Implement the 'first' filter #100

ipmb · 2025-07-26T19:09:06Z

Full disclosure upfront, this was generated with Claude Code. I'm excited about this project and some of this porting work felt like a good fit for an LLM. If this isn't welcome, feel free to close it. I'm happy to iterate on it if there is feedback.

Implementation Details

The filter returns the first item of a sequence (list, tuple, string) or empty string for empty sequences. It matches Django's behavior, including error handling.

Django References

Filter implementation: django/template/defaultfilters.py#L599-L605
Django tests: tests/template_tests/filter_tests/test_first.py

Changes

Added FirstFilter struct and enum variant
Implemented parsing for the 'first' filter (no arguments)
Implemented ResolveFilter trait with Django-compatible behavior
- Returns first item for sequences
- Returns empty string for empty sequences
- Raises TypeError for None and non-subscriptable types
- Uses exact Django error messages
Added comprehensive test suite porting all Django test cases plus additional edge cases

Testing

The implementation handles:

Lists, tuples, strings (including empty)
Falsy values (0, False) correctly returned
Autoescape on/off behavior
Safe strings preservation
Type errors for None and integers

- Add FirstFilter struct and enum variant - Implement parsing logic for 'first' filter - Implement ResolveFilter trait with Django-compatible behavior - Returns first item of sequences (lists, tuples, strings) - Returns empty string for empty sequences - Raises TypeError for None and non-subscriptable types - Matches Django's exact error messages - Add comprehensive test suite porting all Django test cases - Test with lists, tuples, strings - Test empty sequences - Test autoescape on/off behavior - Test with mark_safe strings - Test error conditions (None, integers) - Add additional edge case tests refs LilyFirefly#1

codecov · 2025-07-27T12:46:42Z

Codecov Report

❌ Patch coverage is 92.36111% with 22 lines in your changes missing coverage. Please review.

Files with missing lines	Patch %	Lines
src/render/filters.rs	80.35%	21 Missing and 1 partial ⚠️

📢 Thoughts on this report? Let us know!

LilyFirefly

Thanks for this! I have a few suggestions, but this is looking really good already.

src/render/filters.rs

tests/test_filters_first.py

- Added custom RenderError::NotSubscriptable for better error messages - Used inline isinstance check for PyIndexError - Implemented horizontal space reduction with early returns - Used str.split("") for efficient character iteration - Fixed all clippy warnings and formatting issues - Moved tests to tests/filters/test_first.py following project structure - All 18 tests pass with 100% Django compatibility The filter returns the first item of sequences (lists, tuples, strings) or empty string for empty sequences, matching Django's behavior exactly including error handling for non-subscriptable types. Co-Authored-By: Claude <[email protected]>

src/render/filters.rs

src/template.rs

src/render/filters.rs

Following feedback on the first filter implementation, this commit adds proper source location tracking to the FirstFilter struct. The location is captured during parsing and used when reporting NotSubscriptable errors for non-sequence types (int and float). Changes: - Add `at: SourceSpan` field to FirstFilter struct - Add constructor method `FirstFilter::new(at: SourceSpan)` - Pass filter location from parser when creating FirstFilter instances - Use stored location instead of hardcoded (0, 0) in error reporting This provides more accurate error diagnostics by preserving the template position where the filter was applied, though the location is not yet displayed in the Python error messages. 🤖 Generated with [Claude Code](https://claude.ai/code) Co-Authored-By: Claude <[email protected]>

Brings in CenterFilter and Bool content type support while maintaining FirstFilter location tracking

- Update FirstFilter to return Option<Content> to match new API - Add handling for Content::Bool variant in FirstFilter

The IntoOwnedContent trait is already imported at module scope, so the local import in FirstFilter::resolve is unnecessary.

- Inline s.as_raw() call to reduce unnecessary variable - More concise and readable code

- Implement WithSourceCode trait for PyTypeError to show miette error formatting - Update NotSubscriptable error handling to use the new method - Handle Python TypeError in FirstFilter to detect not subscriptable errors - Preserve custom TypeError messages while formatting standard ones - Now shows source code context with location indicators for type errors

Ensures complete coverage of all Content type branches

ipmb · 2025-08-22T21:27:50Z

I think I've resolved all the open comments here 👍

LilyFirefly

Thanks! This is great progress. I've got a few more suggestions:

LilyFirefly · 2025-08-22T21:35:21Z

src/render/filters.rs

+                    Err(e) if e.is_instance_of::<pyo3::exceptions::PyTypeError>(py) => {
+                        // Check if this is the standard "'type' object is not subscriptable" error
+                        let error_msg = e.to_string();
+                        if error_msg.contains("is not subscriptable") {
+                            // Standard not subscriptable error - provide better formatting
+                            let type_name = obj
+                                .get_type()
+                                .name()
+                                .map(|n| n.to_string())
+                                .unwrap_or_else(|_| "unknown".to_string());
+                            Err(RenderError::NotSubscriptable {
+                                type_name,
+                                at: self.at,
+                            }
+                            .into())
+                        } else {
+                            // Custom TypeError - preserve the original message
+                            Err(e.into())
+                        }
+                    }
+                    Err(e) => Err(e.into()),


Since my last review, I've added a PyErr::annotate method which I think we can use here:

Suggested change

Err(e) if e.is_instance_of::<pyo3::exceptions::PyTypeError>(py) => {

// Check if this is the standard "'type' object is not subscriptable" error

let error_msg = e.to_string();

if error_msg.contains("is not subscriptable") {

// Standard not subscriptable error - provide better formatting

let type_name = obj

.get_type()

.name()

.map(|n| n.to_string())

.unwrap_or_else(|_| "unknown".to_string());

Err(RenderError::NotSubscriptable {

type_name,

at: self.at,

}

.into())

} else {

// Custom TypeError - preserve the original message

Err(e.into())

}

}

Err(e) => Err(e.into()),

Err(error) => {

let error = error.annotate(py, self.at, "here", template);

Err(error.into())

}

LilyFirefly · 2025-08-22T21:35:31Z

src/render/filters.rs

+        &self,
+        variable: Option<Content<'t, 'py>>,
+        py: Python<'py>,
+        _template: TemplateString<'t>,


Suggested change

_template: TemplateString<'t>,

template: TemplateString<'t>,

LilyFirefly · 2025-08-22T21:38:38Z

src/filters.rs


+#[derive(Clone, Debug, PartialEq)]
+pub struct FirstFilter {
+    pub at: SourceSpan,


It doesn't really matter, but I'd keep at as a (usize, usize) until using it in RenderError:

Suggested change

pub at: SourceSpan,

pub at: (usize, usize),

The reason is basically just to avoid using the third party type (miette::SourceSpan) until it's necessary.

Actually, I think this change will be necessary to use PyErr::annotate.

LilyFirefly · 2025-08-22T21:41:36Z

src/error.rs

    },
+    #[error("'{type_name}' object is not subscriptable")]
+    NotSubscriptable {
+        type_name: String,


I wonder if we can use &'static str here to avoid an allocation:

Suggested change

type_name: String,

type_name: &'static str,

Not a problem to leave it as is if this doesn't compile.

I checked, and we can do this alongside the PyErr::annotate change.

LilyFirefly · 2025-08-22T21:42:13Z

src/render/filters.rs

+                // Numbers are not sequences, should raise TypeError
+                // Match Django's behavior exactly
+                Err(RenderError::NotSubscriptable {
+                    type_name: "int".to_string(),


If type_name can be &'static str, we can avoid the to_string() call:

Suggested change

type_name: "int".to_string(),

type_name: "int",

LilyFirefly · 2025-08-22T21:48:17Z

tests/filters/test_first.py

+    assert "'NoneType' object is not subscriptable" in str(django_exc.value)
+    assert "'NoneType' object is not subscriptable" in str(rusty_exc.value)


I'd prefer to check the full error messages. Here's an example of what I mean:

https://github.com/LilyAcorn/django-rusty-templates/blob/49b96d65a672d2ec09b48885a83b89f2bea841ef/tests/filters/test_add.py#L25-L48

LilyFirefly · 2025-08-22T21:50:08Z

src/render/filters.rs

+            Content::String(s) => {
+                // For strings, get the first character
+                // Skip the empty string always present at the start after splitting
+                let mut chars = s.as_raw().split("").skip(1);
+                match chars.next() {
+                    None => Ok(Some("".as_content())),
+                    Some(c) => Ok(Some(c.to_string().into_content())),
+                }
+            }
+            Content::Int(_) => {
+                // Numbers are not sequences, should raise TypeError
+                // Match Django's behavior exactly
+                Err(RenderError::NotSubscriptable {
+                    type_name: "int".to_string(),
+                    at: self.at,
+                }
+                .into())
+            }
+            Content::Float(_) => {
+                // Floats are not sequences, should raise TypeError
+                // Match Django's behavior exactly
+                Err(RenderError::NotSubscriptable {
+                    type_name: "float".to_string(),
+                    at: self.at,
+                }
+                .into())
+            }
+            Content::Bool(_) => {
+                // Booleans are not sequences, should raise TypeError
+                // Match Django's behavior exactly
+                Err(RenderError::NotSubscriptable {
+                    type_name: "bool".to_string(),
+                    at: self.at,
+                }
+                .into())
+            }
+        }


We need tests for these.

ipmb force-pushed the implement-first-filter branch from ca856bc to f98bb33 Compare July 26, 2025 19:11

ipmb force-pushed the implement-first-filter branch from f98bb33 to 284297f Compare July 26, 2025 19:14

Bump up test coverage on 'first' filter

d75bb0f

LilyFirefly reviewed Jul 27, 2025

View reviewed changes

ipmb commented Jul 27, 2025

View reviewed changes

src/render/filters.rs Outdated Show resolved Hide resolved

ipmb commented Jul 27, 2025

View reviewed changes

src/render/filters.rs Outdated Show resolved Hide resolved

LilyFirefly reviewed Jul 28, 2025

View reviewed changes

ipmb and others added 6 commits August 22, 2025 14:40

hoist use call

959d160

Merge branch 'main' into implement-first-filter

15d356a

Brings in CenterFilter and Bool content type support while maintaining FirstFilter location tracking

Fix merge compilation errors

f9f63bb

- Update FirstFilter to return Option<Content> to match new API - Add handling for Content::Bool variant in FirstFilter

Remove redundant local import of IntoOwnedContent

632c758

The IntoOwnedContent trait is already imported at module scope, so the local import in FirstFilter::resolve is unnecessary.

Simplify FirstFilter string handling

2cf99a1

- Inline s.as_raw() call to reduce unnecessary variable - More concise and readable code

ipmb force-pushed the implement-first-filter branch from a5acdf3 to 2cf99a1 Compare August 22, 2025 21:06

ipmb added 2 commits August 22, 2025 15:16

Add test for first filter with boolean values

d36981f

Ensures complete coverage of all Content type branches

LilyFirefly reviewed Aug 22, 2025

View reviewed changes

		assert "'NoneType' object is not subscriptable" in str(django_exc.value)
		assert "'NoneType' object is not subscriptable" in str(rusty_exc.value)

Implement the 'first' filter #100

Are you sure you want to change the base?

Implement the 'first' filter #100

Uh oh!

Conversation

ipmb commented Jul 26, 2025

Implementation Details

Django References

Changes

Testing

Uh oh!

codecov bot commented Jul 27, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Codecov Report

Uh oh!

LilyFirefly left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

ipmb commented Aug 22, 2025

Uh oh!

LilyFirefly left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

codecov bot commented Jul 27, 2025 •

edited

Loading