--- /dev/null Thu Jan 01 00:00:00 1970 +0000
+++ b/ucx/json.c Sun Jan 05 22:00:39 2025 +0100
@@ -0,0 +1,1212 @@
+/*
+ * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS HEADER.
+ *
+ * Copyright 2024 Mike Becker, Olaf Wintermann All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ *
+ * 1. Redistributions of source code must retain the above copyright
+ * notice, this list of conditions and the following disclaimer.
+ *
+ * 2. Redistributions in binary form must reproduce the above copyright
+ * notice, this list of conditions and the following disclaimer in the
+ * documentation and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+ * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+ * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
+ * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE
+ * LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
+ * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
+ * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
+ * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
+ * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
+ * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
+ * POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include "cx/json.h"
+#include "cx/compare.h"
+
+#include <string.h>
+#include <ctype.h>
+#include <assert.h>
+#include <stdio.h>
+#include <errno.h>
+#include <inttypes.h>
+
+/*
+ * RFC 8259
+ * https://tools.ietf.org/html/rfc8259
+ */
+
+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 CxJsonObjValue *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);
+ size_t index = cx_array_binary_search(
+ obj->value.object.values,
+ obj->value.object.values_size,
+ sizeof(CxJsonObjValue),
+ &kv_dummy,
+ json_cmp_objvalue
+ );
+ if (index == obj->value.object.values_size) {
+ return NULL;
+ } else {
+ return &obj->value.object.values[index];
+ }
+}
+
+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;
+ }
+ }
+
+ // 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);
+ }
+}
+
+static int num_isexp(const char *content, size_t length, size_t pos) {
+ if (pos >= length) {
+ return 0;
+ }
+
+ int ok = 0;
+ for (size_t i = pos; i < length; i++) {
+ char c = content[i];
+ if (isdigit(c)) {
+ ok = 1;
+ } else if (i == pos) {
+ if (!(c == '+' || c == '-')) {
+ return 0;
+ }
+ } else {
+ return 0;
+ }
+ }
+
+ return ok;
+}
+
+static CxJsonTokenType token_numbertype(const char *content, size_t length) {
+ if (length == 0) return CX_JSON_TOKEN_ERROR;
+
+ if (content[0] != '-' && !isdigit(content[0])) {
+ return CX_JSON_TOKEN_ERROR;
+ }
+
+ CxJsonTokenType type = CX_JSON_TOKEN_INTEGER;
+ for (size_t i = 1; i < length; i++) {
+ if (content[i] == '.') {
+ if (type == CX_JSON_TOKEN_NUMBER) {
+ return CX_JSON_TOKEN_ERROR; // more than one decimal separator
+ }
+ type = CX_JSON_TOKEN_NUMBER;
+ } else if (content[i] == 'e' || content[i] == 'E') {
+ return num_isexp(content, length, i + 1) ? CX_JSON_TOKEN_NUMBER : CX_JSON_TOKEN_ERROR;
+ } else if (!isdigit(content[i])) {
+ return CX_JSON_TOKEN_ERROR; // char is not a digit, decimal separator or exponent sep
+ }
+ }
+
+ return type;
+}
+
+static CxJsonToken token_create(CxJson *json, bool isstring, size_t start, size_t end) {
+ cxmutstr str = cx_mutstrn(json->buffer.space + start, end - start);
+ bool allocated = false;
+ if (json->uncompleted.tokentype != CX_JSON_NO_TOKEN) {
+ allocated = true;
+ str = cx_strcat_m(json->uncompleted.content, 1, str);
+ if (str.ptr == NULL) { // LCOV_EXCL_START
+ return (CxJsonToken){CX_JSON_NO_TOKEN, false, {NULL, 0}};
+ } // LCOV_EXCL_STOP
+ }
+ json->uncompleted = (CxJsonToken){0};
+ CxJsonTokenType ttype;
+ if (isstring) {
+ ttype = CX_JSON_TOKEN_STRING;
+ } else {
+ cxstring s = cx_strcast(str);
+ if (!cx_strcmp(s, CX_STR("true")) || !cx_strcmp(s, CX_STR("false"))
+ || !cx_strcmp(s, CX_STR("null"))) {
+ ttype = CX_JSON_TOKEN_LITERAL;
+ } else {
+ ttype = token_numbertype(str.ptr, str.length);
+ }
+ }
+ if (ttype == CX_JSON_TOKEN_ERROR) {
+ if (allocated) {
+ cx_strfree(&str);
+ }
+ return (CxJsonToken){CX_JSON_TOKEN_ERROR, false, {NULL, 0}};
+ }
+ return (CxJsonToken){ttype, allocated, str};
+}
+
+static CxJsonTokenType char2ttype(char c) {
+ switch (c) {
+ case '[': {
+ return CX_JSON_TOKEN_BEGIN_ARRAY;
+ }
+ case '{': {
+ return CX_JSON_TOKEN_BEGIN_OBJECT;
+ }
+ case ']': {
+ return CX_JSON_TOKEN_END_ARRAY;
+ }
+ case '}': {
+ return CX_JSON_TOKEN_END_OBJECT;
+ }
+ case ':': {
+ return CX_JSON_TOKEN_NAME_SEPARATOR;
+ }
+ case ',': {
+ return CX_JSON_TOKEN_VALUE_SEPARATOR;
+ }
+ case '"': {
+ return CX_JSON_TOKEN_STRING;
+ }
+ default: {
+ if (isspace(c)) {
+ return CX_JSON_TOKEN_SPACE;
+ }
+ }
+ }
+ return CX_JSON_NO_TOKEN;
+}
+
+static enum cx_json_status token_parse_next(CxJson *json, CxJsonToken *result) {
+ // check if there is data in the buffer
+ if (cxBufferEof(&json->buffer)) {
+ return json->uncompleted.tokentype == CX_JSON_NO_TOKEN ?
+ CX_JSON_NO_DATA : CX_JSON_INCOMPLETE_DATA;
+ }
+
+ // current token type and start index
+ CxJsonTokenType ttype = json->uncompleted.tokentype;
+ size_t token_start = json->buffer.pos;
+
+ for (size_t i = json->buffer.pos; i < json->buffer.size; i++) {
+ char c = json->buffer.space[i];
+ if (ttype != CX_JSON_TOKEN_STRING) {
+ // currently non-string token
+ CxJsonTokenType ctype = char2ttype(c); // start of new token?
+ if (ttype == CX_JSON_NO_TOKEN) {
+ if (ctype == CX_JSON_TOKEN_SPACE) {
+ json->buffer.pos++;
+ continue;
+ } else if (ctype == CX_JSON_TOKEN_STRING) {
+ // begin string
+ ttype = CX_JSON_TOKEN_STRING;
+ token_start = i;
+ } else if (ctype != CX_JSON_NO_TOKEN) {
+ // single-char token
+ json->buffer.pos = i + 1;
+ *result = (CxJsonToken){ctype, false, {NULL, 0}};
+ return CX_JSON_NO_ERROR;
+ } else {
+ ttype = CX_JSON_TOKEN_LITERAL; // number or literal
+ token_start = i;
+ }
+ } else {
+ // finish token
+ if (ctype != CX_JSON_NO_TOKEN) {
+ *result = token_create(json, false, token_start, i);
+ if (result->tokentype == CX_JSON_NO_TOKEN) {
+ return CX_JSON_BUFFER_ALLOC_FAILED; // LCOV_EXCL_LINE
+ }
+ if (result->tokentype == CX_JSON_TOKEN_ERROR) {
+ return CX_JSON_FORMAT_ERROR_NUMBER;
+ }
+ json->buffer.pos = i;
+ return CX_JSON_NO_ERROR;
+ }
+ }
+ } else {
+ // currently inside a string
+ if (json->tokenizer_escape) {
+ json->tokenizer_escape = false;
+ } else {
+ if (c == '"') {
+ *result = token_create(json, true, token_start, i + 1);
+ if (result->tokentype == CX_JSON_NO_TOKEN) {
+ return CX_JSON_BUFFER_ALLOC_FAILED; // LCOV_EXCL_LINE
+ }
+ json->buffer.pos = i + 1;
+ return CX_JSON_NO_ERROR;
+ } else if (c == '\\') {
+ json->tokenizer_escape = true;
+ }
+ }
+ }
+ }
+
+ if (ttype != CX_JSON_NO_TOKEN) {
+ // uncompleted token
+ size_t uncompleted_len = json->buffer.size - token_start;
+ if (json->uncompleted.tokentype == CX_JSON_NO_TOKEN) {
+ // current token is uncompleted
+ // save current token content
+ CxJsonToken uncompleted = {
+ ttype, true,
+ cx_strdup(cx_strn(json->buffer.space + token_start, uncompleted_len))
+ };
+ if (uncompleted.content.ptr == NULL) {
+ return CX_JSON_BUFFER_ALLOC_FAILED; // LCOV_EXCL_LINE
+ }
+ json->uncompleted = uncompleted;
+ } else {
+ // previously we also had an uncompleted token
+ // combine the uncompleted token with the current token
+ assert(json->uncompleted.allocated);
+ cxmutstr str = cx_strcat_m(json->uncompleted.content, 1,
+ cx_strn(json->buffer.space + token_start, uncompleted_len));
+ if (str.ptr == NULL) {
+ return CX_JSON_BUFFER_ALLOC_FAILED; // LCOV_EXCL_LINE
+ }
+ json->uncompleted.content = str;
+ }
+ // advance the buffer position - we saved the stuff in the uncompleted token
+ json->buffer.pos += uncompleted_len;
+ }
+
+ return CX_JSON_INCOMPLETE_DATA;
+}
+
+static cxmutstr unescape_string(const CxAllocator *a, cxmutstr str) {
+ // TODO: support more escape sequences
+ // we know that the unescaped string will be shorter by at least 2 chars
+ cxmutstr result;
+ result.length = 0;
+ result.ptr = cxMalloc(a, str.length - 1);
+ if (result.ptr == NULL) return result; // LCOV_EXCL_LINE
+
+ bool u = false;
+ for (size_t i = 1; i < str.length - 1; i++) {
+ char c = str.ptr[i];
+ if (u) {
+ u = false;
+ if (c == 'n') {
+ c = '\n';
+ } else if (c == 't') {
+ c = '\t';
+ }
+ result.ptr[result.length++] = c;
+ } else {
+ if (c == '\\') {
+ u = true;
+ } else {
+ result.ptr[result.length++] = c;
+ }
+ }
+ }
+ result.ptr[result.length] = 0;
+
+ return result;
+}
+
+static CxJsonValue* create_json_value(CxJson *json, CxJsonValueType type) {
+ CxJsonValue *v = cxCalloc(json->allocator, 1, sizeof(CxJsonValue));
+ if (v == NULL) return NULL; // LCOV_EXCL_LINE
+
+ // initialize the value
+ 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
+ } 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
+ }
+
+ // add the new value to a possible parent
+ if (json->vbuf_size > 0) {
+ CxJsonValue *parent = json->vbuf[json->vbuf_size - 1];
+ 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)) {
+ 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)) {
+ goto create_json_value_exit_error; // LCOV_EXCL_LINE
+ }
+ json->uncompleted_member.name = (cxmutstr) {NULL, 0};
+ } else {
+ assert(false); // LCOV_EXCL_LINE
+ }
+ }
+
+ // add the new value to the stack, if it is an array or object
+ if (type == CX_JSON_ARRAY || type == CX_JSON_OBJECT) {
+ CxArrayReallocator vbuf_realloc = cx_array_reallocator(NULL, json->vbuf_internal);
+ if (cx_array_simple_add_a(&vbuf_realloc, json->vbuf, v)) {
+ goto create_json_value_exit_error; // LCOV_EXCL_LINE
+ }
+ }
+
+ // if currently no value is parsed, this is now the value of interest
+ if (json->parsed == NULL) {
+ json->parsed = v;
+ }
+
+ return v;
+ // LCOV_EXCL_START
+create_json_value_exit_error:
+ cxJsonValueFree(v);
+ return NULL;
+ // LCOV_EXCL_STOP
+}
+
+#define JP_STATE_VALUE_BEGIN 0
+#define JP_STATE_VALUE_END 10
+#define JP_STATE_VALUE_BEGIN_OBJ 1
+#define JP_STATE_OBJ_SEP_OR_CLOSE 11
+#define JP_STATE_VALUE_BEGIN_AR 2
+#define JP_STATE_ARRAY_SEP_OR_CLOSE 12
+#define JP_STATE_OBJ_NAME_OR_CLOSE 5
+#define JP_STATE_OBJ_NAME 6
+#define JP_STATE_OBJ_COLON 7
+
+void cxJsonInit(CxJson *json, const CxAllocator *allocator) {
+ if (allocator == NULL) {
+ allocator = cxDefaultAllocator;
+ }
+
+ memset(json, 0, sizeof(CxJson));
+ json->allocator = allocator;
+
+ json->states = json->states_internal;
+ json->states_capacity = cx_nmemb(json->states_internal);
+ json->states[0] = JP_STATE_VALUE_BEGIN;
+ json->states_size = 1;
+
+ json->vbuf = json->vbuf_internal;
+ json->vbuf_capacity = cx_nmemb(json->vbuf_internal);
+}
+
+void cxJsonDestroy(CxJson *json) {
+ cxBufferDestroy(&json->buffer);
+ if (json->states != json->states_internal) {
+ free(json->states);
+ }
+ if (json->vbuf != json->vbuf_internal) {
+ free(json->vbuf);
+ }
+ 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};
+ }
+}
+
+int cxJsonFilln(CxJson *json, const char *buf, size_t size) {
+ if (cxBufferEof(&json->buffer)) {
+ // reinitialize the buffer
+ cxBufferDestroy(&json->buffer);
+ cxBufferInit(&json->buffer, (char*) buf, size,
+ NULL, CX_BUFFER_AUTO_EXTEND | CX_BUFFER_COPY_ON_WRITE);
+ json->buffer.size = size;
+ return 0;
+ } else {
+ return size != cxBufferAppend(buf, 1, size, &json->buffer);
+ }
+}
+
+static void json_add_state(CxJson *json, int state) {
+ // we have guaranteed the necessary space with cx_array_simple_reserve()
+ // therefore, we can safely add the state in the simplest way possible
+ json->states[json->states_size++] = state;
+}
+
+#define return_rec(code) \
+ token_destroy(&token); \
+ return code
+
+static enum cx_json_status json_parse(CxJson *json) {
+ // Reserve a pointer for a possibly read value
+ CxJsonValue *vbuf = NULL;
+
+ // grab the next token
+ CxJsonToken token;
+ {
+ enum cx_json_status ret = token_parse_next(json, &token);
+ if (ret != CX_JSON_NO_ERROR) {
+ return ret;
+ }
+ }
+
+ // pop the current state
+ assert(json->states_size > 0);
+ int state = json->states[--json->states_size];
+
+ // guarantee that at least two more states fit on the stack
+ CxArrayReallocator state_realloc = cx_array_reallocator(NULL, json->states_internal);
+ if (cx_array_simple_reserve_a(&state_realloc, json->states, 2)) {
+ return CX_JSON_BUFFER_ALLOC_FAILED; // LCOV_EXCL_LINE
+ }
+
+
+ // 0 JP_STATE_VALUE_BEGIN value begin
+ // 10 JP_STATE_VALUE_END expect value end
+ // 1 JP_STATE_VALUE_BEGIN_OBJ value begin (inside object)
+ // 11 JP_STATE_OBJ_SEP_OR_CLOSE object, expect separator, objclose
+ // 2 JP_STATE_VALUE_BEGIN_AR value begin (inside array)
+ // 12 JP_STATE_ARRAY_SEP_OR_CLOSE array, expect separator or arrayclose
+ // 5 JP_STATE_OBJ_NAME_OR_CLOSE object, expect name or objclose
+ // 6 JP_STATE_OBJ_NAME object, expect name
+ // 7 JP_STATE_OBJ_COLON object, expect ':'
+
+ if (state < 3) {
+ // push expected end state to the stack
+ json_add_state(json, 10 + state);
+ switch (token.tokentype) {
+ case CX_JSON_TOKEN_BEGIN_ARRAY: {
+ if (create_json_value(json, CX_JSON_ARRAY) == NULL) {
+ return_rec(CX_JSON_VALUE_ALLOC_FAILED); // LCOV_EXCL_LINE
+ }
+ json_add_state(json, JP_STATE_VALUE_BEGIN_AR);
+ return_rec(CX_JSON_NO_ERROR);
+ }
+ case CX_JSON_TOKEN_BEGIN_OBJECT: {
+ if (create_json_value(json, CX_JSON_OBJECT) == NULL) {
+ return_rec(CX_JSON_VALUE_ALLOC_FAILED); // LCOV_EXCL_LINE
+ }
+ json_add_state(json, JP_STATE_OBJ_NAME_OR_CLOSE);
+ return_rec(CX_JSON_NO_ERROR);
+ }
+ case CX_JSON_TOKEN_STRING: {
+ if ((vbuf = create_json_value(json, CX_JSON_STRING)) == NULL) {
+ return_rec(CX_JSON_VALUE_ALLOC_FAILED); // LCOV_EXCL_LINE
+ }
+ cxmutstr str = unescape_string(json->allocator, token.content);
+ if (str.ptr == NULL) {
+ return_rec(CX_JSON_VALUE_ALLOC_FAILED); // LCOV_EXCL_LINE
+ }
+ vbuf->value.string = str;
+ return_rec(CX_JSON_NO_ERROR);
+ }
+ case CX_JSON_TOKEN_INTEGER:
+ case CX_JSON_TOKEN_NUMBER: {
+ int type = token.tokentype == CX_JSON_TOKEN_INTEGER ? CX_JSON_INTEGER : CX_JSON_NUMBER;
+ if (NULL == (vbuf = create_json_value(json, type))) {
+ return_rec(CX_JSON_VALUE_ALLOC_FAILED); // LCOV_EXCL_LINE
+ }
+ if (type == CX_JSON_INTEGER) {
+ if (cx_strtoi64(token.content, &vbuf->value.integer, 10)) {
+ return_rec(CX_JSON_FORMAT_ERROR_NUMBER);
+ }
+ } else {
+ if (cx_strtod(token.content, &vbuf->value.number)) {
+ return_rec(CX_JSON_FORMAT_ERROR_NUMBER);
+ }
+ }
+ return_rec(CX_JSON_NO_ERROR);
+ }
+ case CX_JSON_TOKEN_LITERAL: {
+ if ((vbuf = create_json_value(json, CX_JSON_LITERAL)) == NULL) {
+ 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;
+ } else if (0 == cx_strcmp(cx_strcast(token.content), cx_str("false"))) {
+ vbuf->value.literal = CX_JSON_FALSE;
+ } else {
+ vbuf->value.literal = CX_JSON_NULL;
+ }
+ return_rec(CX_JSON_NO_ERROR);
+ }
+ default: {
+ return_rec(CX_JSON_FORMAT_ERROR_UNEXPECTED_TOKEN);
+ }
+ }
+ } else if (state == JP_STATE_ARRAY_SEP_OR_CLOSE) {
+ // expect ',' or ']'
+ if (token.tokentype == CX_JSON_TOKEN_VALUE_SEPARATOR) {
+ json_add_state(json, JP_STATE_VALUE_BEGIN_AR);
+ return_rec(CX_JSON_NO_ERROR);
+ } else if (token.tokentype == CX_JSON_TOKEN_END_ARRAY) {
+ // discard the array from the value buffer
+ json->vbuf_size--;
+ return_rec(CX_JSON_NO_ERROR);
+ } else {
+ return_rec(CX_JSON_FORMAT_ERROR_UNEXPECTED_TOKEN);
+ }
+ } else if (state == JP_STATE_OBJ_NAME_OR_CLOSE || state == JP_STATE_OBJ_NAME) {
+ if (state == JP_STATE_OBJ_NAME_OR_CLOSE && token.tokentype == CX_JSON_TOKEN_END_OBJECT) {
+ // discard the obj from the value buffer
+ json->vbuf_size--;
+ return_rec(CX_JSON_NO_ERROR);
+ } else {
+ // expect string
+ if (token.tokentype != CX_JSON_TOKEN_STRING) {
+ return_rec(CX_JSON_FORMAT_ERROR_UNEXPECTED_TOKEN);
+ }
+
+ // add new entry
+ cxmutstr name = unescape_string(json->allocator, token.content);
+ 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->vbuf_size > 0);
+
+ // next state
+ json_add_state(json, JP_STATE_OBJ_COLON);
+ return_rec(CX_JSON_NO_ERROR);
+ }
+ } else if (state == JP_STATE_OBJ_COLON) {
+ // expect ':'
+ if (token.tokentype != CX_JSON_TOKEN_NAME_SEPARATOR) {
+ return_rec(CX_JSON_FORMAT_ERROR_UNEXPECTED_TOKEN);
+ }
+ // next state
+ json_add_state(json, JP_STATE_VALUE_BEGIN_OBJ);
+ return_rec(CX_JSON_NO_ERROR);
+ } else if (state == JP_STATE_OBJ_SEP_OR_CLOSE) {
+ // expect ',' or '}'
+ if (token.tokentype == CX_JSON_TOKEN_VALUE_SEPARATOR) {
+ json_add_state(json, JP_STATE_OBJ_NAME);
+ return_rec(CX_JSON_NO_ERROR);
+ } else if (token.tokentype == CX_JSON_TOKEN_END_OBJECT) {
+ // discard the obj from the value buffer
+ json->vbuf_size--;
+ return_rec(CX_JSON_NO_ERROR);
+ } else {
+ return_rec(CX_JSON_FORMAT_ERROR_UNEXPECTED_TOKEN);
+ }
+ } else {
+ // should be unreachable
+ assert(false);
+ return_rec(-1);
+ }
+}
+
+CxJsonStatus cxJsonNext(CxJson *json, CxJsonValue **value) {
+ // check if buffer has been filled
+ if (json->buffer.space == NULL) {
+ return CX_JSON_NULL_DATA;
+ }
+
+ // initialize output value
+ *value = &cx_json_value_nothing;
+
+ // parse data
+ CxJsonStatus result;
+ do {
+ result = json_parse(json);
+ if (result == CX_JSON_NO_ERROR && json->states_size == 1) {
+ // final state reached
+ assert(json->states[0] == JP_STATE_VALUE_END);
+ assert(json->vbuf_size == 0);
+
+ // write output value
+ *value = json->parsed;
+ json->parsed = NULL;
+
+ // re-initialize state machine
+ json->states[0] = JP_STATE_VALUE_BEGIN;
+
+ return CX_JSON_NO_ERROR;
+ }
+ } while (result == CX_JSON_NO_ERROR);
+
+ // the parser might think there is no data
+ // but when we did not reach the final state,
+ // we know that there must be more to come
+ if (result == CX_JSON_NO_DATA && json->states_size > 1) {
+ return CX_JSON_INCOMPLETE_DATA;
+ }
+
+ return result;
+}
+
+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);
+ break;
+ }
+ case CX_JSON_ARRAY: {
+ CxJsonArray array = value->value.array;
+ for (size_t i = 0; i < array.array_size; i++) {
+ cxJsonValueFree(array.array[i]);
+ }
+ cxFree(value->allocator, array.array);
+ break;
+ }
+ case CX_JSON_STRING: {
+ cxFree(value->allocator, value->value.string.ptr);
+ break;
+ }
+ default: {
+ break;
+ }
+ }
+ cxFree(value->allocator, value);
+}
+
+CxJsonValue* cxJsonCreateObj(const CxAllocator* allocator) {
+ CxJsonValue* v = cxMalloc(allocator, sizeof(CxJsonValue));
+ 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);
+ cxFree(allocator, v);
+ return NULL;
+ // LCOV_EXCL_STOP
+ }
+ return v;
+}
+
+CxJsonValue* cxJsonCreateArr(const CxAllocator* allocator) {
+ CxJsonValue* v = cxMalloc(allocator, sizeof(CxJsonValue));
+ 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; }
+ return v;
+}
+
+CxJsonValue* cxJsonCreateNumber(const CxAllocator* allocator, double num) {
+ CxJsonValue* v = cxMalloc(allocator, sizeof(CxJsonValue));
+ if (v == NULL) return NULL;
+ v->allocator = allocator;
+ v->type = CX_JSON_NUMBER;
+ v->value.number = num;
+ return v;
+}
+
+CxJsonValue* cxJsonCreateInteger(const CxAllocator* allocator, int64_t num) {
+ CxJsonValue* v = cxMalloc(allocator, sizeof(CxJsonValue));
+ if (v == NULL) return NULL;
+ v->allocator = allocator;
+ v->type = CX_JSON_INTEGER;
+ v->value.integer = num;
+ return v;
+}
+
+CxJsonValue* cxJsonCreateString(const CxAllocator* allocator, const char* str) {
+ return cxJsonCreateCxString(allocator, cx_str(str));
+}
+
+CxJsonValue* cxJsonCreateCxString(const CxAllocator* allocator, cxstring str) {
+ CxJsonValue* v = cxMalloc(allocator, sizeof(CxJsonValue));
+ if (v == NULL) return NULL;
+ v->allocator = allocator;
+ 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;
+ return v;
+}
+
+CxJsonValue* cxJsonCreateLiteral(const CxAllocator* allocator, CxJsonLiteral lit) {
+ CxJsonValue* v = cxMalloc(allocator, sizeof(CxJsonValue));
+ if (v == NULL) return NULL;
+ v->allocator = allocator;
+ v->type = CX_JSON_LITERAL;
+ v->value.literal = lit;
+ return v;
+}
+
+// LCOV_EXCL_START
+// never called as long as malloc() does not return NULL
+static void cx_json_arr_free_temp(CxJsonValue** values, size_t count) {
+ for (size_t i = 0; i < count; i++) {
+ if (values[i] == NULL) break;
+ cxJsonValueFree(values[i]);
+ }
+ free(values);
+}
+// LCOV_EXCL_STOP
+
+int cxJsonArrAddNumbers(CxJsonValue* arr, const double* num, size_t count) {
+ CxJsonValue** values = calloc(count, sizeof(CxJsonValue*));
+ if (values == NULL) return -1;
+ for (size_t i = 0; i < count; i++) {
+ values[i] = cxJsonCreateNumber(arr->allocator, num[i]);
+ if (values[i] == NULL) { cx_json_arr_free_temp(values, count); return -1; }
+ }
+ int ret = cxJsonArrAddValues(arr, values, count);
+ free(values);
+ return ret;
+}
+
+int cxJsonArrAddIntegers(CxJsonValue* arr, const int64_t* num, size_t count) {
+ CxJsonValue** values = calloc(count, sizeof(CxJsonValue*));
+ if (values == NULL) return -1;
+ for (size_t i = 0; i < count; i++) {
+ values[i] = cxJsonCreateInteger(arr->allocator, num[i]);
+ if (values[i] == NULL) { cx_json_arr_free_temp(values, count); return -1; }
+ }
+ int ret = cxJsonArrAddValues(arr, values, count);
+ free(values);
+ return ret;
+}
+
+int cxJsonArrAddStrings(CxJsonValue* arr, const char* const* str, size_t count) {
+ CxJsonValue** values = calloc(count, sizeof(CxJsonValue*));
+ if (values == NULL) return -1;
+ for (size_t i = 0; i < count; i++) {
+ values[i] = cxJsonCreateString(arr->allocator, str[i]);
+ if (values[i] == NULL) { cx_json_arr_free_temp(values, count); return -1; }
+ }
+ int ret = cxJsonArrAddValues(arr, values, count);
+ free(values);
+ return ret;
+}
+
+int cxJsonArrAddCxStrings(CxJsonValue* arr, const cxstring* str, size_t count) {
+ CxJsonValue** values = calloc(count, sizeof(CxJsonValue*));
+ if (values == NULL) return -1;
+ for (size_t i = 0; i < count; i++) {
+ values[i] = cxJsonCreateCxString(arr->allocator, str[i]);
+ if (values[i] == NULL) { cx_json_arr_free_temp(values, count); return -1; }
+ }
+ int ret = cxJsonArrAddValues(arr, values, count);
+ free(values);
+ return ret;
+}
+
+int cxJsonArrAddLiterals(CxJsonValue* arr, const CxJsonLiteral* lit, size_t count) {
+ CxJsonValue** values = calloc(count, sizeof(CxJsonValue*));
+ if (values == NULL) return -1;
+ for (size_t i = 0; i < count; i++) {
+ values[i] = cxJsonCreateLiteral(arr->allocator, lit[i]);
+ if (values[i] == NULL) { cx_json_arr_free_temp(values, count); return -1; }
+ }
+ int ret = cxJsonArrAddValues(arr, values, count);
+ free(values);
+ return ret;
+}
+
+int cxJsonArrAddValues(CxJsonValue* arr, CxJsonValue* const* val, size_t count) {
+ 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,
+ val, count
+ );
+}
+
+int cxJsonObjPut(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)) {
+ cx_strfree_a(obj->allocator, &k);
+ return 1;
+ } else {
+ return 0;
+ }
+}
+
+CxJsonValue* cxJsonObjPutObj(CxJsonValue* obj, cxstring name) {
+ CxJsonValue* v = cxJsonCreateObj(obj->allocator);
+ if (v == NULL) return NULL;
+ if (cxJsonObjPut(obj, name, v)) { cxJsonValueFree(v); return NULL; }
+ return v;
+}
+
+CxJsonValue* cxJsonObjPutArr(CxJsonValue* obj, cxstring name) {
+ CxJsonValue* v = cxJsonCreateArr(obj->allocator);
+ if (v == NULL) return NULL;
+ if (cxJsonObjPut(obj, name, v)) { cxJsonValueFree(v); return NULL; }
+ return v;
+}
+
+CxJsonValue* cxJsonObjPutNumber(CxJsonValue* obj, cxstring name, double num) {
+ CxJsonValue* v = cxJsonCreateNumber(obj->allocator, num);
+ if (v == NULL) return NULL;
+ if (cxJsonObjPut(obj, name, v)) { cxJsonValueFree(v); return NULL; }
+ return v;
+}
+
+CxJsonValue* cxJsonObjPutInteger(CxJsonValue* obj, cxstring name, int64_t num) {
+ CxJsonValue* v = cxJsonCreateInteger(obj->allocator, num);
+ if (v == NULL) return NULL;
+ if (cxJsonObjPut(obj, name, v)) { cxJsonValueFree(v); return NULL; }
+ return v;
+}
+
+CxJsonValue* cxJsonObjPutString(CxJsonValue* obj, cxstring name, const char* str) {
+ CxJsonValue* v = cxJsonCreateString(obj->allocator, str);
+ if (v == NULL) return NULL;
+ if (cxJsonObjPut(obj, name, v)) { cxJsonValueFree(v); return NULL; }
+ return v;
+}
+
+CxJsonValue* cxJsonObjPutCxString(CxJsonValue* obj, cxstring name, cxstring str) {
+ CxJsonValue* v = cxJsonCreateCxString(obj->allocator, str);
+ if (v == NULL) return NULL;
+ if (cxJsonObjPut(obj, name, v)) { cxJsonValueFree(v); return NULL; }
+ return v;
+}
+
+CxJsonValue* cxJsonObjPutLiteral(CxJsonValue* obj, cxstring name, CxJsonLiteral lit) {
+ CxJsonValue* v = cxJsonCreateLiteral(obj->allocator, lit);
+ if (v == NULL) return NULL;
+ if (cxJsonObjPut(obj, name, v)) { cxJsonValueFree(v); return NULL;}
+ return v;
+}
+
+CxJsonValue *cxJsonArrGet(const CxJsonValue *value, size_t index) {
+ if (index >= value->value.array.array_size) {
+ return &cx_json_value_nothing;
+ }
+ return value->value.array.array[index];
+}
+
+CxIterator cxJsonArrIter(const CxJsonValue *value) {
+ return cxIteratorPtr(
+ value->value.array.array,
+ value->value.array.array_size
+ );
+}
+
+CxIterator cxJsonObjIter(const CxJsonValue *value) {
+ return cxIterator(
+ value->value.object.values,
+ sizeof(CxJsonObjValue),
+ value->value.object.values_size
+ );
+}
+
+CxJsonValue *cx_json_obj_get_cxstr(const CxJsonValue *value, cxstring name) {
+ CxJsonObjValue *member = json_find_objvalue(value, name);
+ if (member == NULL) {
+ return &cx_json_value_nothing;
+ } else {
+ return member->value;
+ }
+}
+
+static const CxJsonWriter cx_json_writer_default = {
+ false,
+ true,
+ 255,
+ false,
+ 4
+};
+
+CxJsonWriter cxJsonWriterCompact(void) {
+ return cx_json_writer_default;
+}
+
+CxJsonWriter cxJsonWriterPretty(bool use_spaces) {
+ return (CxJsonWriter) {
+ true,
+ true,
+ 255,
+ use_spaces,
+ 4
+ };
+}
+
+static int cx_json_writer_indent(
+ void *target,
+ cx_write_func wfunc,
+ const CxJsonWriter *settings,
+ unsigned int depth
+) {
+ if (depth == 0) return 0;
+
+ // determine the width and characters to use
+ const char* indent; // for 32 prepared chars
+ size_t width = depth;
+ if (settings->indent_space) {
+ if (settings->indent == 0) return 0;
+ width *= settings->indent;
+ indent = " ";
+ } else {
+ indent = "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t";
+ }
+
+ // calculate the number of write calls and write
+ size_t full = width / 32;
+ size_t remaining = width % 32;
+ for (size_t i = 0; i < full; i++) {
+ if (32 != wfunc(indent, 1, 32, target)) return 1;
+ }
+ if (remaining != wfunc(indent, 1, remaining, target)) return 1;
+
+ return 0;
+}
+
+
+int cx_json_write_rec(
+ void *target,
+ const CxJsonValue *value,
+ cx_write_func wfunc,
+ const CxJsonWriter *settings,
+ unsigned int depth
+) {
+ // keep track of written items
+ // the idea is to reduce the number of jumps for error checking
+ size_t actual = 0, expected = 0;
+
+ // small buffer for number to string conversions
+ char numbuf[32];
+
+ // recursively write the values
+ switch (value->type) {
+ case CX_JSON_OBJECT: {
+ const char *begin_obj = "{\n";
+ if (settings->pretty) {
+ actual += wfunc(begin_obj, 1, 2, target);
+ expected += 2;
+ } else {
+ actual += wfunc(begin_obj, 1, 1, target);
+ 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];
+ if (settings->sort_members) {
+ depth++;depth--;
+ }
+
+ // possible indentation
+ if (settings->pretty) {
+ if (cx_json_writer_indent(target, wfunc, settings, depth)) {
+ return 1; // LCOV_EXCL_LINE
+ }
+ }
+
+ // the name
+ actual += wfunc("\"", 1, 1, target);
+ // TODO: escape the string
+ actual += wfunc(member->name.ptr, 1,
+ member->name.length, target);
+ 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;
+ } else {
+ actual += wfunc(obj_name_sep, 1, 1, target);
+ expected += 3 + member->name.length;
+ }
+
+ // 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) {
+ const char *obj_value_sep = ",\n";
+ if (settings->pretty) {
+ actual += wfunc(obj_value_sep, 1, 2, target);
+ expected += 2;
+ } else {
+ actual += wfunc(obj_value_sep, 1, 1, target);
+ expected++;
+ }
+ } else {
+ if (settings->pretty) {
+ actual += wfunc("\n", 1, 1, target);
+ expected ++;
+ }
+ }
+ }
+ depth--;
+ if (settings->pretty) {
+ if (cx_json_writer_indent(target, wfunc, settings, depth)) return 1;
+ }
+ actual += wfunc("}", 1, 1, target);
+ expected++;
+ break;
+ }
+ case CX_JSON_ARRAY: {
+ actual += wfunc("[", 1, 1, target);
+ expected++;
+ CxIterator iter = cxJsonArrIter(value);
+ cx_foreach(CxJsonValue*, element, iter) {
+ if (cx_json_write_rec(
+ target, element,
+ wfunc, settings, depth)
+ ) return 1;
+
+ if (iter.index < iter.elem_count - 1) {
+ const char *arr_value_sep = ", ";
+ if (settings->pretty) {
+ actual += wfunc(arr_value_sep, 1, 2, target);
+ expected += 2;
+ } else {
+ actual += wfunc(arr_value_sep, 1, 1, target);
+ expected++;
+ }
+ }
+ }
+ actual += wfunc("]", 1, 1, target);
+ expected++;
+ break;
+ }
+ case CX_JSON_STRING: {
+ actual += wfunc("\"", 1, 1, target);
+ // TODO: escape the string
+ actual += wfunc(value->value.string.ptr, 1,
+ value->value.string.length, target);
+ actual += wfunc("\"", 1, 1, target);
+ expected += 2 + value->value.string.length;
+ break;
+ }
+ case CX_JSON_NUMBER: {
+ // TODO: locale bullshit
+ // TODO: formatting settings
+ snprintf(numbuf, 32, "%g", value->value.number);
+ size_t len = strlen(numbuf);
+ actual += wfunc(numbuf, 1, len, target);
+ expected += len;
+ break;
+ }
+ case CX_JSON_INTEGER: {
+ snprintf(numbuf, 32, "%" PRIi64, value->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) {
+ actual += wfunc("true", 1, 4, target);
+ expected += 4;
+ } else if (value->value.literal == CX_JSON_FALSE) {
+ actual += wfunc("false", 1, 5, target);
+ expected += 5;
+ } else {
+ actual += wfunc("null", 1, 4, target);
+ expected += 4;
+ }
+ break;
+ }
+ case CX_JSON_NOTHING: {
+ // deliberately supported as an empty string!
+ // users might want to just write the result
+ // of a get operation without testing the value
+ // and therefore this should not blow up
+ break;
+ }
+ default: assert(false); // LCOV_EXCL_LINE
+ }
+
+ return expected != actual;
+}
+
+int cxJsonWrite(
+ void *target,
+ const CxJsonValue *value,
+ cx_write_func wfunc,
+ const CxJsonWriter *settings
+) {
+ if (settings == NULL) {
+ settings = &cx_json_writer_default;
+ }
+ assert(target != NULL);
+ assert(value != NULL);
+ assert(wfunc != NULL);
+
+ return cx_json_write_rec(target, value, wfunc, settings, 0);
+}