asp/keep
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

refactor: extract helper methods to reduce code duplication in TextPlugin

Browse Source 
Co-authored-by: aider (openai/andrew/openrouter/deepseek/deepseek-chat-v3.1) <aider@aider.chat>
This commit is contained in:
Andrew Phillips
2025-08-27 12:21:52 -03:00
parent 89d62c7b11
commit a3ff74c8e9
1 changed files with 116 additions and 76 deletions
Download Patch File Download Diff File Expand all files Collapse all files

188
src/meta_plugin/text.rs
View File

@@ -215,16 +215,15 @@ impl TextMetaPlugin {
(metadata, is_binary_result) (metadata, is_binary_result)
} }
/// Helper method to output word and line counts /// Helper method to process the remaining UTF-8 buffer and finalize text statistics
fn output_word_line_counts(&mut self) -> Vec<crate::meta_plugin::MetaData> { fn process_remaining_utf8_buffer(&mut self) {
let mut metadata = Vec::new();
// Process any remaining data in utf8_buffer
if !self.utf8_buffer.is_empty() { if !self.utf8_buffer.is_empty() {
self.count_text_stats(&[]); self.count_text_stats(&[]);
} }
}
// Handle the last line if tracking line lengths /// Helper method to handle the last line when tracking line lengths
fn handle_last_line_for_length_tracking(&mut self) {
if self.track_line_lengths && self.current_line_length > 0 { if self.track_line_lengths && self.current_line_length > 0 {
// Update max line length for the last line // Update max line length for the last line
if self.current_line_length > self.max_line_length { if self.current_line_length > self.max_line_length {
@@ -240,37 +239,115 @@ impl TextMetaPlugin {
lengths.push(self.current_line_length); lengths.push(self.current_line_length);
} }
} }
}
/// Helper method to output word count metadata
fn output_word_count_metadata(&self) -> Option<crate::meta_plugin::MetaData> {
if self.track_word_count {
crate::meta_plugin::process_metadata_outputs(
"text_word_count",
self.word_count.to_string(),
self.base.outputs()
)
} else {
None
}
}
/// Helper method to output line count metadata
fn output_line_count_metadata(&self) -> Option<crate::meta_plugin::MetaData> {
if self.track_line_count {
crate::meta_plugin::process_metadata_outputs(
"text_line_count",
self.line_count.to_string(),
self.base.outputs()
)
} else {
None
}
}
/// Helper method to output max line length metadata
fn output_max_line_length_metadata(&self) -> Option<crate::meta_plugin::MetaData> {
if self.output_line_max_len && self.line_count_for_stats > 0 {
crate::meta_plugin::process_metadata_outputs(
"text_line_max_len",
self.max_line_length.to_string(),
self.base.outputs()
)
} else {
None
}
}
/// Helper method to output mean line length metadata
fn output_mean_line_length_metadata(&self) -> Option<crate::meta_plugin::MetaData> {
if self.output_line_mean_len && self.line_count_for_stats > 0 {
let mean_len = self.total_line_length as f64 / self.line_count_for_stats as f64;
// Round to nearest integer
let mean_len_int = mean_len.round() as usize;
crate::meta_plugin::process_metadata_outputs(
"text_line_mean_len",
mean_len_int.to_string(),
self.base.outputs()
)
} else {
None
}
}
/// Helper method to output median line length metadata
fn output_median_line_length_metadata(&self) -> Option<crate::meta_plugin::MetaData> {
if self.output_line_median_len {
if let Some(lengths) = &self.line_lengths {
if !lengths.is_empty() {
let mut sorted_lengths = lengths.clone();
sorted_lengths.sort();
let median_len = if lengths.len() % 2 == 0 {
(sorted_lengths[lengths.len() / 2 - 1] + sorted_lengths[lengths.len() / 2]) as f64 / 2.0
} else {
sorted_lengths[lengths.len() / 2] as f64
};
return crate::meta_plugin::process_metadata_outputs(
"text_line_median_len",
median_len.to_string(),
self.base.outputs()
);
}
}
}
None
}
/// Helper method to output word and line counts
fn output_word_line_counts(&mut self) -> Vec<crate::meta_plugin::MetaData> {
let mut metadata = Vec::new();
// Process any remaining data in utf8_buffer
self.process_remaining_utf8_buffer();
// Handle the last line if tracking line lengths
self.handle_last_line_for_length_tracking();
// Debug: check if outputs are configured // Debug: check if outputs are configured
log::debug!("TEXT: Outputs: {:?}", self.base.outputs()); log::debug!("TEXT: Outputs: {:?}", self.base.outputs());
log::debug!("TEXT: Word count: {}, Line count: {}", self.word_count, self.line_count); log::debug!("TEXT: Word count: {}, Line count: {}", self.word_count, self.line_count);
// Output word count if tracked // Output word count if tracked
if self.track_word_count { if let Some(meta_data) = self.output_word_count_metadata() {
if let Some(meta_data) = crate::meta_plugin::process_metadata_outputs( log::debug!("TEXT: Adding word count metadata: {:?}", meta_data);
"text_word_count", metadata.push(meta_data);
self.word_count.to_string(), } else {
self.base.outputs() log::debug!("TEXT: Word count output is disabled or not mapped");
) {
log::debug!("TEXT: Adding word count metadata: {:?}", meta_data);
metadata.push(meta_data);
} else {
log::debug!("TEXT: Word count output is disabled or not mapped");
}
} }
// Output line count if tracked // Output line count if tracked
if self.track_line_count { if let Some(meta_data) = self.output_line_count_metadata() {
if let Some(meta_data) = crate::meta_plugin::process_metadata_outputs( log::debug!("TEXT: Adding line count metadata: {:?}", meta_data);
"text_line_count", metadata.push(meta_data);
self.line_count.to_string(), } else {
self.base.outputs() log::debug!("TEXT: Line count output is disabled or not mapped");
) {
log::debug!("TEXT: Adding line count metadata: {:?}", meta_data);
metadata.push(meta_data);
} else {
log::debug!("TEXT: Line count output is disabled or not mapped");
}
} }
// Output line length statistics if tracked // Output line length statistics if tracked
@@ -279,58 +356,21 @@ impl TextMetaPlugin {
self.output_line_max_len, self.output_line_mean_len, self.output_line_median_len); self.output_line_max_len, self.output_line_mean_len, self.output_line_median_len);
// Calculate and output max line length if enabled // Calculate and output max line length if enabled
if self.output_line_max_len { if let Some(meta_data) = self.output_max_line_length_metadata() {
if let Some(meta_data) = crate::meta_plugin::process_metadata_outputs( log::debug!("TEXT: Adding max line length metadata: {:?}", meta_data);
"text_line_max_len", metadata.push(meta_data);
self.max_line_length.to_string(),
self.base.outputs()
) {
log::debug!("TEXT: Adding max line length metadata: {:?}", meta_data);
metadata.push(meta_data);
}
} }
// Calculate and output mean line length if enabled // Calculate and output mean line length if enabled
if self.output_line_mean_len { if let Some(meta_data) = self.output_mean_line_length_metadata() {
let mean_len = self.total_line_length as f64 / self.line_count_for_stats as f64; log::debug!("TEXT: Adding mean line length metadata: {:?}", meta_data);
// Round to nearest integer metadata.push(meta_data);
let mean_len_int = mean_len.round() as usize;
if let Some(meta_data) = crate::meta_plugin::process_metadata_outputs(
"text_line_mean_len",
mean_len_int.to_string(),
self.base.outputs()
) {
log::debug!("TEXT: Adding mean line length metadata: {:?}", meta_data);
metadata.push(meta_data);
}
} }
// Calculate and output median line length if enabled // Calculate and output median line length if enabled
if self.output_line_median_len { if let Some(meta_data) = self.output_median_line_length_metadata() {
if let Some(lengths) = &self.line_lengths { log::debug!("TEXT: Adding median line length metadata: {:?}", meta_data);
if !lengths.is_empty() { metadata.push(meta_data);
let mut sorted_lengths = lengths.clone();
sorted_lengths.sort();
let median_len = if lengths.len() % 2 == 0 {
(sorted_lengths[lengths.len() / 2 - 1] + sorted_lengths[lengths.len() / 2]) as f64 / 2.0
} else {
sorted_lengths[lengths.len() / 2] as f64
};
if let Some(meta_data) = crate::meta_plugin::process_metadata_outputs(
"text_line_median_len",
median_len.to_string(),
self.base.outputs()
) {
log::debug!("TEXT: Adding median line length metadata: {:?}", meta_data);
metadata.push(meta_data);
}
} else {
log::debug!("TEXT: No line lengths recorded for median calculation");
}
} else {
log::debug!("TEXT: Line lengths tracking is None for median calculation");
}
} }
} else { } else {
log::debug!("TEXT: Line lengths tracking is disabled or no lines processed"); log::debug!("TEXT: Line lengths tracking is disabled or no lines processed");