WIP: Autodetect pasted url and fill in form fields

binderhub/config.py

@@ -10,6 +10,12 @@ def generate_config(self):
             "repo_providers"
         ].items():
             config[repo_provider_class_alias] = repo_provider_class.labels
+            config[repo_provider_class_alias][
+                "display_name"
+            ] = repo_provider_class.display_name
+            config[repo_provider_class_alias][
+                "regex_detect"
+            ] = repo_provider_class.regex_detect
         return config
 
     async def get(self):

binderhub/repoproviders.py

@@ -99,6 +99,10 @@ class RepoProvider(LoggingConfigurable):
         config=True,
     )
 
+    # Not a traitlet because the class property is serialised in
+    # config.ConfigHandler.generate_config()
+    regex_detect = None
+
     unresolved_ref = Unicode()
 
     git_credentials = Unicode(
@@ -192,6 +196,15 @@ def is_valid_sha1(sha1):
 class FakeProvider(RepoProvider):
     """Fake provider for local testing of the UI"""
 
+    name = Unicode("Fake")
+
+    display_name = "Fake GitHub"
+
+    regex_detect = [
+        r"^https://github\.com/(?<repo>[^/]+/[^/]+)(/blob/(?<ref>[^/]+)(/(?<filepath>.+))?)?$",
+        r"^https://github\.com/(?<repo>[^/]+/[^/]+)(/tree/(?<ref>[^/]+)(/(?<urlpath>.+))?)?$",
+    ]
+
     labels = {
         "text": "Fake Provider",
         "tag_text": "Fake Ref",
@@ -627,6 +640,13 @@ def _default_git_credentials(self):
             return rf"username=binderhub\npassword={self.private_token}"
         return ""
 
+    # Gitlab repos can be nested under projects
+    _regex_detect_base = r"^https://gitlab\.com/(?<repo>[^/]+/[^/]+(/[^/-][^/]+)*)"
+    regex_detect = [
+        _regex_detect_base + r"(/-/blob/(?<ref>[^/]+)(/(?<filepath>.+))?)?$",
+        _regex_detect_base + r"(/-/tree/(?<ref>[^/]+)(/(?<urlpath>.+))?)?$",
+    ]
+
     labels = {
         "text": "GitLab.com repository or URL",
         "tag_text": "Git ref (branch, tag, or commit)",
@@ -780,6 +800,11 @@ def _default_git_credentials(self):
                 return rf"username={self.access_token}\npassword=x-oauth-basic"
         return ""
 
+    regex_detect = [
+        r"^https://github\.com/(?<repo>[^/]+/[^/]+)(/blob/(?<ref>[^/]+)(/(?<filepath>.+))?)?$",
+        r"^https://github\.com/(?<repo>[^/]+/[^/]+)(/tree/(?<ref>[^/]+)(/(?<urlpath>.+))?)?$",
+    ]
+
     labels = {
         "text": "GitHub repository name or URL",
         "tag_text": "Git ref (branch, tag, or commit)",
@@ -973,6 +998,10 @@ class GistRepoProvider(GitHubRepoProvider):
         help="Flag for allowing usages of secret Gists.  The default behavior is to disallow secret gists.",
     )
 
+    regex_detect = [
+        r"^https://gist\.github\.com/(?<repo>[^/]+/[^/]+)(/(?<ref>[^/]+))?$"
+    ]
+
     labels = {
         "text": "Gist ID (username/gistId) or URL",
         "tag_text": "Git commit SHA",

binderhub/static/js/index.js

@@ -19,7 +19,7 @@ import "../index.css";
 import { setUpLog } from "./src/log";
 import { updateUrls } from "./src/urls";
 import { getBuildFormValues } from "./src/form";
-import { updateRepoText } from "./src/repo";
+import { detectPastedRepo, updateRepoText } from "./src/repo";
 
 /**
  * @type {URL}
@@ -166,7 +166,19 @@ function indexMain() {
   updatePathText();
   updateRepoText(BASE_URL);
 
-  $("#repository").on("keyup paste change", function () {
+  // If the user pastes a URL into the repository field try to autodetect
+  // In all other cases don't do anything to avoid overwriting the user's input
+  // We need to wait for the paste to complete before we can read the input field
+  // https://stackoverflow.com/questions/10972954/javascript-onpaste/10972973#10972973
+  $("#repository").on("paste", () => {
+    setTimeout(() => {
+      detectPastedRepo(BASE_URL).then(() => {
+        updateUrls(BADGE_BASE_URL);
+      });
+    }, 0);
+  });
+
+  $("#repository").on("keyup change", function () {
     updateUrls(BADGE_BASE_URL);
   });
 

binderhub/static/js/src/repo.js

@@ -1,9 +1,10 @@
+import { detect, getRepoProviders } from "@jupyterhub/binderhub-client";
+import { updatePathText } from "./path";
+
 /**
- * Dict holding cached values of API request to _config endpoint
+ * @param {Object} configDict Dict holding cached values of API request to _config endpoint
  */
-let configDict = {};
-
-function setLabels() {
+function setLabels(configDict) {
   const provider = $("#provider_prefix").val();
   const text = configDict[provider]["text"];
   const tagText = configDict[provider]["tag_text"];
@@ -23,15 +24,42 @@ function setLabels() {
  * @param {URL} baseUrl Base URL to use for constructing path to _config endpoint
  */
 export function updateRepoText(baseUrl) {
-  if (Object.keys(configDict).length === 0) {
-    const configUrl = new URL("_config", baseUrl);
-    fetch(configUrl).then((resp) => {
-      resp.json().then((data) => {
-        configDict = data;
-        setLabels();
-      });
-    });
-  } else {
-    setLabels();
+  getRepoProviders(baseUrl).then(setLabels);
+}
+
+/**
+ * Attempt to fill in all fields by parsing a pasted repository URL
+ *
+ * @param {URL} baseUrl Base URL to use for constructing path to _config endpoint
+ */
+export async function detectPastedRepo(baseUrl) {
+  const repoField = $("#repository").val().trim();
+  const fields = await detect(baseUrl, repoField);
+  // Special case: The BinderHub UI supports https://git{hub,lab}.com/ in the
+  // repository (it's stripped out later in the UI).
+  // To keep the UI consistent insert it back if it was originally included.
+  console.log(fields);
+  if (fields) {
+    let repo = fields.repository;
+    if (repoField.startsWith("https://github.com/")) {
+      repo = "https://github.com/" + repo;
+    }
+    if (repoField.startsWith("https://gitlab.com/")) {
+      repo = "https://gitlab.com/" + repo;
+    }
+    $("#provider_prefix-selected").text(fields.providerName);
+    $("#provider_prefix").val(fields.providerPrefix);
+    $("#repository").val(repo);
+    if (fields.ref) {
+      $("#ref").val(fields.ref);
+    }
+    if (fields.path) {
+      $("#filepath").val(fields.path);
+      $("#url-or-file-selected").text(
+        fields.pathType === "filepath" ? "File" : "URL",
+      );
+    }
+    updatePathText();
+    updateRepoText(baseUrl);
   }
 }

binderhub/tests/test_repoproviders.py

@@ -539,3 +539,114 @@ def test_gist_secret():
 
     provider = GistRepoProvider(spec=spec, allow_secret_gist=True)
     assert IOLoop().run_sync(provider.get_resolved_ref) is not None
+
+
+@pytest.mark.parametrize(
+    "provider,url,groupdict",
+    [
+        (
+            GitHubRepoProvider,
+            "https://github.com/binder-examples/conda",
+            {"repo": "binder-examples/conda", "filepath": None, "ref": None},
+        ),
+        (
+            GitHubRepoProvider,
+            "https://github.com/binder-examples/conda/blob/main/index.ipynb",
+            {"repo": "binder-examples/conda", "ref": "main", "filepath": "index.ipynb"},
+        ),
+        (
+            GitHubRepoProvider,
+            "https://github.com/binder-examples/conda/tree/main/.github/workflows",
+            {
+                "repo": "binder-examples/conda",
+                "ref": "main",
+                "urlpath": ".github/workflows",
+            },
+        ),
+        (GitHubRepoProvider, "https://github.com/binder-examples/conda/pulls", None),
+        (
+            GitLabRepoProvider,
+            "https://gitlab.com/owner/repo",
+            {
+                "repo": "owner/repo",
+                "ref": None,
+                "filepath": None,
+            },
+        ),
+        (
+            GitLabRepoProvider,
+            "https://gitlab.com/owner/repo/-/tree/branch/folder?ref_type=heads",
+            {"repo": "owner/repo", "ref": "branch", "urlpath": "folder?ref_type=heads"},
+        ),
+        (
+            GitLabRepoProvider,
+            "https://gitlab.com/owner/repo/-/blob/branch/README.md?ref_type=heads",
+            {
+                "repo": "owner/repo",
+                "ref": "branch",
+                "filepath": "README.md?ref_type=heads",
+            },
+        ),
+        (
+            GitLabRepoProvider,
+            "https://gitlab.com/owner/project/repo",
+            {
+                "repo": "owner/project/repo",
+                "ref": None,
+                "filepath": None,
+            },
+        ),
+        (
+            GitLabRepoProvider,
+            "https://gitlab.com/owner/project/repo/-/tree/branch/folder?ref_type=heads",
+            {
+                "repo": "owner/project/repo",
+                "ref": "branch",
+                "urlpath": "folder?ref_type=heads",
+            },
+        ),
+        (
+            GitLabRepoProvider,
+            "https://gitlab.com/owner/project/repo/-/blob/branch/README.md?ref_type=heads",
+            {
+                "repo": "owner/project/repo",
+                "ref": "branch",
+                "filepath": "README.md?ref_type=heads",
+            },
+        ),
+        (
+            GitLabRepoProvider,
+            "https://gitlab.com/owner/repo/-/merge_requests/123",
+            None,
+        ),
+        (
+            GistRepoProvider,
+            "https://gist.github.com/owner/0123456789abcde0123456789abcde01",
+            {
+                "repo": "owner/0123456789abcde0123456789abcde01",
+                "ref": None,
+            },
+        ),
+        (
+            GistRepoProvider,
+            "https://gist.github.com/owner/0123456789abcde0123456789abcde01/sha",
+            {
+                "repo": "owner/0123456789abcde0123456789abcde01",
+                "ref": "sha",
+            },
+        ),
+        (GistRepoProvider, "https://gist.github.com/owner", None),
+    ],
+)
+def test_provider_regex_detect(provider, url, groupdict):
+    regex_js = provider.regex_detect
+    regex_py = [r.replace("(?<", "(?P<") for r in regex_js]
+    m = None
+    for r in regex_py:
+        m = re.match(r, url)
+        if m:
+            break
+    if groupdict:
+        assert m.groupdict() == groupdict
+    else:
+        assert not m

js/packages/binderhub-client/lib/autodetect.js

@@ -0,0 +1,60 @@
+/**
+ * Dict holding cached values of API request to _config endpoint for base URL
+ */
+let repoProviders = {};
+
+/**
+ * Get the repo provider configurations supported by the BinderHub instance
+ *
+ * @param {URL} baseUrl Base URL to use for constructing path to _config endpoint
+ */
+export async function getRepoProviders(baseUrl) {
+  if (!repoProviders[baseUrl]) {
+    const configUrl = new URL("_config", baseUrl);
+    const resp = await fetch(configUrl);
+    repoProviders[baseUrl] = resp.json();
+  }
+  return repoProviders[baseUrl];
+}
+
+/**
+ * Attempt to parse a string (typically a repository URL) into a BinderHub
+ * provider/repository/reference/path
+ *
+ * @param {URL} baseUrl Base URL to use for constructing path to _config endpoint
+ * @param {string} text Repository URL or similar to parse
+ * @returns {Object} An object if the repository could be parsed with fields
+ *   - providerPrefix Prefix denoting what provider was selected
+ *   - repository Repository to build
+ *   - ref Ref in this repo to build (optional)
+ *   - path Path to launch after this repo has been built (optional)
+ *   - pathType Type of thing to open path with (raw url, notebook file) (optional)
+ *   - providerName User friendly display name of the provider (optional)
+ *   null otherwise
+ */
+export async function detect(baseUrl, text) {
+  const config = await getRepoProviders(baseUrl);
+
+  for (const provider in config) {
+    const regex_detect = config[provider].regex_detect || [];
+    for (const regex of regex_detect) {
+      const m = text.match(regex);
+      if (m?.groups.repo) {
+        return {
+          providerPrefix: provider,
+          repository: m.groups.repo,
+          ref: m.groups.ref || null,
+          path: m.groups.filepath || m.groups.urlpath || null,
+          pathType: m.groups.filepath
+            ? "filepath"
+            : m.groups.urlpath
+            ? "urlpath"
+            : null,
+          providerName: config[provider].display_name,
+        };
+      }
+    }
+  }
+
+  return null;
+}

js/packages/binderhub-client/lib/index.js

@@ -1,6 +1,8 @@
 import { NativeEventSource, EventSourcePolyfill } from "event-source-polyfill";
 import { EventIterator } from "event-iterator";
 
+import { detect, getRepoProviders } from "./autodetect";
+
 // Use native browser EventSource if available, and use the polyfill if not available
 const EventSource = NativeEventSource || EventSourcePolyfill;
 
@@ -211,3 +213,5 @@ export function makeBadgeMarkup(publicBaseUrl, url, syntax) {
     );
   }
 }
+
+export { detect, getRepoProviders };