Add --xml option to save the parsed code into XML, using node.start_byte() and node.end_byte()

Cargo.lock

@@ -272,6 +272,15 @@ version = "0.3.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "9b919933a397b79c37e33b77bb2aa3dc8eb6e165ad809e58ff75bc7db2e34574"
 
+[[package]]
+name = "html-escape"
+version = "0.2.6"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "d348900ce941b7474395ba922ed3735a517df4546a2939ddb416ce85eeaa988e"
+dependencies = [
+ "utf8-width",
+]
+
 [[package]]
 name = "idna"
 version = "0.1.5"
@@ -841,6 +850,7 @@ dependencies = [
  "difference",
  "dirs",
  "glob",
+ "html-escape",
  "lazy_static",
  "libloading",
  "log",
@@ -940,6 +950,12 @@ version = "1.0.2"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "796f7e48bef87609f7ade7e06495a87d5cd06c7866e6a5cbfceffc558a243737"
 
+[[package]]
+name = "utf8-width"
+version = "0.1.4"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "9071ac216321a4470a69fb2b28cfc68dcd1a39acd877c8be8e014df6772d8efa"
+
 [[package]]
 name = "vec_map"
 version = "0.8.1"

cli/Cargo.toml

@@ -36,6 +36,7 @@ serde_derive = "1.0"
 smallbitvec = "2.3.0"
 tiny_http = "0.6"
 webbrowser = "0.5.1"
+html-escape = "0.2.6"
 
 [dependencies.tree-sitter]
 version = ">= 0.17.0"

cli/src/main.rs

@@ -63,6 +63,7 @@ fn run() -> error::Result<()> {
                 .arg(Arg::with_name("scope").long("scope").takes_value(true))
                 .arg(Arg::with_name("debug").long("debug").short("d"))
                 .arg(Arg::with_name("debug-graph").long("debug-graph").short("D"))
+                .arg(Arg::with_name("debug-xml").long("xml").short("x"))
                 .arg(Arg::with_name("quiet").long("quiet").short("q"))
                 .arg(Arg::with_name("stat").long("stat").short("s"))
                 .arg(Arg::with_name("time").long("time").short("t"))
@@ -222,6 +223,7 @@ fn run() -> error::Result<()> {
     } else if let Some(matches) = matches.subcommand_matches("parse") {
         let debug = matches.is_present("debug");
         let debug_graph = matches.is_present("debug-graph");
+        let debug_xml = matches.is_present("debug-xml");
         let quiet = matches.is_present("quiet");
         let time = matches.is_present("time");
         let edits = matches
@@ -257,6 +259,7 @@ fn run() -> error::Result<()> {
                 timeout,
                 debug,
                 debug_graph,
+                debug_xml,
                 Some(&cancellation_flag),
             )?;
 

cli/src/parse.rs

@@ -40,6 +40,7 @@ pub fn parse_file_at_path(
     timeout: u64,
     debug: bool,
     debug_graph: bool,
+    debug_xml: bool,
     cancellation_flag: Option<&AtomicUsize>,
 ) -> Result<bool> {
     let mut _log_session = None;
@@ -151,6 +152,60 @@ pub fn parse_file_at_path(
             println!("");
         }
 
+        if debug_xml {
+            let mut needs_newline = false;
+            let mut indent_level = 0;
+            let mut did_visit_children = false;
+            let mut tags: Vec<&str> = Vec::new();
+            loop {
+                let node = cursor.node();
+                let is_named = node.is_named();
+                if did_visit_children {
+                    if is_named {
+                        let tag = tags.pop();
+                        write!(&mut stdout, "</{}>\n", tag.expect("there is a tag"))?;
+                        needs_newline = true;
+                    }
+                    if cursor.goto_next_sibling() {
+                        did_visit_children = false;
+                    } else if cursor.goto_parent() {
+                        did_visit_children = true;
+                        indent_level -= 1;
+                    } else {
+                        break;
+                    }
+                } else {
+                    if is_named {
+                        if needs_newline {
+                            stdout.write(b"\n")?;
+                        }
+                        for _ in 0..indent_level {
+                            stdout.write(b"  ")?;
+                        }
+                        write!(&mut stdout, "<{}", node.kind())?;
+                        if let Some(field_name) = cursor.field_name() {
+                            write!(&mut stdout, " type=\"{}\"", field_name)?;
+                        }
+                        write!(&mut stdout, ">")?;
+                        tags.push(node.kind());
+                        needs_newline = true;
+                    }
+                    if cursor.goto_first_child() {
+                        did_visit_children = false;
+                        indent_level += 1;
+                    } else {
+                        did_visit_children = true;
+                        let start = node.start_byte();
+                        let end = node.end_byte();
+                        let value = std::str::from_utf8(&source_code[start..end]).expect("has a string");
+                        write!(&mut stdout, "{}", html_escape::encode_text(value))?;
+                    }
+                }
+            }
+            cursor.reset(tree.root_node());
+            println!("");
+        }
+
         let mut first_error = None;
         loop {
             let node = cursor.node();