+ }
+ }
+ return res;
+}
+
+/** @brief Splits a string into a dynar of strings, taking quotes into account
+ *
+ * It basically does the same argument separation than the shell, where white
+ * spaces can be escaped and where arguments are never split within a
+ * quote group.
+ * Several subsequent spaces are ignored (unless within quotes, of course).
+ * You may want to trim the input string, if you want to avoid empty entries
+ *
+ */
+
+xbt_dynar_t xbt_str_split_quoted(const char *s)
+{
+ xbt_dynar_t res = xbt_dynar_new(sizeof(char *), &xbt_free_ref);
+ xbt_dynar_t parsed;
+ char *str_to_free; /* we have to copy the string before, to handle backslashes */
+ unsigned int cursor;
+ char *p;
+
+ if (s[0] == '\0')
+ return res;
+ str_to_free = xbt_strdup(s);
+
+ parsed = xbt_str_split_quoted_in_place(str_to_free);
+ xbt_dynar_foreach(parsed,cursor,p) {
+ char *q=xbt_strdup(p);
+ xbt_dynar_push(res,&q);
+ }
+ free(str_to_free);
+ xbt_dynar_shrink(res, 0);
+ xbt_dynar_free(&parsed);
+ return res;
+}
+
+/** @brief Join a set of strings as a single string */
+char *xbt_str_join(xbt_dynar_t dyn, const char *sep)
+{
+ int len = 1, dyn_len = xbt_dynar_length(dyn);
+ unsigned int cpt;
+ char *cursor;
+ char *res, *p;
+
+ if (!dyn_len)
+ return xbt_strdup("");
+
+ /* compute the length */
+ xbt_dynar_foreach(dyn, cpt, cursor) {
+ len += strlen(cursor);
+ }
+ len += strlen(sep) * dyn_len;
+ /* Do the job */
+ res = xbt_malloc(len);
+ p = res;
+ xbt_dynar_foreach(dyn, cpt, cursor) {
+ if ((int) cpt < dyn_len - 1)
+ p += sprintf(p, "%s%s", cursor, sep);
+ else
+ p += sprintf(p, "%s", cursor);
+ }
+ return res;
+}
+/** @brief Join a set of strings as a single string
+ *
+ * The parameter must be a NULL-terminated array of chars,
+ * just like xbt_dynar_to_array() produces
+ */
+char *xbt_str_join_array(const char *const *strs, const char *sep)
+{
+ char *res,*q;
+ int amount_strings=0;
+ int len=0;
+ int i;
+
+ if ((!strs) || (!strs[0]))
+ return xbt_strdup("");
+
+ /* compute the length before malloc */
+ for (i=0;strs[i];i++) {
+ len += strlen(strs[i]);
+ amount_strings++;
+ }
+ len += strlen(sep) * amount_strings;
+
+ /* Do the job */
+ q = res = xbt_malloc(len);
+ for (i=0;strs[i];i++) {
+ if (i!=0) { // not first loop
+ q += sprintf(q, "%s%s", sep, strs[i]);
+ } else {
+ q += sprintf(q,"%s",strs[i]);
+ }
+ }
+ return res;
+}
+
+/** @brief Get a single line from the stream (reimplementation of the GNU getline)
+ *
+ * This is a reimplementation of the GNU getline function, so that our code don't depends on the GNU libc.
+ *
+ * xbt_getline() reads an entire line from stream, storing the address of the
+ * buffer containing the text into *buf. The buffer is null-terminated and
+ * includes the newline character, if one was found.
+ *
+ * If *buf is NULL, then xbt_getline() will allocate a buffer for storing the
+ * line, which should be freed by the user program.
+ *
+ * Alternatively, before calling xbt_getline(), *buf can contain a pointer to a
+ * malloc()-allocated buffer *n bytes in size. If the buffer is not large
+ * enough to hold the line, xbt_getline() resizes it with realloc(), updating
+ * *buf and *n as necessary.
+ *
+ * In either case, on a successful call, *buf and *n will be updated to reflect
+ * the buffer address and allocated size respectively.
+ */
+ssize_t xbt_getline(char **buf, size_t *n, FILE *stream)
+{
+ ssize_t i;
+ int ch;
+
+ ch = getc(stream);
+ if (ferror(stream) || feof(stream))
+ return -1;
+
+ if (!*buf) {
+ *n = 512;
+ *buf = xbt_malloc(*n);
+ }
+
+ i = 0;
+ do {
+ if (i == *n)
+ *buf = xbt_realloc(*buf, *n += 512);
+ (*buf)[i++] = ch;
+ } while (ch != '\n' && (ch = getc(stream)) != EOF);
+
+ if (i == *n)
+ *buf = xbt_realloc(*buf, *n += 1);
+ (*buf)[i] = '\0';
+
+ return i;
+}
+
+/*
+ * Diff related functions
+ *
+ * Implementation of the algorithm described in "An O(NP) Sequence Comparison
+ * Algorithm", by Sun Wu, Udi Manber, Gene Myers, and Webb Miller (Information
+ * Processing Letters 35(6):317-323, 1990), with the linear-space
+ * divide-and-conquer strategy described in "An O(ND) Difference Algorithm and
+ * Its Variations", by Eugene W. Myers (Algorithmica 1:251-266, 1986).
+ */
+
+struct subsequence {
+ int x, y; /* starting coordinates */
+ int len; /* length */
+};
+
+static XBT_INLINE
+void diff_snake(const char *vec_a[], int a0, int len_a,
+ const char *vec_b[], int b0, int len_b,
+ struct subsequence *seqs, int *fp, int k, int limit)
+{
+ int record_seq;
+ int x, y;
+ int fp_left = fp[k - 1] + 1;
+ int fp_right = fp[k + 1];
+ if (fp_left > fp_right) {
+ x = fp_left;
+ record_seq = k - 1;
+ } else {
+ x = fp_right;
+ record_seq = k + 1;
+ }
+ y = x - k;
+ if (x + y <= limit) {
+ seqs[k].x = x;
+ seqs[k].y = y;
+ record_seq = k;
+ } else {
+ seqs[k] = seqs[record_seq];
+ }
+ while (x < len_a && y < len_b && !strcmp(vec_a[a0 + x], vec_b[b0 + y]))
+ ++x, ++y;
+ fp[k] = x;
+ if (record_seq == k)
+ seqs[k].len = x - seqs[k].x;
+}
+
+/* Returns the length of a shortest edit script, and a common
+ * subsequence from the middle.
+ */
+static int diff_middle_subsequence(const char *vec_a[], int a0, int len_a,
+ const char *vec_b[], int b0, int len_b,
+ struct subsequence *subseq,
+ struct subsequence *seqs, int *fp)
+{
+ const int delta = len_a - len_b;
+ const int limit = (len_a + len_b) / 2;
+ int kmin;
+ int kmax;
+ int k;
+ int p = -1;
+
+ if (delta >= 0) {
+ kmin = 0;
+ kmax = delta;
+ } else {
+ kmin = delta;
+ kmax = 0;
+ }
+ for (k = kmin; k <= kmax; k++)
+ fp[k] = -1;
+ do {
+ p++;
+ fp[kmin - p - 1] = fp[kmax + p + 1] = -1;
+ for (k = kmax + p; k > delta; k--)
+ diff_snake(vec_a, a0, len_a, vec_b, b0, len_b, seqs, fp, k, limit);
+ for (k = kmin - p; k <= delta; k++)
+ diff_snake(vec_a, a0, len_a, vec_b, b0, len_b, seqs, fp, k, limit);
+ } while (fp[delta] != len_a);
+
+ subseq->x = a0 + seqs[delta].x;
+ subseq->y = b0 + seqs[delta].y;
+ subseq->len = seqs[delta].len;
+ return abs(delta) + 2 * p;;
+}
+
+/* Finds a longest common subsequence.
+ * Returns its length.
+ */
+static int diff_compute_lcs(const char *vec_a[], int a0, int len_a,
+ const char *vec_b[], int b0, int len_b,
+ xbt_dynar_t common_sequence,
+ struct subsequence *seqs, int *fp)
+{
+ if (len_a > 0 && len_b > 0) {
+ struct subsequence subseq;
+ int ses_len = diff_middle_subsequence(vec_a, a0, len_a, vec_b, b0, len_b,
+ &subseq, seqs, fp);
+ int lcs_len = (len_a + len_b - ses_len) / 2;
+ if (lcs_len == 0) {
+ return 0;
+ } else if (ses_len > 1) {
+ int lcs_len1 = subseq.len;
+ if (lcs_len1 < lcs_len)
+ lcs_len1 += diff_compute_lcs(vec_a, a0, subseq.x - a0,
+ vec_b, b0, subseq.y - b0,
+ common_sequence, seqs, fp);
+ if (subseq.len > 0)
+ xbt_dynar_push(common_sequence, &subseq);
+ if (lcs_len1 < lcs_len) {
+ int u = subseq.x + subseq.len;
+ int v = subseq.y + subseq.len;
+ diff_compute_lcs(vec_a, u, a0 + len_a - u, vec_b, v, b0 + len_b - v,
+ common_sequence, seqs, fp);
+ }
+ } else {
+ int len = MIN(len_a, len_b) - subseq.len;
+ if (subseq.x == a0 && subseq.y == b0) {
+ if (subseq.len > 0)
+ xbt_dynar_push(common_sequence, &subseq);
+ if (len > 0) {
+ struct subsequence subseq0 = {a0 + len_a - len,
+ b0 + len_b - len, len};
+ xbt_dynar_push(common_sequence, &subseq0);
+ }
+ } else {
+ if (len > 0) {
+ struct subsequence subseq0 = {a0, b0, len};
+ xbt_dynar_push(common_sequence, &subseq0);
+ }
+ if (subseq.len > 0)
+ xbt_dynar_push(common_sequence, &subseq);
+ }
+ }
+ return lcs_len;
+ } else {
+ return 0;
+ }
+}
+
+static int diff_member(const char *s, const char *vec[], int from, int to)
+{
+ for ( ; from < to ; from++)
+ if (!strcmp(s, vec[from]))
+ return 1;
+ return 0;
+}
+
+/* Extract common prefix.
+ */
+static void diff_easy_prefix(const char *vec_a[], int *a0_p, int *len_a_p,
+ const char *vec_b[], int *b0_p, int *len_b_p,
+ xbt_dynar_t common_sequence)
+{
+ int a0 = *a0_p;
+ int b0 = *b0_p;
+ int len_a = *len_a_p;
+ int len_b = *len_b_p;
+
+ while (len_a > 0 && len_b > 0) {
+ struct subsequence subseq = {a0, b0, 0};
+ while (len_a > 0 && len_b > 0 && !strcmp(vec_a[a0], vec_b[b0])) {
+ a0++, len_a--;
+ b0++, len_b--;
+ subseq.len++;
+ }
+ if (subseq.len > 0)
+ xbt_dynar_push(common_sequence, &subseq);
+ if (len_a > 0 && len_b > 0 &&
+ !diff_member(vec_a[a0], vec_b, b0 + 1, b0 + len_b)) {
+ a0++, len_a--;
+ } else {
+ break;
+ }
+ }
+
+ *a0_p = a0;
+ *b0_p = b0;
+ *len_a_p = len_a;
+ *len_b_p = len_b;
+}
+
+/* Extract common suffix.
+ */
+static void diff_easy_suffix(const char *vec_a[], int *a0_p, int *len_a_p,
+ const char *vec_b[], int *b0_p, int *len_b_p,
+ xbt_dynar_t common_suffix)
+{
+ int a0 = *a0_p;
+ int b0 = *b0_p;
+ int len_a = *len_a_p;
+ int len_b = *len_b_p;
+
+ while (len_a > 0 && len_b > 0){
+ struct subsequence subseq;
+ subseq.len = 0;
+ while (len_a > 0 && len_b > 0 &&
+ !strcmp(vec_a[a0 + len_a - 1], vec_b[b0 + len_b - 1])) {
+ len_a--;
+ len_b--;
+ subseq.len++;
+ }
+ if (subseq.len > 0) {
+ subseq.x = a0 + len_a;
+ subseq.y = b0 + len_b;
+ xbt_dynar_push(common_suffix, &subseq);
+ }
+ if (len_a > 0 && len_b > 0 &&
+ !diff_member(vec_b[b0 + len_b - 1], vec_a, a0, a0 + len_a - 1)) {
+ len_b--;
+ } else {
+ break;
+ }
+ }
+
+ *a0_p = a0;
+ *b0_p = b0;
+ *len_a_p = len_a;
+ *len_b_p = len_b;
+}
+
+/** @brief Compute the unified diff of two strings */
+char *xbt_str_diff(const char *a, const char *b)
+{
+ xbt_dynar_t da = xbt_str_split(a, "\n");
+ xbt_dynar_t db = xbt_str_split(b, "\n");
+ xbt_dynar_t common_sequence, common_suffix;
+ size_t len;
+ const char **vec_a, **vec_b;
+ int a0, b0;
+ int len_a, len_b;
+ int max;
+ int *fp_base, *fp;
+ struct subsequence *seqs_base, *seqs;
+ struct subsequence subseq;
+ xbt_dynar_t diff;
+ char *res;
+ int x, y;
+ unsigned s;
+
+ /* Clean empty lines at the end of da and db */
+ len = strlen(a);
+ if (len > 0 && a[len - 1] == '\n')
+ xbt_dynar_pop(da, NULL);
+ len = strlen(b);
+ if (len > 0 && b[len - 1] == '\n')
+ xbt_dynar_pop(db, NULL);
+
+ /* Various initializations */
+ /* Assume that dynar's content is contiguous */
+ a0 = 0;
+ len_a = xbt_dynar_length(da);
+ vec_a = len_a ? xbt_dynar_get_ptr(da, 0) : NULL;
+ b0 = 0;
+ len_b = xbt_dynar_length(db);
+ vec_b = len_b ? xbt_dynar_get_ptr(db, 0) : NULL;
+ max = MAX(len_a, len_b) + 1;
+ fp_base = xbt_new(int, 2 * max + 1);
+ fp = fp_base + max; /* indexes in [-max..max] */
+ seqs_base = xbt_new(struct subsequence, 2 * max + 1);
+ seqs = seqs_base + max; /* indexes in [-max..max] */
+ common_sequence = xbt_dynar_new(sizeof(struct subsequence), NULL);
+ common_suffix = xbt_dynar_new(sizeof(struct subsequence), NULL);
+
+ /* Add a sentinel a the end of the sequence */
+ subseq.x = len_a;
+ subseq.y = len_b;
+ subseq.len = 0;
+ xbt_dynar_push(common_suffix, &subseq);
+
+ /* Compute the Longest Common Subsequence */
+ diff_easy_prefix(vec_a, &a0, &len_a, vec_b, &b0, &len_b, common_sequence);
+ diff_easy_suffix(vec_a, &a0, &len_a, vec_b, &b0, &len_b, common_suffix);
+ diff_compute_lcs(vec_a, a0, len_a, vec_b, b0, len_b, common_sequence, seqs, fp);
+ while (!xbt_dynar_is_empty(common_suffix)) {
+ xbt_dynar_pop(common_suffix, &subseq);
+ xbt_dynar_push(common_sequence, &subseq);
+ }
+
+ /* Build a Shortest Edit Script, and the final result */
+ diff = xbt_dynar_new(sizeof(char *), &xbt_free_ref);
+ x = 0;
+ y = 0;
+ xbt_dynar_foreach(common_sequence, s, subseq) {
+ while (x < subseq.x) {
+ char *topush = bprintf("- %s", vec_a[x++]);
+ xbt_dynar_push_as(diff, char*, topush);
+ }
+ while (y < subseq.y) {
+ char *topush = bprintf("+ %s", vec_b[y++]);
+ xbt_dynar_push_as(diff, char*, topush);
+ }
+ while (x < subseq.x + subseq.len) {
+ char *topush = bprintf(" %s", vec_a[x++]);
+ xbt_dynar_push_as(diff, char*, topush);
+ y++;
+ }
+ }
+ res = xbt_str_join(diff, "\n");
+
+ xbt_free(fp_base);
+ xbt_free(seqs_base);
+ xbt_dynar_free(&db);
+ xbt_dynar_free(&da);
+ xbt_dynar_free(&common_sequence);
+ xbt_dynar_free(&common_suffix);
+ xbt_dynar_free(&diff);
+
+ return res;
+}
+
+
+/** @brief creates a new string containing what can be read on a fd
+ *
+ */
+char *xbt_str_from_file(FILE * file)
+{
+ xbt_strbuff_t buff = xbt_strbuff_new();
+ char *res;
+ char bread[1024];
+ memset(bread, 0, 1024);
+
+ while (!feof(file)) {
+ int got = fread(bread, 1, 1023, file);
+ bread[got] = '\0';
+ xbt_strbuff_append(buff, bread);