First Alpha Release for Version 4.5

With this change, Instaloader is capable of resuming a previously-aborted download loop. To do so, it creates a JSON file within the target directory when interrupted, that contains all the necessary information to later resume that operation.

Resuming an interrupted download is supported for most, but not all targets. It is supported for:

- Regular profile posts,
- IGTV posts
- Saved posts,
- Tagged posts,
- Explore posts.

.github/ISSUE_TEMPLATE/bug_report.md

@@ -24,4 +24,4 @@ If applicable, add error messages and tracebacks to help explain your problem.
 **Additional context**
 Add any other context about the problem here.
 
-(please also see https://instaloader.github.io/contributing.html for how to report a bug)
+<!-- please also see https://instaloader.github.io/contributing.html for how to report a bug -->

.github/ISSUE_TEMPLATE/config.yml

@@ -0,0 +1,11 @@
+blank_issues_enabled: false
+contact_links:
+  - name: Contributing to Instaloader
+    url: https://instaloader.github.io/contributing.html
+    about: Guidelines for how to report an issue.
+  - name: Instaloader Troubleshooting
+    url: https://instaloader.github.io/troubleshooting.html
+    about: Many common problems and their workarounds are described here.
+  - name: Instaloader Documentation
+    url: https://instaloader.github.io/
+    about: Many questions might be already answered here.

.github/ISSUE_TEMPLATE/feature_request.md

@@ -15,10 +15,10 @@ A clear and concise description of what you want to happen.
 A clear and concise description of any alternative solutions or features you've considered.
 
 **If the feature request is accepted, would you be willing to submit a PR?**
-
-Yes / No _(Help can be provided if you need assistance submitting a PR)_
+Yes / No
+<!-- Help can be provided if you need assistance submitting a PR -->
 
 **Additional context**
 Add any other context about the feature request here.
 
-(please also see https://instaloader.github.io/contributing.html for how to suggest a feature)
+<!-- please also see https://instaloader.github.io/contributing.html for how to suggest a feature -->

.github/ISSUE_TEMPLATE/question.md

@@ -6,3 +6,7 @@ labels: question
 ---
 
 Your question here...
+
+<!-- Please see our documentation: https://instaloader.github.io/ -->
+
+<!-- Basic questions regarding Instaloader's usage might be more appropriate to ask on Stack Overflow -->

.github/pull_request_template.md

@@ -0,0 +1,17 @@
+
+
+<!--
+Please describe:
+
+- A motivation for this change, e.g.
+  - Fixes # .
+  - More general: What problem does the pull request solve?
+
+- The changes proposed in this pull request
+
+- The completeness of this change
+  - Is it just a proof of concept?
+  - Is the documentation updated (if appropriate)?
+  - Do you consider it ready to be merged or is it a draft?
+  - Can we help you at some point?
+-->

.github/workflows/lint.yml

@@ -0,0 +1,30 @@
+name: PyLint and MyPy
+
+on: [push, pull_request]
+
+jobs:
+  lint:
+    name: PyLint and MyPy
+    runs-on: ubuntu-latest
+    strategy:
+      matrix:
+        python-version: [3.8]
+    steps:
+      - name: Checkout Instaloader Repository
+        uses: actions/checkout@v2
+        with:
+          fetch-depth: 0  # needed for building docs
+      - name: Setup Python
+        uses: actions/setup-python@v2
+        with:
+          python-version: ${{ matrix.python-version }}
+      - name: Install Dependencies
+        run: |
+          python -m pip install pipenv==2020.6.2
+          pipenv sync --dev
+      - name: PyLint
+        run: pipenv run pylint instaloader
+      - name: MyPy
+        run: pipenv run mypy -m instaloader
+      - name: Build Documentation
+        run: pipenv run make -C docs html SPHINXOPTS="-W -n"

.github/workflows/stale.yml

@@ -2,14 +2,25 @@ name: Mark stale issues and pull requests
 
 on:
   schedule:
-  - cron: "15 16 * * *"
+  - cron: "10 14 * * *"
 
 jobs:
   stale:
 
+    name: Mark stale issues and pull requests
     runs-on: ubuntu-latest
     
     steps:
+    - uses: actions/stale@v3
+      with:
+        repo-token: ${{ secrets.GITHUB_TOKEN }}
+        only-labels: 'question'
+        stale-issue-message: 'There has been no activity on this question for an extended period of time. This issue will be closed after further 14 days of inactivity.'
+        stale-issue-label: 'stale'
+        exempt-issue-labels: 'leave open'
+        days-before-stale: 15
+        days-before-close: -1
+        remove-stale-when-updated: false
     - uses: actions/stale@v1
       with:
         repo-token: ${{ secrets.GITHUB_TOKEN }}
@@ -21,4 +32,3 @@ jobs:
         exempt-pr-label: 'leave open'
         days-before-stale: 90
         days-before-close: 14
-        operations-per-run: 7

.github/workflows/windows_exe.yml

@@ -4,6 +4,7 @@ on: [push, pull_request]
 
 jobs:
   build:
+    name: Windows EXE build
     runs-on: windows-latest
     steps:
       - name: Checkout Instaloader repository

.travis.yml

@@ -27,7 +27,7 @@ deploy:
     repo: instaloader/instaloader.github.io
     target-branch: master
     on:
-      tags: true
+      branch: master
       python: 3.6
   - provider: script
     script: deploy/arch/deploy.sh $TRAVIS_TAG

README.rst

@@ -49,7 +49,9 @@
   directory accordingly,
 
 - allows **fine-grained customization** of filters and where to store
-  downloaded media.
+  downloaded media,
+
+- automatically **resumes previously-interrupted** download iterations.
 
 ::
 

deploy/windows/create_exe.py

@@ -8,33 +8,18 @@ import sys
 # copy the required files into repo root
 shutil.copy('docs/favicon.ico', '.')
 shutil.copy('deploy/windows/instaloader.spec', '.')
-shutil.unpack_archive('deploy/windows/ps', '.', 'xztar')
 
 code = """
+import contextlib
 import psutil
 import subprocess
 
 def __main():
-    grandparentpid = psutil.Process(os.getppid()).ppid()
-    grandparentpidsearchstring = ' ' + str(grandparentpid) + ' '
-    if hasattr(sys, "_MEIPASS"):
-        ps = os.path.join(sys._MEIPASS, 'tasklist.exe')
-    else:
-        ps = 'tasklist'
-    popen = subprocess.Popen(ps, stdout=subprocess.PIPE, universal_newlines=True)
-    for examine in iter(popen.stdout.readline, ""):
-        if grandparentpidsearchstring in examine:
-            pname = examine
-            break
-    popen.stdout.close()
-    return_code = popen.wait()
-    if return_code:
-        raise subprocess.CalledProcessError(return_code, ps)
-    if pname[0:12] == 'explorer.exe':
-        subprocess.Popen("cmd /K \\\"{0}\\\"".format(os.path.splitext(os.path.basename(sys.argv[0]))[0]))
-    else:
-        main()
-
+    with contextlib.suppress(AttributeError, psutil.Error):
+        if psutil.Process().parent().parent().name() == "explorer.exe":
+            subprocess.Popen("powershell -NoExit -Command \\\"& {0}\\\"".format(sys.argv[0]))
+            return
+    main()
 
 if __name__ == "__main__":
     __main()

deploy/windows/instaloader.spec

@@ -5,7 +5,7 @@ block_cipher = None
 
 a = Analysis(['__main__.py'],
              binaries=[],
-             datas=[('tasklist.exe', '.'), ('framedyn.dll', '.')],
+             datas=[],
              hiddenimports=[],
              hookspath=[],
              runtime_hooks=[],

docs/_static/instaloader.css

@@ -18,14 +18,17 @@ code {
     padding: 0.7em;
     color: #fff; }
   .highlight .c1 {
-    color: #008d06; }
+    color: #666; }
   .highlight .k, .highlight .kn, .highlight .ow {
-    color: #f48400;
-    font-weight: bold; }
-  .highlight .nb {
+    color: #008d06; }
+  .highlight .nb, .highlight .ne, .highlight .nf, .highlight .vm {
     color: #f48400; }
-  .highlight .s1, .highlight .s2, .highlight .mi {
+  .highlight .s1, .highlight .s2, .highlight .sa, .highlight .se, .highlight .si, .highlight .mi {
     color: #ff7efd; }
+  .highlight .c1, .highlight .nf, .highlight .sa, .highlight .se, .highlight .si {
+    font-style: italic; }
+  .highlight .k, .highlight .kn, .highlight .ow {
+    font-weight: bold; }
 
 .doc-content {
   padding-top: 0.5em;

docs/_static/instaloader.scss

@@ -37,21 +37,28 @@ code {
   }
 
   .c1 {
-    color: $color_instaloader_main
+    color: #666;
   }
 
   .k, .kn, .ow {
-    color: $color_instaloader_accent;
-    font-weight: bold
+    color: $color_instaloader_main
   }
 
-  .nb {
+  .nb, .ne, .nf, .vm {
     color: $color_instaloader_accent
   }
 
-  .s1, .s2, .mi {
+  .s1, .s2, .sa, .se, .si, .mi {
     color: $color_instaloader_hyper
   }
+
+  .c1, .nf, .sa, .se, .si {
+    font-style: italic;
+  }
+
+  .k, .kn, .ow {
+    font-weight: bold
+  }
 }
 
 .doc-content {

docs/as-module.rst

@@ -226,8 +226,32 @@ Exceptions
 
 .. autoexception:: TooManyRequestsException
 
+Resumable Iterations
+^^^^^^^^^^^^^^^^^^^^
+
+.. versionadded:: 4.5
+
+.. autoclass:: NodeIterator
+   :no-show-inheritance:
+
+.. autoclass:: FrozenNodeIterator
+   :no-show-inheritance:
+
+.. autofunction:: resumable_iteration
+
 ``InstaloaderContext`` (Low-level functions)
 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
 
+``InstaloaderContext``
+""""""""""""""""""""""
+
 .. autoclass:: InstaloaderContext
    :no-show-inheritance:
+
+``RateController``
+""""""""""""""""""
+
+.. autoclass:: RateController
+   :no-show-inheritance:
+
+   .. versionadded:: 4.5

docs/basic-usage.rst

@@ -51,7 +51,7 @@ invoke it with
     instaloader --login=your_username profile [profile ...]
 
 When logging in, Instaloader **stores the session cookies** in a file in your
-temporary directory, which will be reused later the next time :option:`--login`
+home directory, which will be reused later the next time :option:`--login`
 is given.  So you can download private profiles **non-interactively** when you
 already have a valid session cookie file.
 
@@ -111,8 +111,12 @@ Instaloader supports the following targets:
    ``profile`` (requires :option:`--login`).
 
 - ``-post``
-   The single **post** with the given shortcode. Must be preceeded by ``--`` in
-   the argument list to not be mistaken as an option flag.
+   Replace **post** with the post's shortcode to download single post. Must be preceeded by ``--`` in
+   the argument list to not be mistaken as an option flag::
+
+    instaloader -- -B_K4CykAOtf
+
+
 
    .. versionadded:: 4.1
 
@@ -291,17 +295,18 @@ Instaloader as Cronjob
 Instaloader is suitable for running as a cronjob to periodically update your
 personal Instagram archive. The :option:`--quiet` option disables user
 interactions and logging of non-error messages. To non-interactively use
-Instaloader logged-in, create a session file, e.g. in your home directory::
+Instaloader logged-in, create a session file::
 
-   instaloader --login=your_username --sessionfile=~/.instaloader-session
+   instaloader --login=your_username
 
-Then use the same parameters in your cronjob to load the session and download
+Then use the same parameter in your cronjob to load the session and download
 the given targets::
 
-   instaloader --login=your_username --sessionfile=~/.instaloader-session --quiet <target> [...]
+   instaloader --login=your_username --quiet <target> [...]
 
-Without :option:`--sessionfile` option, Instaloader saves the session file in
-a path within your temporary directory.
+Instaloader saves the session file to
+``~/.config/instaloader/session-YOUR-USERNAME``. See
+:option:`--sessionfile` option for how to override this path.
 
 Programming Instaloader
 ^^^^^^^^^^^^^^^^^^^^^^^

docs/cli-options.rst

@@ -193,9 +193,8 @@ Instaloader to login.
 
 .. option:: --sessionfile SESSIONFILE, -f SESSIONFILE
 
-   Path for loading and storing session key file.  Defaults to a path within
-   your temporary directory, encoding your local username and your Instagram
-   profile name.
+   Path for loading and storing session key file.  Defaults to
+   ``~/.config/instaloader/session-YOUR-USERNAME``.
 
 .. option:: --password YOUR-PASSWORD, -p YOUR-PASSWORD
 
@@ -224,6 +223,29 @@ How to Download
    ``#hashtag`` or the profile name. Defaults to ``{date_utc}_UTC``.
    See :ref:`filename-specification` for a list of supported tokens.
 
+.. option:: --resume-prefix prefix
+
+   For many targets, Instaloader is capable of resuming a previously-aborted
+   download loop.  To do so, it creates a JSON file within the target directory
+   when interrupted.  This option controls the prefix for filenames that are
+   used to save the information to resume an interrupted download.  The default
+   prefix is ``iterator``.
+
+   Resuming an interrupted download is supported for most, but not all targets.
+   JSON files with resume information are always compressed, regardless of
+   :option:`--no-compress-json`.
+
+   This feature is turned off entirely with :option:`--no-resume`.
+
+   .. versionadded:: 4.5
+
+.. option:: --no-resume
+
+   Do not resume a previously-aborted download iteration, and do not save such
+   information when interrupted.
+
+   .. versionadded:: 4.5
+
 .. option:: --user-agent USER_AGENT
 
    User Agent to use for HTTP requests. Per default, Instaloader pretends being
@@ -243,6 +265,11 @@ How to Download
 
    .. versionadded:: 4.2
 
+   .. warning::
+      We discourage from using the "commit mode" due to its bugs when being used
+      with other command line options (e.g. :issue:`257` with
+      :option:`--fast-update` and :issue:`483` with :option:`--tagged`).
+
 .. option:: --request-timeout N
 
    Seconds to wait before timing out a connection request.

docs/codesnippets.rst

@@ -28,8 +28,9 @@ Download Posts in a Specific Period
 -----------------------------------
 
 To only download Instagram pictures (and metadata) that are within a specific
-period, you can play around with :func:`~itertools.dropwhile` and
-:func:`~itertools.takewhile` from :mod:`itertools` like in this snippet.
+period, you can simply use :func:`~itertools.dropwhile` and
+:func:`~itertools.takewhile` from :mod:`itertools` on a generator that returns
+Posts in **exact chronological order**, such as :meth:`Profile.get_posts`.
 
 .. literalinclude:: codesnippets/121_since_until.py
 
@@ -37,6 +38,14 @@ See also :class:`Post`, :meth:`Instaloader.download_post`.
 
 Discussed in :issue:`121`.
 
+The code example with :func:`~itertools.dropwhile` and
+:func:`~itertools.takewhile` makes the assumption that the post iterator returns
+posts in exact chronological order.  As discussed in :issue:`666`, the following
+approach fits for an **almost chronological order**, where up to *k* older posts
+are inserted into an otherwise chronological order, such as an Hashtag feed.
+
+.. literalinclude:: codesnippets/666_historical_hashtag_data.py
+
 Likes of a Profile / Ghost Followers
 ------------------------------------
 

docs/codesnippets/121_since_until.py

@@ -5,13 +5,11 @@ import instaloader
 
 L = instaloader.Instaloader()
 
-posts = instaloader.Hashtag.from_name(L.context, 'urbanphotography').get_posts()
-# or
-# posts = instaloader.Profile.from_username(L.context, PROFILE).get_posts()
+posts = instaloader.Profile.from_username(L.context, "instagram").get_posts()
 
 SINCE = datetime(2015, 5, 1)
 UNTIL = datetime(2015, 3, 1)
 
 for post in takewhile(lambda p: p.date > UNTIL, dropwhile(lambda p: p.date > SINCE, posts)):
     print(post.date)
-    L.download_post(post, '#urbanphotography')
+    L.download_post(post, "instagram")

docs/codesnippets/615_import_firefox_session.py

@@ -0,0 +1,53 @@
+from argparse import ArgumentParser
+from glob import glob
+from os.path import expanduser
+from platform import system
+from sqlite3 import OperationalError, connect
+
+try:
+    from instaloader import ConnectionException, Instaloader
+except ModuleNotFoundError:
+    raise SystemExit("Instaloader not found.\n  pip install [--user] instaloader")
+
+
+def get_cookiefile():
+    default_cookiefile = {
+        "Windows": "~/AppData/Roaming/Mozilla/Firefox/Profiles/*/cookies.sqlite",
+        "Darwin": "~/Library/Application Support/Firefox/Profiles/*/cookies.sqlite",
+    }.get(system(), "~/.mozilla/firefox/*/cookies.sqlite")
+    cookiefiles = glob(expanduser(default_cookiefile))
+    if not cookiefiles:
+        raise SystemExit("No Firefox cookies.sqlite file found. Use -c COOKIEFILE.")
+    return cookiefiles[0]
+
+
+def import_session(cookiefile, sessionfile):
+    print("Using cookies from {}.".format(cookiefile))
+    conn = connect(cookiefile)
+    try:
+        cookie_data = conn.execute(
+            "SELECT name, value FROM moz_cookies WHERE baseDomain='instagram.com'"
+        )
+    except OperationalError:
+        cookie_data = conn.execute(
+            "SELECT name, value FROM moz_cookies WHERE host LIKE '%instagram.com'"
+        )
+    instaloader = Instaloader(max_connection_attempts=1)
+    instaloader.context._session.cookies.update(cookie_data)
+    username = instaloader.test_login()
+    if not username:
+        raise SystemExit("Not logged in. Are you logged in successfully in Firefox?")
+    print("Imported session cookie for {}.".format(username))
+    instaloader.context.username = username
+    instaloader.save_session_to_file(sessionfile)
+
+
+if __name__ == "__main__":
+    p = ArgumentParser()
+    p.add_argument("-c", "--cookiefile")
+    p.add_argument("-f", "--sessionfile")
+    args = p.parse_args()
+    try:
+        import_session(args.cookiefile or get_cookiefile(), args.sessionfile)
+    except (ConnectionException, OperationalError) as e:
+        raise SystemExit("Cookie import failed: {}".format(e))

docs/codesnippets/666_historical_hashtag_data.py

@@ -0,0 +1,30 @@
+from datetime import datetime
+import instaloader
+
+L = instaloader.Instaloader()
+
+posts = instaloader.Hashtag.from_name(L.context, "urbanphotography").get_posts()
+
+SINCE = datetime(2020, 5, 10)  # further from today, inclusive
+UNTIL = datetime(2020, 5, 11)  # closer to today, not inclusive
+
+k = 0  # initiate k
+k_list = []  # uncomment this to tune k
+
+for post in posts:
+    postdate = post.date
+
+    if postdate > UNTIL:
+        continue
+    elif postdate <= SINCE:
+        k += 1
+        if k == 50:
+            break
+        else:
+            continue
+    else:
+        L.download_post(post, "#urbanphotography")
+        k = 0  # set k to 0
+        # if you want to tune k, uncomment below to get your k max
+        #k_list.append(k)
+#max(k_list)

docs/codesnippets/92_import_firefox_session.py

@@ -1,23 +0,0 @@
-from glob import glob
-from os.path import expanduser
-from sqlite3 import connect
-
-from instaloader import ConnectionException, Instaloader
-
-# FIREFOXCOOKIEFILE = "/home/alex/.mozilla/firefox/l96w6b90.default/cookies.sqlite"
-FIREFOXCOOKIEFILE = glob(expanduser("~/.mozilla/firefox/*.default/cookies.sqlite"))[0]
-
-instaloader = Instaloader(max_connection_attempts=1)
-instaloader.context._session.cookies.update(connect(FIREFOXCOOKIEFILE)
-                                            .execute("SELECT name, value FROM moz_cookies "
-                                                     "WHERE baseDomain='instagram.com'"))
-
-try:
-    username = instaloader.test_login()
-    if not username:
-        raise ConnectionException()
-except ConnectionException:
-    raise SystemExit("Cookie import failed. Are you logged in successfully in Firefox?")
-
-instaloader.context.username = username
-instaloader.save_session_to_file()

docs/conf.py

@@ -18,6 +18,7 @@
 # documentation root, use os.path.abspath to make it absolute, like shown here.
 #
 import os
+import platform
 import subprocess
 import sys
 import typing
@@ -49,7 +50,25 @@ autodoc_member_order = 'bysource'
 intersphinx_mapping = {'python': ('https://docs.python.org/3', None),
                        'requests': ('https://requests.kennethreitz.org/en/master/', None)}
 
-extlinks = {'issue': ('https://github.com/instaloader/instaloader/issues/%s', 'issue #')}
+nitpick_ignore = [('py:class', 'typing.Tuple')]
+
+current_release = subprocess.check_output(["git", "describe", "--abbrev=0"]).decode("ascii")[1:-1]
+date_format = "%e %b %Y" if platform.system() != "Windows" else "%d %b %Y"
+current_release_date = subprocess.check_output(
+    ["git", "log", "-1", "--tags", "--format=%ad", "--date=format:"+date_format]).decode("ascii")[:-1]
+
+html_context = {'current_release': current_release, 'current_release_date': current_release_date}
+
+extlinks = {
+    'issue': (
+        'https://github.com/instaloader/instaloader/issues/%s',
+        'Issue #'
+    ),
+    'example': (
+        'https://raw.githubusercontent.com/instaloader/instaloader/v' + current_release + '/docs/codesnippets/%s',
+        'Example '
+    ),
+}
 
 # Add any paths that contain templates here, relative to this directory.
 templates_path = ['_templates']
@@ -69,7 +88,7 @@ master_doc = 'index'
 
 # General information about the project.
 project = 'Instaloader'
-copyright = '2017, Alexander Graf and Andre Koch-Kramer'
+copyright = '2017-2020, Alexander Graf and Andre Koch-Kramer'
 author = 'Alexander Graf and Andre Koch-Kramer'
 
 # The version info for the project you're documenting, acts as replacement for
@@ -353,12 +372,6 @@ texinfo_documents = [
 # texinfo_no_detailmenu = False
 
 
-current_release = subprocess.check_output(["git", "describe", "--abbrev=0"]).decode("ascii")[1:-1]
-current_release_date = subprocess.check_output(["git", "log", "-1", "--tags", "--format=%ad", "--date=format:%e %b %Y"]).decode("ascii")[:-1]
-
-html_context = {'current_release': current_release, 'current_release_date': current_release_date}
-
-
 def setup(app):
     typing.TYPE_CHECKING = True
     app.add_stylesheet("instaloader.css")

docs/contributing.rst

@@ -35,8 +35,8 @@ reporting a problem, please keep the following in mind:
 - If not obvious, describe **which behavior you expected**
   instead of what actually happened.
 
-- If you think an issue has been closed accidentally or inappropriately, feel
-  free to reopen it.
+- If we have closed an issue apparently inadvertently or inappropriately, please
+  let us know.
 
 Writing Code or Improving the Documentation
 -------------------------------------------
@@ -45,28 +45,25 @@ Changes of the Instaloader source can be proposed as a
 `Pull Request <https://github.com/instaloader/instaloader/pulls>`__. There are only
 few things to consider:
 
-- Sometimes, the most current code is not in the ``master`` branch. Check that
-  you forked from the most recent branch.
+- Base your Pull Request on the ``master`` branch if it fixes a bug,
+  or the ``upcoming/v4.X`` branch (if it exists at the moment of submitting the PR)
+  otherwise.
 
 - We use `Pylint <https://www.pylint.org/>`__ for error and syntax checking of
-  the source. The file ``.travis.yml`` in the project's root directory
-  shows how it is invoked. Note that sometimes it might be better to disable a
-  warning rather than adapting the code to Pylint's desires.
+  the source and `MyPy <https://github.com/python/mypy>`__ for type checking.
+  Beware that sometimes it might be better to disable a warning rather than
+  adapting the code to a tool's desires.
 
 - The documentation source is located in the ``docs`` folder. The file
   ``cli-options.rst`` is merely an RST-formatted copy of ``instaloader --help``
   output, of which the source is in ``instaloader/__main__.py``.
 
-- Feel free to contact us, even if you "only" have Proof-of-Concepts or
-  not-fully integrated changes. They already might be an advance for the
-  project.
-
-Suggesting Features
--------------------
+Proposing Features
+------------------
 
 .. goal-start
 
-Instaloader's goal is to mimick the browser's behavior to access the data that
+Instaloader's goal is to mimic the browser's behavior to access the data that
 is available through the Instagram web interface, pack this data into complete
 and easily-(re)usable python objects, and provide a user interface for the most
 common downloading and metadata collection tasks, without changing any of the
@@ -74,10 +71,9 @@ online data.
 
 .. goal-end
 
-If you have an idea of how Instaloader should be enhanced, but do not want to
-implement the feature yourself, feel free to open a ticket in our 
-`Issue Tracker <https://github.com/instaloader/instaloader/issues>`__.
-Please consider the following:
+Prior spending effort on implementing a new feature, it might be appropriate to
+clarify how it could fit into the project's scope or discuss implementation
+details. If you feel the need to do so, please create a "feature suggestion".
 
 - Instaloader already has plenty of features. **Check the documentation**
   beforehand to ensure your desired suggestion is not already implemented.
@@ -86,16 +82,21 @@ Please consider the following:
   an issue suggesting the same or a similar feature, share your thoughts in a
   comment there, instead of opening a new issue.
 
-- If possible, provide us a **use case of the feature**: How could the user
-  invoke the new function? Which problem would it solve? If new information is
-  obtained, how would it be further processed?
+- **Motivate the feature**, i.e.
 
-- If not obvious, briefly motivate how your suggested feature **conforms with
-  Instaloader's project goal**.
+  - Provide us a **use case of the feature**: How could the user
+    invoke the new function? Which problem would it solve? If new information is
+    obtained, how would it be further processed?
 
-- **Be patient**. Naturally, bugs and pull requests have a higher priority than
-  feature suggestions. Keep in mind that this is a free software project, and
-  unfortunately we only have limited time to work on it.
+  - Describe already-working **alternatives of the feature** and how they
+    compare to your proposed feature.
+
+  - Briefly describe how your suggested feature **conforms with Instaloader's
+    project goal**.
+
+- Explain your **solution ideas**. Describe your ideas on how the feature could
+  be implemented and the underlying problem could be solved. Also **describe
+  alternatives** that you have considered.
 
 Donations
 ---------

docs/index.rst

@@ -40,6 +40,8 @@ See :ref:`install` for more options on how to install Instaloader.
 - allows **fine-grained customization** of filters and where to store
   downloaded media,
 
+- automatically **resumes previously-interrupted** download iterations,
+
 - is free `open source <https://github.com/instaloader/instaloader>`__
   software written in Python.
 

docs/troubleshooting.rst

@@ -27,12 +27,26 @@ expired. In case the retry continuously fails for some reason, which should not
 happen in normal conditions, consider adjusting the
 :option:`--max-connection-attempts` option.
 
+There have been observations that services, that in their nature offer
+promiscious IP addresses, such as cloud, VPN and public proxy services, might be
+subject to significantly stricter limits for anonymous access. However,
+logged-in accesses (see :option:`--login`) do not seem to be affected.
+
+Too many queries in the last time
+---------------------------------
+
 **"Too many queries in the last time"** is not an error. It is a notice that the
 rate limit has almost been reached, according to Instaloader's own rate
 accounting mechanism. We regularly adjust this mechanism to match Instagram's
 current rate limiting.
 
-Login error
+Private but not followed
+------------------------
+
+You have to follow a private account to access most of its associated
+information.
+
+Login Error
 -----------
 
 Instaloader's login *should* work fine, both with and without
@@ -40,21 +54,31 @@ Two-Factor-Authentication. It also supports handling the *checkpoint challenge*,
 issued when Instagram suspects authentication activity on your account, by
 pointing the user to an URL to be opened in a browser.
 
-Nevertheless, in :issue:`92` users report problems with logging in. To still use
-Instaloader's logged-in functionality, you may use the following script to
-workaround login problems by importing the session cookies from Firefox and
-bypassing Instaloader's login.
+Nevertheless, in :issue:`92` and :issue:`615` users reported problems with
+logging in. We recommend to always keep the sessionfile which Instaloader
+creates when using :option:`--login`. If a sessionfile is present,
+:option:`--login` does not make make use of the failure-prone login procedure.
+Also, sessionfiles usually do not expire and can be copied between different
+computers without any problems.
 
-.. literalinclude:: codesnippets/92_import_firefox_session.py
+If you do not have a sessionfile present, you may use the following script
+(:example:`615_import_firefox_session.py`) to workaround login problems by
+importing the session cookies from Firefox and bypassing Instaloader's login and
+so still use Instaloader's logged-in functionality.
 
-To use this,
+.. literalinclude:: codesnippets/615_import_firefox_session.py
 
-#. login to Instagram in Firefox, 
+To use this script,
 
-#. execute the snippet,
+#. Download the script: :example:`615_import_firefox_session.py`,
 
-#. then, ``instaloader -l USERNAME`` should work fine.
+#. Login to Instagram in Firefox,
 
-If you do not use your default firefox profile, or your operating system has the
-paths differently set up, you may have to alter the ``FIREFOXCOOKIEFILE``
-variable first.
+#. Execute the snippet, e.g. with ``python 615_import_firefox_session.py``,
+
+#. Then, ``instaloader -l USERNAME`` should work fine.
+
+This script also supports specifying a cookiefile path, which may be useful if
+you use multiple Firefox profiles or if your operating system has the directory
+structure differently set up. Also, you can specify an alternative sessionfile
+path.

instaloader/__init__.py

@@ -1,7 +1,7 @@
 """Download pictures (or videos) along with their captions and other metadata from Instagram."""
 
 
-__version__ = '4.4b1'
+__version__ = '4.5a1'
 
 
 try:
@@ -14,6 +14,7 @@ else:
 
 from .exceptions import *
 from .instaloader import Instaloader
-from .instaloadercontext import InstaloaderContext
+from .instaloadercontext import InstaloaderContext, RateController
+from .nodeiterator import NodeIterator, FrozenNodeIterator, resumable_iteration
 from .structures import (Hashtag, Highlight, Post, PostSidecarNode, PostComment, PostCommentAnswer, PostLocation,
                          Profile, Story, StoryItem, TopSearchResults, load_structure_from_file, save_structure_to_file)

instaloader/__main__.py

@@ -82,6 +82,8 @@ def _main(instaloader: Instaloader, targetlist: List[str],
         instaloader.context.log('Only download storyitems with property "{}".'.format(storyitem_filter_str))
     # Login, if desired
     if username is not None:
+        if not re.match(r"^[A-Za-z0-9._]+$", username):
+            instaloader.context.error("Warning: Parameter \"{}\" for --login is not a valid username.".format(username))
         try:
             instaloader.load_session_from_file(username, sessionfile)
         except FileNotFoundError as err:
@@ -143,7 +145,7 @@ def _main(instaloader: Instaloader, targetlist: List[str],
                     for followee in profile.get_followees():
                         instaloader.save_profile_id(followee)
                         profiles.add(followee)
-                elif re.match(r"^#[A-Za-z0-9]+$", target):
+                elif re.match(r"^#\w+$", target):
                     instaloader.download_hashtag(hashtag=target[1:], max_count=max_count, fast_update=fast_update,
                                                  post_filter=post_filter,
                                                  profile_pic=download_profile_pic, posts=download_posts)
@@ -196,8 +198,8 @@ def _main(instaloader: Instaloader, targetlist: List[str],
         if len(profiles) > 1:
             instaloader.context.log("Downloading {} profiles: {}".format(len(profiles),
                                                                          ' '.join([p.username for p in profiles])))
-        if profiles and download_profile_pic and not instaloader.context.is_logged_in:
-            instaloader.context.error("Warning: Use --login to download HD version of profile pictures.")
+        if profiles and (download_profile_pic or download_posts) and not instaloader.context.is_logged_in:
+            instaloader.context.error("Warning: Use --login to download higher-quality versions of pictures.")
         instaloader.download_profiles(profiles,
                                       download_profile_pic, download_posts, download_tagged, download_igtv,
                                       download_highlights, download_stories,
@@ -349,6 +351,13 @@ def main():
                             '--dirname-pattern. {profile} is replaced by the profile name,'
                             '{target} is replaced by the target you specified, i.e. either :feed'
                             '#hashtag or the profile name. Defaults to \'{date_utc}_UTC\'')
+    g_how.add_argument('--resume-prefix', metavar='PREFIX',
+                       help='Prefix for filenames that are used to save the information to resume an interrupted '
+                            'download.')
+    g_how.add_argument('--no-resume', action='store_true',
+                       help='Do not resume a previously-aborted download iteration, and do not save such information '
+                            'when interrupted.')
+    g_how.add_argument('--use-aged-resume-files', action='store_true', help=SUPPRESS)
     g_how.add_argument('--user-agent',
                        help='User Agent to use for HTTP requests. Defaults to \'{}\'.'.format(default_user_agent()))
     g_how.add_argument('-S', '--no-sleep', action='store_true', help=SUPPRESS)
@@ -356,10 +365,7 @@ def main():
                        help='Maximum number of connection attempts until a request is aborted. Defaults to 3. If a '
                             'connection fails, it can be manually skipped by hitting CTRL+C. Set this to 0 to retry '
                             'infinitely.')
-    g_how.add_argument('--commit-mode', action='store_true',
-                       help='Tries to ensure downloaded images avoid corruption in case of unexpected interruption. '
-                       'If the last picture is corrupted, Instaloader will fix the picture the next time it is run. '
-                       'Requires the JSON metadata to be saved.')
+    g_how.add_argument('--commit-mode', action='store_true', help=SUPPRESS)
     g_how.add_argument('--request-timeout', metavar='N', type=float,
                        help='seconds to wait before timing out a connection request')
 
@@ -395,6 +401,10 @@ def main():
                 raise SystemExit("--no-captions and --post-metadata-txt or --storyitem-metadata-txt given; "
                                  "That contradicts.")
 
+        if args.no_resume and args.resume_prefix:
+            raise SystemExit("--no-resume and --resume-prefix given; That contradicts.")
+        resume_prefix = (args.resume_prefix if args.resume_prefix else 'iterator') if not args.no_resume else None
+
         if args.no_pictures and args.fast_update:
             raise SystemExit('--no-pictures and --fast-update cannot be used together.')
 
@@ -403,9 +413,6 @@ def main():
         download_posts = not (args.no_posts or args.stories_only or args.profile_pic_only)
         download_stories = args.stories or args.stories_only
 
-        if args.commit_mode and args.no_metadata_json:
-            raise SystemExit('--commit-mode requires JSON metadata to be saved.')
-
         loader = Instaloader(sleep=not args.no_sleep, quiet=args.quiet, user_agent=args.user_agent,
                              dirname_pattern=args.dirname_pattern, filename_pattern=args.filename_pattern,
                              download_pictures=not args.no_pictures,
@@ -417,7 +424,8 @@ def main():
                              storyitem_metadata_txt_pattern=storyitem_metadata_txt_pattern,
                              max_connection_attempts=args.max_connection_attempts,
                              request_timeout=args.request_timeout,
-                             commit_mode=args.commit_mode)
+                             resume_prefix=resume_prefix,
+                             check_resume_bbd=not args.use_aged_resume_files)
         _main(loader,
               args.profile,
               username=args.login.lower() if args.login is not None else None,

instaloader/instaloader.py

@@ -1,6 +1,5 @@
 import getpass
 import json
-import lzma
 import os
 import platform
 import re
@@ -20,13 +19,28 @@ import requests
 import urllib3  # type: ignore
 
 from .exceptions import *
-from .instaloadercontext import InstaloaderContext
+from .instaloadercontext import InstaloaderContext, RateController
+from .nodeiterator import NodeIterator, resumable_iteration
 from .structures import (Hashtag, Highlight, JsonExportable, Post, PostLocation, Profile, Story, StoryItem,
-                         save_structure_to_file, load_structure_from_file)
+                         load_structure_from_file, save_structure_to_file)
 
 
 def get_default_session_filename(username: str) -> str:
     """Returns default session filename for given username."""
+    sessionfilename = "session-{}".format(username)
+    if platform.system() == "Windows":
+        # on Windows, use %LOCALAPPDATA%\Instaloader\session-USERNAME
+        localappdata = os.getenv("LOCALAPPDATA")
+        if localappdata is not None:
+            return os.path.join(localappdata, "Instaloader", sessionfilename)
+        # legacy fallback - store in temp dir if %LOCALAPPDATA% is not set
+        return os.path.join(tempfile.gettempdir(), ".instaloader-" + getpass.getuser(), sessionfilename)
+    # on Unix, use ~/.config/instaloader/session-USERNAME
+    return os.path.join(os.getenv("XDG_CONFIG_HOME", os.path.expanduser("~/.config")), "instaloader", sessionfilename)
+
+
+def get_legacy_session_filename(username: str) -> str:
+    """Returns legacy (until v4.4.3) default session filename for given username."""
     dirname = tempfile.gettempdir() + "/" + ".instaloader-" + getpass.getuser()
     filename = dirname + "/" + "session-" + username
     return filename.lower()
@@ -103,22 +117,20 @@ class _ArbitraryItemFormatter(string.Formatter):
 
 
 class _PostPathFormatter(_ArbitraryItemFormatter):
-    def vformat(self, format_string, args, kwargs):
-        """Override :meth:`string.Formatter.vformat` for character substitution in paths for Windows, see issue #84."""
-        ret = super().vformat(format_string, args, kwargs)
-        if platform.system() == 'Windows':
-            ret = ret.replace(':', '\ua789').replace('<', '\ufe64').replace('>', '\ufe65').replace('\"', '\uff02')
-            ret = ret.replace('\\', '\ufe68').replace('|', '\uff5c').replace('?', '\ufe16').replace('*', '\uff0a')
-        return ret
-
     def get_value(self, key, args, kwargs):
-        """Replaces '/' with similar looking Division Slash and on windows newline with space"""
         ret = super().get_value(key, args, kwargs)
         if not isinstance(ret, str):
             return ret
+        return self.sanitize_path(ret)
+
+    @staticmethod
+    def sanitize_path(ret: str) -> str:
+        """Replaces '/' with similar looking Division Slash and some other illegal filename characters on Windows."""
         ret = ret.replace('/', '\u2215')
         if platform.system() == 'Windows':
-            ret = ret.replace('\n', ' ')
+            ret = ret.replace(':', '\uff1a').replace('<', '\ufe64').replace('>', '\ufe65').replace('\"', '\uff02')
+            ret = ret.replace('\\', '\ufe68').replace('|', '\uff5c').replace('?', '\ufe16').replace('*', '\uff0a')
+            ret = ret.replace('\n', ' ').replace('\r', ' ')
         return ret
 
 
@@ -141,8 +153,10 @@ class Instaloader:
        txt file.
     :param storyitem_metadata_txt_pattern: :option:`--storyitem-metadata-txt`, default is empty (=none)
     :param max_connection_attempts: :option:`--max-connection-attempts`
-    :param commit_mode: :option:`--commit-mode`
     :param request_timeout: :option:`--request-timeout`, set per-request timeout (seconds)
+    :param rate_controller: Generator for a :class:`RateController` to override rate controlling behavior
+    :param resume_prefix: :option:`--resume-prefix`, or None for :option:`--no-resume`.
+    :param check_resume_bbd: Whether to check the date of expiry of resume files and reject them if expired.
 
     .. attribute:: context
 
@@ -166,9 +180,12 @@ class Instaloader:
                  storyitem_metadata_txt_pattern: str = None,
                  max_connection_attempts: int = 3,
                  request_timeout: Optional[float] = None,
-                 commit_mode: bool = False):
+                 rate_controller: Optional[Callable[[InstaloaderContext], RateController]] = None,
+                 resume_prefix: Optional[str] = "iterator",
+                 check_resume_bbd: bool = True):
 
-        self.context = InstaloaderContext(sleep, quiet, user_agent, max_connection_attempts, request_timeout)
+        self.context = InstaloaderContext(sleep, quiet, user_agent, max_connection_attempts,
+                                          request_timeout, rate_controller)
 
         # configuration parameters
         self.dirname_pattern = dirname_pattern or "{target}"
@@ -184,12 +201,8 @@ class Instaloader:
             else post_metadata_txt_pattern
         self.storyitem_metadata_txt_pattern = '' if storyitem_metadata_txt_pattern is None \
             else storyitem_metadata_txt_pattern
-        self.commit_mode = commit_mode
-        if self.commit_mode and not self.save_metadata:
-            raise InvalidArgumentException("Commit mode requires JSON metadata to be saved.")
-
-        # Used to keep state in commit mode
-        self._committed = None  # type: Optional[bool]
+        self.resume_prefix = resume_prefix
+        self.check_resume_bbd = check_resume_bbd
 
     @contextmanager
     def anonymous_copy(self):
@@ -211,7 +224,8 @@ class Instaloader:
             storyitem_metadata_txt_pattern=self.storyitem_metadata_txt_pattern,
             max_connection_attempts=self.context.max_connection_attempts,
             request_timeout=self.context.request_timeout,
-            commit_mode=self.commit_mode)
+            resume_prefix=self.resume_prefix,
+            check_resume_bbd=self.check_resume_bbd)
         yield new_loader
         self.context.error_log.extend(new_loader.context.error_log)
         new_loader.context.error_log = []  # avoid double-printing of errors
@@ -237,15 +251,9 @@ class Instaloader:
         if filename_suffix is not None:
             filename += '_' + filename_suffix
         filename += '.' + file_extension
-        # A post is considered "commited" if the json file exists and is not malformed.
-        if self.commit_mode:
-            if self._committed and os.path.isfile(filename):
-                self.context.log(filename + ' exists', end=' ', flush=True)
-                return False
-        else:
-            if os.path.isfile(filename):
-                self.context.log(filename + ' exists', end=' ', flush=True)
-                return False
+        if os.path.isfile(filename):
+            self.context.log(filename + ' exists', end=' ', flush=True)
+            return False
         self.context.get_and_write_raw(url, filename)
         os.utime(filename, (datetime.now().timestamp(), mtime.timestamp()))
         return True
@@ -357,6 +365,24 @@ class Instaloader:
         os.utime(filename, (datetime.now().timestamp(), mtime.timestamp()))
         self.context.log('geo', end=' ', flush=True)
 
+    def format_filename_within_target_path(self,
+                                           target: Union[str, Path],
+                                           owner_profile: Optional[Profile],
+                                           identifier: str,
+                                           name_suffix: str,
+                                           extension: str):
+        """Returns a filename within the target path.
+
+        .. versionadded:: 4.5"""
+        if ((format_string_contains_key(self.dirname_pattern, 'profile') or
+             format_string_contains_key(self.dirname_pattern, 'target'))):
+            profile_str = owner_profile.username.lower() if owner_profile is not None else target
+            return os.path.join(self.dirname_pattern.format(profile=profile_str, target=target),
+                                '{0}_{1}.{2}'.format(identifier, name_suffix, extension))
+        else:
+            return os.path.join(self.dirname_pattern.format(),
+                                '{0}_{1}_{2}.{3}'.format(target, identifier, name_suffix, extension))
+
     @_retry_on_connection_error
     def download_title_pic(self, url: str, target: Union[str, Path], name_suffix: str, owner_profile: Optional[Profile],
                            _attempt: int = 1) -> None:
@@ -377,16 +403,7 @@ class Instaloader:
         else:
             pic_bytes = http_response.content
             pic_identifier = md5(pic_bytes).hexdigest()[:16]
-        pic_extension = 'jpg'
-        if ((format_string_contains_key(self.dirname_pattern, 'profile') or
-             format_string_contains_key(self.dirname_pattern, 'target'))):
-            profile_str = owner_profile.username.lower() if owner_profile is not None else target
-            filename = '{0}/{1}_{2}.{3}'.format(self.dirname_pattern.format(profile=profile_str,
-                                                                            target=target),
-                                                pic_identifier, name_suffix, pic_extension)
-        else:
-            filename = '{0}/{1}_{2}_{3}.{4}'.format(self.dirname_pattern.format(), target,
-                                                    pic_identifier, name_suffix, pic_extension)
+        filename = self.format_filename_within_target_path(target, owner_profile, pic_identifier, name_suffix, 'jpg')
         content_length = http_response.headers.get('Content-Length', None)
         if os.path.isfile(filename) and (not self.context.is_logged_in or
                                          (content_length is not None and
@@ -443,6 +460,8 @@ class Instaloader:
         """
         if filename is None:
             filename = get_default_session_filename(username)
+            if not os.path.exists(filename):
+                filename = get_legacy_session_filename(username)
         with open(filename, 'rb') as sessionfile:
             self.context.load_session_from_file(username, sessionfile)
             self.context.log("Loaded session from %s." % filename)
@@ -487,12 +506,11 @@ class Instaloader:
         """
 
         dirname = _PostPathFormatter(post).format(self.dirname_pattern, target=target)
-        filename = dirname + '/' + self.format_filename(post, target=target)
+        filename = os.path.join(dirname, self.format_filename(post, target=target))
         os.makedirs(os.path.dirname(filename), exist_ok=True)
 
         # Download the image(s) / video thumbnail and videos within sidecars if desired
         downloaded = True
-        self._committed = self.check_if_committed(filename)
         if self.download_pictures:
             if post.typename == 'GraphSidecar':
                 edge_number = 1
@@ -616,7 +634,7 @@ class Instaloader:
 
         date_local = item.date_local
         dirname = _PostPathFormatter(item).format(self.dirname_pattern, target=target)
-        filename = dirname + '/' + self.format_filename(item, target=target)
+        filename = os.path.join(dirname, self.format_filename(item, target=target))
         os.makedirs(os.path.dirname(filename), exist_ok=True)
         downloaded = False
         if not item.is_video or self.download_video_thumbnails is True:
@@ -681,7 +699,8 @@ class Instaloader:
             name = user_highlight.owner_username
             highlight_target = (filename_target
                                 if filename_target
-                                else Path(name) / Path(user_highlight.title))  # type: Union[str, Path]
+                                else (Path(_PostPathFormatter.sanitize_path(name)) /
+                                      _PostPathFormatter.sanitize_path(user_highlight.title)))  # type: Union[str, Path]
             self.context.log("Retrieving highlights \"{}\" from profile {}".format(user_highlight.title, name))
             self.download_highlight_cover(user_highlight, highlight_target)
             totalcount = user_highlight.itemcount
@@ -704,59 +723,75 @@ class Instaloader:
                             fast_update: bool = False,
                             post_filter: Optional[Callable[[Post], bool]] = None,
                             max_count: Optional[int] = None,
-                            total_count: Optional[int] = None) -> None:
+                            total_count: Optional[int] = None,
+                            owner_profile: Optional[Profile] = None) -> None:
         """
         Download the Posts returned by given Post Iterator.
 
         ..versionadded:: 4.4
 
+        ..versionchanged:: 4.5
+           Transparently resume an aborted operation if `posts` is a :class:`NodeIterator`.
+
         :param posts: Post Iterator to loop through.
-        :param target: Target name
-        :param fast_update: :option:`--fast-update`
-        :param post_filter: :option:`--post-filter`
-        :param max_count: Maximum count of Posts to download (:option:`--count`)
-        :param total_count: Total number of posts returned by given iterator
+        :param target: Target name.
+        :param fast_update: :option:`--fast-update`.
+        :param post_filter: :option:`--post-filter`.
+        :param max_count: Maximum count of Posts to download (:option:`--count`).
+        :param total_count: Total number of posts returned by given iterator.
+        :param owner_profile: Associated profile, if any.
         """
-        for number, post in enumerate(posts):
-            if max_count is not None and number >= max_count:
-                break
-            if total_count is not None:
-                self.context.log("[{0:{w}d}/{1:{w}d}] ".format(number + 1, total_count,
-                                                               w=len(str(total_count))),
-                                 end="", flush=True)
-            else:
-                if max_count is not None:
-                    self.context.log("[{0:{w}d}/{1:{w}d}] ".format(number + 1, max_count,
-                                                                   w=len(str(max_count))),
+        displayed_count = (max_count if total_count is None or max_count is not None and max_count < total_count
+                           else total_count)
+        with resumable_iteration(
+                context=self.context,
+                iterator=posts,
+                load=load_structure_from_file,
+                save=save_structure_to_file,
+                format_path=lambda magic: self.format_filename_within_target_path(
+                    target, owner_profile, self.resume_prefix or '', magic, 'json.xz'
+                ),
+                check_bbd=self.check_resume_bbd,
+                enabled=self.resume_prefix is not None
+        ) as resume_info:
+            is_resuming, start_index = resume_info
+            for number, post in enumerate(posts):
+                if max_count is not None and number + start_index >= max_count:
+                    break
+                if displayed_count is not None:
+                    self.context.log("[{0:{w}d}/{1:{w}d}] ".format(number + start_index + 1, displayed_count,
+                                                                   w=len(str(displayed_count))),
                                      end="", flush=True)
                 else:
-                    self.context.log("[{:3d}] ".format(number + 1), end="", flush=True)
-            if post_filter is not None:
-                try:
-                    if not post_filter(post):
-                        self.context.log("{} skipped".format(post))
-                        continue
-                except (InstaloaderException, KeyError, TypeError) as err:
-                    self.context.error("{} skipped. Filter evaluation failed: {}".format(post, err))
-                    continue
-            with self.context.error_catcher("Download {} of {}".format(post, target)):
-                # The PostChangedException gets raised if the Post's id/shortcode changed while obtaining
-                # additional metadata. This is most likely the case if a HTTP redirect takes place while
-                # resolving the shortcode URL.
-                # The `post_changed` variable keeps the fast-update functionality alive: A Post which is
-                # obained after a redirect has probably already been downloaded as a previous Post of the
-                # same Profile.
-                # Observed in issue #225: https://github.com/instaloader/instaloader/issues/225
-                post_changed = False
-                while True:
+                    self.context.log("[{:3d}] ".format(number + start_index + 1), end="", flush=True)
+                if post_filter is not None:
                     try:
-                        downloaded = self.download_post(post, target=target)
-                        break
-                    except PostChangedException:
-                        post_changed = True
+                        if not post_filter(post):
+                            self.context.log("{} skipped".format(post))
+                            continue
+                    except (InstaloaderException, KeyError, TypeError) as err:
+                        self.context.error("{} skipped. Filter evaluation failed: {}".format(post, err))
                         continue
-                if fast_update and not downloaded and not post_changed:
-                    break
+                with self.context.error_catcher("Download {} of {}".format(post, target)):
+                    # The PostChangedException gets raised if the Post's id/shortcode changed while obtaining
+                    # additional metadata. This is most likely the case if a HTTP redirect takes place while
+                    # resolving the shortcode URL.
+                    # The `post_changed` variable keeps the fast-update functionality alive: A Post which is
+                    # obained after a redirect has probably already been downloaded as a previous Post of the
+                    # same Profile.
+                    # Observed in issue #225: https://github.com/instaloader/instaloader/issues/225
+                    post_changed = False
+                    while True:
+                        try:
+                            downloaded = self.download_post(post, target=target)
+                            break
+                        except PostChangedException:
+                            post_changed = True
+                            continue
+                    if fast_update and not downloaded and not post_changed:
+                        # disengage fast_update for first post when resuming
+                        if not is_resuming or number > 0:
+                            break
 
     @_requires_login
     def get_feed_posts(self) -> Iterator[Post]:
@@ -816,8 +851,10 @@ class Instaloader:
         """
         self.context.log("Retrieving saved posts...")
         assert self.context.username is not None  # safe due to @_requires_login; required by typechecker
-        self.posts_download_loop(Profile.from_username(self.context, self.context.username).get_saved_posts(), ":saved",
-                                 fast_update, post_filter, max_count=max_count)
+        node_iterator = Profile.from_username(self.context, self.context.username).get_saved_posts()
+        self.posts_download_loop(node_iterator, ":saved",
+                                 fast_update, post_filter,
+                                 max_count=max_count, total_count=node_iterator.count)
 
     @_requires_login
     def get_location_posts(self, location: str) -> Iterator[Post]:
@@ -872,18 +909,20 @@ class Instaloader:
                                  max_count=max_count)
 
     @_requires_login
-    def get_explore_posts(self) -> Iterator[Post]:
+    def get_explore_posts(self) -> NodeIterator[Post]:
         """Get Posts which are worthy of exploring suggested by Instagram.
 
         :return: Iterator over Posts of the user's suggested posts.
+        :rtype: NodeIterator[Post]
         :raises LoginRequiredException: If called without being logged in.
         """
-        data = self.context.get_json('explore/', {})
-        yield from (Post(self.context, node)
-                    for node in self.context.graphql_node_list("df0dcc250c2b18d9fd27c5581ef33c7c",
-                                                               {}, 'https://www.instagram.com/explore/',
-                                                               lambda d: d['data']['user']['edge_web_discover_media'],
-                                                               data.get('rhx_gis')))
+        return NodeIterator(
+            self.context,
+            'df0dcc250c2b18d9fd27c5581ef33c7c',
+            lambda d: d['data']['user']['edge_web_discover_media'],
+            lambda n: Post(self.context, n),
+            query_referer='https://www.instagram.com/explore/',
+        )
 
     def get_hashtag_posts(self, hashtag: str) -> Iterator[Post]:
         """Get Posts associated with a #hashtag.
@@ -942,8 +981,10 @@ class Instaloader:
         .. versionadded:: 4.1"""
         self.context.log("Retrieving tagged posts for profile {}.".format(profile.username))
         self.posts_download_loop(profile.get_tagged_posts(),
-                                 target if target else Path(profile.username) / Path(':tagged'), fast_update,
-                                 post_filter)
+                                 target if target
+                                 else (Path(_PostPathFormatter.sanitize_path(profile.username)) /
+                                       _PostPathFormatter.sanitize_path(':tagged')),
+                                 fast_update, post_filter)
 
     def download_igtv(self, profile: Profile, fast_update: bool = False,
                       post_filter: Optional[Callable[[Post], bool]] = None) -> None:
@@ -952,15 +993,17 @@ class Instaloader:
         .. versionadded:: 4.3"""
         self.context.log("Retrieving IGTV videos for profile {}.".format(profile.username))
         self.posts_download_loop(profile.get_igtv_posts(), profile.username, fast_update, post_filter,
-                                 total_count=profile.igtvcount)
+                                 total_count=profile.igtvcount, owner_profile=profile)
 
     def _get_id_filename(self, profile_name: str) -> str:
         if ((format_string_contains_key(self.dirname_pattern, 'profile') or
              format_string_contains_key(self.dirname_pattern, 'target'))):
-            return '{0}/id'.format(self.dirname_pattern.format(profile=profile_name.lower(),
-                                                               target=profile_name.lower()))
+            return os.path.join(self.dirname_pattern.format(profile=profile_name.lower(),
+                                                            target=profile_name.lower()),
+                                'id')
         else:
-            return '{0}/{1}_id'.format(self.dirname_pattern.format(), profile_name.lower())
+            return os.path.join(self.dirname_pattern.format(),
+                                '{0}_id'.format(profile_name.lower()))
 
     def save_profile_id(self, profile: Profile):
         """
@@ -1074,9 +1117,9 @@ class Instaloader:
 
                 # Save metadata as JSON if desired.
                 if self.save_metadata:
-                    json_filename = '{0}/{1}_{2}'.format(self.dirname_pattern.format(profile=profile_name,
-                                                                                     target=profile_name),
-                                                         profile_name, profile.userid)
+                    json_filename = os.path.join(self.dirname_pattern.format(profile=profile_name,
+                                                                             target=profile_name),
+                                                 '{0}_{1}'.format(profile_name, profile.userid))
                     self.save_metadata_json(json_filename, profile)
 
                 # Catch some errors
@@ -1105,7 +1148,7 @@ class Instaloader:
                 if posts:
                     self.context.log("Retrieving posts from profile {}.".format(profile_name))
                     self.posts_download_loop(profile.get_posts(), profile_name, fast_update, post_filter,
-                                             total_count=profile.mediacount)
+                                             total_count=profile.mediacount, owner_profile=profile)
 
         if stories and profiles:
             with self.context.error_catcher("Download stories"):
@@ -1185,26 +1228,7 @@ class Instaloader:
         # Iterate over pictures and download them
         self.context.log("Retrieving posts from profile {}.".format(profile_name))
         self.posts_download_loop(profile.get_posts(), profile_name, fast_update, post_filter,
-                                 total_count=profile.mediacount)
-
-    def check_if_committed(self, filename: str) -> bool:
-        """Checks to see if the current post has been committed.
-
-        A post is considered committed if its json metadata file exists and is not malformed.
-
-        .. versionadded:: 4.2
-        """
-        if os.path.isfile(filename + '.json.xz'):
-            filename += '.json.xz'
-        elif os.path.isfile(filename + '.json'):
-            filename += '.json'
-        else:
-            return False
-        try:
-            load_structure_from_file(self.context, filename)
-            return True
-        except (FileNotFoundError, lzma.LZMAError, json.decoder.JSONDecodeError):
-            return False
+                                 total_count=profile.mediacount, owner_profile=profile)
 
     def interactive_login(self, username: str) -> None:
         """Logs in and internally stores session, asking user for password interactively.

instaloader/instaloadercontext.py

@@ -1,5 +1,6 @@
 import hashlib
 import json
+import os
 import pickle
 import random
 import re
@@ -52,7 +53,8 @@ class InstaloaderContext:
     """
 
     def __init__(self, sleep: bool = True, quiet: bool = False, user_agent: Optional[str] = None,
-                 max_connection_attempts: int = 3, request_timeout: Optional[float] = None):
+                 max_connection_attempts: int = 3, request_timeout: Optional[float] = None,
+                 rate_controller: Optional[Callable[["InstaloaderContext"], "RateController"]] = None):
 
         self.user_agent = user_agent if user_agent is not None else default_user_agent()
         self.request_timeout = request_timeout
@@ -68,9 +70,7 @@ class InstaloaderContext:
         # error log, filled with error() and printed at the end of Instaloader.main()
         self.error_log = []                      # type: List[str]
 
-        # For the adaption of sleep intervals (rate control)
-        self._graphql_query_timestamps = dict()  # type: Dict[str, List[float]]
-        self._graphql_earliest_next_request_time = 0.0
+        self._rate_controller = rate_controller(self) if rate_controller is not None else RateController(self)
 
         # Can be set to True for testing, disables supression of InstaloaderContext._error_catcher
         self.raise_all_errors = False
@@ -113,7 +113,7 @@ class InstaloaderContext:
     def close(self):
         """Print error log and close session"""
         if self.error_log and not self.quiet:
-            print("\nErrors occurred:", file=sys.stderr)
+            print("\nErrors or warnings occurred:", file=sys.stderr)
             for err in self.error_log:
                 print(err, file=sys.stderr)
         self._session.close()
@@ -215,8 +215,11 @@ class InstaloaderContext:
         session.headers.update({'X-CSRFToken': csrf_token})
         # Not using self.get_json() here, because we need to access csrftoken cookie
         self.do_sleep()
+        # Workaround credits to pgrimaud.
+        # See: https://github.com/pgrimaud/instagram-user-feed/commit/96ad4cf54d1ad331b337f325c73e664999a6d066
+        enc_password = '#PWD_INSTAGRAM_BROWSER:0:{}:{}'.format(int(datetime.now().timestamp()), passwd)
         login = session.post('https://www.instagram.com/accounts/login/ajax/',
-                             data={'password': passwd, 'username': user}, allow_redirects=True)
+                             data={'enc_password': enc_password, 'username': user}, allow_redirects=True)
         try:
             resp_json = login.json()
         except json.decoder.JSONDecodeError:
@@ -291,77 +294,6 @@ class InstaloaderContext:
         if self.sleep:
             time.sleep(min(random.expovariate(0.7), 5.0))
 
-    def _dump_query_timestamps(self, current_time: float):
-        """Output the number of GraphQL queries grouped by their query_hash within the last time."""
-        windows = [10, 11, 15, 20, 30, 60]
-        print("GraphQL requests:", file=sys.stderr)
-        for query_hash, times in self._graphql_query_timestamps.items():
-            print("  {}".format(query_hash), file=sys.stderr)
-            for window in windows:
-                reqs_in_sliding_window = sum(t > current_time - window * 60 for t in times)
-                print("    last {} minutes: {} requests".format(window, reqs_in_sliding_window), file=sys.stderr)
-
-    def _graphql_request_count_per_sliding_window(self, query_hash: str) -> int:
-        """Return how many GraphQL requests can be done within the sliding window."""
-        if self.is_logged_in:
-            max_reqs = {'1cb6ec562846122743b61e492c85999f': 20,
-                        '33ba35852cb50da46f5b5e889df7d159': 20,
-                        'iphone': 100,
-                        'other': 100}
-        else:
-            max_reqs = {'1cb6ec562846122743b61e492c85999f': 200,
-                        '33ba35852cb50da46f5b5e889df7d159': 200,
-                        'other': 200}
-        return max_reqs.get(query_hash) or min(max_reqs.values())
-
-    def _graphql_query_waittime(self, query_hash: str, current_time: float, untracked_queries: bool = False) -> float:
-        """Calculate time needed to wait before GraphQL query can be executed."""
-        sliding_window = 660
-        if query_hash not in self._graphql_query_timestamps:
-            self._graphql_query_timestamps[query_hash] = []
-        self._graphql_query_timestamps[query_hash] = list(filter(lambda t: t > current_time - 60 * 60,
-                                                                 self._graphql_query_timestamps[query_hash]))
-        reqs_in_sliding_window = list(filter(lambda t: t > current_time - sliding_window,
-                                             self._graphql_query_timestamps[query_hash]))
-        count_per_sliding_window = self._graphql_request_count_per_sliding_window(query_hash)
-        if len(reqs_in_sliding_window) < count_per_sliding_window and not untracked_queries:
-            return max(0, self._graphql_earliest_next_request_time - current_time)
-        next_request_time = min(reqs_in_sliding_window) + sliding_window + 6
-        if untracked_queries:
-            self._graphql_earliest_next_request_time = next_request_time
-        return round(max(next_request_time, self._graphql_earliest_next_request_time) - current_time)
-
-    def _ratecontrol_graphql_query(self, query_hash: str, untracked_queries: bool = False):
-        """Called before a GraphQL query is made in order to stay within Instagram's rate limits.
-
-        :param query_hash: The query_hash parameter of the query.
-        :param untracked_queries: True, if 429 has been returned to apply 429 logic.
-        """
-        if not untracked_queries:
-            waittime = self._graphql_query_waittime(query_hash, time.monotonic(), untracked_queries)
-            assert waittime >= 0
-            if waittime > 10:
-                self.log('\nToo many queries in the last time. Need to wait {} seconds, until {:%H:%M}.'
-                         .format(waittime, datetime.now() + timedelta(seconds=waittime)))
-            time.sleep(waittime)
-            if query_hash not in self._graphql_query_timestamps:
-                self._graphql_query_timestamps[query_hash] = [time.monotonic()]
-            else:
-                self._graphql_query_timestamps[query_hash].append(time.monotonic())
-        else:
-            text_for_429 = ("HTTP error code 429 was returned because too many queries occurred in the last time. "
-                            "Please do not use Instagram in your browser or run multiple instances of Instaloader "
-                            "in parallel.")
-            print(textwrap.fill(text_for_429), file=sys.stderr)
-            current_time = time.monotonic()
-            waittime = self._graphql_query_waittime(query_hash, current_time, untracked_queries)
-            assert waittime >= 0
-            if waittime > 10:
-                self.log('The request will be retried in {} seconds, at {:%H:%M}.'
-                         .format(waittime, datetime.now() + timedelta(seconds=waittime)))
-            self._dump_query_timestamps(current_time)
-            time.sleep(waittime)
-
     def get_json(self, path: str, params: Dict[str, Any], host: str = 'www.instagram.com',
                  session: Optional[requests.Session] = None, _attempt=1) -> Dict[str, Any]:
         """JSON request to Instagram.
@@ -382,11 +314,11 @@ class InstaloaderContext:
         try:
             self.do_sleep()
             if is_graphql_query:
-                self._ratecontrol_graphql_query(params['query_hash'])
+                self._rate_controller.wait_before_query(params['query_hash'])
             if is_iphone_query:
-                self._ratecontrol_graphql_query('iphone')
+                self._rate_controller.wait_before_query('iphone')
             if is_other_query:
-                self._ratecontrol_graphql_query('other')
+                self._rate_controller.wait_before_query('other')
             resp = sess.get('https://{0}/{1}'.format(host, path), params=params, allow_redirects=False)
             while resp.is_redirect:
                 redirect_url = resp.headers['location']
@@ -442,12 +374,13 @@ class InstaloaderContext:
                     raise ConnectionException(error_string) from err
             self.error(error_string + " [retrying; skip with ^C]", repeat_at_end=False)
             try:
-                if is_graphql_query and isinstance(err, TooManyRequestsException):
-                    self._ratecontrol_graphql_query(params['query_hash'], untracked_queries=True)
-                if is_iphone_query and isinstance(err, TooManyRequestsException):
-                    self._ratecontrol_graphql_query('iphone', untracked_queries=True)
-                if is_other_query and isinstance(err, TooManyRequestsException):
-                    self._ratecontrol_graphql_query('other', untracked_queries=True)
+                if isinstance(err, TooManyRequestsException):
+                    if is_graphql_query:
+                        self._rate_controller.handle_429(params['query_hash'])
+                    if is_iphone_query:
+                        self._rate_controller.handle_429('iphone')
+                    if is_other_query:
+                        self._rate_controller.handle_429('other')
                 return self.get_json(path=path, params=params, host=host, session=sess, _attempt=_attempt + 1)
             except KeyboardInterrupt:
                 self.error("[skipped by user]", repeat_at_end=False)
@@ -495,7 +428,12 @@ class InstaloaderContext:
                           edge_extractor: Callable[[Dict[str, Any]], Dict[str, Any]],
                           rhx_gis: Optional[str] = None,
                           first_data: Optional[Dict[str, Any]] = None) -> Iterator[Dict[str, Any]]:
-        """Retrieve a list of GraphQL nodes."""
+        """
+        Retrieve a list of GraphQL nodes.
+
+        ..deprecated:: 4.5
+           Use :class:`NodeIterator` instead, which provides more functionality.
+        """
 
         def _query():
             query_variables['first'] = self._graphql_page_length
@@ -533,8 +471,8 @@ class InstaloaderContext:
 
         .. versionadded:: 4.2.1"""
         with copy_session(self._session, self.request_timeout) as tempsession:
-            tempsession.headers['User-Agent'] = 'Instagram 10.3.2 (iPhone7,2; iPhone OS 9_3_3; en_US; en-US; ' \
-                                                'scale=2.00; 750x1334) AppleWebKit/420+'
+            tempsession.headers['User-Agent'] = 'Instagram 123.1.0.26.115 (iPhone12,1; iOS 13_3; en_US; en-US; ' \
+                                                'scale=2.00; 1656x3584; 190542906)'
             for header in ['Host', 'Origin', 'X-Instagram-AJAX', 'X-Requested-With']:
                 tempsession.headers.pop(header, None)
             return self.get_json(path, params, 'i.instagram.com', tempsession)
@@ -544,11 +482,12 @@ class InstaloaderContext:
 
         .. versionadded:: 4.2.1"""
         self.log(filename, end=' ', flush=True)
-        with open(filename, 'wb') as file:
+        with open(filename + '.temp', 'wb') as file:
             if isinstance(resp, requests.Response):
                 shutil.copyfileobj(resp.raw, file)
             else:
                 file.write(resp)
+        os.rename(filename + '.temp', filename)
 
     def get_raw(self, url: str, _attempt=1) -> requests.Response:
         """Downloads a file anonymously.
@@ -590,3 +529,97 @@ class InstaloaderContext:
         if self._root_rhx_gis is None:
             self._root_rhx_gis = self.get_json('', {}).get('rhx_gis', '')
         return self._root_rhx_gis or None
+
+
+class RateController:
+    """
+    Class providing request tracking and rate controlling to stay within rate limits.
+
+    It can be overridden to change Instaloader's behavior regarding rate limits, for example to raise a custom
+    exception when the rate limit is hit::
+
+       import instaloader
+
+       class MyRateController(instaloader.RateController):
+           def sleep(self, secs):
+               raise MyCustomException()
+
+       L = instaloader.Instaloader(rate_controller=lambda ctx: MyRateController(ctx))
+    """
+
+    def __init__(self, context: InstaloaderContext):
+        self._context = context
+        self._graphql_query_timestamps = dict()  # type: Dict[str, List[float]]
+        self._graphql_earliest_next_request_time = 0.0
+
+    def sleep(self, secs: float):
+        """Wait given number of seconds."""
+        # Not static, to allow for the behavior of this method to depend on context-inherent properties, such as
+        # whether we are logged in.
+        # pylint:disable=no-self-use
+        time.sleep(secs)
+
+    def _dump_query_timestamps(self, current_time: float):
+        windows = [10, 11, 15, 20, 30, 60]
+        print("GraphQL requests:", file=sys.stderr)
+        for query_hash, times in self._graphql_query_timestamps.items():
+            print("  {}".format(query_hash), file=sys.stderr)
+            for window in windows:
+                reqs_in_sliding_window = sum(t > current_time - window * 60 for t in times)
+                print("    last {} minutes: {} requests".format(window, reqs_in_sliding_window), file=sys.stderr)
+
+    def count_per_sliding_window(self, query_type: str) -> int:
+        """Return how many GraphQL requests can be done within the sliding window."""
+        # Not static, to allow for the count_per_sliding_window to depend on context-inherent properties, such as
+        # whether we are logged in.
+        # pylint:disable=no-self-use,unused-argument
+        return 200
+
+    def query_waittime(self, query_type: str, current_time: float, untracked_queries: bool = False) -> float:
+        """Calculate time needed to wait before GraphQL query can be executed."""
+        sliding_window = 660
+        if query_type not in self._graphql_query_timestamps:
+            self._graphql_query_timestamps[query_type] = []
+        self._graphql_query_timestamps[query_type] = list(filter(lambda t: t > current_time - 60 * 60,
+                                                                 self._graphql_query_timestamps[query_type]))
+        reqs_in_sliding_window = list(filter(lambda t: t > current_time - sliding_window,
+                                             self._graphql_query_timestamps[query_type]))
+        count_per_sliding_window = self.count_per_sliding_window(query_type)
+        if len(reqs_in_sliding_window) < count_per_sliding_window and not untracked_queries:
+            return max(0.0, self._graphql_earliest_next_request_time - current_time)
+        next_request_time = min(reqs_in_sliding_window) + sliding_window + 6
+        if untracked_queries:
+            self._graphql_earliest_next_request_time = next_request_time
+        return max(next_request_time, self._graphql_earliest_next_request_time) - current_time
+
+    def wait_before_query(self, query_type: str) -> None:
+        """This method is called before a query to Instagram. It calls :meth:`RateController.sleep` to wait
+        until the request can be made."""
+        waittime = self.query_waittime(query_type, time.monotonic(), False)
+        assert waittime >= 0
+        if waittime > 15:
+            self._context.log("\nToo many queries in the last time. Need to wait {} seconds, until {:%H:%M}."
+                              .format(round(waittime), datetime.now() + timedelta(seconds=waittime)))
+        if waittime > 0:
+            self.sleep(waittime)
+        if query_type not in self._graphql_query_timestamps:
+            self._graphql_query_timestamps[query_type] = [time.monotonic()]
+        else:
+            self._graphql_query_timestamps[query_type].append(time.monotonic())
+
+    def handle_429(self, query_type: str) -> None:
+        """This method is called to handle a 429 Too Many Requests response. It calls :meth:`RateController.sleep` to
+         wait until we can repeat the same request."""
+        text_for_429 = ("HTTP error code 429 was returned because too many queries occurred in the last time. "
+                        "Please do not use Instagram in your browser or run multiple instances of Instaloader "
+                        "in parallel.")
+        print(textwrap.fill(text_for_429), file=sys.stderr)
+        current_time = time.monotonic()
+        waittime = self.query_waittime(query_type, current_time, True)
+        assert waittime >= 0
+        if waittime > 15:
+            self._context.log("The request will be retried in {} seconds, at {:%H:%M}."
+                              .format(round(waittime), datetime.now() + timedelta(seconds=waittime)))
+        self._dump_query_timestamps(current_time)
+        if waittime > 0:
+            self.sleep(waittime)

instaloader/nodeiterator.py

@@ -0,0 +1,261 @@
+import base64
+import hashlib
+import json
+import os
+from contextlib import contextmanager
+from datetime import datetime, timedelta
+from lzma import LZMAError
+from typing import Any, Callable, Dict, Iterator, NamedTuple, Optional, Tuple, TypeVar
+
+from .exceptions import InvalidArgumentException, QueryReturnedBadRequestException
+from .instaloadercontext import InstaloaderContext
+
+FrozenNodeIterator = NamedTuple('FrozenNodeIterator',
+                                [('query_hash', str),
+                                 ('query_variables', Dict),
+                                 ('query_referer', Optional[str]),
+                                 ('context_username', Optional[str]),
+                                 ('total_index', int),
+                                 ('best_before', Optional[float]),
+                                 ('remaining_data', Optional[Dict])])
+FrozenNodeIterator.__doc__ = \
+    """A serializable representation of a :class:`NodeIterator` instance, saving its iteration state."""
+FrozenNodeIterator.query_hash.__doc__ = """The GraphQL ``query_hash`` parameter."""
+FrozenNodeIterator.query_variables.__doc__ = """The GraphQL ``query_variables`` parameter."""
+FrozenNodeIterator.query_referer.__doc__ = """The HTTP referer used for the GraphQL query."""
+FrozenNodeIterator.context_username.__doc__ = """The username who created the iterator, or ``None``."""
+FrozenNodeIterator.total_index.__doc__ = """Number of items that have already been returned."""
+FrozenNodeIterator.best_before.__doc__ = """Date when parts of the stored nodes might have expired."""
+FrozenNodeIterator.remaining_data.__doc__ = \
+    """The already-retrieved, yet-unprocessed ``edges`` and the ``page_info`` at time of freezing."""
+
+
+T = TypeVar('T')
+
+
+class NodeIterator(Iterator[T]):
+    """
+    Iterate the nodes within edges in a GraphQL pagination. Instances of this class are returned by many (but not all)
+    of Instaloader's :class:`Post`-returning functions (such as :meth:`Profile.get_posts` etc.).
+
+    What makes this iterator special is its ability to freeze/store its current state, e.g. to interrupt an iteration,
+    and later thaw/resume from where it left off.
+
+    You can freeze a NodeIterator with :meth:`NodeIterator.freeze`::
+
+       post_iterator = profile.get_posts()
+       try:
+           for post in post_iterator:
+               do_something_with(post)
+       except KeyboardInterrupt:
+           save("resume_information.json", post_iterator.freeze())
+
+    and later reuse it with :meth:`NodeIterator.thaw` on an equally-constructed NodeIterator::
+
+       post_iterator = profile.get_posts()
+       post_iterator.thaw(load("resume_information.json"))
+
+    A :class:`FrozenNodeIterator` can only be thawn with a matching NodeIterator, i.e. a NodeIterator instance that has
+    been constructed with the same parameters as the instance that is represented by the :class:`FrozenNodeIterator` in
+    question. This is to ensure that an iteration cannot be resumed in a wrong, unmatching loop. As a quick way to
+    distinguish iterators that are saved e.g. in files, there is the :attr:`NodeIterator.magic` string: Two
+    NodeIterators are matching if they have the same magic.
+
+    See also :func:`resumable_iteration` for a high-level context manager that handles a resumable iteration.
+    """
+
+    _graphql_page_length = 50
+    shelf_life = timedelta(days=29)
+
+    def __init__(self,
+                 context: InstaloaderContext,
+                 query_hash: str,
+                 edge_extractor: Callable[[Dict[str, Any]], Dict[str, Any]],
+                 node_wrapper: Callable[[Dict], T],
+                 query_variables: Optional[Dict[str, Any]] = None,
+                 query_referer: Optional[str] = None,
+                 first_data: Optional[Dict[str, Any]] = None):
+        self._context = context
+        self._query_hash = query_hash
+        self._edge_extractor = edge_extractor
+        self._node_wrapper = node_wrapper
+        self._query_variables = query_variables if query_variables is not None else {}
+        self._query_referer = query_referer
+        self._data = first_data
+        self._page_index = 0
+        self._total_index = 0
+        self._best_before = (None if first_data is None else
+                             datetime.now() + NodeIterator.shelf_life)
+
+    def _query(self, after: Optional[str] = None) -> Dict:
+        pagination_variables = {'first': NodeIterator._graphql_page_length}  # type: Dict[str, Any]
+        if after is not None:
+            pagination_variables['after'] = after
+        try:
+            data = self._edge_extractor(
+                self._context.graphql_query(
+                    self._query_hash, {**self._query_variables, **pagination_variables}, self._query_referer
+                )
+            )
+            self._best_before = datetime.now() + NodeIterator.shelf_life
+            return data
+        except QueryReturnedBadRequestException:
+            new_page_length = int(NodeIterator._graphql_page_length / 2)
+            if new_page_length >= 12:
+                NodeIterator._graphql_page_length = new_page_length
+                self._context.error("HTTP Error 400 (Bad Request) on GraphQL Query. Retrying with shorter page length.",
+                                    repeat_at_end=False)
+                return self._query(after)
+            else:
+                raise
+
+    def __iter__(self):
+        return self
+
+    def __next__(self):
+        if self._data is None:
+            self._data = self._query()
+        if self._page_index < len(self._data['edges']):
+            node = self._data['edges'][self._page_index]['node']
+            page_index, total_index = self._page_index, self._total_index
+            try:
+                self._page_index += 1
+                self._total_index += 1
+            except KeyboardInterrupt:
+                self._page_index, self._total_index = page_index, total_index
+                raise
+            return self._node_wrapper(node)
+        if self._data['page_info']['has_next_page']:
+            query_response = self._query(self._data['page_info']['end_cursor'])
+            page_index, data = self._page_index, self._data
+            try:
+                self._page_index = 0
+                self._data = query_response
+            except KeyboardInterrupt:
+                self._page_index, self._data = page_index, data
+                raise
+            return self.__next__()
+        raise StopIteration()
+
+    @property
+    def count(self) -> Optional[int]:
+        """The ``count`` as returned by Instagram. This is not always the total count this iterator will yield."""
+        return self._data.get('count') if self._data is not None else None
+
+    @property
+    def total_index(self) -> int:
+        """Number of items that have already been returned."""
+        return self._total_index
+
+    @property
+    def magic(self) -> str:
+        """Magic string for easily identifying a matching iterator file for resuming (hash of some parameters)."""
+        if 'blake2b' not in hashlib.algorithms_available:
+            magic_hash = hashlib.new('sha224')
+        else:
+            # Use blake2b when possible, i.e. on Python >= 3.6.
+            magic_hash = hashlib.blake2b(digest_size=6)  # type:ignore  # pylint: disable=no-member
+        magic_hash.update(json.dumps(
+            [self._query_hash, self._query_variables, self._query_referer, self._context.username]
+        ).encode())
+        return base64.urlsafe_b64encode(magic_hash.digest()).decode()
+
+    def freeze(self) -> FrozenNodeIterator:
+        """Freeze the iterator for later resuming."""
+        remaining_data = None
+        if self._data is not None:
+            remaining_data = {**self._data,
+                              'edges': (self._data['edges'][(max(self._page_index - 1, 0)):])}
+        return FrozenNodeIterator(
+            query_hash=self._query_hash,
+            query_variables=self._query_variables,
+            query_referer=self._query_referer,
+            context_username=self._context.username,
+            total_index=max(self.total_index - 1, 0),
+            best_before=self._best_before.timestamp() if self._best_before else None,
+            remaining_data=remaining_data,
+        )
+
+    def thaw(self, frozen: FrozenNodeIterator) -> None:
+        """Use this iterator for resuming from earlier iteration."""
+        if self._total_index or self._page_index:
+            raise InvalidArgumentException("thaw() called on already-used iterator.")
+        if (self._query_hash != frozen.query_hash or
+                self._query_variables != frozen.query_variables or
+                self._query_referer != frozen.query_referer or
+                self._context.username != frozen.context_username):
+            raise InvalidArgumentException("Mismatching resume information.")
+        self._total_index = frozen.total_index
+        self._best_before = datetime.fromtimestamp(frozen.best_before) if frozen.best_before else None
+        self._data = frozen.remaining_data
+
+
+@contextmanager
+def resumable_iteration(context: InstaloaderContext,
+                        iterator: Iterator,
+                        load: Callable[[InstaloaderContext, str], Any],
+                        save: Callable[[FrozenNodeIterator, str], None],
+                        format_path: Callable[[str], str],
+                        check_bbd: bool = True,
+                        enabled: bool = True) -> Iterator[Tuple[bool, int]]:
+    """
+    High-level context manager to handle a resumable iteration that can be interrupted with a KeyboardInterrupt.
+
+    It can be used as follows to automatically load a previously-saved state into the iterator, save the iterator's
+    state when interrupted, and delete the resume file upon completion::
+
+       post_iterator = profile.get_posts()
+       with resumable_iteration(
+               context=L.context,
+               iterator=post_iterator,
+               load=lambda _, path: FrozenNodeIterator(**json.load(open(path))),
+               save=lambda fni, path: json.dump(fni._asdict(), open(path, 'w')),
+               format_path=lambda magic: "resume_info_{}.json".format(magic)
+       ) as resume_info:
+           is_resuming, start_index = resume_info
+           for post in post_iterator:
+               do_something_with(post)
+
+    It yields a tuple (is_resuming, start_index).
+
+    When the passed iterator is not a :class:`NodeIterator`, it behaves as if ``resumable_iteration`` was not used,
+    just executing the inner body.
+
+    :param context: The :class:`InstaloaderContext`.
+    :param iterator: The fresh :class:`NodeIterator`.
+    :param load: Loads a FrozenNodeIterator from given path. The object is ignored if it has a different type.
+    :param save: Saves the given FrozenNodeIterator to the given path.
+    :param format_path: Returns the path to the resume file for the given magic.
+    :param check_bbd: Whether to check the best before date and reject an expired FrozenNodeIterator.
+    :param enabled: Set to False to disable all functionality and simply execute the inner body.
+    """
+    if not enabled or not isinstance(iterator, NodeIterator):
+        yield False, 0
+        return
+    is_resuming = False
+    start_index = 0
+    resume_file_path = format_path(iterator.magic)
+    resume_file_exists = os.path.isfile(resume_file_path)
+    if resume_file_exists:
+        try:
+            fni = load(context, resume_file_path)
+            if not isinstance(fni, FrozenNodeIterator):
+                raise InvalidArgumentException("Invalid type.")
+            if check_bbd and fni.best_before and datetime.fromtimestamp(fni.best_before) < datetime.now():
+                raise InvalidArgumentException("\"Best before\" date exceeded.")
+            iterator.thaw(fni)
+            is_resuming = True
+            start_index = iterator.total_index
+            context.log("Resuming from {}.".format(resume_file_path))
+        except (InvalidArgumentException, LZMAError, json.decoder.JSONDecodeError) as exc:
+            context.error("Warning: Not resuming from {}: {}".format(resume_file_path, exc))
+    try:
+        yield is_resuming, start_index
+    except KeyboardInterrupt:
+        os.makedirs(os.path.dirname(resume_file_path), exist_ok=True)
+        save(iterator.freeze(), resume_file_path)
+        context.log("\nSaved resume information to {}.".format(resume_file_path))
+        raise
+    if resume_file_exists:
+        os.unlink(resume_file_path)
+        context.log("Iteration complete, deleted resume information file {}.".format(resume_file_path))

instaloader/structures.py

@@ -9,7 +9,7 @@ from typing import Any, Dict, Iterator, List, Optional, Union
 from . import __version__
 from .exceptions import *
 from .instaloadercontext import InstaloaderContext
-
+from .nodeiterator import FrozenNodeIterator, NodeIterator
 
 PostSidecarNode = namedtuple('PostSidecarNode', ['is_video', 'display_url', 'video_url'])
 PostSidecarNode.__doc__ = "Item of a Sidecar Post."
@@ -69,8 +69,11 @@ class Post:
         self._node = node
         self._owner_profile = owner_profile
         self._full_metadata_dict = None  # type: Optional[Dict[str, Any]]
-        self._rhx_gis_str = None         # type: Optional[str]
         self._location = None            # type: Optional[PostLocation]
+        self._iphone_struct_ = None
+        if 'iphone_struct' in node:
+            # if loaded from JSON with load_structure_from_file()
+            self._iphone_struct_ = node['iphone_struct']
 
     @classmethod
     def from_shortcode(cls, context: InstaloaderContext, shortcode: str):
@@ -111,6 +114,8 @@ class Post:
             node['owner'] = self.owner_profile._asdict()
         if self._location:
             node['location'] = self._location._asdict()
+        if self._iphone_struct_:
+            node['iphone_struct'] = self._iphone_struct_
         return node
 
     @property
@@ -136,9 +141,11 @@ class Post:
 
     def _obtain_metadata(self):
         if not self._full_metadata_dict:
-            pic_json = self._context.get_json("p/{0}/".format(self.shortcode), params={})
-            self._full_metadata_dict = pic_json['entry_data']['PostPage'][0]['graphql']['shortcode_media']
-            self._rhx_gis_str = pic_json.get('rhx_gis')
+            pic_json = self._context.graphql_query(
+                '2b0673e0dc4580674a88d426fe00ea90',
+                {'shortcode': self.shortcode}
+            )
+            self._full_metadata_dict = pic_json['data']['shortcode_media']
             if self._full_metadata_dict is None:
                 # issue #449
                 self._context.error("Fetching Post metadata failed (issue #449). "
@@ -156,9 +163,13 @@ class Post:
         return self._full_metadata_dict
 
     @property
-    def _rhx_gis(self) -> Optional[str]:
-        self._obtain_metadata()
-        return self._rhx_gis_str
+    def _iphone_struct(self) -> Dict[str, Any]:
+        if not self._context.is_logged_in:
+            raise LoginRequiredException("--login required to access iPhone media info endpoint.")
+        if not self._iphone_struct_:
+            data = self._context.get_iphone_json(path='api/v1/media/{}/info/'.format(self.mediaid), params={})
+            self._iphone_struct_ = data['items'][0]
+        return self._iphone_struct_
 
     def _field(self, *keys) -> Any:
         """Lookups given fields in _node, and if not found in _full_metadata. Raises KeyError if not found anywhere."""
@@ -225,6 +236,13 @@ class Post:
     @property
     def url(self) -> str:
         """URL of the picture / video thumbnail of the post"""
+        if self.typename == "GraphImage" and self._context.is_logged_in:
+            try:
+                orig_url = self._iphone_struct['image_versions2']['candidates'][0]['url']
+                url = re.sub(r'&se=\d+(&?)', r'\1', orig_url)
+                return url
+            except (InstaloaderException, KeyError, IndexError) as err:
+                self._context.error('{} Unable to fetch high quality image version of {}.'.format(err, self))
         return self._node["display_url"] if "display_url" in self._node else self._node["display_src"]
 
     @property
@@ -239,10 +257,18 @@ class Post:
             if any(edge['node']['is_video'] for edge in edges):
                 # video_url is only present in full metadata, issue #558.
                 edges = self._full_metadata['edge_sidecar_to_children']['edges']
-            for edge in edges:
+            for idx, edge in enumerate(edges):
                 node = edge['node']
                 is_video = node['is_video']
-                yield PostSidecarNode(is_video=is_video, display_url=node['display_url'],
+                display_url = node['display_url']
+                if not is_video and self._context.is_logged_in:
+                    try:
+                        carousel_media = self._iphone_struct['carousel_media']
+                        orig_url = carousel_media[idx]['image_versions2']['candidates'][0]['url']
+                        display_url = re.sub(r'&se=\d+(&?)', r'\1', orig_url)
+                    except (InstaloaderException, KeyError, IndexError) as err:
+                        self._context.error('{} Unable to fetch high quality image version of {}.'.format(err, self))
+                yield PostSidecarNode(is_video=is_video, display_url=display_url,
                                       video_url=node['video_url'] if is_video else None)
 
     @property
@@ -362,7 +388,7 @@ class Post:
                                      created_at_utc=datetime.utcfromtimestamp(node['created_at']),
                                      text=node['text'],
                                      owner=Profile(self._context, node['owner']),
-                                     likes_count=node['edge_liked_by']['count'])
+                                     likes_count=node.get('edge_liked_by', {}).get('count', 0))
 
         def _postcommentanswers(node):
             if 'edge_threaded_comments' not in node:
@@ -376,11 +402,14 @@ class Post:
                 # If the answer's metadata already contains all comments, don't do GraphQL requests to obtain them
                 yield from (_postcommentanswer(comment['node']) for comment in answer_edges)
                 return
-            yield from (_postcommentanswer(answer_node) for answer_node in
-                        self._context.graphql_node_list("51fdd02b67508306ad4484ff574a0b62",
-                                                        {'comment_id': node['id']},
-                                                        'https://www.instagram.com/p/' + self.shortcode + '/',
-                                                        lambda d: d['data']['comment']['edge_threaded_comments']))
+            yield from NodeIterator(
+                self._context,
+                '51fdd02b67508306ad4484ff574a0b62',
+                lambda d: d['data']['comment']['edge_threaded_comments'],
+                _postcommentanswer,
+                {'comment_id': node['id']},
+                'https://www.instagram.com/p/{0}/'.format(self.shortcode),
+            )
 
         def _postcomment(node):
             return PostComment(*_postcommentanswer(node),
@@ -388,29 +417,22 @@ class Post:
         if self.comments == 0:
             # Avoid doing additional requests if there are no comments
             return
-        try:
-            comment_edges = self._field('edge_media_to_parent_comment', 'edges')
-            answers_count = sum([edge['node']['edge_threaded_comments']['count'] for edge in comment_edges])
-            threaded_comments_available = True
-        except KeyError:
-            comment_edges = self._field('edge_media_to_comment', 'edges')
-            answers_count = 0
-            threaded_comments_available = False
+
+        comment_edges = self._field('edge_media_to_comment', 'edges')
+        answers_count = sum([edge['node'].get('edge_threaded_comments', {}).get('count', 0) for edge in comment_edges])
 
         if self.comments == len(comment_edges) + answers_count:
             # If the Post's metadata already contains all parent comments, don't do GraphQL requests to obtain them
             yield from (_postcomment(comment['node']) for comment in comment_edges)
             return
-        yield from (_postcomment(node) for node in
-                    self._context.graphql_node_list(
-                        "97b41c52301f77ce508f55e66d17620e" if threaded_comments_available
-                        else "f0986789a5c5d17c2400faebf16efd0d",
-                        {'shortcode': self.shortcode},
-                        'https://www.instagram.com/p/' + self.shortcode + '/',
-                        lambda d:
-                        d['data']['shortcode_media'][
-                            'edge_media_to_parent_comment' if threaded_comments_available else 'edge_media_to_comment'],
-                        self._rhx_gis))
+        yield from NodeIterator(
+            self._context,
+            '97b41c52301f77ce508f55e66d17620e',
+            lambda d: d['data']['shortcode_media']['edge_media_to_parent_comment'],
+            _postcomment,
+            {'shortcode': self.shortcode},
+            'https://www.instagram.com/p/{0}/'.format(self.shortcode),
+        )
 
     def get_likes(self) -> Iterator['Profile']:
         """Iterate over all likes of the post. A :class:`Profile` instance of each likee is yielded."""
@@ -422,11 +444,14 @@ class Post:
             # If the Post's metadata already contains all likes, don't do GraphQL requests to obtain them
             yield from (Profile(self._context, like['node']) for like in likes_edges)
             return
-        yield from (Profile(self._context, node) for node in
-                    self._context.graphql_node_list("1cb6ec562846122743b61e492c85999f", {'shortcode': self.shortcode},
-                                                    'https://www.instagram.com/p/' + self.shortcode + '/',
-                                                    lambda d: d['data']['shortcode_media']['edge_liked_by'],
-                                                    self._rhx_gis))
+        yield from NodeIterator(
+            self._context,
+            '1cb6ec562846122743b61e492c85999f',
+            lambda d: d['data']['shortcode_media']['edge_liked_by'],
+            lambda n: Profile(self._context, n),
+            {'shortcode': self.shortcode},
+            'https://www.instagram.com/p/{0}/'.format(self.shortcode),
+        )
 
     @property
     def is_sponsored(self) -> bool:
@@ -507,7 +532,6 @@ class Profile:
         self._has_public_story = None  # type: Optional[bool]
         self._node = node
         self._has_full_metadata = False
-        self._rhx_gis = None
         self._iphone_struct_ = None
         if 'iphone_struct' in node:
             # if loaded from JSON with load_structure_from_file()
@@ -569,10 +593,9 @@ class Profile:
     def _obtain_metadata(self):
         try:
             if not self._has_full_metadata:
-                metadata = self._context.get_json('{}/'.format(self.username), params={})
+                metadata = self._context.get_json('{}/feed/'.format(self.username), params={})
                 self._node = metadata['entry_data']['ProfilePage'][0]['graphql']['user']
                 self._has_full_metadata = True
-                self._rhx_gis = metadata.get('rhx_gis')
         except (QueryReturnedNotFoundException, KeyError) as err:
             top_search_results = TopSearchResults(self._context, self.username)
             similar_profiles = [profile.username for profile in top_search_results.get_profiles()]
@@ -654,6 +677,16 @@ class Profile:
     def external_url(self) -> Optional[str]:
         return self._metadata('external_url')
 
+    @property
+    def is_business_account(self) -> bool:
+        """.. versionadded:: 4.4"""
+        return self._metadata('is_business_account')
+
+    @property
+    def business_category_name(self) -> str:
+        """.. versionadded:: 4.4"""
+        return self._metadata('business_category_name')
+
     @property
     def biography(self) -> str:
         return self._metadata('biography')
@@ -695,8 +728,7 @@ class Profile:
                                                         'include_reel': False, 'include_suggested_users': False,
                                                         'include_logged_out_extras': True,
                                                         'include_highlight_reels': False},
-                                                       'https://www.instagram.com/{}/'.format(self.username),
-                                                       self._rhx_gis)
+                                                       'https://www.instagram.com/{}/'.format(self.username))
             self._has_public_story = data['data']['user']['has_public_story']
         assert self._has_public_story is not None
         return self._has_public_story
@@ -747,86 +779,110 @@ class Profile:
 	   Use :attr:`profile_pic_url`."""
         return self.profile_pic_url
 
-    def get_posts(self) -> Iterator[Post]:
-        """Retrieve all posts from a profile."""
-        self._obtain_metadata()
-        yield from (Post(self._context, node, self) for node in
-                    self._context.graphql_node_list("472f257a40c653c64c666ce877d59d2b",
-                                                    {'id': self.userid},
-                                                    'https://www.instagram.com/{0}/'.format(self.username),
-                                                    lambda d: d['data']['user']['edge_owner_to_timeline_media'],
-                                                    self._rhx_gis,
-                                                    self._metadata('edge_owner_to_timeline_media')))
+    def get_posts(self) -> NodeIterator[Post]:
+        """Retrieve all posts from a profile.
 
-    def get_saved_posts(self) -> Iterator[Post]:
-        """Get Posts that are marked as saved by the user."""
+        :rtype:NodeIterator[Post]"""
+        self._obtain_metadata()
+        return NodeIterator(
+            self._context,
+            '472f257a40c653c64c666ce877d59d2b',
+            lambda d: d['data']['user']['edge_owner_to_timeline_media'],
+            lambda n: Post(self._context, n, self),
+            {'id': self.userid},
+            'https://www.instagram.com/{0}/'.format(self.username),
+            self._metadata('edge_owner_to_timeline_media'),
+        )
+
+    def get_saved_posts(self) -> NodeIterator[Post]:
+        """Get Posts that are marked as saved by the user.
+
+        :rtype:NodeIterator[Post]"""
 
         if self.username != self._context.username:
             raise LoginRequiredException("--login={} required to get that profile's saved posts.".format(self.username))
 
         self._obtain_metadata()
-        yield from (Post(self._context, node) for node in
-                    self._context.graphql_node_list("f883d95537fbcd400f466f63d42bd8a1",
-                                                    {'id': self.userid},
-                                                    'https://www.instagram.com/{0}/'.format(self.username),
-                                                    lambda d: d['data']['user']['edge_saved_media'],
-                                                    self._rhx_gis,
-                                                    self._metadata('edge_saved_media')))
+        return NodeIterator(
+            self._context,
+            'f883d95537fbcd400f466f63d42bd8a1',
+            lambda d: d['data']['user']['edge_saved_media'],
+            lambda n: Post(self._context, n),
+            {'id': self.userid},
+            'https://www.instagram.com/{0}/'.format(self.username),
+            self._metadata('edge_saved_media'),
+        )
 
-    def get_tagged_posts(self) -> Iterator[Post]:
+    def get_tagged_posts(self) -> NodeIterator[Post]:
         """Retrieve all posts where a profile is tagged.
 
+        :rtype: NodeIterator[Post]
+
         .. versionadded:: 4.0.7"""
         self._obtain_metadata()
-        yield from (Post(self._context, node, self if int(node['owner']['id']) == self.userid else None) for node in
-                    self._context.graphql_node_list("e31a871f7301132ceaab56507a66bbb7",
-                                                    {'id': self.userid},
-                                                    'https://www.instagram.com/{0}/'.format(self.username),
-                                                    lambda d: d['data']['user']['edge_user_to_photos_of_you'],
-                                                    self._rhx_gis))
+        return NodeIterator(
+            self._context,
+            'e31a871f7301132ceaab56507a66bbb7',
+            lambda d: d['data']['user']['edge_user_to_photos_of_you'],
+            lambda n: Post(self._context, n, self if int(n['owner']['id']) == self.userid else None),
+            {'id': self.userid},
+            'https://www.instagram.com/{0}/'.format(self.username),
+        )
 
-    def get_igtv_posts(self) -> Iterator[Post]:
+    def get_igtv_posts(self) -> NodeIterator[Post]:
         """Retrieve all IGTV posts.
 
+        :rtype: NodeIterator[Post]
+
         .. versionadded:: 4.3"""
         self._obtain_metadata()
-        yield from (Post(self._context, node, self) for node in
-                    self._context.graphql_node_list('bc78b344a68ed16dd5d7f264681c4c76',
-                                                    {'id': self.userid},
-                                                    'https://www.instagram.com/{0}/channel/'.format(self.username),
-                                                    lambda d: d['data']['user']['edge_felix_video_timeline'],
-                                                    self._rhx_gis,
-                                                    self._metadata('edge_felix_video_timeline')))
+        return NodeIterator(
+            self._context,
+            'bc78b344a68ed16dd5d7f264681c4c76',
+            lambda d: d['data']['user']['edge_felix_video_timeline'],
+            lambda n: Post(self._context, n, self),
+            {'id': self.userid},
+            'https://www.instagram.com/{0}/channel/'.format(self.username),
+            self._metadata('edge_felix_video_timeline'),
+        )
 
-    def get_followers(self) -> Iterator['Profile']:
+    def get_followers(self) -> NodeIterator['Profile']:
         """
         Retrieve list of followers of given profile.
         To use this, one needs to be logged in and private profiles has to be followed.
+
+        :rtype:NodeIterator[Profile]
         """
         if not self._context.is_logged_in:
             raise LoginRequiredException("--login required to get a profile's followers.")
         self._obtain_metadata()
-        yield from (Profile(self._context, node) for node in
-                    self._context.graphql_node_list("37479f2b8209594dde7facb0d904896a",
-                                                    {'id': str(self.userid)},
-                                                    'https://www.instagram.com/' + self.username + '/',
-                                                    lambda d: d['data']['user']['edge_followed_by'],
-                                                    self._rhx_gis))
+        return NodeIterator(
+            self._context,
+            '37479f2b8209594dde7facb0d904896a',
+            lambda d: d['data']['user']['edge_followed_by'],
+            lambda n: Profile(self._context, n),
+            {'id': str(self.userid)},
+            'https://www.instagram.com/{0}/'.format(self.username),
+        )
 
-    def get_followees(self) -> Iterator['Profile']:
+    def get_followees(self) -> NodeIterator['Profile']:
         """
         Retrieve list of followees (followings) of given profile.
         To use this, one needs to be logged in and private profiles has to be followed.
+
+        :rtype:NodeIterator[Profile]
         """
         if not self._context.is_logged_in:
             raise LoginRequiredException("--login required to get a profile's followees.")
         self._obtain_metadata()
-        yield from (Profile(self._context, node) for node in
-                    self._context.graphql_node_list("58712303d941c6855d4e888c5f0cd22f",
-                                                    {'id': str(self.userid)},
-                                                    'https://www.instagram.com/' + self.username + '/',
-                                                    lambda d: d['data']['user']['edge_follow'],
-                                                    self._rhx_gis))
+        return NodeIterator(
+            self._context,
+            '58712303d941c6855d4e888c5f0cd22f',
+            lambda d: d['data']['user']['edge_follow'],
+            lambda n: Profile(self._context, n),
+            {'id': str(self.userid)},
+            'https://www.instagram.com/{0}/'.format(self.username),
+        )
 
     def get_similar_accounts(self) -> Iterator['Profile']:
         """
@@ -841,8 +897,8 @@ class Profile:
         yield from (Profile(self._context, edge["node"]) for edge in
                     self._context.graphql_query("ad99dd9d3646cc3c0dda65debcd266a7",
                                                 {"user_id": str(self.userid), "include_chaining": True},
-                                                "https://www.instagram.com/{0}/".format(self.username),
-                                                self._rhx_gis)["data"]["user"]["edge_chaining"]["edges"])
+                                                "https://www.instagram.com/{0}/"
+                                                .format(self.username))["data"]["user"]["edge_chaining"]["edges"])
 
 
 class StoryItem:
@@ -1381,7 +1437,7 @@ class TopSearchResults:
         return self._searchstring
 
 
-JsonExportable = Union[Post, Profile, StoryItem, Hashtag]
+JsonExportable = Union[Post, Profile, StoryItem, Hashtag, FrozenNodeIterator]
 
 
 def save_structure_to_file(structure: JsonExportable, filename: str) -> None:
@@ -1430,6 +1486,8 @@ def load_structure_from_file(context: InstaloaderContext, filename: str) -> Json
             return StoryItem(context, json_structure['node'])
         elif node_type == "Hashtag":
             return Hashtag(context, json_structure['node'])
+        elif node_type == "FrozenNodeIterator":
+            return FrozenNodeIterator(**json_structure['node'])
         else:
             raise InvalidArgumentException("{}: Not an Instaloader JSON.".format(filename))
     elif 'shortcode' in json_structure:

test/instaloader_unittests.py

@@ -5,7 +5,7 @@ import shutil
 import tempfile
 import unittest
 from itertools import islice
-from typing import Dict, List
+from typing import Optional
 
 import instaloader
 
@@ -23,8 +23,7 @@ PRIVATE_PROFILE_ID = 1706625676
 EMPTY_PROFILE = "not_public"
 EMPTY_PROFILE_ID = 1928659031
 
-# Preserve query timestamps (rate control) between tests to not get rate limited
-instaloadercontext_query_timestamps = dict()  # type: Dict[str, List[float]]
+ratecontroller = None  # type: Optional[instaloader.RateController]
 
 
 class TestInstaloaderAnonymously(unittest.TestCase):
@@ -37,13 +36,15 @@ class TestInstaloaderAnonymously(unittest.TestCase):
                                          download_comments=True,
                                          save_metadata=True)
         self.L.context.raise_all_errors = True
-        # pylint:disable=protected-access
-        self.L.context._graphql_query_timestamps = instaloadercontext_query_timestamps.copy()
+        if ratecontroller is not None:
+            # pylint:disable=protected-access
+            ratecontroller._context = self.L.context
+            self.L.context._rate_controller = ratecontroller
 
     def tearDown(self):
         # pylint:disable=global-statement,protected-access
-        global instaloadercontext_query_timestamps
-        instaloadercontext_query_timestamps = self.L.context._graphql_query_timestamps.copy()
+        global ratecontroller
+        ratecontroller = self.L.context._rate_controller
         self.L.close()
         os.chdir('/')
         print("Removing {}".format(self.dir))
@@ -94,6 +95,10 @@ class TestInstaloaderAnonymously(unittest.TestCase):
         self.assertEqual(EMPTY_PROFILE.lower(),
                          instaloader.Profile.from_id(self.L.context, EMPTY_PROFILE_ID).username)
 
+    def test_get_username_by_name_empty(self):
+        self.assertEqual(EMPTY_PROFILE_ID,
+                         instaloader.Profile.from_username(self.L.context, EMPTY_PROFILE).userid)
+
     def test_post_from_mediaid(self):
         for post in instaloader.Profile.from_username(self.L.context, PUBLIC_PROFILE).get_posts():
             post2 = instaloader.Post.from_mediaid(self.L.context, post.mediaid)