Merge pull request #2319 from dgquintas/str_join_with_sep

Added convenience functions gpr_strjoin_sep and grp_strsplit

src/core/support/string.c

@@ -38,6 +38,7 @@
 #include <string.h>
 
 #include <grpc/support/alloc.h>
+#include <grpc/support/log.h>
 #include <grpc/support/port_platform.h>
 #include <grpc/support/useful.h>
 
@@ -174,6 +175,12 @@ int gpr_ltoa(long value, char *string) {
 }
 
 char *gpr_strjoin(const char **strs, size_t nstrs, size_t *final_length) {
+  return gpr_strjoin_sep(strs, nstrs, "", final_length);
+}
+
+char *gpr_strjoin_sep(const char **strs, size_t nstrs, const char *sep,
+                      size_t *final_length) {
+  const size_t sep_len = strlen(sep);
   size_t out_length = 0;
   size_t i;
   char *out;
@@ -181,10 +188,17 @@ char *gpr_strjoin(const char **strs, size_t nstrs, size_t *final_length) {
     out_length += strlen(strs[i]);
   }
   out_length += 1;  /* null terminator */
+  if (nstrs > 0) {
+    out_length += sep_len * (nstrs - 1);  /* separators */
+  }
   out = gpr_malloc(out_length);
   out_length = 0;
   for (i = 0; i < nstrs; i++) {
-    size_t slen = strlen(strs[i]);
+    const size_t slen = strlen(strs[i]);
+    if (i != 0) {
+      memcpy(out + out_length, sep, sep_len);
+      out_length += sep_len;
+    }
     memcpy(out + out_length, strs[i], slen);
     out_length += slen;
   }
@@ -195,6 +209,52 @@ char *gpr_strjoin(const char **strs, size_t nstrs, size_t *final_length) {
   return out;
 }
 
+/** Finds the initial (\a begin) and final (\a end) offsets of the next
+ * substring from \a str + \a read_offset until the next \a sep or the end of \a
+ * str.
+ *
+ * Returns 1 and updates \a begin and \a end. Returns 0 otherwise. */
+static int slice_find_separator_offset(const gpr_slice str,
+                                       const char *sep,
+                                       const size_t read_offset,
+                                       size_t *begin,
+                                       size_t *end) {
+  size_t i;
+  const gpr_uint8 *str_ptr = GPR_SLICE_START_PTR(str) + read_offset;
+  const size_t str_len = GPR_SLICE_LENGTH(str) - read_offset;
+  const size_t sep_len = strlen(sep);
+  if (str_len < sep_len) {
+    return 0;
+  }
+
+  for (i = 0; i <= str_len - sep_len; i++) {
+    if (memcmp(str_ptr + i, sep, sep_len) == 0) {
+      *begin = read_offset;
+      *end = read_offset + i;
+      return 1;
+    }
+  }
+  return 0;
+}
+
+void gpr_slice_split(gpr_slice str, const char *sep, gpr_slice_buffer *dst) {
+  const size_t sep_len = strlen(sep);
+  size_t begin, end;
+
+  GPR_ASSERT(sep_len > 0);
+
+  if (slice_find_separator_offset(str, sep, 0, &begin, &end) != 0) {
+    do {
+      gpr_slice_buffer_add_indexed(dst, gpr_slice_sub(str, begin, end));
+    } while (slice_find_separator_offset(str, sep, end + sep_len, &begin,
+                                         &end) != 0);
+    gpr_slice_buffer_add_indexed(
+        dst, gpr_slice_sub(str, end + sep_len, GPR_SLICE_LENGTH(str)));
+  } else { /* no sep found, add whole input */
+    gpr_slice_buffer_add_indexed(dst, gpr_slice_ref(str));
+  }
+}
+
 void gpr_strvec_init(gpr_strvec *sv) {
   memset(sv, 0, sizeof(*sv));
 }

src/core/support/string.h

@@ -37,6 +37,7 @@
 #include <stddef.h>
 
 #include <grpc/support/port_platform.h>
+#include <grpc/support/slice_buffer.h>
 #include <grpc/support/slice.h>
 
 #ifdef __cplusplus
@@ -77,6 +78,16 @@ void gpr_reverse_bytes(char *str, int len);
    if it is non-null. */
 char *gpr_strjoin(const char **strs, size_t nstrs, size_t *total_length);
 
+/* Join a set of strings using a separator, returning the resulting string.
+   Total combined length (excluding null terminator) is returned in total_length
+   if it is non-null. */
+char *gpr_strjoin_sep(const char **strs, size_t nstrs, const char *sep,
+                      size_t *total_length);
+
+/** Split \a str by the separator \a sep. Results are stored in \a dst, which
+ * should be a properly initialized instance. */
+void gpr_slice_split(gpr_slice str, const char *sep, gpr_slice_buffer *dst);
+
 /* A vector of strings... for building up a final string one piece at a time */
 typedef struct {
   char **strs;

test/core/support/string_test.c

@@ -173,6 +173,119 @@ static void test_asprintf(void) {
   }
 }
 
+static void test_strjoin(void) {
+  const char *parts[4] = {"one", "two", "three", "four"};
+  size_t joined_len;
+  char *joined;
+
+  LOG_TEST_NAME("test_strjoin");
+
+  joined = gpr_strjoin(parts, 4, &joined_len);
+  GPR_ASSERT(0 == strcmp("onetwothreefour", joined));
+  gpr_free(joined);
+
+  joined = gpr_strjoin(parts, 0, &joined_len);
+  GPR_ASSERT(0 == strcmp("", joined));
+  gpr_free(joined);
+
+  joined = gpr_strjoin(parts, 1, &joined_len);
+  GPR_ASSERT(0 == strcmp("one", joined));
+  gpr_free(joined);
+}
+
+static void test_strjoin_sep(void) {
+  const char *parts[4] = {"one", "two", "three", "four"};
+  size_t joined_len;
+  char *joined;
+
+  LOG_TEST_NAME("test_strjoin_sep");
+
+  joined = gpr_strjoin_sep(parts, 4, ", ", &joined_len);
+  GPR_ASSERT(0 == strcmp("one, two, three, four", joined));
+  gpr_free(joined);
+
+  /* empty separator */
+  joined = gpr_strjoin_sep(parts, 4, "", &joined_len);
+  GPR_ASSERT(0 == strcmp("onetwothreefour", joined));
+  gpr_free(joined);
+
+  /* degenerated case specifying zero input parts */
+  joined = gpr_strjoin_sep(parts, 0, ", ", &joined_len);
+  GPR_ASSERT(0 == strcmp("", joined));
+  gpr_free(joined);
+
+  /* single part should have no separator */
+  joined = gpr_strjoin_sep(parts, 1, ", ", &joined_len);
+  GPR_ASSERT(0 == strcmp("one", joined));
+  gpr_free(joined);
+}
+
+static void test_strsplit(void) {
+  gpr_slice_buffer* parts;
+  gpr_slice str;
+
+  LOG_TEST_NAME("test_strsplit");
+
+  parts = gpr_malloc(sizeof(gpr_slice_buffer));
+  gpr_slice_buffer_init(parts);
+
+  str = gpr_slice_from_copied_string("one, two, three, four");
+  gpr_slice_split(str, ", ", parts);
+  GPR_ASSERT(4 == parts->count);
+  GPR_ASSERT(0 == gpr_slice_str_cmp(parts->slices[0], "one"));
+  GPR_ASSERT(0 == gpr_slice_str_cmp(parts->slices[1], "two"));
+  GPR_ASSERT(0 == gpr_slice_str_cmp(parts->slices[2], "three"));
+  GPR_ASSERT(0 == gpr_slice_str_cmp(parts->slices[3], "four"));
+  gpr_slice_buffer_reset_and_unref(parts);
+  gpr_slice_unref(str);
+
+  /* separator not present in string */
+  str = gpr_slice_from_copied_string("one two three four");
+  gpr_slice_split(str, ", ", parts);
+  GPR_ASSERT(1 == parts->count);
+  GPR_ASSERT(0 == gpr_slice_str_cmp(parts->slices[0], "one two three four"));
+  gpr_slice_buffer_reset_and_unref(parts);
+  gpr_slice_unref(str);
+
+  /* separator at the end */
+  str = gpr_slice_from_copied_string("foo,");
+  gpr_slice_split(str, ",", parts);
+  GPR_ASSERT(2 == parts->count);
+  GPR_ASSERT(0 == gpr_slice_str_cmp(parts->slices[0], "foo"));
+  GPR_ASSERT(0 == gpr_slice_str_cmp(parts->slices[1], ""));
+  gpr_slice_buffer_reset_and_unref(parts);
+  gpr_slice_unref(str);
+
+  /* separator at the beginning */
+  str = gpr_slice_from_copied_string(",foo");
+  gpr_slice_split(str, ",", parts);
+  GPR_ASSERT(2 == parts->count);
+  GPR_ASSERT(0 == gpr_slice_str_cmp(parts->slices[0], ""));
+  GPR_ASSERT(0 == gpr_slice_str_cmp(parts->slices[1], "foo"));
+  gpr_slice_buffer_reset_and_unref(parts);
+  gpr_slice_unref(str);
+
+  /* standalone separator */
+  str = gpr_slice_from_copied_string(",");
+  gpr_slice_split(str, ",", parts);
+  GPR_ASSERT(2 == parts->count);
+  GPR_ASSERT(0 == gpr_slice_str_cmp(parts->slices[0], ""));
+  GPR_ASSERT(0 == gpr_slice_str_cmp(parts->slices[1], ""));
+  gpr_slice_buffer_reset_and_unref(parts);
+  gpr_slice_unref(str);
+
+  /* empty input */
+  str = gpr_slice_from_copied_string("");
+  gpr_slice_split(str, ", ", parts);
+  GPR_ASSERT(1 == parts->count);
+  GPR_ASSERT(0 == gpr_slice_str_cmp(parts->slices[0], ""));
+  gpr_slice_buffer_reset_and_unref(parts);
+  gpr_slice_unref(str);
+
+  gpr_slice_buffer_destroy(parts);
+  gpr_free(parts);
+}
+
 int main(int argc, char **argv) {
   grpc_test_init(argc, argv);
   test_strdup();
@@ -180,5 +293,8 @@ int main(int argc, char **argv) {
   test_dump_slice();
   test_parse_uint32();
   test_asprintf();
+  test_strjoin();
+  test_strjoin_sep();
+  test_strsplit();
   return 0;
 }