har2tree 1.31.0__tar.gz → 1.31.2__tar.gz

{har2tree-1.31.0 → har2tree-1.31.2}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.3
 Name: har2tree
-Version: 1.31.0
+Version: 1.31.2
 Summary: HTTP Archive (HAR) to ETE Toolkit generator
 License: BSD-3-Clause
 Author: Raphaël Vinot
@@ -27,8 +27,8 @@ Requires-Dist: filetype (>=1.2.0)
 Requires-Dist: legacy-cgi (>=2.6.3) ; python_version >= "3.13,<4.0"
 Requires-Dist: numpy (<2.1) ; python_version < "3.10"
 Requires-Dist: numpy (<2.3) ; python_version < "3.11"
-Requires-Dist: numpy (>=2.3.0) ; python_version >= "3.11"
-Requires-Dist: publicsuffixlist (>=1.0.2.20250611)
+Requires-Dist: numpy (>=2.3.1) ; python_version >= "3.11"
+Requires-Dist: publicsuffixlist (>=1.0.2.20250627)
 Requires-Dist: six (>=1.17.0) ; extra == "docs"
 Requires-Dist: tinycss2 (>=1.4.0)
 Requires-Dist: w3lib (>=2.3.1)

{har2tree-1.31.0 → har2tree-1.31.2}/har2tree/helper.py

@@ -283,7 +283,12 @@ def make_soup(html: bytes) -> BeautifulSoup:
     if doc_as_str.startswith('<?xml'):
         return BeautifulSoup(doc_as_str, 'lxml-xml')
     else:
-        return BeautifulSoup(doc_as_str, 'lxml')
+        try:
+            return BeautifulSoup(doc_as_str, 'lxml')
+        except Exception as e:
+            logger.warning(f'Unable to parse doc with lxml, try again with the default parser: {e}')
+            # Fallback to the default python parser
+            return BeautifulSoup(doc_as_str, 'html.parser')
 
 # The CSS processing bit is taked out of WeasyPrint
 # https://github.com/Kozea/WeasyPrint/blob/main/weasyprint/css/__init__.py#L876 -> preprocess_stylesheet
@@ -380,8 +385,13 @@ def find_external_ressources(mimetype: str, data: bytes, base_url: str, all_requ
             else:
                 external_ressources['css'].append(url)
     else:
-        soup = make_soup(data)
-        string_soup = str(soup)
+        try:
+            soup = make_soup(data)
+            string_soup = str(soup)
+        except Exception as e:
+            logger.warning(f'Unable to parse HTML blob: {e}')
+            string_soup = ''
+
         if not string_soup:
             # Empty HTML document, nothing to do
             return external_ressources, embedded_ressources

{har2tree-1.31.0 → har2tree-1.31.2}/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "har2tree"
-version = "1.31.0"
+version = "1.31.2"
 description = "HTTP Archive (HAR) to ETE Toolkit generator"
 authors = [
     {name="Raphaël Vinot", email="raphael.vinot@circl.lu"}
@@ -14,13 +14,13 @@ dynamic = [ "classifiers" ]
 dependencies = [
     "ete3 (>=3.1.3)",
     "beautifulsoup4[charset-normalizer,lxml] (>=4.13.4)",
-    "publicsuffixlist (>=1.0.2.20250611)",
+    "publicsuffixlist (>=1.0.2.20250627)",
     "filetype (>=1.2.0)",
     # poetry up fails with the version of numpy forced for python < 3.10.
     # The work around is to comment it, run poetry up, uncomment it. and run poetry update.
     "numpy (<2.1) ; python_version < \"3.10\"",
     "numpy (<2.3) ; python_version < \"3.11\"",
-    "numpy (>=2.3.0) ; python_version >= \"3.11\"",
+    "numpy (>=2.3.1) ; python_version >= \"3.11\"",
     "w3lib (>=2.3.1)",
     "tinycss2 (>=1.4.0)",
     "legacy-cgi (>=2.6.3) ; python_version >= \"3.13,<4.0\"",
@@ -45,9 +45,9 @@ classifiers = [
 docs = ["Sphinx (>=8.2.3) ; python_version >= \"3.11\"", "six (>=1.17.0)"]
 
 [tool.poetry.group.dev.dependencies]
-mypy = "^1.16.0"
+mypy = "^1.16.1"
 pytest-cov = "^6.2.1"
-coverage = "^7.9.0"
+coverage = "^7.9.1"
 types-beautifulsoup4 = "^4.12.0.20250516"
 
 [build-system]