From e2bef42a557730be52f4b39c6c74faa720d66387 Mon Sep 17 00:00:00 2001
From: Andrew Phillips <andrew.phillips2@canada.ca>
Date: Tue, 26 Aug 2025 19:37:21 -0300
Subject: [PATCH] feat: fix text plugin word and line count tracking

Co-authored-by: aider (openai/andrew/openrouter/qwen/qwen3-coder) <aider@aider.chat>
---
 src/meta_plugin/text.rs | 71 ++++++++++++++++++++++-------------------
 1 file changed, 39 insertions(+), 32 deletions(-)

diff --git a/src/meta_plugin/text.rs b/src/meta_plugin/text.rs
index ff94123..27043d6 100644
--- a/src/meta_plugin/text.rs
+++ b/src/meta_plugin/text.rs
@@ -120,31 +120,21 @@ impl MetaPlugin for TextMetaPlugin {
             };
         }
         
-        // If we've already determined it's binary, stop processing
-        if self.is_binary_content == Some(true) {
-            return MetaPluginResponse {
-                metadata: Vec::new(),
-                is_finalized: false, // We might still want to finalize later
-            };
-        }
-        
         let mut metadata = Vec::new();
         
-        // Calculate how much data we can still accept
-        let remaining_capacity = self.max_buffer_size.saturating_sub(self.buffer.len());
-        if remaining_capacity > 0 {
-            // Determine how much data to copy
+        // If we haven't determined if content is binary yet, build buffer and check
+        if self.is_binary_content.is_none() {
+            // Add data to our buffer up to max_buffer_size
+            let remaining_capacity = self.max_buffer_size.saturating_sub(self.buffer.len());
             let bytes_to_take = std::cmp::min(data.len(), remaining_capacity);
-
-            // Add data to our buffer
             self.buffer.extend_from_slice(&data[..bytes_to_take]);
             
             // If we have enough data to make a binary determination, do it now
-            if self.buffer.len() >= std::cmp::min(1024, self.max_buffer_size) && self.is_binary_content.is_none() {
+            if self.buffer.len() >= std::cmp::min(1024, self.max_buffer_size) {
                 let is_binary_result = is_binary(&self.buffer);
                 self.is_binary_content = Some(is_binary_result);
                 
-                // Output text and binary status immediately
+                // Output text and binary status
                 let text_value = if is_binary_result { "false".to_string() } else { "true".to_string() };
                 let binary_value = if is_binary_result { "true".to_string() } else { "false".to_string() };
                 
@@ -173,24 +163,34 @@ impl MetaPlugin for TextMetaPlugin {
                         is_finalized: true,
                     };
                 }
-            }
-            
-            // If content is text, count words and lines
-            if self.is_binary_content == Some(false) {
+                
+                // If it's text, count words and lines for this chunk and stop buffering
+                self.count_text_stats(&data[..bytes_to_take]);
+                
+                // If we've reached our buffer limit, we're finalized
+                if self.buffer.len() >= self.max_buffer_size {
+                    self.is_finalized = true;
+                }
+            } else {
+                // Still building up buffer, count words and lines for this chunk
                 self.count_text_stats(&data[..bytes_to_take]);
             }
+        } else if self.is_binary_content == Some(false) {
+            // We've already determined it's text, just count words and lines
+            self.count_text_stats(data);
         }
-
-        // If we've reached our buffer limit and haven't finalized yet
-        if self.buffer.len() >= self.max_buffer_size && !self.is_finalized {
-            // Mark as finalized but don't output word/line counts here
+        // If is_binary_content == Some(true), we should have already finalized, but just in case:
+        else if self.is_binary_content == Some(true) {
             self.is_finalized = true;
+            return MetaPluginResponse {
+                metadata: Vec::new(),
+                is_finalized: true,
+            };
         }
 
-        let is_finalized = self.is_finalized;
         MetaPluginResponse {
             metadata,
-            is_finalized,
+            is_finalized: self.is_finalized,
         }
     }
 
@@ -205,7 +205,7 @@ impl MetaPlugin for TextMetaPlugin {
         
         let mut metadata = Vec::new();
         
-        // If we haven't determined binary status yet, do it now
+        // If we haven't determined binary status yet, do it now with whatever we have
         if self.is_binary_content.is_none() && !self.buffer.is_empty() {
             let is_binary_result = is_binary(&self.buffer);
             self.is_binary_content = Some(is_binary_result);
@@ -229,18 +229,25 @@ impl MetaPlugin for TextMetaPlugin {
             ) {
                 metadata.push(meta_data);
             }
+            
+            // If it's binary, we're done
+            if is_binary_result {
+                self.is_finalized = true;
+                return MetaPluginResponse {
+                    metadata,
+                    is_finalized: true,
+                };
+            }
         }
         
-        // If content is text and we have some data, output word and line counts
-        if self.is_binary_content == Some(false) && (!self.buffer.is_empty() || self.word_count > 0 || self.line_count > 0) {
+        // If content is text, output word and line counts
+        if self.is_binary_content == Some(false) {
             // Process any remaining data in utf8_buffer
             if !self.utf8_buffer.is_empty() {
                 self.count_text_stats(&[]);
             }
             
-            // If we're still in a word at the end of the stream, we've counted it correctly
-            // No special handling needed for this case
-            
+            // Output word and line counts
             if let Some(meta_data) = crate::meta_plugin::process_metadata_outputs(
                 "text_word_count", 
                 self.word_count.to_string(),