llm/SIA
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Fix superfluous CDATA tokens

Browse Source
This commit is contained in:
geens
2024-11-11 14:23:08 +01:00
parent 3a1d8a449c
commit 59f482bc0e
17 changed files with 102 additions and 324 deletions
Download Patch File Download Diff File Expand all files Collapse all files

8
test/background_entry_test.py
View File

@@ -44,7 +44,7 @@ class BackgroundEntryTest(unittest.TestCase):
context = entry.generate_context()
# Initial context should have script but no pid or exit_code
self.assertEqual(context.text, "<![CDATA[echo test]]>")
self.assertEqual(context.text, "echo test")
self.assertIsNone(context.get("pid"))
self.assertIsNone(context.get("exit_code"))
@@ -60,8 +60,8 @@ class BackgroundEntryTest(unittest.TestCase):
# Verify output
context = entry.generate_context()
self.assertTrue("<![CDATA[test" in context.find("stdout").text)
self.assertEqual(context.find("stderr").text, "<![CDATA[]]>")
self.assertEqual(context.find("stdout").text, "test\n")
self.assertEqual(context.find("stderr").text, "")
def test_continuous_output(self):
"""Test process that generates continuous output"""
@@ -96,7 +96,7 @@ class BackgroundEntryTest(unittest.TestCase):
# Verify error captured
context = entry.generate_context()
self.assertTrue(len(context.find("stderr").text) > 0)
self.assertEqual(context.find("stdout").text, "<![CDATA[]]>")
self.assertEqual(context.find("stdout").text, "")
def test_cleanup_running_process(self):
"""Test cleanup of running process"""

4
test/base_agent_test.py
View File

@@ -162,7 +162,7 @@ class BaseAgentTest(unittest.TestCase):
self.assertIsNotNone(reasoning_elem)
self.assertEqual(reasoning_elem.get("id"), "test-id")
# Strip whitespace before comparing
self.assertEqual(reasoning_elem.text.strip(), "<![CDATA[test reasoning]]>")
self.assertEqual(reasoning_elem.text.strip(), "test reasoning")
def test_multiple_entries(self):
"""Test handling multiple entries in working memory."""
@@ -186,4 +186,4 @@ class BaseAgentTest(unittest.TestCase):
self.assertEqual(len(reasoning_elems), 3)
for i, elem in enumerate(reasoning_elems):
self.assertEqual(elem.get("id"), f"id-{i}")
self.assertEqual(elem.text.strip(), f"<![CDATA[test {i}]]>")
self.assertEqual(elem.text.strip(), f"test {i}")

12
test/parse_error_entry_test.py
View File

@@ -48,12 +48,12 @@ class ParseErrorEntryTest(unittest.TestCase):
# Verify error element
error_elem = element.find("error")
self.assertIsNotNone(error_elem)
self.assertEqual(error_elem.text, f"<![CDATA[{error}]]>")
self.assertEqual(error_elem.text, f"{error}")
# Verify content element
content_elem = element.find("content")
self.assertIsNotNone(content_elem)
self.assertEqual(content_elem.text, f"<![CDATA[{content}]]>")
self.assertEqual(content_elem.text, f"{content}")
def test_special_characters(self):
"""Test handling content and errors with special characters"""
@@ -62,13 +62,13 @@ class ParseErrorEntryTest(unittest.TestCase):
entry = ParseErrorEntry(content, error, self.test_id, self.test_timestamp)
element = entry.generate_context()
self.assertEqual(element.find("content").text, f"<![CDATA[{content}]]>")
self.assertEqual(element.find("error").text, f"<![CDATA[{error}]]>")
self.assertEqual(element.find("content").text, f"{content}")
self.assertEqual(element.find("error").text, f"{error}")
def test_empty_content_and_error(self):
"""Test handling empty content and error messages"""
entry = ParseErrorEntry("", "", self.test_id, self.test_timestamp)
element = entry.generate_context()
self.assertEqual(element.find("content").text, "<![CDATA[]]>")
self.assertEqual(element.find("error").text, "<![CDATA[]]>")
self.assertEqual(element.find("content").text, "")
self.assertEqual(element.find("error").text, "")

4
test/read_entry_test.py
View File

@@ -94,7 +94,7 @@ class ReadEntryTest(unittest.TestCase):
# Verify XML structure
self.assertEqual(element.tag, "read_stdin")
self.assertEqual(element.get("id"), self.test_id)
self.assertEqual(element.text, f"<![CDATA[{test_input}]]>")
self.assertEqual(element.text, f"{test_input}")
def test_unicode_content(self):
"""Test reading Unicode content"""
@@ -120,4 +120,4 @@ class ReadEntryTest(unittest.TestCase):
# Verify XML escaping for multiline content
element = entry.generate_context()
self.assertEqual(element.text, f"<![CDATA[{test_input}]]>")
self.assertEqual(element.text, f"{test_input}")

10
test/reasoning_entry_test.py
View File

@@ -42,7 +42,7 @@ class ReasoningEntryTest(unittest.TestCase):
# Verify XML structure
self.assertEqual(element.tag, "reasoning")
self.assertEqual(element.get("id"), self.test_id)
self.assertEqual(element.text, f"<![CDATA[{content}]]>")
self.assertEqual(element.text, f"{content}")
def test_special_characters(self):
"""Test handling reasoning text with special characters"""
@@ -50,14 +50,14 @@ class ReasoningEntryTest(unittest.TestCase):
entry = ReasoningEntry(content, self.test_id, self.test_timestamp)
element = entry.generate_context()
self.assertEqual(element.text, f"<![CDATA[{content}]]>")
self.assertEqual(element.text, f"{content}")
def test_empty_content(self):
"""Test handling empty reasoning text"""
entry = ReasoningEntry("", self.test_id, self.test_timestamp)
element = entry.generate_context()
self.assertEqual(element.text, "<![CDATA[]]>")
self.assertEqual(element.text, "")
def test_large_content(self):
"""Test handling large reasoning text"""
@@ -65,7 +65,7 @@ class ReasoningEntryTest(unittest.TestCase):
entry = ReasoningEntry(content, self.test_id, self.test_timestamp)
element = entry.generate_context()
self.assertEqual(element.text, f"<![CDATA[{content}]]>")
self.assertEqual(element.text, f"{content}")
def test_multiline_content(self):
"""Test handling multiline reasoning text"""
@@ -75,4 +75,4 @@ class ReasoningEntryTest(unittest.TestCase):
entry = ReasoningEntry(content, self.test_id, self.test_timestamp)
element = entry.generate_context()
self.assertEqual(element.text, f"<![CDATA[{content}]]>")
self.assertEqual(element.text, f"{content}")

11
test/repeat_entry_test.py
View File

@@ -95,19 +95,14 @@ class RepeatEntryTest(unittest.TestCase):
entry = RepeatEntry(script, self.test_id, self.test_timestamp)
entry.update()
element = entry.generate_context()
# Verify XML structure
self.assertEqual(element.tag, "repeat")
self.assertEqual(element.get("id"), self.test_id)
self.assertEqual(element.text, f"<![CDATA[{script}]]>")
self.assertEqual(element.text, f"{script}")
self.assertEqual(element.get("exit_code"), "0")
# Get stdout and remove trailing newline for comparison
stdout_text = element.find("stdout").text
self.assertTrue(stdout_text.startswith("<![CDATA[test"))
self.assertTrue(stdout_text.endswith("]]>"))
self.assertEqual(element.find("stderr").text, "<![CDATA[]]>")
self.assertEqual(stdout_text, "test\n")
self.assertEqual(element.find("stderr").text, "")
def test_changing_output(self):
"""Test handling of changing command output"""

10
test/single_entry_test.py
View File

@@ -98,7 +98,7 @@ class SingleEntryTest(unittest.TestCase):
# Verify XML structure
self.assertEqual(element.tag, "single")
self.assertEqual(element.get("id"), self.test_id)
self.assertEqual(element.text, "<![CDATA[echo test]]>")
self.assertEqual(element.text, "echo test")
# Verify no output elements exist
self.assertIsNone(element.find("stdout"))
@@ -116,14 +116,12 @@ class SingleEntryTest(unittest.TestCase):
# Verify XML structure
self.assertEqual(element.tag, "single")
self.assertEqual(element.get("id"), self.test_id)
self.assertEqual(element.text, f"<![CDATA[{script}]]>")
self.assertEqual(element.text, f"{script}")
# Get stdout and remove trailing newline for comparison
stdout_text = element.find("stdout").text
self.assertTrue(stdout_text.startswith("<![CDATA[test"))
self.assertTrue(stdout_text.endswith("]]>"))
self.assertEqual(element.find("stderr").text, "<![CDATA[]]>")
self.assertEqual(stdout_text, "test\n")
self.assertEqual(element.find("stderr").text, "")
self.assertEqual(element.get("exit_code"), "0")
def test_multiple_commands(self):

80
test/util_test.py
View File

@@ -36,86 +36,6 @@ class UtilTest(unittest.TestCase):
result = list(util.stop_before_value(create_iterator(input_sequence), 'STOP'))
self.assertEqual(result, [])
def test_none_value(self):
"""Test that None values are converted to empty strings"""
self.assertEqual(util.escape_text_for_xml(None), '')
def test_number_values(self):
"""Test that numbers are properly converted to strings and wrapped in CDATA"""
self.assertEqual(util.escape_text_for_xml(42), '<![CDATA[42]]>')
self.assertEqual(util.escape_text_for_xml(3.14), '<![CDATA[3.14]]>')
self.assertEqual(util.escape_text_for_xml(-100), '<![CDATA[-100]]>')
self.assertEqual(util.escape_text_for_xml(0), '<![CDATA[0]]>')
def test_simple_strings(self):
"""Test that simple strings are wrapped in CDATA"""
self.assertEqual(util.escape_text_for_xml('hello'), '<![CDATA[hello]]>')
self.assertEqual(util.escape_text_for_xml(''), '<![CDATA[]]>')
self.assertEqual(util.escape_text_for_xml(' '), '<![CDATA[ ]]>')
def test_strings_with_special_characters(self):
"""Test strings containing XML special characters but no CDATA end sequence"""
self.assertEqual(util.escape_text_for_xml('<hello>'), '<![CDATA[<hello>]]>')
self.assertEqual(util.escape_text_for_xml('a & b'), '<![CDATA[a & b]]>')
self.assertEqual(util.escape_text_for_xml('"quoted"'), '<![CDATA["quoted"]]>')
self.assertEqual(util.escape_text_for_xml("'single'"), "<![CDATA['single']]>")
self.assertEqual(util.escape_text_for_xml('<![CDATA['), '<![CDATA[<![CDATA[]]>')
def test_strings_with_cdata_end_sequence(self):
"""Test strings containing CDATA end sequence are properly escaped"""
# Simple case with just the CDATA end sequence
self.assertEqual(util.escape_text_for_xml(']]>'), ']]&gt;')
# CDATA end sequence with other special characters
self.assertEqual(
util.escape_text_for_xml('Text with ]]> and <tags> & ampersands'),
'Text with ]]&gt; and &lt;tags&gt; &amp; ampersands'
)
# Multiple CDATA end sequences
self.assertEqual(
util.escape_text_for_xml('Multiple ]]> end ]]> sequences'),
'Multiple ]]&gt; end ]]&gt; sequences'
)
def test_multiline_strings(self):
"""Test multiline strings are properly handled"""
multiline = """Line 1
Line 2
Line 3"""
self.assertEqual(util.escape_text_for_xml(multiline), '<![CDATA[Line 1\n Line 2\n Line 3]]>')
multiline_with_cdata = """Line 1
Line ]]> 2
Line 3"""
self.assertEqual(
util.escape_text_for_xml(multiline_with_cdata),
'Line 1\n Line ]]&gt; 2\n Line 3'
)
def test_edge_cases(self):
"""Test edge cases and unusual inputs"""
# Empty string
self.assertEqual(util.escape_text_for_xml(''), '<![CDATA[]]>')
# String with only whitespace
self.assertEqual(util.escape_text_for_xml('\n\t '), '<![CDATA[\n\t ]]>')
# String with Unicode characters
self.assertEqual(util.escape_text_for_xml('Hello 世界'), '<![CDATA[Hello 世界]]>')
# String with control characters
self.assertEqual(util.escape_text_for_xml('Hello\0World'), '<![CDATA[Hello\0World]]>')
# Very long string
long_string = 'x' * 1000
self.assertEqual(util.escape_text_for_xml(long_string), f'<![CDATA[{long_string}]]>')
def test_boolean_values(self):
"""Test boolean values are properly converted"""
self.assertEqual(util.escape_text_for_xml(True), '<![CDATA[True]]>')
self.assertEqual(util.escape_text_for_xml(False), '<![CDATA[False]]>')
def test_pretty_print_element_basic(self):
"""Test basic element printing"""
elem = ET.Element('root')

6
test/write_entry_test.py
View File

@@ -85,14 +85,14 @@ class WriteEntryTest(unittest.TestCase):
# Verify XML structure
self.assertEqual(element.tag, "write_stdout")
self.assertEqual(element.get("id"), self.test_id)
self.assertEqual(element.text, f"<![CDATA[{content}]]>")
self.assertEqual(element.text, f"{content}")
# Write content and verify context remains the same
entry.update()
element_after = entry.generate_context()
self.assertEqual(element_after.tag, "write_stdout")
self.assertEqual(element_after.get("id"), self.test_id)
self.assertEqual(element_after.text, f"<![CDATA[{content}]]>")
self.assertEqual(element_after.text, f"{content}")
def test_unicode_content(self):
"""Test writing Unicode content"""
@@ -114,4 +114,4 @@ class WriteEntryTest(unittest.TestCase):
# Verify XML escaping for multiline content
element = entry.generate_context()
self.assertEqual(element.text, f"<![CDATA[{content}]]>")
self.assertEqual(element.text, f"{content}")