Fix superfluous CDATA tokens

This commit is contained in:
2024-11-11 14:23:08 +01:00
parent 3a1d8a449c
commit 59f482bc0e
17 changed files with 102 additions and 324 deletions

View File

@@ -4,7 +4,6 @@ import xml.etree.ElementTree as ET
from typing import Optional
from .entry import Entry
from .util import escape_text_for_xml
class BackgroundEntry(Entry):
"""
@@ -66,13 +65,10 @@ class BackgroundEntry(Entry):
"""
if self._process is not None:
try:
# Close file handles if they're open
if self._process.stdout:
self._process.stdout.close()
if self._process.stderr:
self._process.stderr.close()
# Terminate process if it's still running
if self._process.poll() is None:
self._process.terminate()
try:
@@ -90,65 +86,48 @@ class BackgroundEntry(Entry):
Start the process if not running and collect any new output.
Updates stdout and stderr with any new output.
"""
try:
# Start process if not running
if self._process is None and self._exit_code is None:
self._process = subprocess.Popen(
self._script,
shell=True,
stdout=subprocess.PIPE,
stderr=subprocess.PIPE,
text=True,
bufsize=1 # Line buffered
)
return # Return after starting to allow process to generate output
if self._process is None:
return # Process already completed
# Check if process has finished
exit_code = self._process.poll()
if exit_code is not None:
# Process has terminated, collect remaining output
try:
remaining_out, remaining_err = self._process.communicate(timeout=0.1)
self._stdout += remaining_out
self._stderr += remaining_err
except subprocess.TimeoutExpired:
pass # Process didn't finish communicating, try again next update
self._exit_code = exit_code
self.cleanup()
return
if self._process is None and self._exit_code is None:
self._process = subprocess.Popen(
self._script,
shell=True,
stdout=subprocess.PIPE,
stderr=subprocess.PIPE,
text=True,
bufsize=1 # Line buffered
)
return
if self._process is None:
return
exit_code = self._process.poll()
if exit_code is not None:
try:
remaining_out, remaining_err = self._process.communicate(timeout=0.1)
self._stdout += remaining_out
self._stderr += remaining_err
except subprocess.TimeoutExpired:
pass # Process didn't finish communicating, try again next update
# Read stdout if available
if self._process.stdout:
while True:
try:
line = self._process.stdout.readline()
if not line:
break
self._stdout += line
except:
break
# Read stderr if available
if self._process.stderr:
while True:
try:
line = self._process.stderr.readline()
if not line:
break
self._stderr += line
except:
break
except Exception as e:
# Handle any errors
error_msg = f"Error handling background process: {str(e)}"
self._stderr += f"\n{error_msg}"
self._exit_code = -1
self._exit_code = exit_code
self.cleanup()
return
if self._process.stdout:
while True:
try:
line = self._process.stdout.readline()
if not line:
break
self._stdout += line
except:
break
if self._process.stderr:
while True:
try:
line = self._process.stderr.readline()
if not line:
break
self._stderr += line
except:
break
def generate_context(self) -> ET.Element:
"""
@@ -157,23 +136,15 @@ class BackgroundEntry(Entry):
Returns:
ET.Element: XML element containing the entry's data
"""
# Create root element with appropriate status
element = ET.Element("background", {"id": self._id})
if self._process is not None:
element.set("pid", str(self._process.pid))
elif self._exit_code is not None:
element.set("exit_code", str(self._exit_code))
# Add script as CDATA or escaped text
element.text = escape_text_for_xml(self._script)
# Add stdout element with accumulated output
element.text = self._script
stdout_elem = ET.SubElement(element, "stdout")
stdout_elem.text = escape_text_for_xml(self._stdout)
# Add stderr element with accumulated output
stdout_elem.text = self._stdout
stderr_elem = ET.SubElement(element, "stderr")
stderr_elem.text = escape_text_for_xml(self._stderr)
stderr_elem.text = self._stderr
return element

View File

@@ -2,7 +2,6 @@ from datetime import datetime
import xml.etree.ElementTree as ET
from .entry import Entry
from .util import escape_text_for_xml
class ParseErrorEntry(Entry):
"""
@@ -43,15 +42,10 @@ class ParseErrorEntry(Entry):
Returns:
ET.Element: XML element containing the entry's data
"""
# Create root element
element = ET.Element("parse_error", {"id": self._id})
# Add error subelement
error_elem = ET.SubElement(element, "error")
error_elem.text = escape_text_for_xml(self._error)
# Add content subelement
error_elem.text = self._error
content_elem = ET.SubElement(element, "content")
content_elem.text = escape_text_for_xml(self._content)
content_elem.text = self._content
return element

View File

@@ -2,7 +2,6 @@ from datetime import datetime
import xml.etree.ElementTree as ET
from .entry import Entry
from .util import escape_text_for_xml
class ReadEntry(Entry):
"""
@@ -51,11 +50,8 @@ class ReadEntry(Entry):
Returns:
ET.Element: XML element containing the entry's data
"""
# Create root element
element = ET.Element("read_stdin", {"id": self._id})
# Add content as CDATA or escaped text if content has been read
if self._read:
element.text = escape_text_for_xml(self._content)
element.text = self._content
return element

View File

@@ -2,7 +2,6 @@ from datetime import datetime
import xml.etree.ElementTree as ET
from .entry import Entry
from .util import escape_text_for_xml
class ReasoningEntry(Entry):
"""
@@ -42,10 +41,7 @@ class ReasoningEntry(Entry):
Returns:
ET.Element: XML element containing the entry's data
"""
# Create root element
element = ET.Element("reasoning", {"id": self._id})
# Add content as CDATA or escaped text
element.text = escape_text_for_xml(self._content)
element.text = self._content
return element

View File

@@ -4,7 +4,6 @@ import xml.etree.ElementTree as ET
from typing import Optional
from .entry import Entry
from .util import escape_text_for_xml
class RepeatEntry(Entry):
"""
@@ -51,31 +50,15 @@ class RepeatEntry(Entry):
Execute the script and update the output.
Captures stdout, stderr and exit code from each execution.
"""
try:
# Run script and capture output
process = subprocess.run(
self._script,
shell=True,
capture_output=True,
text=True
)
# Store results
self._stdout = process.stdout
self._stderr = process.stderr
self._exit_code = process.returncode
except subprocess.SubprocessError as e:
# Handle subprocess errors by capturing the error message
self._stdout = ""
self._stderr = str(e)
self._exit_code = -1
except Exception as e:
# Handle any other errors
self._stdout = ""
self._stderr = f"Error executing script: {str(e)}"
self._exit_code = -1
process = subprocess.run(
self._script,
shell=True,
capture_output=True,
text=True
)
self._stdout = process.stdout
self._stderr = process.stderr
self._exit_code = process.returncode
def generate_context(self) -> ET.Element:
"""
@@ -84,21 +67,14 @@ class RepeatEntry(Entry):
Returns:
ET.Element: XML element containing the entry's data
"""
# Create root element
element = ET.Element("repeat", {
"id": self.id,
"exit_code": str(self._exit_code) if self._exit_code is not None else "-1"
})
# Add script as CDATA or escaped text
element.text = escape_text_for_xml(self.script)
# Add stdout element
element.text = self.script
stdout_elem = ET.SubElement(element, "stdout")
stdout_elem.text = escape_text_for_xml(self._stdout)
# Add stderr element
stdout_elem.text = self._stdout
stderr_elem = ET.SubElement(element, "stderr")
stderr_elem.text = escape_text_for_xml(self._stderr)
stderr_elem.text = self._stderr
return element

View File

@@ -4,7 +4,6 @@ import xml.etree.ElementTree as ET
from typing import Optional
from .entry import Entry
from .util import escape_text_for_xml
class SingleEntry(Entry):
"""
@@ -61,33 +60,15 @@ class SingleEntry(Entry):
"""
if self._executed:
return
try:
# Run script and capture output
process = subprocess.run(
self.script,
shell=True,
capture_output=True,
text=True
)
# Store results
self._stdout = process.stdout
self._stderr = process.stderr
self._exit_code = process.returncode
except subprocess.SubprocessError as e:
# Handle subprocess errors by capturing the error message
self._stdout = ""
self._stderr = str(e)
self._exit_code = -1
except Exception as e:
# Handle any other errors
self._stdout = ""
self._stderr = f"Error executing script: {str(e)}"
self._exit_code = -1
process = subprocess.run(
self.script,
shell=True,
capture_output=True,
text=True
)
self._stdout = process.stdout
self._stderr = process.stderr
self._exit_code = process.returncode
self._executed = True
def generate_context(self) -> ET.Element:
@@ -97,25 +78,14 @@ class SingleEntry(Entry):
Returns:
ET.Element: XML element containing the entry's data
"""
# Create root element
element = ET.Element("single", {
"id": self.id,
})
# Add script as CDATA or escaped text
element.text = escape_text_for_xml(self.script)
# Only add output elements if script has executed
element.text = self.script
if self._executed:
# Add stdout element if there is output
stdout_elem = ET.SubElement(element, "stdout")
stdout_elem.text = escape_text_for_xml(self._stdout)
# Add stderr element if there are errors
stdout_elem.text = self._stdout
stderr_elem = ET.SubElement(element, "stderr")
stderr_elem.text = escape_text_for_xml(self._stderr)
# Add exit code attribute after execution
stderr_elem.text = self._stderr
element.set("exit_code", str(self._exit_code))
return element

View File

@@ -23,12 +23,6 @@ def stop_before_value(iterator: Iterator[str], stop_value: str) -> Iterator[str]
break
yield item
def escape_text_for_xml(text: str) -> str:
"""Convert any value to string representation for XML storage."""
if text is None:
return ''
return str(text)
def pretty_print_element(elem: ET.Element, level: int = 0, max_line: int = 80) -> str:
"""Convert ElementTree element to pretty-printed string with custom formatting."""
indent = ' ' * level
@@ -72,31 +66,4 @@ def pretty_print_element(elem: ET.Element, level: int = 0, max_line: int = 80) -
parts.append(f'{indent} {child.tail}')
parts.append(f'{indent}</{tag}>')
return '\n'.join(parts)
def escape_text_for_xml(text: str) -> str:
"""
Convert any text value to a properly escaped XML string.
Args:
text: Text to escape, can be string, number, or None
Returns:
str: XML-escaped string representation
"""
if text is None:
return ''
# Convert numbers to strings
text_str = str(text)
# Standard XML escaping if string contains CDATA end sequence
if ']]>' in text_str:
return text_str.replace('&', '&amp;') \
.replace('<', '&lt;') \
.replace('>', '&gt;') \
.replace('"', '&quot;') \
.replace("'", '&apos;')
# Otherwise wrap in CDATA
return f'<![CDATA[{text_str}]]>'
return '\n'.join(parts)

View File

@@ -2,7 +2,6 @@ from datetime import datetime
import xml.etree.ElementTree as ET
from .entry import Entry
from .util import escape_text_for_xml
class WriteEntry(Entry):
"""
@@ -45,10 +44,6 @@ class WriteEntry(Entry):
Returns:
ET.Element: XML element containing the entry's data
"""
# Create root element
element = ET.Element("write_stdout", {"id": self.id})
# Add content as CDATA or escaped text
element.text = escape_text_for_xml(self.content)
element.text = self.content
return element