har2tree 1.34.2__tar.gz → 1.35.0__tar.gz

{har2tree-1.34.2 → har2tree-1.35.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: har2tree
-Version: 1.34.2
+Version: 1.35.0
 Summary: HTTP Archive (HAR) to ETE Toolkit generator
 License-Expression: BSD-3-Clause
 License-File: LICENSE
@@ -28,7 +28,7 @@ Requires-Dist: json-stream (>=2.3.3,<3.0.0)
 Requires-Dist: legacy-cgi (>=2.6.4) ; python_version >= "3.13" and python_version < "4.0"
 Requires-Dist: multipart (>=1.3.0,<2.0.0)
 Requires-Dist: numpy (>=2.2,<2.3) ; python_version < "3.11"
-Requires-Dist: numpy (>=2.3.4) ; python_version >= "3.11" and python_version < "4.0"
+Requires-Dist: numpy (>=2.3.5) ; python_version >= "3.11" and python_version < "4.0"
 Requires-Dist: publicsuffixlist (>=1.0.2.20251115)
 Requires-Dist: requests-toolbelt (>=1.0.0,<2.0.0)
 Requires-Dist: six (>=1.17.0) ; extra == "docs"

{har2tree-1.34.2 → har2tree-1.35.0}/har2tree/har2tree.py

@@ -426,12 +426,20 @@ class Har2Tree:
 
         self.url_tree = self._nodes_list.pop(0)
 
+    def _url_to_local_only_content(self, url: str | None) -> bool:
+        return (url is None
+                or url in ['about:blank', 'about:srcdoc', '']  # not loading anything remotely
+                or url.startswith('data')  # base64 encoded content
+                or url.startswith('chrome-error')  # not in the HAR/tree
+                or url.startswith('blob')  # blobs aren't URLs
+                )
+
     def _load_iframes(self, current: URLNode, frames: FramesResponse) -> None:
         if not frames.get('content') or frames['content'] is None:
             # NOTE: debug stuff, no content makes it pretty useless.
             if frames.get('url'):
-                if frames['url'] == "about:blank":
-                    self.logger.info('Got a frame to about:blank with no content.')
+                if self._url_to_local_only_content(frames['url']):
+                    self.logger.info('Got an empty frame to local content.')
                 else:
                     u = unquote_plus(frames['url'])
                     self.logger.warning(f'Got a url ({u}) for the frame, but no content')
@@ -439,12 +447,7 @@ class Har2Tree:
                 self.logger.info('Got a frame, but no content.')
             return
 
-        if (frames.get('url')
-                and not (frames['url'] in ['about:blank']  # not loading anything, same as empty
-                         or frames['url'].startswith('data')  # base64 encoded content
-                         or frames['url'].startswith('chrome-error')  # not in the HAR/tree
-                         or frames['url'].startswith('blob')  # blobs aren't URLs
-                         )):
+        if frames.get('url') and not self._url_to_local_only_content(frames['url']):
             u = unquote_plus(frames['url'])
             possible_child_name = {u, u.split('#', 1)[0]}
             # this url should be in a node directly attached to that one
@@ -821,11 +824,7 @@ class Har2Tree:
 
     def all_real_urls_in_children(self, frame: FramesResponse) -> Iterator[str]:
         # from a frame, search all the real urls in each of the children, stop at the first one
-        if (frame.get('url') and frame['url'] is not None
-                and not (frame['url'] in ['about:blank', 'about:srcdoc']  # not loading anything, same as empty
-                         or frame['url'].startswith('data')  # base64 encoded content
-                         or frame['url'].startswith('chrome-error')  # not in the HAR/tree
-                         or frame['url'].startswith('blob'))):  # blobs aren't URLs
+        if (frame.get('url') and frame['url'] is not None and not self._url_to_local_only_content(frame['url'])):
             yield frame['url']
         else:
             # got no real URL, try the children

{har2tree-1.34.2 → har2tree-1.35.0}/har2tree/helper.py

@@ -72,7 +72,7 @@ def make_hhhash(entry: dict[str, Any]) -> str:
     # We need the HTTP version used for the query:
     # * The HTTP Header names in HTTP 1.1 can have uppercase characters
     # * The HTTP Header names in HTTP 2 *must* be lowercase: https://www.rfc-editor.org/rfc/rfc7540#section-8.1.2
-    if entry['httpVersion'].lower() in ["http/1.1", "http/1.0"]:
+    if entry['httpVersion'].lower() in ["http/1.1", "http/1.0", "1.1"]:
         return f'hhh:1:{sha256}'
     if entry['httpVersion'].lower() == "http/2.0":
         return f'hhh:2:{sha256}'

{har2tree-1.34.2 → har2tree-1.35.0}/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "har2tree"
-version = "1.34.2"
+version = "1.35.0"
 description = "HTTP Archive (HAR) to ETE Toolkit generator"
 authors = [
     {name="Raphaël Vinot", email="raphael.vinot@circl.lu"}
@@ -19,7 +19,7 @@ dependencies = [
     # poetry up fails with the version of numpy forced for python < 3.11.
     # The work around is to comment it, run poetry up, uncomment it. and run poetry update.
     "numpy (>=2.2,<2.3) ; python_version < '3.11'",
-    "numpy (>=2.3.4) ; python_version >= \"3.11\" and python_version < \"4.0\"",
+    "numpy (>=2.3.5) ; python_version >= \"3.11\" and python_version < \"4.0\"",
     "w3lib (>=2.3.1)",
     "tinycss2 (>=1.4.0)",
     "legacy-cgi (>=2.6.4) ; python_version >= \"3.13\" and python_version < \"4.0\"",