Add convert_0.2-0.3.sh.

[paraslash.git] / string.c

/*
 * Copyright (C) 2004-2007 Andre Noll <maan@systemlinux.org>
 *
 * Licensed under the GPL v2. For licencing details see COPYING.
 */

/** \file string.c memory allocation and string handling functions */

#include "para.h"
#include "string.h"

#include <sys/time.h> /* gettimeofday */
#include <pwd.h>
#include <sys/utsname.h> /* uname() */
#include <string.h>

#include "error.h"

/**
 * paraslash's version of realloc()
 *
 * \param p pointer to the memory block, may be NULL
 * \param size desired new size
 *
 * A wrapper for realloc(3). It calls \p exit(\p EXIT_FAILURE) on errors,
 * i.e. there is no need to check the return value in the caller.
 *
 * \return A pointer to  the newly allocated memory, which is suitably aligned
 * for any kind of variable and may be different from \p p.
 *
 * \sa realloc(3)
 */
__must_check __malloc void *para_realloc(void *p, size_t size)
{
        /*
         * No need to check for NULL pointers: If p is NULL, the  call
         * to realloc is equivalent to malloc(size)
         */
        assert(size);
        if (!(p = realloc(p, size))) {
                PARA_EMERG_LOG("realloc failed (size = %zu), aborting\n",
                        size);
                exit(EXIT_FAILURE);
        }
        return p;
}

/**
 * paraslash's version of malloc()
 *
 * \param size desired new size
 *
 * A wrapper for malloc(3) which exits on errors.
 *
 * \return A pointer to the allocated memory, which is suitably aligned for any
 * kind  of variable.
 *
 * \sa malloc(3)
 */
__must_check __malloc void *para_malloc(size_t size)
{
        assert(size);
        void *p = malloc(size);

        if (!p) {
                PARA_EMERG_LOG("%s", "malloc failed, aborting\n");
                exit(EXIT_FAILURE);
        }
        return p;
}

/**
 * paraslash's version of calloc()
 *
 * \param size desired new size
 *
 * A wrapper for calloc(3) which exits on errors.
 *
 * \return A pointer to the allocated and zeroed-out memory, which is suitably
 * aligned for any kind  of variable.
 *
 * \sa calloc(3)
 */
__must_check __malloc void *para_calloc(size_t size)
{
        void *ret = para_malloc(size);

        memset(ret, 0, size);
        return ret;
}

/**
 * paraslash's version of strdup()
 *
 * \param s string to be duplicated
 *
 * A wrapper for strdup(3). It calls \p exit(EXIT_FAILURE) on errors, i.e.
 * there is no need to check the return value in the caller.
 *
 * \return A pointer to the duplicated string. If \p s was the NULL pointer,
 * an pointer to an empty string is returned.
 *
 * \sa strdup(3)
 */
__must_check __malloc char *para_strdup(const char *s)
{
        char *ret;

        if ((ret = strdup(s? s: "")))
                return ret;
        PARA_EMERG_LOG("%s", "strdup failed, aborting\n");
        exit(EXIT_FAILURE);
}

/**
 * allocate a sufficiently large string and print into it
 *
 * \param fmt usual format string
 *
 * Produce output according to \p fmt. No artificial bound on the length of the
 * resulting string is imposed.
 *
 * \return This function either returns a pointer to a string that must be
 * freed by the caller or aborts without returning.
 *
 * \sa printf(3)
 */
__must_check __printf_1_2 __malloc char *make_message(const char *fmt, ...)
{
        char *msg;

        PARA_VSPRINTF(fmt, msg);
        return msg;
}

/**
 * paraslash's version of strcat()
 *
 * \param a string to be appended to
 * \param b string to append
 *
 * Append \p b to \p a.
 *
 * \return If \p a is NULL, return a pointer to a copy of \p b, i.e.
 * para_strcat(NULL, b) is equivalent to para_strdup(b). If \p b is NULL,
 * return \p a without making a copy of \p a.  Otherwise, construct the
 * concatenation \p c, free \p a (but not \p b) and return \p c.
 *
 * \sa strcat(3)
 */
__must_check __malloc char *para_strcat(char *a, const char *b)
{
        char *tmp;

        if (!a)
                return para_strdup(b);
        if (!b)
                return a;
        tmp = make_message("%s%s", a, b);
        free(a);
        return tmp;
}

/**
 * paraslash's version of dirname()
 *
 * \param name pointer to the full path
 *
 * Compute the directory component of \p name
 *
 * \return If \p name is \þ NULL or the empty string, return \p NULL.
 * Otherwise, Make a copy of \p name and return its directory component. Caller
 * is responsible to free the result.
 */
__must_check __malloc char *para_dirname(const char *name)
{
        char *p, *ret;

        if (!name || !*name)
                return NULL;
        ret = para_strdup(name);
        p = strrchr(ret, '/');
        if (!p)
                *ret = '\0';
        else
                *p = '\0';
        return ret;
}

/**
 * paraslash's version of basename()
 *
 * \param name Pointer to the full path
 *
 * Compute the filename component of \p name
 *
 * \return If \p name is \p NULL or the empty string, return \p NULL,
 * Otherwise, make a copy of \p name and return its filename component. Caller
 * is responsible to free the result.
 */
__must_check __malloc char *para_basename(const char *name)
{
        char *p;

        if (!name || !*name)
                return NULL;
        p = strrchr(name, '/');
        if (!p)
                return para_strdup(name);
        p++;
        if (!*p)
                return NULL;
        return para_strdup(p);
}

/**
 * cut trailing newline
 *
 * \param buf the string to be chopped.
 *
 * Replace the last character in \p buf by zero if it is euqal to
 * the newline character.
 */
void chop(char *buf)
{
        int n = strlen(buf);
        if (!n)
                return;
        if (buf[n - 1] == '\n')
                buf[n - 1] = '\0';
}

/**
 * get a random filename
 *
 * This is by no means a secure way to create temporary files in a hostile
 * direcory like \p /tmp. However, it is OK to use for temp files, fifos,
 * sockets that are created in ~/.paraslash. Result must be freed by the
 * caller.
 *
 * \return a pointer to a random filename.
 */
__must_check __malloc char *para_tmpname(void)
{
        struct timeval now;
        unsigned int seed;

        gettimeofday(&now, NULL);
        seed = now.tv_usec;
        srand(seed);
        return make_message("%08i", rand());
}

/**
 * create unique temporary file
 *
 * \param template the template to be passed to mkstemp()
 * \param mode the desired mode of the tempfile
 *
 * This wrapper for mkstemp additionally uses fchmod() to
 * set the given mode of the tempfile if mkstemp() returned success.
 *
 * \return The file descriptor of the temp file just created on success.
 * On errors, -E_MKSTEMP or -E_FCHMOD is returned.
 */
__must_check int para_mkstemp(char *template, mode_t mode)
{
        int tmp, fd = mkstemp(template);

        if (fd < 0)
                return -E_MKSTEMP;
        tmp = fchmod(fd, mode);
        if (tmp >= 0)
                return fd;
        close(fd);
        unlink(template);
        return -E_FCHMOD;
}

/**
 * get the logname of the current user
 *
 * \return A dynammically allocated string that must be freed by the caller. On
 * errors, the string "unknown user" is returned, i.e. this function never
 * returns NULL.
 *
 * \sa getpwuid(3)
 */
__must_check __malloc char *para_logname(void)
{
        struct passwd *pw = getpwuid(getuid());
        return para_strdup(pw? pw->pw_name : "unknown_user");
}

/**
 * get the home directory of the current user
 *
 * \return A dynammically allocated string that must be freed by the caller. If
 * the home directory could not be found, this function returns "/tmp".
 */
__must_check __malloc char *para_homedir(void)
{
        struct passwd *pw = getpwuid(getuid());
        return para_strdup(pw? pw->pw_dir : "/tmp");
}

/**
 * split string and return pointers to its parts.
 *
 * \param args the string to be split
 * \param argv_ptr  pointer to the list of substrings
 * \param delim delimiter
 *
 * This function modifies \a args by replacing each occurance of \a delim by
 * zero. A \p NULL-terminated array of pointers to char* is allocated dynamically
 * and these pointers are initialized to point to the broken-up substrings
 * within \a args. A pointer to this array is returned via \a argv_ptr. It's OK
 * to call this function with \a args \a == \p NULL.
 *
 * \return The number of substrings found in \a args.
 */
__must_check unsigned split_args(char *args, char *** const argv_ptr, const char *delim)
{
        char *p = args;
        char **argv;
        size_t n = 0, i, j;

        p = args + strspn(args, delim);
        for (;;) {
                i = strcspn(p, delim);
                if (!i)
                        break;
                p += i;
                n++;
                p += strspn(p, delim);
        }
        *argv_ptr = para_malloc((n + 1) * sizeof(char *));
        argv = *argv_ptr;
        i = 0;
        p = args + strspn(args, delim);
        while (p) {
                argv[i] = p;
                j = strcspn(p, delim);
                if (!j)
                        break;
                p += strcspn(p, delim);
                if (*p) {
                        *p = '\0';
                        p++;
                        p += strspn(p, delim);
                }
                i++;
        }
        argv[n] = NULL;
        return n;
}

/**
 * ensure that file descriptors 0, 1, and 2 are valid
 *
 * Common approach that opens /dev/null until it gets a file descriptor greater
 * than two.
 *
 * \sa okir's Black Hats Manual.
 */
void valid_fd_012(void)
{
        while (1) {
        int     fd;

                fd = open("/dev/null", O_RDWR);
                if (fd < 0)
                        exit(EXIT_FAILURE);
                if (fd > 2) {
                        close(fd);
                        break;
                }
        }
}

/**
 * get the own hostname
 *
 * \return A dynammically allocated string containing the hostname.
 *
 * \sa uname(2)
 */
__malloc char *para_hostname(void)
{
        struct utsname u;

        uname(&u);
        return para_strdup(u.nodename);
}

enum for_each_line_modes{LINE_MODE_RO, LINE_MODE_RW};

static int for_each_complete_line(enum for_each_line_modes mode, char *buf,
                size_t size, line_handler_t *line_handler, void *private_data)
{
        char *start = buf, *end;
        int ret, i, num_lines = 0;

//      PARA_NOTICE_LOG("buf: %s\n", buf);
        while (start < buf + size) {
                char *next_null;
                char *next_cr;

                next_cr = memchr(start, '\n', buf + size - start);
                next_null = memchr(start, '\0', buf + size - start);
                if (!next_cr && !next_null)
                        break;
                if (next_cr && next_null) {
                        end = next_cr < next_null? next_cr : next_null;
                } else if (next_null) {
                        end = next_null;
                } else
                        end = next_cr;
                num_lines++;
                if (!line_handler) {
                        start = ++end;
                        continue;
                }
                if (mode == LINE_MODE_RO) {
                        size_t s = end - start;
                        char *b = para_malloc(s + 1);
                        memcpy(b, start, s);
                        b[s] = '\0';
//                      PARA_NOTICE_LOG("b: %s, start: %s\n", b, start);
                        ret = line_handler(b, private_data);
                        free(b);
                } else {
                        *end = '\0';
                        ret = line_handler(start, private_data);
                }
                if (ret < 0)
                        return ret;
                start = ++end;
        }
        if (!line_handler || mode == LINE_MODE_RO)
                return num_lines;
        i = buf + size - start;
        if (i && i != size)
                memmove(buf, start, i);
        return i;
}

/**
 * call a custom function for each complete line
 *
 * \param buf the buffer containing data seperated by newlines
 * \param size the number of bytes in \a buf
 * \param line_handler the custom function
 * \param private_data pointer to data which is passed to \a line_handler
 *
 * If \p line_handler is \p NULL, the function returns the number of complete
 * lines in \p buf.  Otherwise, \p line_handler is called for each complete
 * line in \p buf.  The first argument to \p line_handler is the current line,
 * and \p private_data is passed as the second argument.  The function returns
 * if \p line_handler returns a negative value or no more lines are in the
 * buffer.  The rest of the buffer (last chunk containing an incomplete line)
 * is moved to the beginning of the buffer.
 *
 * \return If \p line_handler is not \p NULL, this function returns the number
 * of bytes not handled to \p line_handler on success, or the negative return
 * value of the \p line_handler on errors.
 *
 * \sa for_each_line_ro()
 */
int for_each_line(char *buf, size_t size, line_handler_t *line_handler,
                void *private_data)
{
        return for_each_complete_line(LINE_MODE_RW, buf, size, line_handler,
                private_data);
}

/**
 * call a custom function for each complete line
 *
 * \param buf same meaning as in \p for_each_line
 * \param size same meaning as in \p for_each_line
 * \param line_handler same meaning as in \p for_each_line
 * \param private_data same meaning as in \p for_each_line
 *
 * This function behaves like \p for_each_line() with the following differences:
 *
 * - buf is left unchanged
 *
 * \return On success, the function returns the number of complete lines in \p
 * buf, otherwise the (negative) return value of \p line_handler is returned.
 *
 * \sa for_each_line()
 */
int for_each_line_ro(char *buf, size_t size, line_handler_t *line_handler,
                void *private_data)
{
        return for_each_complete_line(LINE_MODE_RO, buf, size, line_handler,
                private_data);
}

/**
 * Safely print into a buffer at a given offset
 *
 * \param b Determines the buffer, its size, and the offset.
 * \param fmt The format string.
 *
 * This function prints into the buffer given by \a b at the offset which is
 * also given by \a b. If there is not enough space to hold the result, the
 * buffer size is doubled until the underlying call to vsnprintf() succeeds.
 * Upon return, the offset of \a b is adjusted accordingly so that subsequent
 * calls to this function append data to what is already contained in the
 * buffer.
 *
 * It's OK to call this function with \p b->buf being \p NULL. In this case, an
 * initial buffer is allocated.
 *
 * \return The number of bytes printed into the buffer (not including the
 * therminating \p NULL byte).
 *
 * \sa make_message(), vsnprintf(3).
 */
__printf_2_3 int para_printf(struct para_buffer *b, const char *fmt, ...)
{
        int ret;

        if (!b->buf) {
                b->buf = para_malloc(128);
                b->size = 128;
                b->offset = 0;
        } else if (b->size <= b->offset + 1) {
                b->size *= 2;
                b->buf = para_realloc(b->buf, b->size);
        }
        while (1) {
                char *p = b->buf + b->offset;
                size_t size = b->size - b->offset;
                va_list ap;
                va_start(ap, fmt);
                ret = vsnprintf(p, size, fmt, ap);
                va_end(ap);
                if (ret > -1 && ret < size) { /* success */
                        b->offset += ret;
                        break;
                }
                /* try again with more space */
                b->size *= 2;
                b->buf = para_realloc(b->buf, b->size);
        }
        return ret;
}