CST-agent/tests/test_llm_body.c

#include "cJSON.h"
#include "cst_agent.h"

#include <stdio.h>
#include <stdlib.h>
#include <string.h>

static int nfail;

#define EXPECT(cond, msg)                                                      \
  do {                                                                         \
    if (!(cond)) {                                                             \
      fprintf(stderr, "FAIL %s:%d: %s\n", __FILE__, __LINE__, msg);            \
      nfail++;                                                                 \
    }                                                                          \
  } while (0)

static cst_message_t msg(cst_role_t role, const char *content) {
  cst_message_t m;
  memset(&m, 0, sizeof(m));
  m.role = role;
  m.content = (char *)content;
  return m;
}

static cJSON *build(const cst_message_t *msgs, int n, int tools,
                    size_t tool_cap) {
  cst_llm_request_t r;
  memset(&r, 0, sizeof(r));
  r.model = "test-model";
  r.msgs = msgs;
  r.nmsgs = n;
  r.enable_tools = tools;
  r.tool_out_cap = tool_cap;
  char *body = cst_llm_build_body(&r);
  EXPECT(body != NULL, "body built");
  cJSON *j = body ? cJSON_Parse(body) : NULL;
  EXPECT(j != NULL, "body is valid JSON");
  free(body);
  return j;
}

static const char *content_at(const cJSON *root, int i) {
  const cJSON *msgs = cJSON_GetObjectItemCaseSensitive(root, "messages");
  const cJSON *m = cJSON_GetArrayItem(msgs, i);
  const cJSON *c = cJSON_GetObjectItemCaseSensitive(m, "content");
  return cJSON_IsString(c) ? c->valuestring : NULL;
}

static const char *role_at(const cJSON *root, int i) {
  const cJSON *msgs = cJSON_GetObjectItemCaseSensitive(root, "messages");
  const cJSON *m = cJSON_GetArrayItem(msgs, i);
  const cJSON *r = cJSON_GetObjectItemCaseSensitive(m, "role");
  return cJSON_IsString(r) ? r->valuestring : "";
}

static int nmsgs(const cJSON *root) {
  return cJSON_GetArraySize(cJSON_GetObjectItemCaseSensitive(root, "messages"));
}

/* TAB / quotes / backslash / control bytes / CJK must survive verbatim
 * (the old hand-rolled escaper turned a Makefile TAB into HTTP 400). */
static void test_escaping(void) {
  const char *text = "all:\n\tgcc -o \"a b\" x.c \\\n\x01\x1f end 中文";
  cst_message_t m[2] = {msg(CST_ROLE_SYSTEM, "sys"), msg(CST_ROLE_USER, text)};
  cJSON *j = build(m, 2, 0, 0);
  const char *c = content_at(j, 1);
  EXPECT(c && strcmp(c, text) == 0, "user content roundtrips exactly");
  EXPECT(!cJSON_GetObjectItemCaseSensitive(j, "tools"), "no tools when off");
  const cJSON *so = cJSON_GetObjectItemCaseSensitive(j, "stream_options");
  EXPECT(cJSON_IsTrue(cJSON_GetObjectItemCaseSensitive(so, "include_usage")),
         "stream_options.include_usage");
  cJSON_Delete(j);
}

static void test_invalid_utf8(void) {
  const char text[] = "ok \xff\xfe bad \xe4\xb8 trunc";
  cst_message_t m[1] = {msg(CST_ROLE_USER, text)};
  cJSON *j = build(m, 1, 0, 0);
  const char *c = content_at(j, 0);
  EXPECT(c != NULL, "content present");
  EXPECT(c && strlen(c) == strlen(text), "sanitize keeps byte length");
  EXPECT(c && strstr(c, "ok ") == c, "valid prefix kept");
  cJSON_Delete(j);
}

static void test_tool_calls_and_null_content(void) {
  cst_message_t m[3];
  m[0] = msg(CST_ROLE_USER, "go");
  m[1] = msg(CST_ROLE_ASSISTANT, "");
  m[1].tool_calls_json =
      "[{\"id\":\"c1\",\"type\":\"function\",\"function\":{\"name\":"
      "\"read_file\",\"arguments\":\"{\\\"path\\\":\\\"a\\tb.c\\\"}\"}}]";
  m[2] = msg(CST_ROLE_TOOL, "[tool read_file id=c1 exit=0]\nhello");
  snprintf(m[2].tool_call_id, sizeof(m[2].tool_call_id), "c1");
  cJSON *j = build(m, 3, 1, 0);
  const cJSON *msgs = cJSON_GetObjectItemCaseSensitive(j, "messages");
  const cJSON *a = cJSON_GetArrayItem(msgs, 1);
  EXPECT(cJSON_IsNull(cJSON_GetObjectItemCaseSensitive(a, "content")),
         "assistant content null with tool_calls");
  const cJSON *tc = cJSON_GetObjectItemCaseSensitive(a, "tool_calls");
  EXPECT(cJSON_IsArray(tc) && cJSON_GetArraySize(tc) == 1, "tool_calls kept");
  const cJSON *t = cJSON_GetArrayItem(msgs, 2);
  EXPECT(strcmp(role_at(j, 2), "tool") == 0, "tool role");
  const cJSON *tid = cJSON_GetObjectItemCaseSensitive(t, "tool_call_id");
  EXPECT(cJSON_IsString(tid) && strcmp(tid->valuestring, "c1") == 0,
         "tool_call_id");
  const cJSON *tools = cJSON_GetObjectItemCaseSensitive(j, "tools");
  EXPECT(cJSON_IsArray(tools) && cJSON_GetArraySize(tools) >= 4,
         "tools[] schema present");
  const cJSON *choice = cJSON_GetObjectItemCaseSensitive(j, "tool_choice");
  EXPECT(cJSON_IsString(choice) && strcmp(choice->valuestring, "auto") == 0,
         "tool_choice auto");
  cJSON_Delete(j);
}

static void test_merge_and_orphan_tool(void) {
  cst_message_t m[3];
  m[0] = msg(CST_ROLE_USER, "first");
  m[1] = msg(CST_ROLE_TOOL, "orphan output"); /* no id -> user text */
  m[2] = msg(CST_ROLE_USER, "third");
  cJSON *j = build(m, 3, 0, 0);
  EXPECT(nmsgs(j) == 1, "consecutive user messages merged");
  const char *c = content_at(j, 0);
  EXPECT(c && strstr(c, "first") && strstr(c, "orphan output") &&
             strstr(c, "third"),
         "merged content keeps all parts in order");
  cJSON_Delete(j);
}

static void test_tool_cap_head_tail(void) {
  size_t n = 100000;
  char *big = malloc(n + 1);
  memset(big, 'x', n);
  memcpy(big, "HEAD", 4);
  memcpy(big + n - 4, "TAIL", 4);
  big[n] = 0;
  cst_message_t m[3];
  m[0] = msg(CST_ROLE_USER, "go");
  m[1] = msg(CST_ROLE_ASSISTANT, "");
  m[1].tool_calls_json = "[{\"id\":\"c1\",\"type\":\"function\",\"function\":"
                         "{\"name\":\"run_shell\",\"arguments\":\"{}\"}}]";
  m[2] = msg(CST_ROLE_TOOL, big);
  snprintf(m[2].tool_call_id, sizeof(m[2].tool_call_id), "c1");
  cJSON *j = build(m, 3, 1, 1000);
  const char *c = content_at(j, 2);
  EXPECT(c && strlen(c) < 1200, "tool output capped");
  EXPECT(c && strncmp(c, "HEAD", 4) == 0, "head kept");
  EXPECT(c && strcmp(c + strlen(c) - 4, "TAIL") == 0, "tail kept");
  EXPECT(c && strstr(c, "omitted"), "omission note");
  cJSON_Delete(j);

  /* user text (e.g. text-protocol result frames) is never capped */
  cst_message_t u[1] = {msg(CST_ROLE_USER, big)};
  j = build(u, 1, 0, 1000);
  c = content_at(j, 0);
  EXPECT(c && strlen(c) == n, "user text not capped");
  cJSON_Delete(j);
  free(big);
}

int main(void) {
  test_escaping();
  test_invalid_utf8();
  test_tool_calls_and_null_content();
  test_merge_and_orphan_tool();
  test_tool_cap_head_tail();
  if (nfail) {
    fprintf(stderr, "test_llm_body: %d failed\n", nfail);
    return 1;
  }
  printf("test_llm_body: OK\n");
  return 0;
}