[FIX] Prevent cst duplicate imports

[mohammedahmed18]

This fixes the issue here: misrasaurabh1/recce@16f0638

Basically, from recce.adapter.base import BaseAdapter got added even though it was already in the file.

my guess is that AddImportsVisitor.add_needed_import stops looking for imports once it hits something that’s not an import. In this case, it hit the try/except block within lines 39:46, so it never saw the existing BaseAdapter import at line 50 and just added it again.

The fix was to manually collect all top-level imports in a normalized dotted format, so duplicates don’t slip through anymore.

---

Description

• Replace conditional import collector with dotted importer

• Prevent adding duplicate imports in extraction logic

• Update import normalization and alias handling

• Add test for duplicate import removal

---

Diagram Walkthrough

flowchart LR
  parse["Parse destination module"] --> collect["DottedImportCollector.collect imports"]
  gather["Gather needed imports"] --> filter["Filter out existing imports"]
  filter --> add["AddImportsVisitor.add_needed_import"]
  filter --> remove["RemoveImportsVisitor.remove_unused_import"]

Loading

File Walkthrough

	Relevant files
Bug fix	code_extractor.py Refactor import collector and dedupe logic                              codeflash/code_utils/code_extractor.py Rename ConditionalImportCollector to DottedImportCollector Normalize top-level imports to dotted format Replace collector usage and invert duplicate import checks Invoke block import collection in visit_Module +31/-17
Tests	test_add_needed_imports_from_module.py Add test for duplicate import removal                                        tests/test_add_needed_imports_from_module.py Add import re for regex assertions Introduce test_duplicated_imports for dedupe check Assert only one BaseAdapter import remains +127/-0

---

[github-actions]

PR Reviewer Guide 🔍

Here are some key observations to aid the review process:

⏱️ Estimated effort to review: 3 🔵🔵🔵⚪⚪

🧪 PR contains tests

🔒 No security concerns identified

⚡ Recommended focus areas for review Type mismatch The check isinstance(asname, cst.Attribute) will never be true because asname is assigned a Python string. Update the logic to correctly detect alias nodes or remove the unreachable branch. if isinstance(asname, cst.Attribute): self.imports.add(module) else: self.imports.add(module if module == asname else f"{module}.{asname}") Collector invocation visit_Module calls _collect_imports_from_block(node) but _collect_imports_from_block expects a cst.IndentedBlock. Ensure top-level imports are collected by iterating over node.body or updating the helper signature. def visit_Module(self, node: cst.Module) -> None: self.depth = 0 self._collect_imports_from_block(node)

[github-actions]

PR Code Suggestions ✨

Explore these optional code suggestions:

Category	Suggestion	Impact
Possible issue	Correct alias name extraction The current asname is always a string, so the isinstance(asname, cst.Attribute) branch never executes and will raise when alias.name is an Attribute. Derive the alias name from the AST node for both Name and Attribute and then build the fully qualified import string. codeflash/code_utils/code_extractor.py [231-234] -if isinstance(asname, cst.Attribute): - self.imports.add(module) -else: - self.imports.add(module if module == asname else f"{module}.{asname}") +alias_name = ( + alias.asname.name.value + if alias.asname + else ( + alias.name.attr.value + if isinstance(alias.name, cst.Attribute) + else alias.name.value + ) +) +import_fqn = module if module == alias_name else f"{module}.{alias_name}" +self.imports.add(import_fqn) Suggestion importance[1-10]: 9 __ Why: The branch using isinstance(asname, cst.Attribute) never executes because asname is always a string, so the proposed extraction fixes a real bug by handling AST attributes correctly.	High

[misrasaurabh1]

i would prefer a direct string comparison over the entirety of the newly replaced string. this way we have more of a guarantee than searching for particular patterns.