+static int find_urls(const char *tweet, int **pranges)
+{
+ /*
+ * magic obtained from
+ * http://www.geekpedia.com/KB65_How-to-validate-an-URL-using-RegEx-in-Csharp.html
+ */
+ static const char *re_magic =
+ "(([a-zA-Z][0-9a-zA-Z+\\-\\.]*:)/{1,3}"
+ "[0-9a-zA-Z;/~?:@&=+$\\.\\-_'()%]+)"
+ "(#[0-9a-zA-Z;/?:@&=+$\\.\\-_!~*'()%]+)?";
+ pcre *re;
+ const char *errptr;
+ int erroffset;
+ int ovector[10] = {0,};
+ const size_t ovsize = sizeof(ovector)/sizeof(*ovector);
+ int startoffset, tweetlen;
+ int i, rc;
+ int rbound = 10;
+ int rcount = 0;
+ int *ranges = malloc(sizeof(int) * rbound);
+
+ re = pcre_compile(re_magic,
+ PCRE_NO_AUTO_CAPTURE,
+ &errptr, &erroffset, NULL);
+ if (!re) {
+ fprintf(stderr, "pcre_compile @%u: %s\n", erroffset, errptr);
+ exit(1);
+ }
+
+ tweetlen = strlen(tweet);
+ for (startoffset = 0; startoffset < tweetlen; ) {
+
+ rc = pcre_exec(re, NULL, tweet, strlen(tweet), startoffset, 0,
+ ovector, ovsize);
+ if (rc == PCRE_ERROR_NOMATCH)
+ break;
+
+ if (rc < 0) {
+ fprintf(stderr, "pcre_exec @%u: %s\n",
+ erroffset, errptr);
+ exit(1);
+ }
+
+ for (i = 0; i < rc; i += 2) {
+ if ((rcount+2) == rbound) {
+ rbound *= 2;
+ ranges = realloc(ranges, sizeof(int) * rbound);
+ }
+
+ ranges[rcount++] = ovector[i];
+ ranges[rcount++] = ovector[i+1];
+ }
+
+ startoffset = ovector[1];
+ }
+
+ pcre_free(re);
+
+ *pranges = ranges;
+ return rcount;
+}
+
+/**
+ * bidirectional popen() call
+ *
+ * @param rwepipe - int array of size three
+ * @param exe - program to run
+ * @param argv - argument list
+ * @return pid or -1 on error
+ *
+ * The caller passes in an array of three integers (rwepipe), on successful
+ * execution it can then write to element 0 (stdin of exe), and read from
+ * element 1 (stdout) and 2 (stderr).
+ */
+static int popenRWE(int *rwepipe, const char *exe, const char *const argv[])
+{
+ int in[2];
+ int out[2];
+ int err[2];
+ int pid;
+ int rc;
+
+ rc = pipe(in);
+ if (rc < 0)
+ goto error_in;
+
+ rc = pipe(out);
+ if (rc < 0)
+ goto error_out;
+
+ rc = pipe(err);
+ if (rc < 0)
+ goto error_err;
+
+ pid = fork();
+ if (pid > 0) {
+ /* parent */
+ close(in[0]);
+ close(out[1]);
+ close(err[1]);
+ rwepipe[0] = in[1];
+ rwepipe[1] = out[0];
+ rwepipe[2] = err[0];
+ return pid;
+ } else if (pid == 0) {
+ /* child */
+ close(in[1]);
+ close(out[0]);
+ close(err[0]);
+ close(0);
+ rc = dup(in[0]);
+ close(1);
+ rc = dup(out[1]);
+ close(2);
+ rc = dup(err[1]);
+
+ execvp(exe, (char **)argv);
+ exit(1);
+ } else
+ goto error_fork;
+
+ return pid;
+
+error_fork:
+ close(err[0]);
+ close(err[1]);
+error_err:
+ close(out[0]);
+ close(out[1]);
+error_out:
+ close(in[0]);
+ close(in[1]);
+error_in:
+ return -1;
+}
+
+static int pcloseRWE(int pid, int *rwepipe)
+{
+ int rc, status;
+ close(rwepipe[0]);
+ close(rwepipe[1]);
+ close(rwepipe[2]);
+ rc = waitpid(pid, &status, 0);
+ return status;
+}
+
+static char *shrink_one_url(int *rwepipe, char *big)
+{
+ int biglen = strlen(big);
+ char *small;
+ int smalllen;
+ int rc;
+
+ rc = dprintf(rwepipe[0], "%s\n", big);
+ if (rc < 0)
+ return big;
+
+ smalllen = biglen + 128;
+ small = malloc(smalllen);
+ if (!small)
+ return big;
+
+ rc = read(rwepipe[1], small, smalllen);
+ if (rc < 0 || rc > biglen)
+ goto error_free_small;
+
+ if (strncmp(small, "http://", 7))
+ goto error_free_small;
+
+ smalllen = rc;
+ while (smalllen && isspace(small[smalllen-1]))
+ small[--smalllen] = 0;
+
+ free(big);
+ return small;
+
+error_free_small:
+ free(small);
+ return big;
+}
+
+static char *shrink_urls(char *text)
+{
+ int *ranges;
+ int rcount;
+ int i;
+ int inofs = 0;
+ int outofs = 0;
+ const char *const shrink_args[] = {
+ "bti-shrink-urls",
+ NULL
+ };
+ int shrink_pid;
+ int shrink_pipe[3];
+ int inlen = strlen(text);
+
+ dbg("before len=%u\n", inlen);
+
+ shrink_pid = popenRWE(shrink_pipe, shrink_args[0], shrink_args);
+ if (shrink_pid < 0)
+ return text;
+
+ rcount = find_urls(text, &ranges);
+ if (!rcount)
+ return text;
+
+ for (i = 0; i < rcount; i += 2) {
+ int url_start = ranges[i];
+ int url_end = ranges[i+1];
+ int long_url_len = url_end - url_start;
+ char *url = strndup(text + url_start, long_url_len);
+ int short_url_len;
+ int not_url_len = url_start - inofs;
+
+ dbg("long url[%u]: %s\n", long_url_len, url);
+ url = shrink_one_url(shrink_pipe, url);
+ short_url_len = url ? strlen(url) : 0;
+ dbg("short url[%u]: %s\n", short_url_len, url);
+
+ if (!url || short_url_len >= long_url_len) {
+ /* The short url ended up being too long
+ * or unavailable */
+ if (inofs) {
+ strncpy(text + outofs, text + inofs,
+ not_url_len + long_url_len);
+ }
+ inofs += not_url_len + long_url_len;
+ outofs += not_url_len + long_url_len;
+
+ } else {
+ /* copy the unmodified block */
+ strncpy(text + outofs, text + inofs, not_url_len);
+ inofs += not_url_len;
+ outofs += not_url_len;
+
+ /* copy the new url */
+ strncpy(text + outofs, url, short_url_len);
+ inofs += long_url_len;
+ outofs += short_url_len;
+ }
+
+ free(url);
+ }
+
+ /* copy the last block after the last match */
+ if (inofs) {
+ int tail = inlen - inofs;
+ if (tail) {
+ strncpy(text + outofs, text + inofs, tail);
+ outofs += tail;
+ }
+ }
+
+ free(ranges);
+
+ (void)pcloseRWE(shrink_pid, shrink_pipe);
+
+ text[outofs] = 0;
+ dbg("after len=%u\n", outofs);
+ return text;
+}
+
ToastFreeware / gregoa / bti.git / blobdiff