/// <summary>
/// Handles a url starting with api/kPrefix by stripping off that prefix, searching for the file
/// named in the remainder of the url, and opening it in some browser (passing on any anchor specified).
/// </summary>
public static void HandleRequest(ApiRequest request)
{
//NB: be careful not to lose case, as at least chrome is case-sensitive with anchors (e.g. #ChoiceOfTopic)
var localPath = Regex.Replace(request.LocalPath(), "api/"+ kPrefix+"/", "", RegexOptions.IgnoreCase);
var langCode = LocalizationManager.UILanguageId;
var completeEnglishPath = FileLocator.GetFileDistributedWithApplication(localPath);
var completeUiLangPath = completeEnglishPath.Replace("-en.htm", "-" + langCode + ".htm");
string url;
if (langCode != "en" && RobustFile.Exists(completeUiLangPath))
url = completeUiLangPath;
else
url = completeEnglishPath;
var cleanUrl = url.Replace("\\", "/"); // allows jump to file to work
//we would like to get something like foo.htm#secondPart but the browser strips off that fragment part
//so we require that to be written as foo.htm?fragment=secondPart so it gets to us, then we convert
//it back to the normal format before sending it to a parser
if (!string.IsNullOrEmpty(request.Parameters["fragment"]))
{
cleanUrl += "#" + request.Parameters["fragment"];
request.Parameters.Remove("fragment");
}
string browser = string.Empty;
if (SIL.PlatformUtilities.Platform.IsLinux)
{
// REVIEW: This opens HTML files in the browser. Do we have any non-html
// files that this code needs to open in the browser? Currently they get
// opened in whatever application the user has selected for that file type
// which might well be an editor.
browser = "xdg-open";
}
else
{
// If we don't provide the path of the browser, i.e. Process.Start(url + queryPart), we get file not found exception.
// If we prepend "file:///", the anchor part of the link (#xxx) is not sent unless we provide the browser path too.
// This is the same behavior when simply typing a url into the Run command on Windows.
// If we fail to get the browser path for some reason, we still load the page, just without navigating to the anchor.
string defaultBrowserPath;
if (TryGetDefaultBrowserPathWindowsOnly(out defaultBrowserPath))
{
browser = defaultBrowserPath;
}
}
//Note, we don't currently use this, since this is only used for our own html. I added it for completeness... maybe
//someday when we are running the BloomLibrary locally for the user, we'll have links that require a query part.
var queryPart = "";
if (request.Parameters.Count > 0)
{
//reconstruct the query part, this time minus any fragment parameter (which we removed previously, if it was there)
queryPart = "?" + request.Parameters.AllKeys.Aggregate("", (total, key) => total + key + "=" + request.Parameters.Get(key) + "&");
queryPart = queryPart.TrimEnd(new[] { '&' });
}
if (!string.IsNullOrEmpty(browser))
{
try
{
Process.Start(browser, "\"file:///" + cleanUrl + queryPart + "\"");
request.SucceededDoNotNavigate();
return;
}
catch (Exception)
{
Debug.Fail("Jumping to browser with anchor failed.");
// Don't crash Bloom because we can't open an external file.
}
}
// If the above failed, either for lack of default browser or exception, try this:
Process.Start("\"" + cleanUrl + "\"");
request.SucceededDoNotNavigate();
}
