From 740f542b892f391fb4ed020d02e8fd87c7657101 Mon Sep 17 00:00:00 2001
From: Andrea Cecchi <andrea.cecchi85@gmail.com>
Date: Mon, 23 Sep 2024 10:11:43 +0200
Subject: [PATCH] Find broken links view (#115)

* Add view find-broken-links that return a csv file with a list of contents with broken links in blocks

* flake 8

* code cleanup
---
 CHANGES.rst                                   |   3 +-
 README.rst                                    |   4 +
 src/redturtle/volto/browser/configure.zcml    |   8 +-
 .../volto/browser/find_broken_links.py        | 130 ++++++++++++++++++
 .../tests/test_find_broken_links_view.py      |  78 +++++++++++
 5 files changed, 221 insertions(+), 2 deletions(-)
 create mode 100644 src/redturtle/volto/browser/find_broken_links.py
 create mode 100644 src/redturtle/volto/tests/test_find_broken_links_view.py

diff --git a/CHANGES.rst b/CHANGES.rst
index e3c9980..d4ebaaf 100644
--- a/CHANGES.rst
+++ b/CHANGES.rst
@@ -4,7 +4,8 @@ Changelog
 5.5.3 (unreleased)
 ------------------
 
-- Nothing changed yet.
+- Add view "find-broken-links" that return a csv file with a list of contents with broken links in blocks.
+  [cekk]
 
 
 5.5.2 (2024-09-05)
diff --git a/README.rst b/README.rst
index 22a08b0..dfe18c1 100644
--- a/README.rst
+++ b/README.rst
@@ -280,6 +280,10 @@ Find blocks
 
 There is a view **@@find-blocks** that will return contents that have at least one block of the given type.
 
+Find broken links
+=================
+
+There is a view **@@find-broken-links** that will return a csv file with a list of contents with broken internal links in blocks.
 
 Stringinterp adapters
 =====================
diff --git a/src/redturtle/volto/browser/configure.zcml b/src/redturtle/volto/browser/configure.zcml
index 98cdbe8..126261b 100644
--- a/src/redturtle/volto/browser/configure.zcml
+++ b/src/redturtle/volto/browser/configure.zcml
@@ -57,7 +57,13 @@
       permission="cmf.ManagePortal"
       layer="redturtle.volto.interfaces.IRedturtleVoltoLayer"
       />
-
+  <browser:page
+      name="find-broken-links"
+      for="*"
+      class=".find_broken_links.View"
+      permission="cmf.ManagePortal"
+      layer="redturtle.volto.interfaces.IRedturtleVoltoLayer"
+      />
   <browser:page
       name="sitemap.xml.gz"
       for="plone.app.layout.navigation.interfaces.INavigationRoot"
diff --git a/src/redturtle/volto/browser/find_broken_links.py b/src/redturtle/volto/browser/find_broken_links.py
new file mode 100644
index 0000000..29ce08f
--- /dev/null
+++ b/src/redturtle/volto/browser/find_broken_links.py
@@ -0,0 +1,130 @@
+from Acquisition import aq_base
+from plone import api
+from plone.dexterity.utils import iterSchemata
+from plone.restapi.serializer.utils import uid_to_url
+from Products.Five import BrowserView
+from six import StringIO
+from zope.schema import getFieldsInOrder
+
+
+try:
+    from collective.volto.blocksfield.field import BlocksField
+
+    HAS_BLOCKSFIELD = True
+except ImportError:
+    HAS_BLOCKSFIELD = False
+
+import csv
+import logging
+
+
+logger = logging.getLogger(__name__)
+
+
+class View(BrowserView):
+    def __call__(self):
+        """
+        Check all contents if there are some internal links with resolveuid broken
+        """
+
+        results = self.check_links()
+        self.request.response.setHeader("Content-type", "application/csv")
+        self.request.response.setHeader(
+            "Content-disposition", "attachment; filename=incarichi.csv"
+        )
+
+        sbuf = StringIO()
+        writer = csv.writer(sbuf, delimiter=" ", quoting=csv.QUOTE_ALL)
+        writer.writerow(["url"])
+        for row in results:
+            writer.writerow([row])
+
+        res = sbuf.getvalue()
+        sbuf.close()
+        return res
+
+    def check_links(self):
+        """
+        Check on root and all contents
+        """
+        res = []
+        # first of all, check them on root
+        res.extend(self.check_links_on_root())
+        # then check on contents
+        res.extend(self.check_links_on_contents())
+        return res
+
+    def check_links_on_root(self):
+        """
+        Check root blocks
+        """
+        logger.info("## Check broken links on Site Root ##")
+        portal = api.portal.get()
+        blocks = getattr(portal, "blocks", {})
+
+        if self.check_blocks_broken_links(data=blocks):
+            return [portal.portal_url()]
+        return []
+
+    def check_links_on_contents(self):
+        """
+        Iterate over site contents
+        """
+        logger.info("## Check broken links on Content-types ##")
+        portal_catalog = api.portal.get_tool("portal_catalog")
+        brains = portal_catalog()
+        tot = len(brains)
+        i = 0
+        res = []
+        for brain in brains:
+            i += 1
+            if (i + 1) % 200 == 0:
+                logger.info(f" - Progress {i}/{tot}")
+            item = brain.getObject()
+            aq_base_obj = aq_base(item)
+
+            blocks = getattr(item, "blocks", {})
+            if blocks:
+                if self.check_blocks_broken_links(data=blocks):
+                    res.append(brain.getURL())
+                    continue
+            if not HAS_BLOCKSFIELD:
+                continue
+            has_broken_links = False
+            for schemata in iterSchemata(aq_base_obj):
+                for name, field in getFieldsInOrder(schemata):
+                    if not isinstance(field, BlocksField):
+                        continue
+                    value = field.get(item)
+                    if not value:
+                        continue
+                    blocks = value.get("blocks", {})
+                    if self.check_blocks_broken_links(data=blocks):
+                        has_broken_links = True
+                        break
+            if has_broken_links:
+                res.append(brain.getURL())
+        return res
+
+    def check_blocks_broken_links(self, data):
+        """
+        Recursive method that check if there is a broken resolveuid in a block prop
+        """
+        if isinstance(data, str):
+            if "resolveuid" not in data:
+                return False
+            if uid_to_url(data) == data:
+                return True
+            return False
+        if isinstance(data, list):
+            for child in data:
+                res = self.check_blocks_broken_links(data=child)
+                if res:
+                    return True
+            return False
+        if isinstance(data, dict):
+            for child in data.values():
+                res = self.check_blocks_broken_links(data=child)
+                if res:
+                    return True
+        return False
diff --git a/src/redturtle/volto/tests/test_find_broken_links_view.py b/src/redturtle/volto/tests/test_find_broken_links_view.py
new file mode 100644
index 0000000..dee455e
--- /dev/null
+++ b/src/redturtle/volto/tests/test_find_broken_links_view.py
@@ -0,0 +1,78 @@
+# -*- coding: utf-8 -*-
+from plone import api
+from plone.app.testing import setRoles
+from plone.app.testing import TEST_USER_ID
+from redturtle.volto.testing import REDTURTLE_VOLTO_INTEGRATION_TESTING
+
+import unittest
+
+
+class TestFindBrokenLinksView(unittest.TestCase):
+    layer = REDTURTLE_VOLTO_INTEGRATION_TESTING
+    maxDiff = None
+
+    def setUp(self):
+        self.app = self.layer["app"]
+        self.portal = self.layer["portal"]
+        self.request = self.layer["request"]
+        setRoles(self.portal, TEST_USER_ID, ["Manager"])
+
+        self.view = api.content.get_view(
+            name="find-broken-links", context=self.portal, request=self.request
+        )
+
+    def test_view_does_not_return_contents_if_have_working_links(self):
+        page_a = api.content.create(
+            container=self.portal, type="Document", title="Page A"
+        )
+
+        api.content.create(
+            container=self.portal,
+            type="Document",
+            title="Page B",
+            blocks={"xxx": {"foo": f"resolveuid/{page_a.UID()}"}},
+        )
+
+        self.assertEqual([], self.view.check_links())
+
+    def test_view_return_contents_with_broken_links(self):
+        page_a = api.content.create(
+            container=self.portal, type="Document", title="Page A"
+        )
+
+        page_b = api.content.create(
+            container=self.portal,
+            type="Document",
+            title="Page B",
+            blocks={"xxx": {"foo": f"resolveuid/{page_a.UID()}"}},
+        )
+        api.content.delete(obj=page_a, check_linkintegrity=False)
+
+        res = self.view.check_links()
+        self.assertEqual(len(res), 1)
+        self.assertIn(page_b.absolute_url(), res)
+
+    def test_view_can_check_several_patterns(self):
+        page_a = api.content.create(
+            container=self.portal, type="Document", title="Page A"
+        )
+
+        page_b = api.content.create(
+            container=self.portal,
+            type="Document",
+            title="Page B",
+            blocks={"xxx": {"bar": f"../resolveuid/{page_a.UID()}"}},
+        )
+
+        page_c = api.content.create(
+            container=self.portal,
+            type="Document",
+            title="Page C",
+            blocks={"xxx": {"baz": f"../resolveuid/{page_a.UID()}/asd"}},
+        )
+        api.content.delete(obj=page_a, check_linkintegrity=False)
+
+        res = self.view.check_links()
+        self.assertEqual(len(res), 2)
+        self.assertIn(page_b.absolute_url(), res)
+        self.assertIn(page_c.absolute_url(), res)