--- a/ucx/json.c Thu Dec 11 22:58:02 2025 +0100
+++ b/ucx/json.c Fri Dec 12 10:42:53 2025 +0100
@@ -27,6 +27,7 @@
*/
#include "cx/json.h"
+#include "cx/kv_list.h"
#include <string.h>
#include <assert.h>
@@ -41,90 +42,10 @@
static CxJsonValue cx_json_value_nothing = {.type = CX_JSON_NOTHING};
-static int json_cmp_objvalue(const void *l, const void *r) {
- const CxJsonObjValue *left = l;
- const CxJsonObjValue *right = r;
- return cx_strcmp(cx_strcast(left->name), cx_strcast(right->name));
-}
-
-static size_t json_find_objvalue(const CxJsonValue *obj, cxstring name) {
- assert(obj->type == CX_JSON_OBJECT);
- CxJsonObjValue kv_dummy;
- kv_dummy.name = cx_mutstrn((char*) name.ptr, name.length);
- return cx_array_binary_search(
- obj->value.object.values,
- obj->value.object.values_size,
- sizeof(CxJsonObjValue),
- &kv_dummy,
- json_cmp_objvalue
- );
-}
-
-static int json_add_objvalue(CxJsonValue *objv, CxJsonObjValue member) {
- assert(objv->type == CX_JSON_OBJECT);
- const CxAllocator * const al = objv->allocator;
- CxJsonObject *obj = &(objv->value.object);
-
- // determine the index where we need to insert the new member
- size_t index = cx_array_binary_search_sup(
- obj->values,
- obj->values_size,
- sizeof(CxJsonObjValue),
- &member, json_cmp_objvalue
- );
-
- // is the name already present?
- if (index < obj->values_size && 0 == json_cmp_objvalue(&member, &obj->values[index])) {
- // free the original value
- cx_strfree_a(al, &obj->values[index].name);
- cxJsonValueFree(obj->values[index].value);
- // replace the item
- obj->values[index] = member;
-
- // nothing more to do
- return 0;
- }
-
- // determine the old capacity and reserve for one more element
- CxArrayReallocator arealloc = cx_array_reallocator(al, NULL);
- size_t oldcap = obj->values_capacity;
- if (cx_array_simple_reserve_a(&arealloc, obj->values, 1)) return 1;
-
- // check the new capacity, if we need to realloc the index array
- size_t newcap = obj->values_capacity;
- if (newcap > oldcap) {
- if (cxReallocateArray(al, &obj->indices, newcap, sizeof(size_t))) {
- return 1; // LCOV_EXCL_LINE
- }
- }
-
- // check if append or insert
- if (index < obj->values_size) {
- // move the other elements
- memmove(
- &obj->values[index+1],
- &obj->values[index],
- (obj->values_size - index) * sizeof(CxJsonObjValue)
- );
- // increase indices for the moved elements
- for (size_t i = 0; i < obj->values_size ; i++) {
- if (obj->indices[i] >= index) {
- obj->indices[i]++;
- }
- }
- }
-
- // insert the element and set the index
- obj->values[index] = member;
- obj->indices[obj->values_size] = index;
- obj->values_size++;
-
- return 0;
-}
-
static void token_destroy(CxJsonToken *token) {
if (token->allocated) {
cx_strfree(&token->content);
+ token->allocated = false;
}
}
@@ -307,7 +228,9 @@
}
}
- if (ttype != CX_JSON_NO_TOKEN) {
+ if (ttype == CX_JSON_NO_TOKEN) {
+ return CX_JSON_NO_DATA;
+ } else {
// uncompleted token
size_t uncompleted_len = json->buffer.size - token_part_start;
if (json->uncompleted.tokentype == CX_JSON_NO_TOKEN) {
@@ -334,9 +257,8 @@
}
// advance the buffer position - we saved the stuff in the uncompleted token
json->buffer.pos += uncompleted_len;
+ return CX_JSON_INCOMPLETE_DATA;
}
-
- return CX_JSON_INCOMPLETE_DATA;
}
// converts a Unicode codepoint to utf8
@@ -473,7 +395,7 @@
return result;
}
-static cxmutstr escape_string(cxmutstr str, bool escape_slash) {
+static cxmutstr escape_string(cxstring str, bool escape_slash) {
// note: this function produces the string without enclosing quotes
// the reason is that we don't want to allocate memory just for that
CxBuffer buf = {0};
@@ -519,11 +441,27 @@
cxBufferPut(&buf, c);
}
}
- if (!all_printable) {
- str = cx_mutstrn(buf.space, buf.size);
+ cxmutstr ret;
+ if (all_printable) {
+ // don't copy the string when we don't need to escape anything
+ ret = cx_mutstrn((char*)str.ptr, str.length);
+ } else {
+ ret = cx_mutstrn(buf.space, buf.size);
}
cxBufferDestroy(&buf);
- return str;
+ return ret;
+}
+
+static CxJsonObject json_create_object_map(const CxAllocator *allocator) {
+ // TODO: we might want to add a comparator that is sorting the elements by their key
+ CxMap *map = cxKvListCreateAsMap(allocator, NULL, CX_STORE_POINTERS);
+ if (map == NULL) return NULL; // LCOV_EXCL_LINE
+ cxDefineDestructor(map, cxJsonValueFree);
+ return map;
+}
+
+static void json_free_object_map(CxJsonObject obj) {
+ cxMapFree(obj);
}
static CxJsonValue* json_create_value(CxJson *json, CxJsonValueType type) {
@@ -534,14 +472,11 @@
v->type = type;
v->allocator = json->allocator;
if (type == CX_JSON_ARRAY) {
- cx_array_initialize_a(json->allocator, v->value.array.array, 16);
- if (v->value.array.array == NULL) goto create_json_value_exit_error; // LCOV_EXCL_LINE
+ cx_array_initialize_a(json->allocator, v->array.data, 16);
+ if (v->array.data == NULL) goto create_json_value_exit_error; // LCOV_EXCL_LINE
} else if (type == CX_JSON_OBJECT) {
- cx_array_initialize_a(json->allocator, v->value.object.values, 16);
- v->value.object.indices = cxCalloc(json->allocator, 16, sizeof(size_t));
- if (v->value.object.values == NULL ||
- v->value.object.indices == NULL)
- goto create_json_value_exit_error; // LCOV_EXCL_LINE
+ v->object = json_create_object_map(json->allocator);
+ if (v->object == NULL) goto create_json_value_exit_error; // LCOV_EXCL_LINE
}
// add the new value to a possible parent
@@ -550,17 +485,17 @@
assert(parent != NULL);
if (parent->type == CX_JSON_ARRAY) {
CxArrayReallocator value_realloc = cx_array_reallocator(json->allocator, NULL);
- if (cx_array_simple_add_a(&value_realloc, parent->value.array.array, v)) {
+ if (cx_array_simple_add_a(&value_realloc, parent->array.data, v)) {
goto create_json_value_exit_error; // LCOV_EXCL_LINE
}
} else if (parent->type == CX_JSON_OBJECT) {
// the member was already created after parsing the name
- assert(json->uncompleted_member.name.ptr != NULL);
- json->uncompleted_member.value = v;
- if (json_add_objvalue(parent, json->uncompleted_member)) {
+ // store the pointer of the uncompleted value in the map
+ assert(json->uncompleted_member_name.ptr != NULL);
+ if (cxMapPut(parent->object, json->uncompleted_member_name, v)) {
goto create_json_value_exit_error; // LCOV_EXCL_LINE
}
- json->uncompleted_member.name = (cxmutstr) {NULL, 0};
+ cx_strfree_a(json->allocator, &json->uncompleted_member_name);
} else {
assert(false); // LCOV_EXCL_LINE
}
@@ -624,10 +559,8 @@
}
cxJsonValueFree(json->parsed);
json->parsed = NULL;
- if (json->uncompleted_member.name.ptr != NULL) {
- cx_strfree_a(json->allocator, &json->uncompleted_member.name);
- json->uncompleted_member = (CxJsonObjValue){{NULL, 0}, NULL};
- }
+ token_destroy(&json->uncompleted);
+ cx_strfree_a(json->allocator, &json->uncompleted_member_name);
}
void cxJsonReset(CxJson *json) {
@@ -720,7 +653,7 @@
if (str.ptr == NULL) {
return_rec(CX_JSON_VALUE_ALLOC_FAILED); // LCOV_EXCL_LINE
}
- vbuf->value.string = str;
+ vbuf->string = str;
return_rec(CX_JSON_NO_ERROR);
}
case CX_JSON_TOKEN_INTEGER:
@@ -730,11 +663,11 @@
return_rec(CX_JSON_VALUE_ALLOC_FAILED); // LCOV_EXCL_LINE
}
if (type == CX_JSON_INTEGER) {
- if (cx_strtoi64(token.content, &vbuf->value.integer, 10)) {
+ if (cx_strtoi64(token.content, &vbuf->integer, 10)) {
return_rec(CX_JSON_FORMAT_ERROR_NUMBER);
}
} else {
- if (cx_strtod(token.content, &vbuf->value.number)) {
+ if (cx_strtod(token.content, &vbuf->number)) {
// TODO: at the moment this is unreachable, because the tokenizer is already stricter than cx_strtod()
return_rec(CX_JSON_FORMAT_ERROR_NUMBER); // LCOV_EXCL_LINE
}
@@ -746,11 +679,11 @@
return_rec(CX_JSON_VALUE_ALLOC_FAILED); // LCOV_EXCL_LINE
}
if (0 == cx_strcmp(cx_strcast(token.content), cx_str("true"))) {
- vbuf->value.literal = CX_JSON_TRUE;
+ vbuf->literal = CX_JSON_TRUE;
} else if (0 == cx_strcmp(cx_strcast(token.content), cx_str("false"))) {
- vbuf->value.literal = CX_JSON_FALSE;
+ vbuf->literal = CX_JSON_FALSE;
} else {
- vbuf->value.literal = CX_JSON_NULL;
+ vbuf->literal = CX_JSON_NULL;
}
return_rec(CX_JSON_NO_ERROR);
}
@@ -786,8 +719,8 @@
if (name.ptr == NULL) {
return_rec(CX_JSON_VALUE_ALLOC_FAILED); // LCOV_EXCL_LINE
}
- assert(json->uncompleted_member.name.ptr == NULL);
- json->uncompleted_member.name = name;
+ assert(json->uncompleted_member_name.ptr == NULL);
+ json->uncompleted_member_name = name;
assert(json->vbuf_size > 0);
// next state
@@ -860,29 +793,54 @@
return result;
}
+CxJsonStatus cx_json_from_string(const CxAllocator *allocator,
+ cxstring str, CxJsonValue **value) {
+ *value = &cx_json_value_nothing;
+ CxJson parser;
+ cxJsonInit(&parser, allocator);
+ if (cxJsonFill(&parser, str)) {
+ // LCOV_EXCL_START
+ cxJsonDestroy(&parser);
+ return CX_JSON_BUFFER_ALLOC_FAILED;
+ // LCOV_EXCL_STOP
+ }
+ CxJsonStatus status = cxJsonNext(&parser, value);
+ // check if we consume the total string
+ CxJsonValue *chk_value = NULL;
+ CxJsonStatus chk_status = CX_JSON_NO_DATA;
+ if (status == CX_JSON_NO_ERROR) {
+ chk_status = cxJsonNext(&parser, &chk_value);
+ }
+ cxJsonDestroy(&parser);
+ if (chk_status == CX_JSON_NO_DATA) {
+ return status;
+ } else {
+ cxJsonValueFree(*value);
+ // if chk_value is nothing, the free is harmless
+ cxJsonValueFree(chk_value);
+ *value = &cx_json_value_nothing;
+ return CX_JSON_FORMAT_ERROR_UNEXPECTED_TOKEN;
+ }
+
+}
+
void cxJsonValueFree(CxJsonValue *value) {
if (value == NULL || value->type == CX_JSON_NOTHING) return;
switch (value->type) {
case CX_JSON_OBJECT: {
- CxJsonObject obj = value->value.object;
- for (size_t i = 0; i < obj.values_size; i++) {
- cxJsonValueFree(obj.values[i].value);
- cx_strfree_a(value->allocator, &obj.values[i].name);
- }
- cxFree(value->allocator, obj.values);
- cxFree(value->allocator, obj.indices);
+ json_free_object_map(value->object);
break;
}
case CX_JSON_ARRAY: {
- CxJsonArray array = value->value.array;
- for (size_t i = 0; i < array.array_size; i++) {
- cxJsonValueFree(array.array[i]);
+ CxJsonArray array = value->array;
+ for (size_t i = 0; i < array.data_size; i++) {
+ cxJsonValueFree(array.data[i]);
}
- cxFree(value->allocator, array.array);
+ cxFree(value->allocator, array.data);
break;
}
case CX_JSON_STRING: {
- cxFree(value->allocator, value->value.string.ptr);
+ cxFree(value->allocator, value->string.ptr);
break;
}
default: {
@@ -898,15 +856,8 @@
if (v == NULL) return NULL;
v->allocator = allocator;
v->type = CX_JSON_OBJECT;
- cx_array_initialize_a(allocator, v->value.object.values, 16);
- if (v->value.object.values == NULL) { // LCOV_EXCL_START
- cxFree(allocator, v);
- return NULL;
- // LCOV_EXCL_STOP
- }
- v->value.object.indices = cxCalloc(allocator, 16, sizeof(size_t));
- if (v->value.object.indices == NULL) { // LCOV_EXCL_START
- cxFree(allocator, v->value.object.values);
+ v->object = json_create_object_map(allocator);
+ if (v->object == NULL) { // LCOV_EXCL_START
cxFree(allocator, v);
return NULL;
// LCOV_EXCL_STOP
@@ -920,8 +871,8 @@
if (v == NULL) return NULL;
v->allocator = allocator;
v->type = CX_JSON_ARRAY;
- cx_array_initialize_a(allocator, v->value.array.array, 16);
- if (v->value.array.array == NULL) { cxFree(allocator, v); return NULL; }
+ cx_array_initialize_a(allocator, v->array.data, 16);
+ if (v->array.data == NULL) { cxFree(allocator, v); return NULL; }
return v;
}
@@ -931,7 +882,7 @@
if (v == NULL) return NULL;
v->allocator = allocator;
v->type = CX_JSON_NUMBER;
- v->value.number = num;
+ v->number = num;
return v;
}
@@ -941,7 +892,7 @@
if (v == NULL) return NULL;
v->allocator = allocator;
v->type = CX_JSON_INTEGER;
- v->value.integer = num;
+ v->integer = num;
return v;
}
@@ -953,7 +904,7 @@
v->type = CX_JSON_STRING;
cxmutstr s = cx_strdup_a(allocator, str);
if (s.ptr == NULL) { cxFree(allocator, v); return NULL; }
- v->value.string = s;
+ v->string = s;
return v;
}
@@ -963,7 +914,7 @@
if (v == NULL) return NULL;
v->allocator = allocator;
v->type = CX_JSON_LITERAL;
- v->value.literal = lit;
+ v->literal = lit;
return v;
}
@@ -1042,24 +993,14 @@
CxArrayReallocator value_realloc = cx_array_reallocator(arr->allocator, NULL);
assert(arr->type == CX_JSON_ARRAY);
return cx_array_simple_copy_a(&value_realloc,
- arr->value.array.array,
- arr->value.array.array_size,
+ arr->array.data,
+ arr->array.data_size,
val, count
);
}
int cx_json_obj_put(CxJsonValue* obj, cxstring name, CxJsonValue* child) {
- cxmutstr k = cx_strdup_a(obj->allocator, name);
- if (k.ptr == NULL) return -1;
- CxJsonObjValue kv = {k, child};
- if (json_add_objvalue(obj, kv)) {
- // LCOV_EXCL_START
- cx_strfree_a(obj->allocator, &k);
- return 1;
- // LCOV_EXCL_STOP
- } else {
- return 0;
- }
+ return cxMapPut(obj->object, name, child);
}
CxJsonValue* cx_json_obj_put_obj(CxJsonValue* obj, cxstring name) {
@@ -1105,98 +1046,84 @@
}
CxJsonValue *cxJsonArrGet(const CxJsonValue *value, size_t index) {
- if (index >= value->value.array.array_size) {
+ if (index >= value->array.data_size) {
return &cx_json_value_nothing;
}
- return value->value.array.array[index];
+ return value->array.data[index];
}
CxJsonValue *cxJsonArrRemove(CxJsonValue *value, size_t index) {
- if (index >= value->value.array.array_size) {
+ if (index >= value->array.data_size) {
return NULL;
}
- CxJsonValue *ret = value->value.array.array[index];
+ CxJsonValue *ret = value->array.data[index];
// TODO: replace with a low level cx_array_remove()
- size_t count = value->value.array.array_size - index - 1;
+ size_t count = value->array.data_size - index - 1;
if (count > 0) {
- memmove(value->value.array.array + index, value->value.array.array + index + 1, count * sizeof(CxJsonValue*));
+ memmove(value->array.data + index, value->array.data + index + 1, count * sizeof(CxJsonValue*));
}
- value->value.array.array_size--;
+ value->array.data_size--;
return ret;
}
char *cxJsonAsString(const CxJsonValue *value) {
- return value->value.string.ptr;
+ return value->string.ptr;
}
cxstring cxJsonAsCxString(const CxJsonValue *value) {
- return cx_strcast(value->value.string);
+ return cx_strcast(value->string);
}
cxmutstr cxJsonAsCxMutStr(const CxJsonValue *value) {
- return value->value.string;
+ return value->string;
}
double cxJsonAsDouble(const CxJsonValue *value) {
if (value->type == CX_JSON_INTEGER) {
- return (double) value->value.integer;
+ return (double) value->integer;
} else {
- return value->value.number;
+ return value->number;
}
}
int64_t cxJsonAsInteger(const CxJsonValue *value) {
if (value->type == CX_JSON_INTEGER) {
- return value->value.integer;
+ return value->integer;
} else {
- return (int64_t) value->value.number;
+ return (int64_t) value->number;
}
}
CxIterator cxJsonArrIter(const CxJsonValue *value) {
return cxIteratorPtr(
- value->value.array.array,
- value->value.array.array_size,
+ value->array.data,
+ value->array.data_size,
true // arrays need to keep order
);
}
-CxIterator cxJsonObjIter(const CxJsonValue *value) {
- return cxIterator(
- value->value.object.values,
- sizeof(CxJsonObjValue),
- value->value.object.values_size,
- true // TODO: objects do not always need to keep order
- );
+CxMapIterator cxJsonObjIter(const CxJsonValue *value) {
+ return cxMapIterator(value->object);
}
CxJsonValue *cx_json_obj_get(const CxJsonValue *value, cxstring name) {
- size_t index = json_find_objvalue(value, name);
- if (index >= value->value.object.values_size) {
+ CxJsonValue *v = cxMapGet(value->object, name);
+ if (v == NULL) {
return &cx_json_value_nothing;
} else {
- return value->value.object.values[index].value;
+ return v;
}
}
CxJsonValue *cx_json_obj_remove(CxJsonValue *value, cxstring name) {
- size_t index = json_find_objvalue(value, name);
- if (index >= value->value.object.values_size) {
- return NULL;
- } else {
- CxJsonObjValue kv = value->value.object.values[index];
- cx_strfree_a(value->allocator, &kv.name);
- // TODO: replace with cx_array_remove() / cx_array_remove_fast()
- value->value.object.values_size--;
- memmove(value->value.object.values + index, value->value.object.values + index + 1, (value->value.object.values_size - index) * sizeof(CxJsonObjValue));
- return kv.value;
- }
+ CxJsonValue *v = NULL;
+ cxMapRemoveAndGet(value->object, name, &v);
+ return v;
}
CxJsonWriter cxJsonWriterCompact(void) {
return (CxJsonWriter) {
false,
- true,
6,
false,
4,
@@ -1207,7 +1134,6 @@
CxJsonWriter cxJsonWriterPretty(bool use_spaces) {
return (CxJsonWriter) {
true,
- true,
6,
use_spaces,
4,
@@ -1272,14 +1198,8 @@
expected++;
}
depth++;
- size_t elem_count = value->value.object.values_size;
- for (size_t look_idx = 0; look_idx < elem_count; look_idx++) {
- // get the member either via index array or directly
- size_t elem_idx = settings->sort_members
- ? look_idx
- : value->value.object.indices[look_idx];
- CxJsonObjValue *member = &value->value.object.values[elem_idx];
-
+ CxMapIterator member_iter = cxJsonObjIter(value);
+ cx_foreach(const CxMapEntry *, member, member_iter) {
// possible indentation
if (settings->pretty) {
if (cx_json_writer_indent(target, wfunc, settings, depth)) {
@@ -1289,26 +1209,27 @@
// the name
actual += wfunc("\"", 1, 1, target);
- cxmutstr name = escape_string(member->name, settings->escape_slash);
+ cxstring key = cx_strn(member->key->data, member->key->len);
+ cxmutstr name = escape_string(key, settings->escape_slash);
actual += wfunc(name.ptr, 1, name.length, target);
- if (name.ptr != member->name.ptr) {
- cx_strfree(&name);
- }
actual += wfunc("\"", 1, 1, target);
const char *obj_name_sep = ": ";
if (settings->pretty) {
actual += wfunc(obj_name_sep, 1, 2, target);
- expected += 4 + member->name.length;
+ expected += 4 + name.length;
} else {
actual += wfunc(obj_name_sep, 1, 1, target);
- expected += 3 + member->name.length;
+ expected += 3 + name.length;
+ }
+ if (name.ptr != key.ptr) {
+ cx_strfree(&name);
}
// the value
if (cx_json_write_rec(target, member->value, wfunc, settings, depth)) return 1;
// end of object-value
- if (look_idx < elem_count - 1) {
+ if (member_iter.index < member_iter.elem_count - 1) {
const char *obj_value_sep = ",\n";
if (settings->pretty) {
actual += wfunc(obj_value_sep, 1, 2, target);
@@ -1361,13 +1282,14 @@
}
case CX_JSON_STRING: {
actual += wfunc("\"", 1, 1, target);
- cxmutstr str = escape_string(value->value.string, settings->escape_slash);
+ cxmutstr str = escape_string(cx_strcast(value->string),
+ settings->escape_slash);
actual += wfunc(str.ptr, 1, str.length, target);
- if (str.ptr != value->value.string.ptr) {
+ actual += wfunc("\"", 1, 1, target);
+ expected += 2 + str.length;
+ if (str.ptr != value->string.ptr) {
cx_strfree(&str);
}
- actual += wfunc("\"", 1, 1, target);
- expected += 2 + value->value.string.length;
break;
}
case CX_JSON_NUMBER: {
@@ -1375,7 +1297,7 @@
// because of the way how %g is defined, we need to
// double the precision and truncate ourselves
precision = 1 + (precision > 15 ? 30 : 2 * precision);
- snprintf(numbuf, 40, "%.*g", precision, value->value.number);
+ snprintf(numbuf, 40, "%.*g", precision, value->number);
char *dot, *exp;
unsigned char max_digits;
// find the decimal separator and hope that it's one of . or ,
@@ -1439,17 +1361,17 @@
break;
}
case CX_JSON_INTEGER: {
- snprintf(numbuf, 32, "%" PRIi64, value->value.integer);
+ snprintf(numbuf, 32, "%" PRIi64, value->integer);
size_t len = strlen(numbuf);
actual += wfunc(numbuf, 1, len, target);
expected += len;
break;
}
case CX_JSON_LITERAL: {
- if (value->value.literal == CX_JSON_TRUE) {
+ if (value->literal == CX_JSON_TRUE) {
actual += wfunc("true", 1, 4, target);
expected += 4;
- } else if (value->value.literal == CX_JSON_FALSE) {
+ } else if (value->literal == CX_JSON_FALSE) {
actual += wfunc("false", 1, 5, target);
expected += 5;
} else {