feat(agent): scrape linked pages to ground answers in real content

Adds `_fetch_page_content()`, which downloads a URL with a browser-like User-Agent and uses BeautifulSoup with the lxml parser to strip scripts, navigation, and other non-content markup, returning up to 3000 characters of readable text. The response validator now also compares generated replies against this scraped text, flagging named entities and claims that do not appear in the retrieved page content. Dependencies: adds beautifulsoup4==4.12.3 and pins lxml==5.3.0 for parser compatibility.
2026-10-26 16:48:29 -03:00
parent 9eed02645b
commit 723ca1506b
231 changed files with 20508 additions and 5330 deletions
@@ -32,5 +32,59 @@ cdef class _MemDebug:
            raise MemoryError()
        return tree.xmlDictSize(c_dict)

+    def dump(self, output_file=None, byte_count=None):
+        """dump(self, output_file=None, byte_count=None)
+
+        Dumps the current memory blocks allocated by libxml2 to a file.
+
+        The optional parameter 'output_file' specifies the file path.  It defaults
+        to the file ".memorylist" in the current directory.
+
+        The optional parameter 'byte_count' limits the number of bytes in the dump.
+        Note that this parameter is ignored when lxml is compiled against a libxml2
+        version before 2.7.0.
+        """
+        cdef Py_ssize_t c_count
+        if output_file is None:
+            output_file = b'.memorylist'
+        elif isinstance(output_file, unicode):
+            output_file.encode(sys.getfilesystemencoding())
+
+        f = stdio.fopen(output_file, "w")
+        if f is NULL:
+            raise IOError(f"Failed to create file {output_file.decode(sys.getfilesystemencoding())}")
+        try:
+            if byte_count is None:
+                tree.xmlMemDisplay(f)
+            else:
+                c_count = byte_count
+                tree.xmlMemDisplayLast(f, c_count)
+        finally:
+            stdio.fclose(f)
+
+    def show(self, output_file=None, block_count=None):
+        """show(self, output_file=None, block_count=None)
+
+        Dumps the current memory blocks allocated by libxml2 to a file.
+        The output file format is suitable for line diffing.
+
+        The optional parameter 'output_file' specifies the file path.  It defaults
+        to the file ".memorydump" in the current directory.
+
+        The optional parameter 'block_count' limits the number of blocks
+        in the dump.
+        """
+        if output_file is None:
+            output_file = b'.memorydump'
+        elif isinstance(output_file, unicode):
+            output_file.encode(sys.getfilesystemencoding())
+
+        f = stdio.fopen(output_file, "w")
+        if f is NULL:
+            raise IOError(f"Failed to create file {output_file.decode(sys.getfilesystemencoding())}")
+        try:
+            tree.xmlMemShow(f, block_count if block_count is not None else tree.xmlMemBlocks())
+        finally:
+            stdio.fclose(f)

 memory_debugger = _MemDebug()