From b861469319ac962394d63b40f8ac97488c157e1a Mon Sep 17 00:00:00 2001
From: Thomas Gideon <cmdln@thecommandline.net>
Date: Sun, 16 Jul 2023 08:22:28 -0400
Subject: [PATCH] Apply block quote to multiline content

---
 Cargo.lock    |  1 +
 Cargo.toml    |  1 +
 src/format.rs | 78 ++++++++++++++++++++++++++++++++++++++++++++++++---
 src/main.rs   | 18 +++++++-----
 4 files changed, 87 insertions(+), 11 deletions(-)

diff --git a/Cargo.lock b/Cargo.lock
index 7d44745..7f1d35a 100644
--- a/Cargo.lock
+++ b/Cargo.lock
@@ -932,6 +932,7 @@ dependencies = [
  "html2md",
  "log",
  "megalodon",
+ "regex",
  "rss",
  "rustyline",
  "tokio",
diff --git a/Cargo.toml b/Cargo.toml
index 945b07c..0d41947 100644
--- a/Cargo.toml
+++ b/Cargo.toml
@@ -13,6 +13,7 @@ env_logger = "0.10.0"
 html2md = "0.2.14"
 log = "0.4.19"
 megalodon = "0.8.3"
+regex = "1.9.1"
 rss = "2.0.4"
 rustyline = "12.0.0"
 tokio = { version = "1.28.2", features = ["default", "full"] }
diff --git a/src/format.rs b/src/format.rs
index 405561b..a6b50c3 100644
--- a/src/format.rs
+++ b/src/format.rs
@@ -6,6 +6,7 @@ use megalodon::{
     response::Response,
     Megalodon,
 };
+use regex::{Captures, Regex};
 use rustyline::DefaultEditor;
 use url::Url;
 use url_open::UrlOpen;
@@ -19,7 +20,7 @@ pub(super) async fn format_status(
         "{}
 > {}{}",
         status.created_at.with_timezone(&Local).format("%H:%M"),
-        parse_html(&status.content).trim(),
+        apply_block_quote(1, (&status.content).trim()),
         format_attachments(1, &status.media_attachments)
     );
     let Response { json, .. } = client.get_status_context(status.id.clone(), None).await?;
@@ -35,7 +36,7 @@ pub(super) async fn format_status(
 > {}
 >> {}{}",
                 status.created_at.with_timezone(&Local).format("%H:%M"),
-                parse_html(&status.content).trim(),
+                apply_block_quote(2, parse_html(&status.content).trim()),
                 format_attachments(2, &status.media_attachments)
             )
         })
@@ -52,8 +53,11 @@ pub(super) async fn format_status(
     ))
 }
 
+fn quote_prefix(depth: usize) -> String {
+    vec![">"; depth].join("")
+}
+
 fn format_attachments(depth: usize, attachments: &[Attachment]) -> String {
-    let prefix = vec![">"; depth].join("");
     if attachments.is_empty() {
         String::default()
     } else {
@@ -69,10 +73,76 @@ fn format_attachments(depth: usize, attachments: &[Attachment]) -> String {
                     } else {
                         a.url.clone()
                     };
-                    format!("{} <img src=\"{}\" />", prefix, src)
+                    format!("{} <img src=\"{}\" />", quote_prefix(depth), src)
                 })
                 .collect::<Vec<String>>()
                 .join("\n")
         )
     }
 }
+
+// without processing the content per line, the block quote markdown is only added to the first
+// line, breaking the desired formatting
+fn apply_block_quote<S: AsRef<str>>(depth: usize, content: S) -> String {
+    let prefix = quote_prefix(depth);
+    let empty = Regex::new("\n\\s").expect("Failed to compiled regex for nested empty lines!");
+    let non_empty =
+        Regex::new("\n(?P<initial>[^>\\s])").expect("Failed to compile regex for nested lines!");
+    // parsing some multiline content adds paragraph tags, convert those back to new lines to apply
+    // the block quote level to each line
+    let content = content.as_ref().replace("</p><p>", "\n\n");
+    let content = content.replace("<p>", "");
+    let content = content.replace("</p>", "");
+    // replace separately to avoid trailing spaces when replacing empty lines with the prefix
+    let content = empty.replace(content.as_ref(), format!("\n{}\n", prefix));
+    let content = non_empty.replace_all(&content, |c: &Captures| {
+        format!("\n{} {}", prefix, c["initial"].to_string())
+    });
+    content.to_string()
+}
+
+#[cfg(test)]
+mod test {
+    use crate::format::apply_block_quote;
+
+    #[test]
+    fn test_apply_bq() {
+        assert_eq!(
+            "Foo
+>
+> Bar"
+                .to_owned(),
+            apply_block_quote(
+                1, "Foo
+
+Bar"
+            )
+        );
+    }
+
+    #[test]
+    fn test_apply_bq_deeper() {
+        assert_eq!(
+            "Foo
+>>
+>> Bar"
+                .to_owned(),
+            apply_block_quote(
+                2, "Foo
+
+Bar"
+            )
+        );
+    }
+
+    #[test]
+    fn test_apply_bq_strip_ps() {
+        assert_eq!(
+            "Foo
+>>
+>> Bar"
+                .to_owned(),
+            apply_block_quote(2, "<p>Foo</p><p>Bar</p>")
+        );
+    }
+}
diff --git a/src/main.rs b/src/main.rs
index f8adbee..be22d03 100644
--- a/src/main.rs
+++ b/src/main.rs
@@ -27,12 +27,12 @@ mod range;
 #[derive(Debug, Parser)]
 #[command()]
 struct Config {
-    #[arg(short, long, env = "MASTODON_URL")]
+    #[arg(short, long, env = "MASTODON_URL", required = true)]
     url: String,
-    #[arg(short, long, env = "MASTODON_ACCESS_TOKEN")]
+    #[arg(short, long, env = "MASTODON_ACCESS_TOKEN", required = true)]
     access_token: String,
     #[arg(short, long)]
-    output_dir: String,
+    output_dir: Option<String>,
     #[arg(required = true)]
     date: String,
     #[arg(short, long)]
@@ -108,10 +108,14 @@ async fn main() -> Result<()> {
         }
     }
     reversed.reverse();
-    let output = format!("{}{}.md", output_dir, date);
-    let mut f = File::options().append(true).open(&output)?;
-    f.write_all(&reversed.join("\n\n").as_bytes())?;
-    println!("Appended matching posts to {}.", output);
+    if let Some(output_dir) = output_dir {
+        let output = format!("{}{}.md", output_dir, date);
+        let mut f = File::options().append(true).open(&output)?;
+        f.write_all(&reversed.join("\n\n").as_bytes())?;
+        println!("Appended matching posts to {}.", output);
+    } else {
+        println!("{}", reversed.join("\n\n"));
+    }
     Ok(())
 }