feat: add "deno coverage" subcommand (#8664)

This commit adds a new subcommand called "coverage" 
which can generate code coverage reports to stdout in 
multiple formats from code coverage profiles collected to disk.

Currently this supports outputting a pretty printed diff and 
the lcov format for interoperability with third-party services and tools.

Code coverage is still collected via other subcommands 
that run and collect code coverage such as 
"deno test --coverage=<directory>" but that command no 
longer prints a pretty printed report at the end of a test 
run with coverage collection enabled.

The restrictions on which files that can be reported on has 
also been relaxed and are fully controllable with the include 
and exclude regular expression flags on the coverage subcommand.

Co-authored-by: Luca Casonato <lucacasonato@yahoo.com>

cli/tools/coverage.rs

@@ -3,6 +3,8 @@
 use crate::ast;
 use crate::ast::TokenOrComment;
 use crate::colors;
+use crate::flags::Flags;
+use crate::fs_util::collect_files;
 use crate::media_type::MediaType;
 use crate::module_graph::TypeLib;
 use crate::program_state::ProgramState;
@@ -13,12 +15,12 @@ use deno_core::serde_json::json;
 use deno_core::url::Url;
 use deno_runtime::inspector::InspectorSession;
 use deno_runtime::permissions::Permissions;
+use regex::Regex;
 use serde::Deserialize;
 use serde::Serialize;
 use sourcemap::SourceMap;
 use std::fs;
 use std::path::PathBuf;
-use std::sync::Arc;
 use swc_common::Span;
 use uuid::Uuid;
 
@@ -34,7 +36,6 @@ impl CoverageCollector {
 
   pub async fn start_collecting(&mut self) -> Result<(), AnyError> {
     self.session.post_message("Debugger.enable", None).await?;
-
     self.session.post_message("Profiler.enable", None).await?;
 
     self
@@ -66,11 +67,6 @@ impl CoverageCollector {
       fs::write(self.dir.join(filename), &json)?;
     }
 
-    self
-      .session
-      .post_message("Profiler.stopPreciseCoverage", None)
-      .await?;
-
     self.session.post_message("Profiler.disable", None).await?;
     self.session.post_message("Debugger.disable", None).await?;
 
@@ -104,7 +100,7 @@ pub struct ScriptCoverage {
   pub functions: Vec<FunctionCoverage>,
 }
 
-#[derive(Debug, Deserialize)]
+#[derive(Debug, Serialize, Deserialize)]
 #[serde(rename_all = "camelCase")]
 struct TakePreciseCoverageResult {
   result: Vec<ScriptCoverage>,
@@ -117,17 +113,264 @@ pub struct GetScriptSourceResult {
   pub bytecode: Option<String>,
 }
 
-pub struct PrettyCoverageReporter {
-  quiet: bool,
+pub enum CoverageReporterKind {
+  Pretty,
+  Lcov,
 }
 
-// TODO(caspervonb) add support for lcov output (see geninfo(1) for format spec).
-impl PrettyCoverageReporter {
-  pub fn new(quiet: bool) -> PrettyCoverageReporter {
-    PrettyCoverageReporter { quiet }
+fn create_reporter(
+  kind: CoverageReporterKind,
+) -> Box<dyn CoverageReporter + Send> {
+  match kind {
+    CoverageReporterKind::Lcov => Box::new(LcovCoverageReporter::new()),
+    CoverageReporterKind::Pretty => Box::new(PrettyCoverageReporter::new()),
+  }
+}
+
+pub trait CoverageReporter {
+  fn visit_coverage(
+    &mut self,
+    script_coverage: &ScriptCoverage,
+    script_source: &str,
+    maybe_source_map: Option<Vec<u8>>,
+    maybe_original_source: Option<String>,
+  );
+
+  fn done(&mut self);
+}
+
+pub struct LcovCoverageReporter {}
+
+impl LcovCoverageReporter {
+  pub fn new() -> LcovCoverageReporter {
+    LcovCoverageReporter {}
+  }
+}
+
+impl CoverageReporter for LcovCoverageReporter {
+  fn visit_coverage(
+    &mut self,
+    script_coverage: &ScriptCoverage,
+    script_source: &str,
+    maybe_source_map: Option<Vec<u8>>,
+    _maybe_original_source: Option<String>,
+  ) {
+    // TODO(caspervonb) cleanup and reduce duplication between reporters, pre-compute line coverage
+    // elsewhere.
+    let maybe_source_map = if let Some(source_map) = maybe_source_map {
+      Some(SourceMap::from_slice(&source_map).unwrap())
+    } else {
+      None
+    };
+
+    let url = Url::parse(&script_coverage.url).unwrap();
+    let file_path = url.to_file_path().unwrap();
+    println!("SF:{}", file_path.to_str().unwrap());
+
+    let mut functions_found = 0;
+    for function in &script_coverage.functions {
+      if function.function_name.is_empty() {
+        continue;
+      }
+
+      let source_line = script_source[0..function.ranges[0].start_offset]
+        .split('\n')
+        .count();
+
+      let line_index = if let Some(source_map) = maybe_source_map.as_ref() {
+        source_map
+          .tokens()
+          .find(|token| token.get_dst_line() as usize == source_line)
+          .map(|token| token.get_src_line() as usize)
+          .unwrap_or(0)
+      } else {
+        source_line
+      };
+
+      let function_name = &function.function_name;
+
+      println!("FN:{},{}", line_index + 1, function_name);
+
+      functions_found += 1;
+    }
+
+    let mut functions_hit = 0;
+    for function in &script_coverage.functions {
+      if function.function_name.is_empty() {
+        continue;
+      }
+
+      let execution_count = function.ranges[0].count;
+      let function_name = &function.function_name;
+
+      println!("FNDA:{},{}", execution_count, function_name);
+
+      if execution_count != 0 {
+        functions_hit += 1;
+      }
+    }
+
+    println!("FNF:{}", functions_found);
+    println!("FNH:{}", functions_hit);
+
+    let mut branches_found = 0;
+    let mut branches_hit = 0;
+    for (block_number, function) in script_coverage.functions.iter().enumerate()
+    {
+      let block_hits = function.ranges[0].count;
+      for (branch_number, range) in function.ranges[1..].iter().enumerate() {
+        let source_line =
+          script_source[0..range.start_offset].split('\n').count();
+
+        let line_index = if let Some(source_map) = maybe_source_map.as_ref() {
+          source_map
+            .tokens()
+            .find(|token| token.get_dst_line() as usize == source_line)
+            .map(|token| token.get_src_line() as usize)
+            .unwrap_or(0)
+        } else {
+          source_line
+        };
+
+        // From https://manpages.debian.org/unstable/lcov/geninfo.1.en.html:
+        //
+        // Block number and branch number are gcc internal IDs for the branch. Taken is either '-'
+        // if the basic block containing the branch was never executed or a number indicating how
+        // often that branch was taken.
+        //
+        // However with the data we get from v8 coverage profiles it seems we can't actually hit
+        // this as appears it won't consider any nested branches it hasn't seen but its here for
+        // the sake of accuracy.
+        let taken = if block_hits > 0 {
+          range.count.to_string()
+        } else {
+          "-".to_string()
+        };
+
+        println!(
+          "BRDA:{},{},{},{}",
+          line_index + 1,
+          block_number,
+          branch_number,
+          taken
+        );
+
+        branches_found += 1;
+        if range.count > 0 {
+          branches_hit += 1;
+        }
+      }
+    }
+
+    println!("BRF:{}", branches_found);
+    println!("BRH:{}", branches_hit);
+
+    let lines = script_source.split('\n').collect::<Vec<_>>();
+    let line_offsets = {
+      let mut offsets: Vec<(usize, usize)> = Vec::new();
+      let mut index = 0;
+
+      for line in &lines {
+        offsets.push((index, index + line.len() + 1));
+        index += line.len() + 1;
+      }
+
+      offsets
+    };
+
+    let line_counts = line_offsets
+      .iter()
+      .map(|(line_start_offset, line_end_offset)| {
+        let mut count = 0;
+
+        // Count the hits of ranges that include the entire line which will always be at-least one
+        // as long as the code has been evaluated.
+        for function in &script_coverage.functions {
+          for range in &function.ranges {
+            if range.start_offset <= *line_start_offset
+              && range.end_offset >= *line_end_offset
+            {
+              count += range.count;
+            }
+          }
+        }
+
+        // Reset the count if any block intersects with the current line has a count of
+        // zero.
+        //
+        // We check for intersection instead of inclusion here because a block may be anywhere
+        // inside a line.
+        for function in &script_coverage.functions {
+          for range in &function.ranges {
+            if range.count > 0 {
+              continue;
+            }
+
+            if (range.start_offset < *line_start_offset
+              && range.end_offset > *line_start_offset)
+              || (range.start_offset < *line_end_offset
+                && range.end_offset > *line_end_offset)
+            {
+              count = 0;
+            }
+          }
+        }
+
+        count
+      })
+      .collect::<Vec<usize>>();
+
+    let found_lines = if let Some(source_map) = maybe_source_map.as_ref() {
+      let mut found_lines = line_counts
+        .iter()
+        .enumerate()
+        .map(|(index, count)| {
+          source_map
+            .tokens()
+            .filter(move |token| token.get_dst_line() as usize == index)
+            .map(move |token| (token.get_src_line() as usize, *count))
+        })
+        .flatten()
+        .collect::<Vec<(usize, usize)>>();
+
+      found_lines.sort_unstable_by_key(|(index, _)| *index);
+      found_lines.dedup_by_key(|(index, _)| *index);
+      found_lines
+    } else {
+      line_counts
+        .iter()
+        .enumerate()
+        .map(|(index, count)| (index, *count))
+        .collect::<Vec<(usize, usize)>>()
+    };
+
+    for (index, count) in &found_lines {
+      println!("DA:{},{}", index + 1, count);
+    }
+
+    let lines_hit = found_lines.iter().filter(|(_, count)| *count != 0).count();
+
+    println!("LH:{}", lines_hit);
+
+    let lines_found = found_lines.len();
+    println!("LF:{}", lines_found);
+
+    println!("end_of_record");
   }
 
-  pub fn visit_coverage(
+  fn done(&mut self) {}
+}
+
+pub struct PrettyCoverageReporter {}
+
+impl PrettyCoverageReporter {
+  pub fn new() -> PrettyCoverageReporter {
+    PrettyCoverageReporter {}
+  }
+}
+
+impl CoverageReporter for PrettyCoverageReporter {
+  fn visit_coverage(
     &mut self,
     script_coverage: &ScriptCoverage,
     script_source: &str,
@@ -163,6 +406,8 @@ impl PrettyCoverageReporter {
       offsets
     };
 
+    // TODO(caspervonb): collect uncovered ranges on the lines so that we can highlight specific
+    // parts of a line in color (word diff style) instead of the entire line.
     let line_counts = line_offsets
       .iter()
       .enumerate()
@@ -241,67 +486,72 @@ impl PrettyCoverageReporter {
       line_counts
     };
 
-    if !self.quiet {
-      print!("cover {} ... ", script_coverage.url);
+    print!("cover {} ... ", script_coverage.url);
 
-      let hit_lines = line_counts
-        .iter()
-        .filter(|(_, count)| *count != 0)
-        .map(|(index, _)| *index);
+    let hit_lines = line_counts
+      .iter()
+      .filter(|(_, count)| *count != 0)
+      .map(|(index, _)| *index);
 
-      let missed_lines = line_counts
-        .iter()
-        .filter(|(_, count)| *count == 0)
-        .map(|(index, _)| *index);
+    let missed_lines = line_counts
+      .iter()
+      .filter(|(_, count)| *count == 0)
+      .map(|(index, _)| *index);
 
-      let lines_found = line_counts.len();
-      let lines_hit = hit_lines.count();
-      let line_ratio = lines_hit as f32 / lines_found as f32;
+    let lines_found = line_counts.len();
+    let lines_hit = hit_lines.count();
+    let line_ratio = lines_hit as f32 / lines_found as f32;
 
-      let line_coverage =
-        format!("{:.3}% ({}/{})", line_ratio * 100.0, lines_hit, lines_found,);
+    let line_coverage =
+      format!("{:.3}% ({}/{})", line_ratio * 100.0, lines_hit, lines_found,);
 
-      if line_ratio >= 0.9 {
-        println!("{}", colors::green(&line_coverage));
-      } else if line_ratio >= 0.75 {
-        println!("{}", colors::yellow(&line_coverage));
-      } else {
-        println!("{}", colors::red(&line_coverage));
-      }
+    if line_ratio >= 0.9 {
+      println!("{}", colors::green(&line_coverage));
+    } else if line_ratio >= 0.75 {
+      println!("{}", colors::yellow(&line_coverage));
+    } else {
+      println!("{}", colors::red(&line_coverage));
+    }
 
-      let mut last_line = None;
-      for line_index in missed_lines {
-        const WIDTH: usize = 4;
-        const SEPERATOR: &str = "|";
+    let mut last_line = None;
+    for line_index in missed_lines {
+      const WIDTH: usize = 4;
+      const SEPERATOR: &str = "|";
 
-        // Put a horizontal separator between disjoint runs of lines
-        if let Some(last_line) = last_line {
-          if last_line + 1 != line_index {
-            let dash = colors::gray(&"-".repeat(WIDTH + 1));
-            println!("{}{}{}", dash, colors::gray(SEPERATOR), dash);
-          }
+      // Put a horizontal separator between disjoint runs of lines
+      if let Some(last_line) = last_line {
+        if last_line + 1 != line_index {
+          let dash = colors::gray(&"-".repeat(WIDTH + 1));
+          println!("{}{}{}", dash, colors::gray(SEPERATOR), dash);
         }
-
-        println!(
-          "{:width$} {} {}",
-          line_index + 1,
-          colors::gray(SEPERATOR),
-          colors::red(&lines[line_index]),
-          width = WIDTH
-        );
-
-        last_line = Some(line_index);
       }
+
+      println!(
+        "{:width$} {} {}",
+        line_index + 1,
+        colors::gray(SEPERATOR),
+        colors::red(&lines[line_index]),
+        width = WIDTH
+      );
+
+      last_line = Some(line_index);
     }
   }
+
+  fn done(&mut self) {}
 }
 
-fn collect_coverages(dir: &PathBuf) -> Result<Vec<ScriptCoverage>, AnyError> {
+fn collect_coverages(
+  files: Vec<PathBuf>,
+  ignore: Vec<PathBuf>,
+) -> Result<Vec<ScriptCoverage>, AnyError> {
   let mut coverages: Vec<ScriptCoverage> = Vec::new();
+  let file_paths = collect_files(&files, &ignore, |file_path| {
+    file_path.extension().map_or(false, |ext| ext == "json")
+  })?;
 
-  let entries = fs::read_dir(dir)?;
-  for entry in entries {
-    let json = fs::read_to_string(entry.unwrap().path())?;
+  for file_path in file_paths {
+    let json = fs::read_to_string(file_path.as_path())?;
     let new_coverage: ScriptCoverage = serde_json::from_str(&json)?;
 
     let existing_coverage =
@@ -344,49 +594,52 @@ fn collect_coverages(dir: &PathBuf) -> Result<Vec<ScriptCoverage>, AnyError> {
 
 fn filter_coverages(
   coverages: Vec<ScriptCoverage>,
-  exclude: Vec<Url>,
+  include: Vec<String>,
+  exclude: Vec<String>,
 ) -> Vec<ScriptCoverage> {
+  let include: Vec<Regex> =
+    include.iter().map(|e| Regex::new(e).unwrap()).collect();
+
+  let exclude: Vec<Regex> =
+    exclude.iter().map(|e| Regex::new(e).unwrap()).collect();
+
   coverages
     .into_iter()
     .filter(|e| {
-      if let Ok(url) = Url::parse(&e.url) {
-        if url.path().ends_with("__anonymous__") {
-          return false;
-        }
+      let is_internal = e.url.starts_with("deno:")
+        || e.url.ends_with("__anonymous__")
+        || e.url.ends_with("$deno$test.ts");
 
-        for module_url in &exclude {
-          if &url == module_url {
-            return false;
-          }
-        }
+      let is_included = include.iter().any(|p| p.is_match(&e.url));
+      let is_excluded = exclude.iter().any(|p| p.is_match(&e.url));
 
-        if let Ok(path) = url.to_file_path() {
-          for module_url in &exclude {
-            if let Ok(module_path) = module_url.to_file_path() {
-              if path.starts_with(module_path.parent().unwrap()) {
-                return true;
-              }
-            }
-          }
-        }
-      }
-
-      false
+      (include.is_empty() || is_included) && !is_excluded && !is_internal
     })
     .collect::<Vec<ScriptCoverage>>()
 }
 
-pub async fn report_coverages(
-  program_state: Arc<ProgramState>,
-  dir: &PathBuf,
-  quiet: bool,
-  exclude: Vec<Url>,
+pub async fn cover_files(
+  flags: Flags,
+  files: Vec<PathBuf>,
+  ignore: Vec<PathBuf>,
+  include: Vec<String>,
+  exclude: Vec<String>,
+  lcov: bool,
 ) -> Result<(), AnyError> {
-  let coverages = collect_coverages(dir)?;
-  let coverages = filter_coverages(coverages, exclude);
+  let program_state = ProgramState::build(flags).await?;
 
-  let mut coverage_reporter = PrettyCoverageReporter::new(quiet);
-  for script_coverage in coverages {
+  let script_coverages = collect_coverages(files, ignore)?;
+  let script_coverages = filter_coverages(script_coverages, include, exclude);
+
+  let reporter_kind = if lcov {
+    CoverageReporterKind::Lcov
+  } else {
+    CoverageReporterKind::Pretty
+  };
+
+  let mut reporter = create_reporter(reporter_kind);
+
+  for script_coverage in script_coverages {
     let module_specifier =
       deno_core::resolve_url_or_path(&script_coverage.url)?;
     program_state
@@ -408,7 +661,7 @@ pub async fn report_coverages(
       .get_source(&module_specifier)
       .map(|f| f.source);
 
-    coverage_reporter.visit_coverage(
+    reporter.visit_coverage(
       &script_coverage,
       &script_source,
       maybe_source_map,
@@ -416,5 +669,7 @@ pub async fn report_coverages(
     );
   }
 
+  reporter.done();
+
   Ok(())
 }