jsbind.c

网络爬虫程序

  url_priv_t *up;
  dllist *ptr;
  JSObject *robj;

  if(argc != 1 || !JSVAL_IS_INT(argv[0]))
  {
    JS_ReportError(cx, "PavukUrl.get_parent: %s\n",
      gettext("bad parent index"));
    return JS_FALSE;
  }

  n = JSVAL_TO_INT(argv[0]);
  up = JS_GetPrivate(cx, obj);

  ptr = dllist_nth(up->urlp->parent_url, n);

  if(ptr)
  {
    rurl = (url *) ptr->data;

    robj = pjs_url_to_jsobj(cx, rurl, NULL);

    *rv = OBJECT_TO_JSVAL(robj);
  }
  else
    *rv = JSVAL_NULL;

  return JS_TRUE;
}

static JSBool pjs_url_check_cond(JSContext *cx, JSObject *obj, uintN argc, jsval *argv, jsval *rv)
{
  url_priv_t *up;
  char *name;
  JSString *s;
  int r, i;
  cond_info_t condp;

  condp.level = 0;
  condp.urlnr = 0;
  condp.size = 0;
  condp.time = 0L;
  condp.mimet = NULL;
  condp.params = NULL;

  up = JS_GetPrivate(cx, obj);

  if(up->condp)
  {
    condp.level = up->condp->level;
    condp.urlnr = up->condp->urlnr;
    condp.size = up->condp->size;
    condp.time = up->condp->time;
    condp.mimet = up->condp->mimet;
    condp.params = up->condp->params;
  }

  if(argc < 1)
  {
    JS_ReportError(cx, "PavukUrl.check_cond: %s\n",
      gettext("not enough parameters"));
    return JS_FALSE;
  }

  s = JS_ValueToString(cx, argv[0]);
  name = JS_GetStringBytes(s);

  for(i = 1; i < argc; i++)
  {
    char *p;

    s = JS_ValueToString(cx, argv[i]);
    p = JS_GetStringBytes(s);

    if(p)
      condp.params = dllist_append(condp.params, p);
  }

  r = url_append_one_condition(name, up->urlp, &condp);

  dllist_free_all(condp.params);

  if(r < 0)
  {
    JS_ReportError(cx, "pjs: PavukUrl.check_cond: %s - \"%s\"",
      gettext("unknown limiting condition"), name);

    return JS_FALSE;
  }

  *rv = r ? JSVAL_TRUE : JSVAL_FALSE;

  return JS_TRUE;
}

static JSObject *pjs_url_class_init(JSContext *cx, JSObject *obj)
{
  JSObject *prot;

  prot = JS_InitClass(cx, obj, NULL, &urlClass, pjs_url_new, 0,
    urlProperties, urlMethods, NULL, NULL);

  return prot;
}

/* PavukFnrules class implementation */
static JSBool pjs_fnrules_get_macro(JSContext *, JSObject *, uintN, jsval *,
  jsval *);
static JSBool pjs_fnrules_get_sub(JSContext *, JSObject *, uintN, jsval *,
  jsval *);
static JSBool pjs_fnrules_get_property(JSContext *, JSObject *, jsval,
  jsval *);

static JSClass fnrulesClass = {
  "PavukFnrules", JSCLASS_HAS_PRIVATE,
  JS_PropertyStub, JS_PropertyStub,
  pjs_fnrules_get_property, JS_PropertyStub,
  JS_EnumerateStub, JS_ResolveStub, JS_ConvertStub, JS_FinalizeStub
};

static JSFunctionSpec fnrulesMethods[] = {
  {"get_macro", pjs_fnrules_get_macro, 1, JSPROP_ENUMERATE},
  {"get_sub", pjs_fnrules_get_sub, 1, JSPROP_ENUMERATE},
  {NULL, NULL, 0, 0}
};

enum _fnrules_prop
{
  PJSFNRULES_URL,
  PJSFNRULES_PATTERN,
  PJSFNRULES_PATTERN_TYPE,

  PJSFNRULES_NULL
};

static JSPropertySpec fnrulesProperties[] = {
  {"url", PJSFNRULES_URL, JSPROP_ENUMERATE | JSPROP_READONLY},
  {"pattern", PJSFNRULES_PATTERN, JSPROP_ENUMERATE | JSPROP_READONLY},
  {"pattern_type", PJSFNRULES_PATTERN_TYPE,
      JSPROP_ENUMERATE | JSPROP_READONLY},
  {NULL, 0, 0},
};

static JSObject *pjs_fnrules_to_jsobj(JSContext *cx, struct lfname_lsp_interp *interp)
{
  JSObject *obj;

  obj = JS_NewObject(cx, &fnrulesClass, NULL, NULL);
  JS_SetPrivate(cx, obj, interp);

  return obj;
}

/* constructor */
static JSBool pjs_fnrules_new(JSContext *cx, JSObject *obj, uintN argc, jsval *argv, jsval *rv)
{
  JS_ReportError(cx,
    gettext("PavukFnrules constructor call prohibited from script\n"));
  return JS_TRUE;
}

static JSBool pjs_fnrules_get_property(JSContext *cx, JSObject *obj, jsval id, jsval *val)
{
  if(JSVAL_IS_INT(id))
  {
    struct lfname_lsp_interp *interp;
    JSObject *uobj;

    interp = JS_GetPrivate(cx, obj);

    switch (JSVAL_TO_INT(id))
    {
    case PJSFNRULES_URL:
      uobj = pjs_url_to_jsobj(cx, interp->urlp, NULL);
      *val = OBJECT_TO_JSVAL(uobj);
      break;
    case PJSFNRULES_PATTERN:
      SET_STR(interp->orig->matchstr);
      break;
    case PJSFNRULES_PATTERN_TYPE:
      *val = INT_TO_JSVAL(interp->orig->type);
      break;
    }
  }
  return JS_TRUE;
}

static JSBool pjs_fnrules_get_macro(JSContext *cx, JSObject *obj, uintN argc, jsval *argv, jsval *val)
{
  struct lfname_lsp_interp *interp;
  const char *v;
  JSString *s;

  if(argc != 1 || !(s = JS_ValueToString(cx, argv[0])))
  {
    JS_ReportError(cx, "PavukFnrules.get_macro: %s\n",
      gettext("bad parameters"));
    return JS_FALSE;
  }

  interp = JS_GetPrivate(cx, obj);
  v = JS_GetStringBytes(s);

  if(v && v[0] == '%' && lfname_check_macro(v[1]) && !v[2])
    v = lfname_interp_get_macro(interp, v[1]);

  SET_STR(v);

  return JS_TRUE;
}

static JSBool pjs_fnrules_get_sub(JSContext *cx, JSObject *obj, uintN argc, jsval *argv, jsval *val)
{
  struct lfname_lsp_interp *interp;
  char *v;
  int32 n;

  if(argc != 1 || !JSVAL_IS_INT(argv[0]))
  {
    JS_ReportError(cx, "PavukFnrules.get_sub: %s\n",
      gettext("bad parameters"));
    return JS_FALSE;
  }

  interp = JS_GetPrivate(cx, obj);

  JS_ValueToInt32(cx, argv[0], &n);

  v = lfname_re_sub(interp->orig, interp->urlstr, n);

  SET_STR(v);

  return JS_TRUE;
}

static JSObject *pjs_fnrules_class_init(JSContext *cx, JSObject *obj)
{
  JSObject *prot;

  prot = JS_InitClass(cx, obj, NULL, &fnrulesClass, pjs_fnrules_new, 0,
    fnrulesProperties, fnrulesMethods, NULL, NULL);

  return prot;
}

static int pjs_func_exist(JSContext *cx, JSObject *obj, char *name)
{
  jsval rv;

  if(!JS_GetProperty(cx, obj, name, &rv) || rv == JSVAL_VOID)
    return FALSE;
  else
    return TRUE;
}

int pjs_init(void)
{
  LOCK_MOZJS;

  if(!rt)
  {
    rt = JS_Init(1000000L);
    cx = JS_NewContext(rt, STACK_CHUNK_SIZE);
    JS_SetErrorReporter(cx, pjs_error);
    go = JS_NewObject(cx, &globalClass, NULL, NULL);
    JS_InitStandardClasses(cx, go);
    JS_DefineFunctions(cx, go, globalMethods);

    pjs_url_class_init(cx, go);
    pjs_fnrules_class_init(cx, go);

    pjs_load_script(priv_cfg.js_script_file);

    pjs_have_pavuk_url_cond_check =
      pjs_func_exist(cx, go, "pavuk_url_cond_check");
  }

  UNLOCK_MOZJS;

  return 0;
}

int pjs_destroy(void)
{
  LOCK_MOZJS;

  if(rt)
  {
    if(script)
    {
      JS_RemoveRoot(cx, &scriptobj);
      JS_DestroyScript(cx, script);
    }
    JS_GC(cx);
    JS_DestroyContext(cx);
    JS_Finish(rt);

    cx = NULL;
    rt = NULL;
    go = NULL;
    script = NULL;
    scriptobj = NULL;
    pjs_have_pavuk_url_cond_check = FALSE;
  }

  UNLOCK_MOZJS;

  return 0;
}

int pjs_load_script(char *name)
{
  jsval rv;

  JS_ClearPendingException(cx);
  if(script)
  {
    JS_RemoveRoot(cx, &scriptobj);
    JS_DestroyScript(cx, script);
    script = NULL;
    scriptobj = NULL;
  }

  if(name)
  {
    script = JS_CompileFile(cx, go, name);

    if(!script)
      return -1;

    scriptobj = JS_NewScriptObject(cx, script);
    JS_AddNamedRoot(cx, &scriptobj, "pjs_script");

    JS_ExecuteScript(cx, go, script, &rv);
  }

  return 0;
}

int pjs_load_script_string(char *str)
{
  jsval rv;

  LOCK_MOZJS;
  JS_ClearPendingException(cx);
  if(script)
  {
    JS_RemoveRoot(cx, &scriptobj);
    JS_DestroyScript(cx, script);
    script = NULL;
    scriptobj = NULL;
  }

  script = JS_CompileScript(cx, go, str, strlen(str), "intbuffer", 0);
  if(!script)
  {
    UNLOCK_MOZJS;
    return -1;
  }

  scriptobj = JS_NewScriptObject(cx, script);
  JS_AddNamedRoot(cx, &scriptobj, "pjs_script");

  JS_ExecuteScript(cx, go, script, &rv);

  UNLOCK_MOZJS;
  return 0;
}

int pjs_execute(char *scr)
{
  jsval rval;
  int rv;

  LOCK_MOZJS;
  rv = JS_EvaluateScript(cx, go, scr, strlen(scr), "pjs", 0, &rval);
  UNLOCK_MOZJS;

  return rv;
}

int pjs_run_cond_check_func(url *urlp, cond_info_t *condp)
{
  jsval param[2];
  jsval rv;
  JSObject *ourl;
  int ret;

  if(!pjs_have_pavuk_url_cond_check || !script)
    return -1;

  LOCK_MOZJS;

  ourl = pjs_url_to_jsobj(cx, urlp, condp);
  param[0] = OBJECT_TO_JSVAL(ourl);
  param[1] = INT_TO_JSVAL(condp->level);

  rv = JSVAL_FALSE;

  if(!JS_CallFunctionName(cx, go, "pavuk_url_cond_check", 2, param, &rv))
    ret = -1;
  else
  {
    if(JSVAL_IS_BOOLEAN(rv))
      ret = (rv == JSVAL_TRUE);
    else
      ret = -1;
  }

  JS_GC(cx);

  UNLOCK_MOZJS;

  return ret;
}

char *pjs_run_fnrules_func(char *name, struct lfname_lsp_interp *interp)
{
  char *ret = NULL;
  JSObject *rl;
  jsval param[1];
  JSString *s;
  jsval rv;

  LOCK_MOZJS;
  if(pjs_func_exist(cx, go, name))
  {
    rl = pjs_fnrules_to_jsobj(cx, interp);
    param[0] = OBJECT_TO_JSVAL(rl);

    rv = JSVAL_FALSE;

    if(JS_CallFunctionName(cx, go, name, 1, param, &rv))
    {
      s = JS_ValueToString(cx, rv);
      if(s)
        ret = tl_strdup(JS_GetStringBytes(s));
    }

    JS_GC(cx);
  }
  else
  {
    xprintf(1, gettext("pjs: There is no \"%s\" JavaScript function!\n"),
      name);
  }
  UNLOCK_MOZJS;

  return ret;
}

#endif /* HAVE_MOZJS */