🎞 [src] image URLs in /posts (#52)

* 🎞  [src] image URLs in /posts

* fixes for code review

Author: chicks-net

src/main.rs

@@ -101,15 +101,32 @@ fn process_file(file_name: &str, dest_dir: &str) {
     // Note: html5ever may report parsing errors, but they typically don't affect extraction
 
     let post_data = extract_post_data(&dom.document);
-    let markdown_content = generate_markdown(&post_data);
 
-    // Generate output filename
+    // Generate output filename and extract date prefix
     let input_filename = file_path
         .file_stem()
         .unwrap_or_else(|| panic!("Failed to extract filename stem from: {}", file_name))
         .to_str()
         .unwrap_or_else(|| panic!("Filename contains invalid UTF-8: {}", file_name));
-    let output_filename = format!("{}.md", format_filename_date(input_filename));
+    let formatted_name = format_filename_date(input_filename);
+    let output_filename = format!("{}.md", formatted_name);
+
+    // Extract date prefix (YYYY-MM-DD) from formatted filename
+    // Check if it matches the expected format: YYYY-MM-DD
+    let date_prefix = if formatted_name.len() >= 10
+        && formatted_name.as_bytes().get(4) == Some(&b'-')
+        && formatted_name.as_bytes().get(7) == Some(&b'-')
+    {
+        &formatted_name[..10]
+    } else {
+        eprintln!(
+            "Warning: Could not extract date prefix from filename '{}' - images will use empty date prefix",
+            formatted_name
+        );
+        ""
+    };
+
+    let markdown_content = generate_markdown(&post_data, date_prefix);
     let output_path = Path::new(dest_dir).join(output_filename);
 
     // Write markdown file

src/markdown.rs

@@ -2,9 +2,32 @@
 
 use crate::models::PostData;
 use crate::utils::{clean_title, escape_toml_string};
+use std::path::Path;
+
+/// Transform an image path to /posts/YYYY-MM-DD-filename.ext format
+///
+/// # Arguments
+/// * `image_path` - The original image path from the HTML
+/// * `date_prefix` - The date prefix (YYYY-MM-DD) from the post filename
+///
+/// # Returns
+/// The transformed path in the format /posts/YYYY-MM-DD-filename.ext
+fn transform_image_path(image_path: &str, date_prefix: &str) -> String {
+    // Extract just the filename from the path
+    let filename = Path::new(image_path)
+        .file_name()
+        .and_then(|name| name.to_str())
+        .unwrap_or(image_path);
+
+    format!("/posts/{}-{}", date_prefix, filename)
+}
 
 /// Generate markdown from post data
-pub fn generate_markdown(post_data: &PostData) -> String {
+///
+/// # Arguments
+/// * `post_data` - The post data to generate markdown from
+/// * `date_prefix` - The date prefix (YYYY-MM-DD) for image paths
+pub fn generate_markdown(post_data: &PostData, date_prefix: &str) -> String {
     let mut markdown = String::new();
 
     // Generate TOML front matter
@@ -101,7 +124,8 @@ pub fn generate_markdown(post_data: &PostData) -> String {
     if !post_data.images.is_empty() {
         markdown.push_str("## Images\n\n");
         for image_url in &post_data.images {
-            markdown.push_str(&format!("![Image]({})\n\n", image_url));
+            let transformed_path = transform_image_path(image_url, date_prefix);
+            markdown.push_str(&format!("![Image]({})\n\n", transformed_path));
         }
     }
 
@@ -147,3 +171,102 @@ pub fn generate_markdown(post_data: &PostData) -> String {
 
     format!("{}\n", markdown.trim_end())
 }
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    #[test]
+    fn test_transform_image_path_standard() {
+        assert_eq!(
+            transform_image_path("../Photos/Photos from posts/pretty/image.jpg", "2011-11-04"),
+            "/posts/2011-11-04-image.jpg"
+        );
+    }
+
+    #[test]
+    fn test_transform_image_path_complex_path() {
+        assert_eq!(
+            transform_image_path(
+                "../Photos/Photos%20from%20posts/Vasquez%20Rocks/183zw3ui6c0yq.jpg",
+                "2012-11-01"
+            ),
+            "/posts/2012-11-01-183zw3ui6c0yq.jpg"
+        );
+    }
+
+    #[test]
+    fn test_transform_image_path_simple_filename() {
+        assert_eq!(
+            transform_image_path("image.jpg", "2013-07-13"),
+            "/posts/2013-07-13-image.jpg"
+        );
+    }
+
+    #[test]
+    fn test_transform_image_path_empty_date() {
+        assert_eq!(transform_image_path("image.jpg", ""), "/posts/-image.jpg");
+    }
+
+    #[test]
+    fn test_transform_image_path_url() {
+        assert_eq!(
+            transform_image_path("http://example.com/path/image.png", "2013-07-13"),
+            "/posts/2013-07-13-image.png"
+        );
+    }
+
+    #[test]
+    fn test_transform_image_path_windows_path() {
+        // On non-Windows systems, Path won't recognize Windows separators
+        // and will treat the entire string as a filename
+        #[cfg(windows)]
+        {
+            assert_eq!(
+                transform_image_path("C:\\Photos\\image.jpg", "2015-01-01"),
+                "/posts/2015-01-01-image.jpg"
+            );
+        }
+        #[cfg(not(windows))]
+        {
+            // On Unix systems, backslashes are valid filename characters, not separators
+            assert_eq!(
+                transform_image_path("C:\\Photos\\image.jpg", "2015-01-01"),
+                "/posts/2015-01-01-C:\\Photos\\image.jpg"
+            );
+        }
+    }
+
+    #[test]
+    fn test_transform_image_path_absolute_path() {
+        assert_eq!(
+            transform_image_path("/var/data/photos/image.jpg", "2016-06-15"),
+            "/posts/2016-06-15-image.jpg"
+        );
+    }
+
+    #[test]
+    fn test_transform_image_path_special_chars() {
+        assert_eq!(
+            transform_image_path("../Photos/my photo (1).jpg", "2017-03-20"),
+            "/posts/2017-03-20-my photo (1).jpg"
+        );
+    }
+
+    #[test]
+    fn test_transform_image_path_no_extension() {
+        assert_eq!(
+            transform_image_path("../Photos/image", "2018-12-25"),
+            "/posts/2018-12-25-image"
+        );
+    }
+
+    #[test]
+    fn test_transform_image_path_fallback_when_no_filename() {
+        // Edge case: if Path can't extract a filename, use original
+        assert_eq!(
+            transform_image_path("..", "2019-01-01"),
+            "/posts/2019-01-01-.."
+        );
+    }
+}