⏪ fix: self healing response parser

dev_gpt/options/generate/chains/fix_based_on_error.py

@@ -0,0 +1,32 @@
+import traceback
+
+
+
+def fix_based_on_error_chain(context_string, content_type, original_content, parser):
+    from dev_gpt.apis.gpt import ask_gpt
+    current_content = original_content
+    for i in range(3):
+        try:
+            return parser(current_content)
+        except Exception:
+            error_message = traceback.format_exc()
+            current_content = ask_gpt(fix_based_on_error_prompt, context_string=context_string, content_type=content_type, current_content=current_content, error_message=error_message)
+    raise Exception(f'Could not fix the content:\n{original_content}')
+
+
+fix_based_on_error_prompt = '''\
+Context:
+{context_string}
+
+Original {content_type}:
+{current_content}
+
+Error message:
+{error_message}
+
+Your task:
+You must return the fixed {content_type}.
+Most importantly, you are not allowed to return something else - only the fixed {content_type}.
+'''
+
+

dev_gpt/options/generate/generator.py

@@ -21,6 +21,7 @@ from dev_gpt.constants import FILE_AND_TAG_PAIRS, NUM_IMPLEMENTATION_STRATEGIES,
     IMPLEMENTATION_FILE_TAG, LANGUAGE_PACKAGES, UNNECESSARY_PACKAGES, DOCKER_BASE_IMAGE_VERSION, SEARCH_PACKAGES, \
     INDICATOR_TO_IMPORT_STATEMENT
 from dev_gpt.options.generate.conversation_logger import Timer
+from dev_gpt.options.generate.parser import json_parser
 from dev_gpt.options.generate.pm.pm import PM
 from dev_gpt.options.generate.templates_user import template_generate_microservice_name, \
     template_generate_possible_packages, \
@@ -439,8 +440,7 @@ pytest
                 self.previous_solutions.append(suggested_solution)
 
     def generate_solution_suggestion(self, summarized_error, all_files_string):
-        suggested_solutions = json.loads(
-            self.generate_and_persist_file(
+        json_string = self.generate_and_persist_file(
                 section_title='Suggest solution for code issue',
                 template=template_suggest_solutions_code_issue,
                 file_name_s=['solutions.json'],
@@ -449,9 +449,10 @@ pytest
                 test_description=self.microservice_specification.test,
                 all_files_string=all_files_string,
                 response_format_example=response_format_suggest_solutions,
-            )['solutions.json']
             )
 
+        suggested_solutions = json_parser(json_string)['solutions.json']
+
         if len(self.previous_errors) > 0:
             was_error_seen_before = json.loads(
                 self.generate_and_persist_file(

dev_gpt/options/generate/parser.py

@@ -1,6 +1,8 @@
 import json
 import re
 
+from dev_gpt.options.generate.chains.fix_based_on_error import fix_based_on_error_chain
+
 
 def identity_parser(x):
     return x
@@ -16,6 +18,10 @@ def boolean_parser(x):
 
 def json_parser(x):
     if '```' in x:
-        pattern = r'```(.+)```'
-        x = re.findall(pattern, x, re.DOTALL)[-1]
+        pattern = r'```(json)?(.+)```'
+        x = re.findall(pattern, x, re.DOTALL)[-1][-1]
     return json.loads(x)
+
+def self_healing_json_parser(original_json_string):
+    return fix_based_on_error_chain('I want to load my JSON string using json.loads(x) but get the following error:', 'JSON',
+                                    original_json_string, parser=json_parser)

dev_gpt/options/generate/templates_user.py

@@ -310,14 +310,22 @@ Output the apt-get packages that need to be placed at {{APT_GET_PACKAGES}} as js
 ```json
 {{"packages": ["<package1>", "<package2>"]}}
 ```
-Example:
-Error is about missing package `libgl1-mesa-glx`.
-The output is:
+
+Example output for an error is about missing package `libgl1-mesa-glx`:
 **apt-get-packages.json**
 ```json
 {{"packages": [libgl1-mesa-glx]}}
 ```
-Only output content of the apt-get-packages.json file. Ensure the response can be parsed by Python json.loads
+
+Negative example1:
+```json
+{{"packages": [libgl1-mesa-glx]}}
+```
+
+Negative example2:
+{{"packages": [libgl1-mesa-glx]}}
+
+Note: Only output content of the apt-get-packages.json file.
 Note: you must not output the content of any other. Especially don't output the Dockerfile or requirements.txt. 
 Note: the first line you output must be: **apt-get-packages.json**
 '''

test/unit/test_response_parsing.py

@@ -1,6 +1,11 @@
+import os
+
 import pytest
 
+from dev_gpt.apis.gpt import GPTSession
 from dev_gpt.options.generate.generator import Generator
+from dev_gpt.options.generate.parser import self_healing_json_parser
+
 
 def create_code_block(with_backticks, asterisks, with_highlight_info, file_name, start_inline, content):
     code_block = f'''
@@ -42,3 +47,21 @@ def test_extract_content_from_result(plain_text, expected1, expected2):
     parsed_result2 = Generator.extract_content_from_result(plain_text, 'test100.json', True, False)
     assert parsed_result2 == expected2
 
+
+def test_self_healing_json_parser(tmpdir):
+    os.environ['VERBOSE'] = 'true'
+    GPTSession(os.path.join(str(tmpdir), 'log.json'), model='gpt-3.5-turbo')
+    json_response = '''\
+solutions.json
+```json
+{
+    "1": "Change line 7 of microservice.py to 'pdf_file = input_dict['pdf_file'].encode('latin-1')' to convert the bytes object to a string before passing it to PyPDF2.",
+    "2": "Change line 7 of microservice.py to 'pdf_file = input_dict['pdf_file'].decode('utf-8')' to decode the bytes object to a string before passing it to PyPDF2.",
+    "3": "Change line 13 of test_microservice.py to 'input_dict = {"pdf_file": 'Sample PDF file content'.encode('latin-1')}' to encode the string to a bytes object before passing it to func.",
+    "4": "Change line 13 of test_microservice.py to 'input_dict = {"pdf_file": 'Sample PDF file content'.decode('utf-8')}' to decode the string to a bytes object before passing it to func."
+}
+```'''
+    parsed_json = self_healing_json_parser(json_response)
+    for key in ['1', '2', '3', '4']:
+        assert key in parsed_json
+        assert 'Change' in parsed_json[key]