOG_ERROR_STR(FileName);
return false;
}
}
dsyslog("removing %s", FileName);
if (remove(FileName) < 0) {
LOG_ERROR_STR(FileName);
return false;
}
}
else if (errno != ENOENT) {
LOG_ERROR_STR(FileName);
return false;
}
return true;
}
bool RemoveEmptyDirectories(const char *DirName, bool RemoveThis, const char *IgnoreFiles[])
{
bool HasIgnoredFiles = false;
cReadDir d(DirName);
if (d.Ok()) {
bool empty = true;
struct dirent *e;
while ((e = d.Next()) != NULL) {
if (strcmp(e->d_name, "lost+found")) {
cString buffer = AddDirectory(DirName, e->d_name);
struct stat st;
if (stat(buffer, &st) == 0) {
if (S_ISDIR(st.st_mode)) {
if (!RemoveEmptyDirectories(buffer, true, IgnoreFiles))
empty = false;
}
else if (RemoveThis && IgnoreFiles && StrInArray(IgnoreFiles, e->d_name))
HasIgnoredFiles = true;
else
empty = false;
}
else {
LOG_ERROR_STR(*buffer);
empty = false;
}
}
}
if (RemoveThis && empty) {
if (HasIgnoredFiles) {
while (*IgnoreFiles) {
cString buffer = AddDirectory(DirName, *IgnoreFiles);
if (access(buffer, F_OK) == 0) {
dsyslog("removing %s", *buffer);
if (remove(buffer) < 0) {
LOG_ERROR_STR(*buffer);
return false;
}
}
IgnoreFiles++;
}
}
dsyslog("removing %s", DirName);
if (remove(DirName) < 0) {
LOG_ERROR_STR(DirName);
return false;
}
}
return empty;
}
else
LOG_ERROR_STR(DirName);
return false;
}
int DirSizeMB(const char *DirName)
{
cReadDir d(DirName);
if (d.Ok()) {
int size = 0;
struct dirent *e;
while (size >= 0 && (e = d.Next()) != NULL) {
cString buffer = AddDirectory(DirName, e->d_name);
struct stat st;
if (stat(buffer, &st) == 0) {
if (S_ISDIR(st.st_mode)) {
int n = DirSizeMB(buffer);
if (n >= 0)
size += n;
else
size = -1;
}
else
size += st.st_size / MEGABYTE(1);
}
else {
LOG_ERROR_STR(*buffer);
size = -1;
}
}
return size;
}
else if (errno != ENOENT)
LOG_ERROR_STR(DirName);
return -1;
}
char *ReadLink(const char *FileName)
{
if (!FileName)
return NULL;
char *TargetName = canonicalize_file_name(FileName);
if (!TargetName) {
if (errno == ENOENT) // file doesn't exist
TargetName = strdup(FileName);
else // some other error occurred
LOG_ERROR_STR(FileName);
}
return TargetName;
}
bool SpinUpDisk(const char *FileName)
{
for (int n = 0; n < 10; n++) {
cString buf;
if (DirectoryOk(FileName))
buf = cString::sprintf("%s/vdr-%06d", *FileName ? FileName : ".", n);
else
buf = cString::sprintf("%s.vdr-%06d", FileName, n);
if (access(buf, F_OK) != 0) { // the file does not exist
timeval tp1, tp2;
gettimeofday(&tp1, NULL);
int f = open(buf, O_WRONLY | O_CREAT, DEFFILEMODE);
// O_SYNC doesn't work on all file systems
if (f >= 0) {
if (fdatasync(f) < 0)
LOG_ERROR_STR(*buf);
close(f);
remove(buf);
gettimeofday(&tp2, NULL);
double seconds = (((long long)tp2.tv_sec * 1000000 + tp2.tv_usec) - ((long long)tp1.tv_sec * 1000000 + tp1.tv_usec)) / 1000000.0;
if (seconds > 0.5)
dsyslog("SpinUpDisk took %.2f seconds", seconds);
return true;
}
else
LOG_ERROR_STR(*buf);
}
}
esyslog("ERROR: SpinUpDisk failed");
return false;
}
void TouchFile(const char *FileName)
{
if (utime(FileName, NULL) == -1 && errno != ENOENT)
LOG_ERROR_STR(FileName);
}
time_t LastModifiedTime(const char *FileName)
{
struct stat fs;
if (stat(FileName, &fs) == 0)
return fs.st_mtime;
return 0;
}
off_t FileSize(const char *FileName)
{
struct stat fs;
if (stat(FileName, &fs) == 0)
return fs.st_size;
return -1;
}
// --- cTimeMs ---------------------------------------------------------------
cTimeMs::cTimeMs(int Ms)
{
if (Ms >= 0)
Set(Ms);
else
begin = 0;
}
uint64_t cTimeMs::Now(void)
{
#if _POSIX_TIMERS > 0 && defined(_POSIX_MONOTONIC_CLOCK)
#define MIN_RESOLUTION 5 // ms
static bool initialized = false;
static bool monotonic = false;
struct timespec tp;
if (!initialized) {
// check if monotonic timer is available and provides enough accurate resolution:
if (clock_getres(CLOCK_MONOTONIC, &tp) == 0) {
long Resolution = tp.tv_nsec;
// require a minimum resolution:
if (tp.tv_sec == 0 && tp.tv_nsec <= MIN_RESOLUTION * 1000000) {
if (clock_gettime(CLOCK_MONOTONIC, &tp) == 0) {
dsyslog("cTimeMs: using monotonic clock (resolution is %ld ns)", Resolution);
monotonic = true;
}
else
esyslog("cTimeMs: clock_gettime(CLOCK_MONOTONIC) failed");
}
else
dsyslog("cTimeMs: not using monotonic clock - resolution is too bad (%ld s %ld ns)", tp.tv_sec, tp.tv_nsec);
}
else
esyslog("cTimeMs: clock_getres(CLOCK_MONOTONIC) failed");
initialized = true;
}
if (monotonic) {
if (clock_gettime(CLOCK_MONOTONIC, &tp) == 0)
return (uint64_t(tp.tv_sec)) * 1000 + tp.tv_nsec / 1000000;
esyslog("cTimeMs: clock_gettime(CLOCK_MONOTONIC) failed");
monotonic = false;
// fall back to gettimeofday()
}
#else
# warning Posix monotonic clock not available
#endif
struct timeval t;
if (gettimeofday(&t, NULL) == 0)
return (uint64_t(t.tv_sec)) * 1000 + t.tv_usec / 1000;
return 0;
}
void cTimeMs::Set(int Ms)
{
begin = Now() + Ms;
}
bool cTimeMs::TimedOut(void) const
{
return Now() >= begin;
}
uint64_t cTimeMs::Elapsed(void) const
{
return Now() - begin;
}
// --- UTF-8 support ---------------------------------------------------------
static uint SystemToUtf8[128] = { 0 };
int Utf8CharLen(const char *s)
{
if (cCharSetConv::SystemCharacterTable())
return 1;
#define MT(s, m, v) ((*(s) & (m)) == (v)) // Mask Test
if (MT(s, 0xE0, 0xC0) && MT(s + 1, 0xC0, 0x80))
return 2;
if (MT(s, 0xF0, 0xE0) && MT(s + 1, 0xC0, 0x80) && MT(s + 2, 0xC0, 0x80))
return 3;
if (MT(s, 0xF8, 0xF0) && MT(s + 1, 0xC0, 0x80) && MT(s + 2, 0xC0, 0x80) && MT(s + 3, 0xC0, 0x80))
return 4;
return 1;
}
uint Utf8CharGet(const char *s, int Length)
{
if (cCharSetConv::SystemCharacterTable())
return (uchar)*s < 128 ? *s : SystemToUtf8[(uchar)*s - 128];
if (!Length)
Length = Utf8CharLen(s);
switch (Length) {
case 2: return ((*s & 0x1F) << 6) | (*(s + 1) & 0x3F);
case 3: return ((*s & 0x0F) << 12) | ((*(s + 1) & 0x3F) << 6) | (*(s + 2) & 0x3F);
case 4: return ((*s & 0x07) << 18) | ((*(s + 1) & 0x3F) << 12) | ((*(s + 2) & 0x3F) << 6) | (*(s + 3) & 0x3F);
default: ;
}
return *s;
}
int Utf8CharSet(uint c, char *s)
{
if (c < 0x80 || cCharSetConv::SystemCharacterTable()) {
if (s)
*s = c;
return 1;
}
if (c < 0x800) {
if (s) {
*s++ = ((c >> 6) & 0x1F) | 0xC0;
*s = (c & 0x3F) | 0x80;
}
return 2;
}
if (c < 0x10000) {
if (s) {
*s++ = ((c >> 12) & 0x0F) | 0xE0;
*s++ = ((c >> 6) & 0x3F) | 0x80;
*s = (c & 0x3F) | 0x80;
}
return 3;
}
if (c < 0x110000) {
if (s) {
*s++ = ((c >> 18) & 0x07) | 0xF0;
*s++ = ((c >> 12) & 0x3F) | 0x80;
*s++ = ((c >> 6) & 0x3F) | 0x80;
*s = (c & 0x3F) | 0x80;
}
return 4;
}
return 0; // can't convert to UTF-8
}
int Utf8SymChars(const char *s, int Symbols)
{
if (cCharSetConv::SystemCharacterTable())
return Symbols;
int n = 0;
while (*s && Symbols--) {
int sl = Utf8CharLen(s);
s += sl;
n += sl;
}
return n;
}
int Utf8StrLen(const char *s)
{
if (cCharSetConv::SystemCharacterTable())
return strlen(s);
int n = 0;
while (*s) {
s += Utf8CharLen(s);
n++;
}
return n;
}
char *Utf8Strn0Cpy(char *Dest, const char *Src, int n)
{
if (cCharSetConv::SystemCharacterTable())
return strn0cpy(Dest, Src, n);
char *d = Dest;
while (*Src) {
int sl = Utf8CharLen(Src);
n -= sl;
if (n > 0) {
while (sl--)
*d++ = *Src++;
}
else
break;
}
*d = 0;
return Dest;
}
int Utf8ToArray(const char *s, uint *a, int Size)
{
int n = 0;
while (*s && --Size > 0) {
if (cCharSetConv::SystemCharacterTable())
*a++ = (uchar)(*s++);
else {
int sl = Utf8CharLen(s);
*a++ = Utf8CharGet(s, sl);
s += sl;
}
n++;
}
if (Size > 0)
*a = 0;
return n;
}
int Utf8FromArray(const uint *a, char *s, int Size, int Max)
{
int NumChars = 0;
int NumSyms = 0;
while (*a && NumChars < Size) {
if (Max >= 0 && NumSyms++ >= Max)
break;
if (cCharSetConv::SystemCharacterTable()) {
*s++ = *a++;
NumChars++;
}
else {
int sl = Utf8CharSet(*a);
if (NumChars + sl <= Size) {
Utf8CharSet(*a, s);
a++;
s += sl;
NumChars += sl;
}
else
break;
}
}
if (NumChars < Size)
*s = 0;
return NumChars;
}
// --- cCharSetConv ----------------------------------------------------------
char *cCharSetConv::systemCharacterTable = NULL;
cCharSetConv::cCharSetConv(const char *FromCode, const char *ToCode)
{
if (!FromCode)
FromCode = systemCharacterTable ? systemCharacterTable : "UTF-8";
if (!ToCode)
ToCode = "UTF-8";
cd = iconv_open(ToCode, FromCode);
result = NULL;
length = 0;
}
cCharSetConv::~cCharSetConv()
{
free(result);
if (cd != (iconv_t)-1)
iconv_close(cd);
}
void cCharSetConv::SetSystemCharacterTable(const char *CharacterTable)
{
free(systemCharacterTable);
systemCharacterTable = NULL;
if (!strcasestr(CharacterTable, "UTF-8")) {
// Set up a map for the character values 128...255:
char buf[129];
for (int i = 0; i < 128; i++)
buf[i] = i + 128;
buf[128] = 0;
cCharSetConv csc(CharacterTable);
const char *s = csc.Convert(buf);
int i = 0;
while (*s) {
int sl = Utf8CharLen(s);
SystemToUtf8[i] = Utf8CharGet(s, sl);
s += sl;
i++;
}
systemCharacterTable = strdup(CharacterTable);
}
}
const char *cCharSetConv::Convert(const char *From, char *To, size_t ToLength)
{
if (cd != (iconv_t)-1 && From && *From) {
char *FromPtr = (char *)From;
size_t FromLength = strlen(From);
char *ToPtr = To;
if (!ToPtr) {
int NewLength = max(length, FromLength * 2); // some reserve to avoid later reallocations
if (char *NewBuffer = (char *)realloc(result, NewLength)) {
length = NewLength;
result = NewBuffer;
}
else {
esyslog("ERROR: out of memory");
return From;
}
ToPtr = result;
ToLength = length;
}
else if (!ToLength)
return From; // can't convert into a zero sized buffer
ToLength--; // save space for terminating 0
char *Converted = ToPtr;
while (FromLength > 0) {
if (iconv(cd, &FromPtr, &FromLength, &ToPtr, &ToLength) == size_t(-1)) {
if (errno == E2BIG || errno == EILSEQ && ToLength < 1) {
if (To)
break; // caller provided a fixed size buffer, but it was too small
// The result buffer is too small, so increase it:
size_t d = ToPtr - result;
size_t r = length / 2;
int NewLength = length + r;
if (char *NewBuffer = (char *)realloc(result, NewLength)) {
length = NewLength;
Converted = result = NewBuffer;
}
else {
esyslog("ERROR: out of memory");
return From;
}
ToLength += r;
ToPtr = result + d;
}
if (errno == EILSEQ) {
// A character can't be converted, so mark it with '?' and proceed:
FromPtr++;
FromLength--;
*ToPtr++ = '?';
ToLength--;
}
else if (errno != E2BIG)
return From; // unknown error, return original string
}
}
*ToPtr = 0;
return Converted;
}
return From;
}
// --- cString ---------------------------------------------------------------
cString::cString(const char *S, bool TakePointer)
{
s = TakePointer ? (char *)S : S ? strdup(S) : NULL;
}
cString::cString(const char *S, const char *To)
{
if (!S)
s = NULL;
else if (!To)
s = strdup(S);
else {
int l = To - S;
s = MALLOC(char, l + 1);
strncpy(s, S, l);
s[l] = 0;
}
}
cString::cString(const cString &String)
{
s = String.s ? strdup(String.s) : NULL;
}
cString::~cString()
{
free(s);
}
cString &cString::operator=(const cString &String)
{
if (this == &String)
return *this;
free(s);
s = String.s ? strdup(String.s) : NULL;
return *this;
}
cString &cString::operator=(const char *String)
{
if (s == String)
return *this;
free(s);
s = String ? strdup(String) : NULL;
return *this;
}
cString &cString::Append(const char *String)
{
int l1 = strlen(s);
int l2 = strlen(String);
char *p = (char *)realloc(s, l1 + l2 + 1);
if (p != s)
strcpy(p, s);
strcpy(p + l1, String);
return *this;
}
cString &cString::Truncate(int Index)
{
int l = strlen(s);
if (Index < 0)
Index = l + Index;
if (Index >= 0 && Index < l)
s[Index] = 0;
return *this;
}
cString &cString::CompactChars(char c)
{
compactchars(s, c);
return *this;
}
cString cString::sprintf(const char *fmt, ...)
{
va_list ap;
va_start(ap, fmt);
char *buffer;
if (!fmt || vasprintf(&buffer, fmt, ap) < 0) {
esyslog("error in vasprintf('%s', ...)", fmt);
buffer = strdup("???");
}
va_end(ap);
return cString(buffer, true);
}
cString cString::vsprintf(const char *fmt, va_list &ap)
{
char *buffer;
if (!fmt || vasprintf(&buffer, fmt, ap) < 0) {
esyslog("error in vasprintf('%s', ...)", fmt);
buffer = strdup("???");
}
return cString(buffer, true);
}
cString WeekDayName(int WeekDay)
{
char buffer[16];
WeekDay = WeekDay == 0 ? 6 : WeekDay - 1; // we start with Monday==0!
if (0 <= WeekDay && WeekDay <= 6) {
// TRANSLATORS: abbreviated weekdays, beginning with monday (must all be 3 letters!)
const char *day = tr("MonTueWedThuFriSatSun");
day += Utf8SymChars(day, WeekDay * 3);
strn0cpy(buffer, day, min(Utf8SymChars(day, 3) + 1, int(sizeof(buffer))));
return buffer;
}
else
return "???";
}
cString WeekDayName(time_t t)
{
struct tm tm_r;
return WeekDayName(localtime_r(&t, &tm_r)->tm_wday);
}
cString WeekDayNameFull(int WeekDay)
{
WeekDay = WeekDay == 0 ? 6 : WeekDay - 1; // we start with Monday==0!
switch (WeekDay) {
case 0: return tr("Monday");
case 1: return tr("Tuesday");
case 2: return tr("Wednesday");
case 3: return tr("Thursday");
case 4: return tr("Friday");
case 5: return tr("Saturday");
case 6: return tr("Sunday");
default: return "???";
}
}
cString WeekDayNameFull(time_t t)
{
struct tm tm_r;
return WeekDayNameFull(localtime_r(&t, &tm_r)->tm_wday);
}
cString DayDateTime(time_t t)
{
char buffer[32];
if (t == 0)
time(&t);
struct tm tm_r;
tm *tm = localtime_r(&t, &tm_r);
snprintf(buffer, sizeof(buffer), "%s %02d.%02d. %02d:%02d", *WeekDayName(tm->tm_wday), tm->tm_mday, tm->tm_mon + 1, tm->tm_hour, tm->tm_min);
return buffer;
}
cString TimeToString(time_t t)
{
char buffer[32];
if (ctime_r(&t, buffer)) {
buffer[strlen(buffer) - 1] = 0; // strip trailing newline
return buffer;
}
return "???";
}
cString DateString(time_t t)
{
char buf[32];
struct tm tm_r;
tm *tm = localtime_r(&t, &tm_r);
char *p = stpcpy(buf, WeekDayName(tm->tm_wday));
*p++ = ' ';
strftime(p, sizeof(buf) - (p - buf), "%d.%m.%Y", tm);
return buf;
}
cString ShortDateString(time_t t)
{
char buf[32];
struct tm tm_r;
tm *tm = localtime_r(&t, &tm_r);
strftime(buf, sizeof(buf), "%d.%m.%y", tm);
return buf;
}
cString TimeString(time_t t)
{
char buf[25];
struct tm tm_r;
strftime(buf, sizeof(buf), "%R", localtime_r(&t, &tm_r));
return buf;
}
// --- RgbToJpeg -------------------------------------------------------------
#define JPEGCOMPRESSMEM 500000
struct tJpegCompressData {
int size;
uchar *mem;
};
static void JpegCompressInitDestination(j_compress_ptr cinfo)
{
tJpegCompressData *jcd = (tJpegCompressData *)cinfo->client_data;
if (jcd) {
cinfo->dest->free_in_buffer = jcd->size = JPEGCOMPRESSMEM;
cinfo->dest->next_output_byte = jcd->mem = MALLOC(uchar, jcd->size);
}
}
static boolean JpegCompressEmptyOutputBuffer(j_compress_ptr cinfo)
{
tJpegCompressData *jcd = (tJpegCompressData *)cinfo->client_data;
if (jcd) {
int Used = jcd->size;
int NewSize = jcd->size + JPEGCOMPRESSMEM;
if (uchar *NewBuffer = (uchar *)realloc(jcd->mem, NewSize)) {
jcd->size = NewSize;
jcd->mem = NewBuffer;
}
else {
esyslog("ERROR: out of memory");
return false;
}
if (jcd->mem) {
cinfo->dest->next_output_byte = jcd->mem + Used;
cinfo->dest->free_in_buffer = jcd->size - Used;
return true;
}
}
return false;
}
static void JpegCompressTermDestination(j_compress_ptr cinfo)
{
tJpegCompressData *jcd = (tJpegCompressData *)cinfo->client_data;
if (jcd) {
int Used = cinfo->dest->next_output_byte - jcd->mem;
if (Used < jcd->size) {
if (uchar *NewBuffer = (uchar *)realloc(jcd->mem, Used)) {
jcd->size = Used;
jcd->mem = NewBuffer;
}
else
esyslog("ERROR: out of memory");
}
}
}
uchar *RgbToJpeg(uchar *Mem, int Width, int Height, int &Size, int Quality)
{
if (Quality < 0)
Quality = 0;
else if (Quality > 100)
Quality = 100;
jpeg_destination_mgr jdm;
jdm.init_destination = JpegCompressInitDestination;
jdm.empty_output_buffer = JpegCompressEmptyOutputBuffer;
jdm.term_destination = JpegCompressTermDestination;
struct jpeg_compress_struct cinfo;
struct jpeg_error_mgr jerr;
cinfo.err = jpeg_std_error(&jerr);
jpeg_create_compress(&cinfo);
cinfo.dest = &jdm;
tJpegCompressData jcd;
cinfo.client_data = &jcd;
cinfo.image_width = Width;
cinfo.image_height = Height;
cinfo.input_components = 3;
cinfo.in_color_space = JCS_RGB;
jpeg_set_defaults(&cinfo);
jpeg_set_quality(&cinfo, Quality, true);
jpeg_start_compress(&cinfo, true);
int rs = Width * 3;
JSAMPROW rp[Height];
for (int k = 0; k < Height; k++)
rp[k] = &Mem[rs * k];
jpeg_write_scanlines(&cinfo, rp, Height);
jpeg_finish_compress(&cinfo);
jpeg_destroy_compress(&cinfo);
Size = jcd.size;
return jcd.mem;
}
// --- cBase64Encoder --------------------------------------------------------
const char *cBase64Encoder::b64 = "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/";
cBase64Encoder::cBase64Encoder(const uchar *Data, int Length, int MaxResult)
{
data = Data;
length = Length;
maxResult = MaxResult;
i = 0;
result = MALLOC(char, maxResult + 1);
}
cBase64Encoder::~cBase64Encoder()
{
free(result);
}
const char *cBase64Encoder::NextLine(void)
{
int r = 0;
while (i < length && r < maxResult - 3) {
result[r++] = b64[(data[i] >> 2) & 0x3F];
uchar c = (data[i] << 4) & 0x3F;
if (++i < length)
c |= (data[i] >> 4) & 0x0F;
result[r++] = b64[c];
if (i < length) {
c = (data[i] << 2) & 0x3F;
if (++i < length)
c |= (data[i] >> 6) & 0x03;
result[r++] = b64[c];
}
else {
i++;
result[r++] = '=';
}
if (i < length) {
c = data[i] & 0x3F;
result[r++] = b64[c];
}
else
result[r++] = '=';
i++;
}
if (r > 0) {
result[r] = 0;
return result;
}
return NULL;
}
// --- cBitStream ------------------------------------------------------------
int cBitStream::GetBit(void)
{
if (index >= length)
return 1;
int r = (data[index >> 3] >> (7 - (index & 7))) & 1;
++index;
return r;
}
uint32_t cBitStream::GetBits(int n)
{
uint32_t r = 0;
while (n--)
r |= GetBit() << n;
return r;
}
void cBitStream::ByteAlign(void)
{
int n = index % 8;
if (n > 0)
SkipBits(8 - n);
}
void cBitStream::WordAlign(void)
{
int n = index % 16;
if (n > 0)
SkipBits(16 - n);
}
bool cBitStream::SetLength(int Length)
{
if (Length > length)
return false;
length = Length;
return true;
}
// --- cReadLine -------------------------------------------------------------
cReadLine::cReadLine(void)
{
size = 0;
buffer = NULL;
}
cReadLine::~cReadLine()
{
free(buffer);
}
char *cReadLine::Read(FILE *f)
{
int n = getline(&buffer, &size, f);
if (n > 0) {
n--;
if (buffer[n] == '\n') {
buffer[n] = 0;
if (n > 0) {
n--;
if (buffer[n] == '\r')
buffer[n] = 0;
}
}
return buffer;
}
return NULL;
}
// --- cPoller ---------------------------------------------------------------
cPoller::cPoller(int FileHandle, bool Out)
{
numFileHandles = 0;
Add(FileHandle, Out);
}
bool cPoller::Add(int FileHandle, bool Out)
{
if (FileHandle >= 0) {
for (int i = 0; i < numFileHandles; i++) {
if (pfd[i].fd == FileHandle && pfd[i].events == (Out ? POLLOUT : POLLIN))
return true;
}
if (numFileHandles < MaxPollFiles) {
pfd[numFileHandles].fd = FileHandle;
pfd[numFileHandles].events = Out ? POLLOUT : POLLIN;
pfd[numFileHandles].revents = 0;
numFileHandles++;
return true;
}
esyslog("ERROR: too many file handles in cPoller");
}
return false;
}
void cPoller::Del(int FileHandle, bool Out)
{
if (FileHandle >= 0) {
for (int i = 0; i < numFileHandles; i++) {
if (pfd[i].fd == FileHandle && pfd[i].events == (Out ? POLLOUT : POLLIN)) {
if (i < numFileHandles - 1)
memmove(&pfd[i], &pfd[i + 1], (numFileHandles - i - 1) * sizeof(pollfd));
numFileHandles--;
}
}
}
}
bool cPoller::Poll(int TimeoutMs)
{
if (numFileHandles) {
if (poll(pfd, numFileHandles, TimeoutMs) != 0)
return true; // returns true even in case of an error, to let the caller
// access the file and thus see the error code
}
return false;
}
// --- cReadDir --------------------------------------------------------------
cReadDir::cReadDir(const char *Directory)
{
directory = opendir(Directory);
}
cReadDir::~cReadDir()
{
if (directory)
closedir(directory);
}
struct dirent *cReadDir::Next(void)
{
if (directory) {
while (readdir_r(directory, &u.d, &result) == 0 && result) {
if (strcmp(result->d_name, ".") && strcmp(result->d_name, ".."))
return result;
}
}
return NULL;
}
// --- cStringList -----------------------------------------------------------
cStringList::~cStringList()
{
Clear();
}
int cStringList::Find(const char *s) const
{
for (int i = 0; i < Size(); i++) {
if (!strcmp(s, At(i)))
return i;
}
return -1;
}
void cStringList::Clear(void)
{
for (int i = 0; i < Size(); i++)
free(At(i));
cVector<char *>::Clear();
}
// --- cFileNameList ---------------------------------------------------------
// TODO better GetFileNames(const char *Directory, cStringList *List)?
cFileNameList::cFileNameList(const char *Directory, bool DirsOnly)
{
Load(Directory, DirsOnly);
}
bool cFileNameList::Load(const char *Directory, bool DirsOnly)
{
Clear();
if (Directory) {
cReadDir d(Directory);
struct dirent *e;
if (d.Ok()) {
while ((e = d.Next()) != NULL) {
if (DirsOnly) {
struct stat ds;
if (stat(AddDirectory(Directory, e->d_name), &ds) == 0) {
if (!S_ISDIR(ds.st_mode))
continue;
}
}
Append(strdup(e->d_name));
}
Sort();
return true;
}
else
LOG_ERROR_STR(Directory);
}
return false;
}
// --- cFile -----------------------------------------------------------------
bool cFile::files[FD_SETSIZE] = { false };
int cFile::maxFiles = 0;
cFile::cFile(void)
{
f = -1;
}
cFile::~cFile()
{
Close();
}
bool cFile::Open(const char *FileName, int Flags, mode_t Mode)
{
if (!IsOpen())
return Open(open(FileName, Flags, Mode));
esyslog("ERROR: attempt to re-open %s", FileName);
return false;
}
bool cFile::Open(int FileDes)
{
if (FileDes >= 0) {
if (!IsOpen()) {
f = FileDes;
if (f >= 0) {
if (f < FD_SETSIZE) {
if (f >= maxFiles)
maxFiles = f + 1;
if (!files[f])
files[f] = true;
else
esyslog("ERROR: file descriptor %d already in files[]", f);
return true;
}
else
esyslog("ERROR: file descriptor %d is larger than FD_SETSIZE (%d)", f, FD_SETSIZE);
}
}
else
esyslog("ERROR: attempt to re-open file descriptor %d", FileDes);
}
return false;
}
void cFile::Close(void)
{
if (f >= 0) {
close(f);
files[f] = false;
f = -1;
}
}
bool cFile::Ready(bool Wait)
{
return f >= 0 && AnyFileReady(f, Wait ? 1000 : 0);
}
bool cFile::AnyFileReady(int FileDes, int TimeoutMs)
{
fd_set set;
FD_ZERO(&set);
for (int i = 0; i < maxFiles; i++) {
if (files[i])
FD_SET(i, &set);
}
if (0 <= FileDes && FileDes < FD_SETSIZE && !files[FileDes])
FD_SET(FileDes, &set); // in case we come in with an arbitrary descriptor
if (TimeoutMs == 0)
TimeoutMs = 10; // load gets too heavy with 0
struct timeval timeout;
timeout.tv_sec = TimeoutMs / 1000;
timeout.tv_usec = (TimeoutMs % 1000) * 1000;
return select(FD_SETSIZE, &set, NULL, NULL, &timeout) > 0 && (FileDes < 0 || FD_ISSET(FileDes, &set));
}
bool cFile::FileReady(int FileDes, int TimeoutMs)
{
fd_set set;
struct timeval timeout;
FD_ZERO(&set);
FD_SET(FileDes, &set);
if (TimeoutMs >= 0) {
if (TimeoutMs < 100)
TimeoutMs = 100;
timeout.tv_sec = TimeoutMs / 1000;
timeout.tv_usec = (TimeoutMs % 1000) * 1000;
}
return select(FD_SETSIZE, &set, NULL, NULL, (TimeoutMs >= 0) ? &timeout : NULL) > 0 && FD_ISSET(FileDes, &set);
}
bool cFile::FileReadyForWriting(int FileDes, int TimeoutMs)
{
fd_set set;
struct timeval timeout;
FD_ZERO(&set);
FD_SET(FileDes, &set);
if (TimeoutMs < 100)
TimeoutMs = 100;
timeout.tv_sec = 0;
timeout.tv_usec = TimeoutMs * 1000;
return select(FD_SETSIZE, NULL, &set, NULL, &timeout) > 0 && FD_ISSET(FileDes, &set);
}
// --- cSafeFile -------------------------------------------------------------
cSafeFile::cSafeFile(const char *FileName)
{
f = NULL;
fileName = ReadLink(FileName);
tempName = fileName ? MALLOC(char, strlen(fileName) + 5) : NULL;
if (tempName)
strcat(strcpy(tempName, fileName), ".$$$");
}
cSafeFile::~cSafeFile()
{
if (f)
fclose(f);
unlink(tempName);
free(fileName);
free(tempName);
}
bool cSafeFile::Open(void)
{
if (!f && fileName && tempName) {
f = fopen(tempName, "w");
if (!f)
LOG_ERROR_STR(tempName);
}
return f != NULL;
}
bool cSafeFile::Close(void)
{
bool result = true;
if (f) {
if (ferror(f) != 0) {
LOG_ERROR_STR(tempName);
result = false;
}
fflush(f);
fsync(fileno(f));
if (fclose(f) < 0) {
LOG_ERROR_STR(tempName);
result = false;
}
f = NULL;
if (result && rename(tempName, fileName) < 0) {
LOG_ERROR_STR(fileName);
result = false;
}
}
else
result = false;
return result;
}
// --- cUnbufferedFile -------------------------------------------------------
#define USE_FADVISE
#define WRITE_BUFFER KILOBYTE(800)
cUnbufferedFile::cUnbufferedFile(void)
{
fd = -1;
}
cUnbufferedFile::~cUnbufferedFile()
{
Close();
}
int cUnbufferedFile::Open(const char *FileName, int Flags, mode_t Mode)
{
Close();
fd = open(FileName, Flags, Mode);
curpos = 0;
#ifdef USE_FADVISE
begin = lastpos = ahead = 0;
cachedstart = 0;
cachedend = 0;
readahead = KILOBYTE(128);
written = 0;
totwritten = 0;
if (fd >= 0)
posix_fadvise(fd, 0, 0, POSIX_FADV_RANDOM); // we could use POSIX_FADV_SEQUENTIAL, but we do our own readahead, disabling the kernel one.
#endif
return fd;
}
int cUnbufferedFile::Close(void)
{
if (fd >= 0) {
#ifdef USE_FADVISE
if (totwritten) // if we wrote anything make sure the data has hit the disk before
fdatasync(fd); // calling fadvise, as this is our last chance to un-cache it.
posix_fadvise(fd, 0, 0, POSIX_FADV_DONTNEED);
#endif
int OldFd = fd;
fd = -1;
return close(OldFd);
}
errno = EBADF;
return -1;
}
// When replaying and going e.g. FF->PLAY the position jumps back 2..8M
// hence we do not want to drop recently accessed data at once.
// We try to handle the common cases such as PLAY->FF->PLAY, small
// jumps, moving editing marks etc.
#define FADVGRAN KILOBYTE(4) // AKA fadvise-chunk-size; PAGE_SIZE or getpagesize(2) would also work.
#define READCHUNK MEGABYTE(8)
void cUnbufferedFile::SetReadAhead(size_t ra)
{
readahead = ra;
}
int cUnbufferedFile::FadviseDrop(off_t Offset, off_t Len)
{
// rounding up the window to make sure that not PAGE_SIZE-aligned data gets freed.
return posix_fadvise(fd, Offset - (FADVGRAN - 1), Len + (FADVGRAN - 1) * 2, POSIX_FADV_DONTNEED);
}
off_t cUnbufferedFile::Seek(off_t Offset, int Whence)
{
if (Whence == SEEK_SET && Offset == curpos)
return curpos;
curpos = lseek(fd, Offset, Whence);
return curpos;
}
ssize_t cUnbufferedFile::Read(void *Data, size_t Size)
{
if (fd >= 0) {
#ifdef USE_FADVISE
off_t jumped = curpos-lastpos; // nonzero means we're not at the last offset
if ((cachedstart < cachedend) && (curpos < cachedstart || curpos > cachedend)) {
// current position is outside the cached window -- invalidate it.
FadviseDrop(cachedstart, cachedend-cachedstart);
cachedstart = curpos;
cachedend = curpos;
}
cachedstart = min(cachedstart, curpos);
#endif
ssize_t bytesRead = safe_read(fd, Data, Size);
if (bytesRead > 0) {
curpos += bytesRead;
#ifdef USE_FADVISE
cachedend = max(cachedend, curpos);
// Read ahead:
// no jump? (allow small forward jump still inside readahead window).
if (jumped >= 0 && jumped <= (off_t)readahead) {
// Trigger the readahead IO, but only if we've used at least
// 1/2 of the previously requested area. This avoids calling
// fadvise() after every read() call.
if (ahead - curpos < (off_t)(readahead / 2)) {
posix_fadvise(fd, curpos, readahead, POSIX_FADV_WILLNEED);
ahead = curpos + readahead;
cachedend = max(cachedend, ahead);
}
if (readahead < Size * 32) { // automagically tune readahead size.
readahead = Size * 32;
}
}
else
ahead = curpos; // jumped -> we really don't want any readahead, otherwise e.g. fast-rewind gets in trouble.
#endif
}
#ifdef USE_FADVISE
if (cachedstart < cachedend) {
if (curpos - cachedstart > READCHUNK * 2) {
// current position has moved forward enough, shrink tail window.
FadviseDrop(cachedstart, curpos - READCHUNK - cachedstart);
cachedstart = curpos - READCHUNK;
}
else if (cachedend > ahead && cachedend - curpos > READCHUNK * 2) {
// current position has moved back enough, shrink head window.
FadviseDrop(curpos + READCHUNK, cachedend - (curpos + READCHUNK));
cachedend = curpos + READCHUNK;
}
}
lastpos = curpos;
#endif
return bytesRead;
}
return -1;
}
ssize_t cUnbufferedFile::Write(const void *Data, size_t Size)
{
if (fd >=0) {
ssize_t bytesWritten = safe_write(fd, Data, Size);
#ifdef USE_FADVISE
if (bytesWritten > 0) {
begin = min(begin, curpos);
curpos += bytesWritten;
written += bytesWritten;
lastpos = max(lastpos, curpos);
if (written > WRITE_BUFFER) {
if (lastpos > begin) {
// Now do three things:
// 1) Start writeback of begin..lastpos range
// 2) Drop the already written range (by the previous fadvise call)
// 3) Handle nonpagealigned data.
// This is why we double the WRITE_BUFFER; the first time around the
// last (partial) page might be skipped, writeback will start only after
// second call; the third call will still include this page and finally
// drop it from cache.
off_t headdrop = min(begin, off_t(WRITE_BUFFER * 2));
posix_fadvise(fd, begin - headdrop, lastpos - begin + headdrop, POSIX_FADV_DONTNEED);
}
begin = lastpos = curpos;
totwritten += written;
written = 0;
// The above fadvise() works when writing slowly (recording), but could
// leave cached data around when writing at a high rate, e.g. when cutting,
// because by the time we try to flush the cached pages (above) the data
// can still be dirty - we are faster than the disk I/O.
// So we do another round of flushing, just like above, but at larger
// intervals -- this should catch any pages that couldn't be released
// earlier.
if (totwritten > MEGABYTE(32)) {
// It seems in some setups, fadvise() does not trigger any I/O and
// a fdatasync() call would be required do all the work (reiserfs with some
// kind of write gathering enabled), but the syncs cause (io) load..
// Uncomment the next line if you think you need them.
//fdatasync(fd);
off_t headdrop = min(off_t(curpos - totwritten), off_t(totwritten * 2));
posix_fadvise(fd, curpos - totwritten - headdrop, totwritten + headdrop, POSIX_FADV_DONTNEED);
totwritten = 0;
}
}
}
#endif
return bytesWritten;
}
return -1;
}
cUnbufferedFile *cUnbufferedFile::Create(const char *FileName, int Flags, mode_t Mode)
{
cUnbufferedFile *File = new cUnbufferedFile;
if (File->Open(FileName, Flags, Mode) < 0) {
delete File;
File = NULL;
}
return File;
}
// --- cLockFile -------------------------------------------------------------
#define LOCKFILENAME ".lock-vdr"
#define LOCKFILESTALETIME 600 // seconds before considering a lock file "stale"
cLockFile::cLockFile(const char *Directory)
{
fileName = NULL;
f = -1;
if (DirectoryOk(Directory))
fileName = strdup(AddDirectory(Directory, LOCKFILENAME));
}
cLockFile::~cLockFile()
{
Unlock();
free(fileName);
}
bool cLockFile::Lock(int WaitSeconds)
{
if (f < 0 && fileName) {
time_t Timeout = time(NULL) + WaitSeconds;
do {
f = open(fileName, O_WRONLY | O_CREAT | O_EXCL, DEFFILEMODE);
if (f < 0) {
if (errno == EEXIST) {
struct stat fs;
if (stat(fileName, &fs) == 0) {
if (abs(time(NULL) - fs.st_mtime) > LOCKFILESTALETIME) {
esyslog("ERROR: removing stale lock file '%s'", fileName);
if (remove(fileName) < 0) {
LOG_ERROR_STR(fileName);
break;
}
continue;
}
}
else if (errno != ENOENT) {
LOG_ERROR_STR(fileName);
break;
}
}
else {
LOG_ERROR_STR(fileName);
break;
}
if (WaitSeconds)
cCondWait::SleepMs(1000);
}
} while (f < 0 && time(NULL) < Timeout);
}
return f >= 0;
}
void cLockFile::Unlock(void)
{
if (f >= 0) {
close(f);
remove(fileName);
f = -1;
}
}
// --- cListObject -----------------------------------------------------------
cListObject::cListObject(void)
{
prev = next = NULL;
}
cListObject::~cListObject()
{
}
void cListObject::Append(cListObject *Object)
{
next = Object;
Object->prev = this;
}
void cListObject::Insert(cListObject *Object)
{
prev = Object;
Object->next = this;
}
void cListObject::Unlink(void)
{
if (next)
next->prev = prev;
if (prev)
prev->next = next;
next = prev = NULL;
}
int cListObject::Index(void) const
{
cListObject *p = prev;
int i = 0;
while (p) {
i++;
p = p->prev;
}
return i;
}
// --- cListBase -------------------------------------------------------------
cListBase::cListBase(void)
{
objects = lastObject = NULL;
count = 0;
}
cListBase::~cListBase()
{
Clear();
}
void cListBase::Add(cListObject *Object, cListObject *After)
{
if (After && After != lastObject) {
After->Next()->Insert(Object);
After->Append(Object);
}
else {
if (lastObject)
lastObject->Append(Object);
else
objects = Object;
lastObject = Object;
}
count++;
}
void cListBase::Ins(cListObject *Object, cListObject *Before)
{
if (Before && Before != objects) {
Before->Prev()->Append(Object);
Before->Insert(Object);
}
else {
if (objects)
objects->Insert(Object);
else
lastObject = Object;
objects = Object;
}
count++;
}
void cListBase::Del(cListObject *Object, bool DeleteObject)
{
if (Object == objects)
objects = Object->Next();
if (Object == lastObject)
lastObject = Object->Prev();
Object->Unlink();
if (DeleteObject)
delete Object;
count--;
}
void cListBase::Move(int From, int To)
{
Move(Get(From), Get(To));
}
void cListBase::Move(cListObject *From, cListObject *To)
{
if (From && To && From != To) {
if (From->Index() < To->Index())
To = To->Next();
if (From == objects)
objects = From->Next();
if (From == lastObject)
lastObject = From->Prev();
From->Unlink();
if (To) {
if (To->Prev())
To->Prev()->Append(From);
From->Append(To);
}
else {
lastObject->Append(From);
lastObject = From;
}
if (!From->Prev())
objects = From;
}
}
void cListBase::Clear(void)
{
while (objects) {
cListObject *object = objects->Next();
delete objects;
objects = object;
}
objects = lastObject = NULL;
count = 0;
}
cListObject *cListBase::Get(int Index) const
{
if (Index < 0)
return NULL;
cListObject *object = objects;
while (object && Index-- > 0)
object = object->Next();
return object;
}
static int CompareListObjects(const void *a, const void *b)
{
const cListObject *la = *(const cListObject **)a;
const cListObject *lb = *(const cListObject **)b;
return la->Compare(*lb);
}
void cListBase::Sort(void)
{
int n = Count();
cListObject *a[n];
cListObject *object = objects;
int i = 0;
while (object && i < n) {
a[i++] = object;
object = object->Next();
}
qsort(a, n, sizeof(cListObject *), CompareListObjects);
objects = lastObject = NULL;
for (i = 0; i < n; i++) {
a[i]->Unlink();
count--;
Add(a[i]);
}
}
// --- cHashBase -------------------------------------------------------------
cHashBase::cHashBase(int Size)
{
size = Size;
hashTable = (cList<cHashObject>**)calloc(size, sizeof(cList<cHashObject>*));
}
cHashBase::~cHashBase(void)
{
Clear();
free(hashTable);
}
void cHashBase::Add(cListObject *Object, unsigned int Id)
{
unsigned int hash = hashfn(Id);
if (!hashTable[hash])
hashTable[hash] = new cList<cHashObject>;
hashTable[hash]->Add(new cHashObject(Object, Id));
}
void cHashBase::Del(cListObject *Object, unsigned int Id)
{
cList<cHashObject> *list = hashTable[hashfn(Id)];
if (list) {
for (cHashObject *hob = list->First(); hob; hob = list->Next(hob)) {
if (hob->object == Object) {
list->Del(hob);
break;
}
}
}
}
void cHashBase::Clear(void)
{
for (int i = 0; i < size; i++) {
delete hashTable[i];
hashTable[i] = NULL;
}
}
cListObject *cHashBase::Get(unsigned int Id) const
{
cList<cHashObject> *list = hashTable[hashfn(Id)];
if (list) {
for (cHashObject *hob = list->First(); hob; hob = list->Next(hob)) {
if (hob->id == Id)
return hob->object;
}
}
return NULL;
}
cList<cHashObject> *cHashBase::GetList(unsigned int Id) const
{
return hashTable[hashfn(Id)];
}