har2tree 1.35.1__py3-none-any.whl → 1.36.1__py3-none-any.whl

har2tree/har2tree.py

@@ -750,23 +750,47 @@ class Har2Tree:
             self.logger.warning(f'Wrong format for the frames ({type(self.har.frames)}), very old capture.')
         return self.url_tree
 
+    def _guess_best_node_for_partial_referer(self, node: URLNode, potential_parents: list[URLNode]) -> URLNode:
+        # we have more than one node with the hostname of the referer *and* content.
+        # 2025-12-17:
+        # 1. find the deepest HTML node in the list
+        for pp in reversed(potential_parents):
+            if 'html' in pp.mimetype:
+                return pp
+        else:
+            # 2. if there are no HTML node anywhere in the list, attach to the deepest node
+            return potential_parents[-1]
+
     @trace_make_subtree_fallback
     def _make_subtree_fallback(self, node: URLNode, dev_debug: bool=False) -> None:
         if hasattr(node, 'referer'):
             # 2022-04-28: the node has a referer, but for some reason, it could't be attached to the tree
             #             Probable reason: the referer is a part of the URL (hostname)
-            # FIXME: this is a very dirty fix, but I'm not sure we can do it any better
             if (referer_hostname := urlparse(node.referer).hostname):
-                # the referer has a hostname
                 if (nodes_with_hostname := self.url_tree.search_nodes(hostname=referer_hostname)):
-                    # the hostname has at least a node in the tree
-                    for node_with_hostname in nodes_with_hostname:
-                        if not node_with_hostname.empty_response:
-                            # we got an non-empty response, breaking
-                            break
-                    # attach to the the first response with something, or to whatever we get.
-                    self._make_subtree(node_with_hostname, [node], fallback=True)
-                    return
+                    attach_to: URLNode
+                    # 2025-12-17: we have at least one node with that hostname.
+                    if len(nodes_with_hostname) == 1:
+                        # That's the only one, use it
+                        attach_to = nodes_with_hostname[0]
+                    else:
+                        # check if there are empty nodes
+                        if (nodes_with_hostname_and_response := [n for n in nodes_with_hostname if not n.empty_response]):
+                            if len(nodes_with_hostname_and_response) == 1:
+                                attach_to = nodes_with_hostname_and_response[0]
+                            else:
+                                # multiple non-empty nodes with that hostname, this is the more difficult one
+                                attach_to = self._guess_best_node_for_partial_referer(node, nodes_with_hostname_and_response)
+                        else:
+                            # more than one node with that hostname, but they're all empty, attach to the first one
+                            attach_to = nodes_with_hostname[0]
+                    return self._make_subtree(attach_to, [node], fallback=True)
+                else:
+                    # no node with that hostname at all, this should not happen
+                    self.logger.warning(f'Unable to find any node with the hostname {referer_hostname}, despites it being set as referer.')
+            else:
+                # the referer has no hostname and it is fascinating
+                self.logger.warning(f'Unable to get hostname out of referer: {node.referer}')
 
         # Sometimes, the har has a list of pages, generally when we have HTTP redirects.
         # IF we have more than one page in the list
@@ -929,7 +953,6 @@ class Har2Tree:
                     if dev_debug:
                         self.logger.warning(f'Found via initiator from {unode.name} to {matching_urls}.')
                     self._make_subtree(unode, matching_urls)
-
             # The node can have a redirect, but also trigger ressources refering to themselves, we need to trigger this code on each node.
             if self.all_initiator_url.get(unode.name):
                 # The URL (unode.name) is in the list of known urls initiating calls

{har2tree-1.35.1.dist-info → har2tree-1.36.1.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: har2tree
-Version: 1.35.1
+Version: 1.36.1
 Summary: HTTP Archive (HAR) to ETE Toolkit generator
 License-Expression: BSD-3-Clause
 License-File: LICENSE
@@ -20,8 +20,8 @@ Classifier: Programming Language :: Python :: 3.14
 Classifier: Topic :: Internet
 Classifier: Topic :: Security
 Provides-Extra: docs
-Requires-Dist: Sphinx (>=8.2.3) ; (python_version >= "3.11") and (extra == "docs")
-Requires-Dist: beautifulsoup4[charset-normalizer,lxml] (>=4.14.2)
+Requires-Dist: Sphinx (>=9.0.4) ; (python_version >= "3.11") and (extra == "docs")
+Requires-Dist: beautifulsoup4[charset-normalizer,lxml] (>=4.14.3)
 Requires-Dist: ete3 (>=3.1.3)
 Requires-Dist: filetype (>=1.2.0)
 Requires-Dist: json-stream (>=2.3.3,<3.0.0)
@@ -29,10 +29,10 @@ Requires-Dist: legacy-cgi (>=2.6.4) ; python_version >= "3.13" and python_versio
 Requires-Dist: multipart (>=1.3.0,<2.0.0)
 Requires-Dist: numpy (>=2.2,<2.3) ; python_version < "3.11"
 Requires-Dist: numpy (>=2.3.5) ; python_version >= "3.11" and python_version < "4.0"
-Requires-Dist: publicsuffixlist (>=1.0.2.20251119)
+Requires-Dist: publicsuffixlist (>=1.0.2.20251217)
 Requires-Dist: requests-toolbelt (>=1.0.0,<2.0.0)
 Requires-Dist: six (>=1.17.0) ; extra == "docs"
-Requires-Dist: tinycss2 (>=1.5.0)
+Requires-Dist: tinycss2 (>=1.5.1)
 Requires-Dist: w3lib (>=2.3.1)
 Project-URL: Documentation, https://har2tree.readthedocs.io/en/latest/
 Project-URL: Repository, https://github.com/Lookyloo/har2tree

{har2tree-1.35.1.dist-info → har2tree-1.36.1.dist-info}/RECORD

@@ -1,10 +1,10 @@
 har2tree/__init__.py,sha256=Na3mxHkUBq3rzYbxiLNJF37DxH5mcghSorjzXw5Teug,422
-har2tree/har2tree.py,sha256=24Puk4dlDXWOVIAPV7SIXNoP-oP-_7ERH2mZPxXiwn8,52762
+har2tree/har2tree.py,sha256=4SYegFw2ycH6kJTkUTTrJb2f6D4ekITEkyZ-4it1PsA,54280
 har2tree/helper.py,sha256=ktX5Fq-K_t4r0VVAXIH4uy7xc-qCjtSaiUvkX_PYxhw,20737
 har2tree/nodes.py,sha256=QWKqEUnuW7J6pASVvzwWAQNqL-_KDzSs2ld6uJl3qbw,37710
 har2tree/parser.py,sha256=4yej1OcVYAIiLfzYZsO9WCw3WyM_ykDTuvpW7UO1ROE,3645
 har2tree/py.typed,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-har2tree-1.35.1.dist-info/METADATA,sha256=SGudpw_P8Wrs_x9uCJmL__fJciNsWVeDlx8rFaVurVM,2239
-har2tree-1.35.1.dist-info/WHEEL,sha256=zp0Cn7JsFoX2ATtOhtaFYIiE2rmFAD4OcMhtUki8W3U,88
-har2tree-1.35.1.dist-info/licenses/LICENSE,sha256=Xa4EVROgJsEo10CW-ISCRiw0TtqdKz1JuM3BBLBM55c,1803
-har2tree-1.35.1.dist-info/RECORD,,
+har2tree-1.36.1.dist-info/METADATA,sha256=QuSZMYKNtpaVwtP8AVjFk_xFS1g0_kmPJ5V78nIQV_8,2239
+har2tree-1.36.1.dist-info/WHEEL,sha256=zp0Cn7JsFoX2ATtOhtaFYIiE2rmFAD4OcMhtUki8W3U,88
+har2tree-1.36.1.dist-info/licenses/LICENSE,sha256=Xa4EVROgJsEo10CW-ISCRiw0TtqdKz1JuM3BBLBM55c,1803
+har2tree-1.36.1.dist-info/RECORD,,