Add str_join

src/base_core.h

@@ -250,7 +250,20 @@ typedef unsigned __int128 U128;
 #define MemMove(dst, src, len)      memmove(dst, src, len)
 #define MemCopy(dst, src, len)       memcpy(dst, src, len)
 #define MemAppend(dst, src, len)    mempcpy(dst, src, len)
+#define MemCompare(first, second, len) memcmp(first, second, len)
 
+static INLINE U8 *MemFind(U8 *mem, U64 mem_len, U8 *needle, U64 needle_len)
+{
+  if (needle_len > mem_len) {
+      return NULL;
+  }
+  for (U64 i = 0 ; i + needle_len <= mem_len; i++) {
+      if (MemCompare(&mem[i], needle, needle_len) == 0) {
+        return &mem[i];
+      }
+  }
+  return NULL;
+}
 // atomics
 #define Atomic(type) _Atomic(type)
 #define AtomicSub(ptr, val) __atomic_fetch_sub((ptr), (val), __ATOMIC_SEQ_CST)

src/main.c

@@ -416,7 +416,7 @@ void string_test_1(void)
 {
     TRACEF("\n");
 
-    Str str1 = S_("Hello");
+    Str str1 = S("Hello");
     printf("str1: %.*s len %d\n", str1.len, str1.str, str1.len);
     U64 str1_index_of_l = str_find_byte(str1, 'l');
     U64 str1_index_of_g = str_find_byte(str1, 'g');
@@ -424,7 +424,7 @@ void string_test_1(void)
 
     MARENA_DEFINE(arena, 512, 1);
     Allocator alloc = marena_allocator(&arena);
-    Str str2 = str_dup(&alloc, S_(" World"));
+    Str str2 = str_dup(&alloc, S(" World"));
     Str str3 = str_concat(&alloc, str1, str2);
     printf("str3: %.*s len %d\n", str3.len, str3.str, str3.len);
 }
@@ -433,8 +433,8 @@ void string_split_test_1(void)
 {
     TRACEF("\n");
 
-    Str str1 = S_("a:bb|ccc");
+    Str str1 = S("a:bb|ccc");
-    Str sep = S_("|:");
+    Str sep = S("|:");
     STRVEC_DEFINE(strings, 16);
 
     str_split(str1, sep, &strings);
@@ -443,6 +443,61 @@ void string_split_test_1(void)
         printf("split %u: '%.*s'\n", i, str->len, str->str);
     }
 }
+void string_test_misc(void)
+{
+    TRACEF("\n");
+
+    Str str1 = S("Hello World");
+    bool prefix1 = str_starts_with(str1, S("He"));
+    bool prefix2 = str_starts_with(str1, S(""));
+    bool prefix3 = str_starts_with(str1, S("false"));
+    bool suffix1 = str_ends_with(str1, S("orld"));
+    bool suffix2 = str_ends_with(str1, S(""));
+    bool suffix3 = str_ends_with(str1, S("false"));
+    printf("matches: %d %d %d %d %d %d\n", prefix1, prefix2, prefix3, suffix1, suffix2, suffix3);
+
+    Str str2 = S(" Hello World\t \n");
+    Str str2_trim = str_ascii_trim(str2);
+    printf("str2_trim: '%.*s'\n", str2_trim.len, str2_trim.str);
+    printf("str2_trim == str1: %d\n", str_eq(str1, str2_trim));
+
+    Str str3 = str_empty();
+    Str str3_trim = str_ascii_trim(str3);
+    printf("str3_trim: '%.*s'\n", str3_trim.len, str3_trim.str);
+
+
+    Str str4 = S("   \t\t    ");
+    Str str4_trim = str_ascii_trim(str4);
+    printf("str4_trim: '%.*s'\n", str4_trim.len, str4_trim.str);
+
+    S64 found = str_substr(str1, S("World"));
+    printf("1. found %ld\n", found);
+    found = str_substr(str_empty(), S("World"));
+    printf("2. found %ld\n", found);
+    found = str_substr(str_empty(), S(""));
+    printf("3. found %ld\n", found);
+    found = str_substr(str1, S(" "));
+    printf("4. found %ld\n", found);
+
+}
+void string_test_join(void)
+{
+    TRACEF("\n");
+    Arena arena;
+    arena_init(&arena, &ArenaOptions_Default);
+    Allocator allocator = arena_allocator(&arena);
+    Str strs[3] = {S("1"), S("two"), S("3-3")};
+    StrVec vec = STRVEC_STATIC_INITIALIZER(strs);
+
+    Str str = str_join(&allocator, vec, S("/"));
+    printf("str_join 1 '%.*s'\n", str.len, str.str);
+    str = str_join(&allocator, vec, S(""));
+    printf("str_join 2 '%.*s'\n", str.len, str.str);
+    vec.len = 0;
+    str = str_join(&allocator, vec, S("/"));
+    printf("str_join 2 '%.*s'\n", str.len, str.str);
+
+}
 
 int main(int argc, char *argv[])
 {
@@ -458,6 +513,8 @@ int main(int argc, char *argv[])
     apool_test_miniarena();
     string_test_1();
     string_split_test_1();
+    string_test_misc();
+    string_test_join();
 
 }
 

src/string.c

@@ -7,7 +7,7 @@ Str str_empty(void)
     return str;
 }
 
-Str str_substr(Str str, U32 start, U32 len)
+Str str_slice(Str str, U32 start, U32 len)
 {
     Assert(start < str.len);
     Assert(len <= str.len);
@@ -16,7 +16,6 @@ Str str_substr(Str str, U32 start, U32 len)
     return new_str;
 }
 
-
 Str str_concat(Allocator *alloc, Str first, Str second)
 {
     Str str;
@@ -32,15 +31,69 @@ Str str_concat(Allocator *alloc, Str first, Str second)
 
 Str str_push(Allocator *alloc, Str str)
 {
-    U8 *mem = (U8 *)allocator_push(alloc, sizeof(Str) + str.len, alignof(Str));
+    U8 *mem = (U8 *)allocator_push(alloc, str.len, 1);
-    Str *new_str = (Str *)mem;
+    MemCopy(mem, str.str, str.len);
-    new_str->len = str.len;
+
-    new_str->str = &mem[sizeof(Str)];
+    return str_create(mem, str.len);
-    if (!str_is_empty(str)) {
+}
-        MemCopy(new_str->str, str.str, str.len);
+
+static INLINE bool char_is_ascii_whitespace(U8 byte)
+{
+    return byte == ' ' || byte == '\t' || byte == '\v' || byte == '\f' || byte == '\r' || byte == '\n';
+
+}
+
+Str str_ascii_trim_left(Str str)
+{
+    U8 *start = str.str;
+    U8 *end = str.str + str.len;
+    while (start != end) {
+        if (!char_is_ascii_whitespace(*start)) {
+            break;
+        }
+        start++;
     }
 
-    return *new_str;
+    return str_create(start, (U32)(end - start));
+}
+
+Str str_ascii_trim_right(Str str)
+{
+    U8 *start = str.str;
+    U8 *end = str.str + str.len;
+    while (start != end) {
+        if (!char_is_ascii_whitespace(*(end - 1))) {
+            break;
+        }
+        end--;
+    }
+
+    return str_create(start, (U32)(end - start));
+}
+
+Str str_ascii_trim(Str str)
+{
+    Str new_str = str_ascii_trim_left(str);
+    new_str = str_ascii_trim_right(new_str);
+    return new_str;
+}
+
+S64 str_substr(Str haystack, Str needle)
+{
+    if (needle.len == 0) {
+        return 0;
+    }
+
+    if (needle.len == 0) {
+        return str_find_byte(haystack, needle.str[0]);
+    }
+
+    U8 *found = MemFind(haystack.str, haystack.len, needle.str, needle.len);
+    if (found != NULL) {
+        return (S64)(found - haystack.str);
+    }
+
+    return -1;
 }
 
 // split @str on any byte in @separators, place split strings in @vec
@@ -58,15 +111,45 @@ void str_split(Str str, Str separators, StrVec *vec)
                 break;
             }
 
-            Str sub_str = str_substr(str, start_pos, i - start_pos);
+            Str sub_str = str_slice(str, start_pos, i - start_pos);
             vec->strings[vec->len] = sub_str;
             vec->len++;
             start_pos = i + 1;
         }
     }
     if (start_pos < str.len && vec->len < vec->max_len) {
-        Str sub_str = str_substr(str, start_pos, str.len - start_pos);
+        Str sub_str = str_slice(str, start_pos, str.len - start_pos);
         vec->strings[vec->len] = sub_str;
         vec->len++;
     }
 }
+
+Str str_join(Allocator *allocator, StrVec strings, Str separator)
+{
+
+    U32 str_len = 0;
+
+    for (U32 i = 0; i < strings.len; i++) {
+        str_len += strings.strings[i].len;
+    }
+    if (strings.len > 0) {
+        str_len += separator.len * (strings.len - 1);
+    }
+
+    U8 *joined_str = allocator_push(allocator, str_len, 1);
+    if (strings.len > 0) {
+        U32 pos = 0;
+        Str *str = &strings.strings[0];
+        MemCopy(&joined_str[pos], str->str, str->len);
+        pos += str->len;
+        for (U32 i = 1; i < strings.len; i++) {
+            str = &strings.strings[i];
+            MemCopy(&joined_str[pos], separator.str, separator.len);
+            pos += separator.len;
+            MemCopy(&joined_str[pos], str->str, str->len);
+            pos += str->len;
+        }
+    }
+
+    return str_create(joined_str, str_len);
+}

src/string.h

@@ -8,15 +8,29 @@ struct Str {
 };
 
 Str str_empty(void);
-Str str_substr(Str str, U32 start, U32 len);
+Str str_slice(Str str, U32 start, U32 len);
+S64 str_substr(Str haystack, Str needle);
+
 Str str_concat(Allocator *alloc, Str first, Str second);
 Str str_push(Allocator *alloc, Str str);
+Str str_ascii_trim(Str str);
+Str str_ascii_trim_left(Str str);
+Str str_ascii_trim_right(Str str);
+
 
 #define  str_from_lit(str) str_create(((U8 *)str), sizeof(str) -1)
-#define  S_(str) str_from_lit(str)
+#define  S(str) str_from_lit(str)
 #define cstr_len(str) (U32)strlen(str)
 #define str_dup(alloc, str) str_push((alloc), (str))
 
+static INLINE bool str_eq(Str first, Str second)
+{
+    if (first.len != second.len) {
+        return false;
+    }
+    return MemCompare(first.str, second.str, first.len) == 0;
+}
+
 static INLINE bool str_is_empty(Str str)
 {
     return str.len == 0;
@@ -51,10 +65,28 @@ static INLINE S64 str_find_byte(Str str, U8 byte)
     return -1;
 }
 
+static INLINE bool str_starts_with(Str str, Str prefix)
+{
+    if (prefix.len > str.len) {
+        return false;
+    }
+
+    return MemCompare(str.str, prefix.str, prefix.len) == 0;
+}
+
+static INLINE bool str_ends_with(Str str, Str suffix)
+{
+    if (suffix.len > str.len) {
+        return false;
+    }
+    U32 start = str.len - suffix.len;
+    return MemCompare(&str.str[start], suffix.str, suffix.len) == 0;
+}
+
 static INLINE Str str_skip(Str str, U32 len)
 {
     U32 skip_len = Min(str.len, len);
-    Str new_str = str_substr(str, skip_len, str.len - skip_len);
+    Str new_str = str_slice(str, skip_len, str.len - skip_len);
 
     return new_str;
 }
@@ -67,12 +99,15 @@ struct StrVec {
     U32 len;
     U32 max_len;
 };
-#define STRVEC_STATIC_INITIALIZER(strings_) STRVEC_INITIALIZER((strings_), ArraySize(strings_))
+#define STRVEC_STATIC_INITIALIZER(strings_) STRVEC_INITIALIZER((strings_), ArraySize(strings_), ArraySize(strings_))
-#define STRVEC_INITIALIZER(strings_, len_) {.strings = (strings_), .len = 0, .max_len = (len_)}
+#define STRVEC_STATIC_INITIALIZER_EMPTY(strings_) STRVEC_INITIALIZER((strings_), 0, ArraySize(strings_))
+#define STRVEC_INITIALIZER(strings_, len_, max_len_) {.strings = (strings_), .len = (len_), .max_len = (max_len_)}
+
 #define STRVEC_DEFINE(var_name, len_)  \
     struct { \
         Str strings[len_];\
     } var_name##_strvec_array; \
-    StrVec var_name = STRVEC_STATIC_INITIALIZER(var_name##_strvec_array.strings)
+    StrVec var_name = STRVEC_STATIC_INITIALIZER_EMPTY(var_name##_strvec_array.strings)
 
 void str_split(Str str, Str separators, StrVec *vec);
+Str str_join(Allocator *allocator, StrVec strings, Str separator);