Make some things more solid

patchparser/src/lib.rs

@@ -6,7 +6,7 @@ pub mod format_binary;
 pub mod from_lines;
 pub mod line;
 pub mod patch;
-pub mod re;
 pub mod reborrow_in;
+pub mod regex_utils;
 pub mod utils;
 pub mod write_to;

patchparser/src/patch/change.rs

@@ -31,7 +31,7 @@ impl<'a, 'h> Change<'a, 'h> {
         'a: 'b,
         'h: 'b,
     {
-        let mut lines = bc::Vec::new_in(bump);
+        let mut remaining_lines = bc::Vec::new_in(bump);
 
         let Self {
             orig_start,
@@ -44,31 +44,34 @@ impl<'a, 'h> Change<'a, 'h> {
             post,
         } = self;
 
-        // Does the header need to be adapted to the following
-        // patterns? As discovered for `split_hunk` -- cj: Let's just
-        // always print the multi-line range format, it should always
-        // work.
-        // @@ -0,0 +1,2 @@
-        // @@ -42 42 @@
-        // @@ -42 +1,2 @@
-        // @@ -0,0 +1 @@
-        let mut content = BString::new_in(bump);
-        content.extend_from_slice(
-            format!(
-                "@@ -{},{} +{},{} ",
-                orig_start, orig_len, patched_start, patched_len
-            )
-            .as_bytes(),
-        );
-        content.extend_from_slice(head_post);
-        lines.push(Line::from_generated_content(content.into_bump_slice()));
+        let head_line = {
+            // Does the header need to be adapted to the following
+            // patterns? As discovered for `split_hunk` -- cj: Let's just
+            // always print the multi-line range format, it should always
+            // work.
+            // @@ -0,0 +1,2 @@
+            // @@ -42 42 @@
+            // @@ -42 +1,2 @@
+            // @@ -0,0 +1 @@
+            let mut content = BString::new_in(bump);
+            content.extend_from_slice(
+                format!(
+                    "@@ -{},{} +{},{} ",
+                    orig_start, orig_len, patched_start, patched_len
+                )
+                .as_bytes(),
+            );
+            content.extend_from_slice(head_post);
+            Line::from_generated_content(content.into_bump_slice())
+        };
 
-        lines.extend_from_slice(pre);
-        lines.extend_from_slice(group);
-        lines.extend_from_slice(post);
+        remaining_lines.extend_from_slice(pre);
+        remaining_lines.extend_from_slice(group);
+        remaining_lines.extend_from_slice(post);
 
         Hunk {
-            lines: BumpaloCow::Owned(lines),
+            head_line,
+            remaining_lines: BumpaloCow::Owned(remaining_lines),
         }
     }
 }

patchparser/src/patch/change_line.rs

@@ -0,0 +1,99 @@
+use anyhow::{anyhow, Result};
+
+use crate::line::Line;
+
+#[derive(Debug, Clone, Copy, PartialEq, Eq)]
+pub(crate) enum ChangeLineKind {
+    /// ' '
+    Context,
+    /// '+'
+    Plus,
+    /// '-'
+    Minus,
+    /// "\ No newline at end of file"
+    Backslash,
+}
+
+#[derive(Debug, Clone, Copy, PartialEq, Eq)]
+pub(crate) enum ChangeTerminator {
+    // "diff "
+    Diff,
+    // "@@ "
+    Hunk,
+}
+
+#[derive(Debug)]
+pub(crate) enum ChangeLineReport {
+    Kind(ChangeLineKind),
+    Terminator(ChangeTerminator),
+    InvalidSyntax(anyhow::Error),
+}
+
+impl From<Line<'_>> for ChangeLineReport {
+    fn from(line: Line) -> Self {
+        use ChangeLineKind::*;
+        use ChangeLineReport::*;
+        use ChangeTerminator::*;
+        match line.first() {
+            Some(c) => match *c {
+                b' ' => Kind(Context),
+                b'+' => Kind(Plus),
+                b'-' => Kind(Minus),
+                b'\\' => Kind(Backslash),
+                _ => {
+                    if line.starts_with(b"@@ ") {
+                        Terminator(Hunk)
+                    } else if line.starts_with(b"diff ") {
+                        Terminator(Diff)
+                    } else {
+                        InvalidSyntax(anyhow!(
+                            "invalid syntax in change on line {line}: unknown line start"
+                        ))
+                    }
+                }
+            },
+            None => InvalidSyntax(anyhow!(
+                "invalid syntax in change on line {line}: empty line"
+            )),
+        }
+    }
+}
+
+impl ChangeLineReport {
+    /// Returns the type expected by `try_take_while` for predicates
+    pub(crate) fn matches_kinds(self, kinds: &[ChangeLineKind]) -> Result<(), ChangeLineReport> {
+        match self {
+            ChangeLineReport::Kind(change_line_kind) => {
+                if kinds.contains(&change_line_kind) {
+                    Ok(())
+                } else {
+                    Err(self)
+                }
+            }
+            _ => Err(self),
+        }
+    }
+
+    pub(crate) fn kind_or_terminator(self) -> Result<ChangeLineReport> {
+        match self {
+            ChangeLineReport::InvalidSyntax(error) => Err(error),
+            t => Ok(t),
+        }
+    }
+}
+
+pub trait SeparateErrors: Sized {
+    type Error;
+    fn separate_errors(self) -> Result<Self, Self::Error>;
+}
+
+impl SeparateErrors for Option<ChangeLineReport> {
+    type Error = anyhow::Error;
+
+    fn separate_errors(self) -> Result<Self, Self::Error> {
+        match self {
+            Some(report) => Ok(Some(report.kind_or_terminator()?)),
+            None => Ok(None),
+        }
+    }
+}

patchparser/src/patch/diff.rs

@@ -256,7 +256,8 @@ fn gather_hunks<'s>(lines: &'s [Line<'s>], bump: &'s Bump) -> bc::Vec<'s, Hunk<'
         lines,
         |line| line.starts_with(b"@@ "),
         |group| Hunk {
-            lines: BumpaloCow::Borrowed(group),
+            head_line: group[0],
+            remaining_lines: BumpaloCow::Borrowed(&group[1..]),
         },
         bump,
     )