From 4d7bc443f144eb3c18b8354587d3d758b9763cb4 Mon Sep 17 00:00:00 2001
From: Dan Davison <dandavison7@gmail.com>
Date: Mon, 22 Nov 2021 20:15:06 -0500
Subject: [PATCH] New option to map raw styles encountered in input

Unify handling of styles parsed from raw line and computed diff
styles. This enables syntax highlighting to be used in color-moved
sections.

Fixes #72
---
 etc/examples/72-color-moved-2.diff |   8 ++
 src/cli.rs                         |   6 ++
 src/config.rs                      |  26 ++++++
 src/options/set.rs                 |   1 +
 src/paint.rs                       | 135 ++++++++++++++++-------------
 src/style.rs                       | 104 ++++++++++++++++++++++
 6 files changed, 221 insertions(+), 59 deletions(-)
 create mode 100644 etc/examples/72-color-moved-2.diff

diff --git a/etc/examples/72-color-moved-2.diff b/etc/examples/72-color-moved-2.diff
new file mode 100644
index 000000000..69ddc5c87
--- /dev/null
+++ b/etc/examples/72-color-moved-2.diff
@@ -0,0 +1,8 @@
+diff --git a/file.py b/file.py
+index f07db74..3cb162d 100644
+--- a/file.py
++++ b/file.py
+@@ -1,2 +1,2 @@
+-class X: pass
+ class Y: pass
++class X: pass
diff --git a/src/cli.rs b/src/cli.rs
index bc928291f..1691b7d38 100644
--- a/src/cli.rs
+++ b/src/cli.rs
@@ -429,6 +429,12 @@ pub struct Opt {
     /// (underline), 'ol' (overline), or the combination 'ul ol'.
     pub hunk_header_decoration_style: String,
 
+    #[structopt(long = "map-styles")]
+    /// A string specifying a mapping styles encountered in raw input to desired
+    /// output styles. An example is --map-styles='black cyan => white magenta,
+    /// red cyan => white blue'
+    pub map_styles: Option<String>,
+
     /// Format string for git blame commit metadata. Available placeholders are
     /// "{timestamp}", "{author}", and "{commit}".
     #[structopt(
diff --git a/src/config.rs b/src/config.rs
index 1c28d060c..efe6adb0b 100644
--- a/src/config.rs
+++ b/src/config.rs
@@ -20,6 +20,7 @@ use crate::git_config::{GitConfig, GitConfigEntry};
 use crate::minusplus::MinusPlus;
 use crate::paint::BgFillMethod;
 use crate::parse_styles;
+use crate::style;
 use crate::style::Style;
 use crate::tests::TESTING;
 use crate::utils::bat::output::PagingMode;
@@ -105,6 +106,7 @@ pub struct Config {
     pub line_numbers_style_minusplus: MinusPlus<Style>,
     pub line_numbers_zero_style: Style,
     pub line_numbers: bool,
+    pub styles_map: Option<HashMap<style::AnsiTermStyleEqualityKey, Style>>,
     pub max_line_distance_for_naively_paired_lines: f64,
     pub max_line_distance: f64,
     pub max_line_length: usize,
@@ -157,6 +159,7 @@ impl Config {
 impl From<cli::Opt> for Config {
     fn from(opt: cli::Opt) -> Self {
         let styles = parse_styles::parse_styles(&opt);
+        let styles_map = make_styles_map(&opt);
 
         let max_line_distance_for_naively_paired_lines =
             env::get_env_var("DELTA_EXPERIMENTAL_MAX_LINE_DISTANCE_FOR_NAIVELY_PAIRED_LINES")
@@ -297,6 +300,7 @@ impl From<cli::Opt> for Config {
             ),
             line_numbers_zero_style: styles["line-numbers-zero-style"],
             line_buffer_size: opt.line_buffer_size,
+            styles_map,
             max_line_distance: opt.max_line_distance,
             max_line_distance_for_naively_paired_lines,
             max_line_length: match (opt.side_by_side, wrap_max_lines_plus1) {
@@ -396,6 +400,28 @@ fn make_blame_palette(blame_palette: Option<String>, is_light_mode: bool) -> Vec
     }
 }
 
+fn make_styles_map(opt: &cli::Opt) -> Option<HashMap<style::AnsiTermStyleEqualityKey, Style>> {
+    if let Some(styles_map_str) = &opt.map_styles {
+        let mut styles_map = HashMap::new();
+        for pair_str in styles_map_str.split(',') {
+            let mut style_strs = pair_str.split("=>").map(|s| s.trim());
+            if let (Some(from_str), Some(to_str)) = (style_strs.next(), style_strs.next()) {
+                let key = style::ansi_term_style_equality_key(
+                    Style::from_str(from_str, None, None, true, opt.git_config.as_ref())
+                        .ansi_term_style,
+                );
+                styles_map.insert(
+                    key,
+                    Style::from_str(to_str, None, None, true, opt.git_config.as_ref()),
+                );
+            }
+        }
+        Some(styles_map)
+    } else {
+        None
+    }
+}
+
 /// Did the user supply `option` on the command line?
 pub fn user_supplied_option(option: &str, arg_matches: &clap::ArgMatches) -> bool {
     arg_matches.occurrences_of(option) > 0
diff --git a/src/options/set.rs b/src/options/set.rs
index b1dd49d03..4cc2395ec 100644
--- a/src/options/set.rs
+++ b/src/options/set.rs
@@ -157,6 +157,7 @@ pub fn set_options(
             inspect_raw_lines,
             keep_plus_minus_markers,
             line_buffer_size,
+            map_styles,
             max_line_distance,
             max_line_length,
             // Hack: minus-style must come before minus-*emph-style because the latter default
diff --git a/src/paint.rs b/src/paint.rs
index 529065f62..fe958e076 100644
--- a/src/paint.rs
+++ b/src/paint.rs
@@ -1,3 +1,4 @@
+use std::collections::HashMap;
 use std::io::Write;
 
 use itertools::Itertools;
@@ -6,7 +7,6 @@ use syntect::highlighting::Style as SyntectStyle;
 use syntect::parsing::{SyntaxReference, SyntaxSet};
 use unicode_segmentation::UnicodeSegmentation;
 
-use crate::ansi;
 use crate::config::{self, delta_unreachable, Config};
 use crate::delta::State;
 use crate::edits;
@@ -18,6 +18,7 @@ use crate::minusplus::*;
 use crate::paint::superimpose_style_sections::superimpose_style_sections;
 use crate::style::Style;
 use crate::wrapping::wrap_minusplus_block;
+use crate::{ansi, style};
 
 pub struct Painter<'p> {
     pub minus_lines: Vec<(String, State)>,
@@ -134,17 +135,9 @@ impl<'p> Painter<'p> {
         }
     }
 
-    /// Remove the initial +/- character of a line that will be emitted unchanged, including any
-    /// ANSI escape sequences.
-    pub fn prepare_raw_line(&self, line: &str) -> String {
-        ansi::ansi_preserving_slice(
-            &self.expand_tabs(line.graphemes(true)),
-            if self.config.keep_plus_minus_markers {
-                0
-            } else {
-                1
-            },
-        )
+    // Remove initial -/+ character, and expand tabs as spaces, retaining ANSI sequences.
+    pub fn prepare_raw_line(&self, raw_line: &str) -> String {
+        ansi::ansi_preserving_slice(&self.expand_tabs(raw_line.graphemes(true)), 1)
     }
 
     /// Expand tabs as spaces.
@@ -461,33 +454,17 @@ impl<'p> Painter<'p> {
             State::HunkMinus(None) | State::HunkMinusWrapped => {
                 (config.minus_style, config.minus_non_emph_style)
             }
-            State::HunkMinus(Some(raw_line)) => {
-                // TODO: This is the second time we are parsing the ANSI sequences
-                if let Some(ansi_term_style) = ansi::parse_first_style(raw_line) {
-                    let style = Style {
-                        ansi_term_style,
-                        ..Style::new()
-                    };
-                    (style, style)
-                } else {
-                    (config.minus_style, config.minus_non_emph_style)
-                }
-            }
             State::HunkZero | State::HunkZeroWrapped => (config.zero_style, config.zero_style),
             State::HunkPlus(None) | State::HunkPlusWrapped => {
                 (config.plus_style, config.plus_non_emph_style)
             }
-            State::HunkPlus(Some(raw_line)) => {
-                // TODO: This is the second time we are parsing the ANSI sequences
-                if let Some(ansi_term_style) = ansi::parse_first_style(raw_line) {
-                    let style = Style {
-                        ansi_term_style,
-                        ..Style::new()
-                    };
-                    (style, style)
+            State::HunkMinus(Some(_)) | State::HunkPlus(Some(_)) => {
+                let style = if diff_sections.len() > 0 {
+                    diff_sections[diff_sections.len() - 1].0
                 } else {
-                    (config.plus_style, config.plus_non_emph_style)
-                }
+                    config.null_style
+                };
+                (style, style)
             }
             State::Blame(_, _) => (diff_sections[0].0, diff_sections[0].0),
             _ => (config.null_style, config.null_style),
@@ -573,18 +550,6 @@ impl<'p> Painter<'p> {
                 ))
             }
         }
-        match state {
-            State::HunkMinus(Some(raw_line)) | State::HunkPlus(Some(raw_line)) => {
-                // This line has been identified as one which should be emitted unchanged,
-                // including any ANSI escape sequences that it has.
-                return (
-                    format!("{}{}", ansi_term::ANSIStrings(&ansi_strings), raw_line),
-                    false,
-                );
-            }
-            _ => {}
-        }
-
         let superimposed = superimpose_style_sections(
             syntax_sections,
             diff_sections,
@@ -682,19 +647,19 @@ impl<'p> Painter<'p> {
     /// Set background styles to represent diff for minus and plus lines in buffer.
     #[allow(clippy::type_complexity)]
     fn get_diff_style_sections<'a>(
-        minus_lines: &'a [(String, State)],
-        plus_lines: &'a [(String, State)],
+        minus_lines_and_states: &'a [(String, State)],
+        plus_lines_and_states: &'a [(String, State)],
         config: &config::Config,
     ) -> (
         Vec<LineSegments<'a, Style>>,
         Vec<LineSegments<'a, Style>>,
         Vec<(Option<usize>, Option<usize>)>,
     ) {
-        let (minus_lines, minus_styles): (Vec<&str>, Vec<Style>) = minus_lines
+        let (minus_lines, minus_styles): (Vec<&str>, Vec<Style>) = minus_lines_and_states
             .iter()
             .map(|(s, t)| (s.as_str(), *config.get_style(t)))
             .unzip();
-        let (plus_lines, plus_styles): (Vec<&str>, Vec<Style>) = plus_lines
+        let (plus_lines, plus_styles): (Vec<&str>, Vec<Style>) = plus_lines_and_states
             .iter()
             .map(|(s, t)| (s.as_str(), *config.get_style(t)))
             .unzip();
@@ -709,23 +674,30 @@ impl<'p> Painter<'p> {
             config.max_line_distance,
             config.max_line_distance_for_naively_paired_lines,
         );
-
         let minus_non_emph_style = if config.minus_non_emph_style != config.minus_emph_style {
             Some(config.minus_non_emph_style)
         } else {
             None
         };
         let mut lines_style_sections = MinusPlus::new(&mut diff_sections.0, &mut diff_sections.1);
-        Self::update_styles(lines_style_sections[Minus], None, minus_non_emph_style);
+        Self::update_styles(
+            &minus_lines_and_states,
+            lines_style_sections[Minus],
+            None,
+            minus_non_emph_style,
+            config,
+        );
         let plus_non_emph_style = if config.plus_non_emph_style != config.plus_emph_style {
             Some(config.plus_non_emph_style)
         } else {
             None
         };
         Self::update_styles(
+            &plus_lines_and_states,
             lines_style_sections[Plus],
             Some(config.whitespace_error_style),
             plus_non_emph_style,
+            config,
         );
         diff_sections
     }
@@ -738,12 +710,25 @@ impl<'p> Painter<'p> {
     ///    sections.
     /// 2. If the line constitutes a whitespace error, then the whitespace error style
     ///    should be applied to the added material.
-    fn update_styles(
-        lines_style_sections: &mut Vec<LineSegments<'_, Style>>,
+    /// 3. If delta recognized the raw line as one containing ANSI colors that
+    ///    are going to be preserved in the output, then replace delta's
+    ///    computed diff styles with these styles from the raw line. (This is
+    ///    how support for git's --color-moved is implemented.)
+    fn update_styles<'a>(
+        lines_and_states: &'a [(String, State)],
+        lines_style_sections: &mut Vec<LineSegments<'a, Style>>,
         whitespace_error_style: Option<Style>,
         non_emph_style: Option<Style>,
+        config: &config::Config,
     ) {
-        for style_sections in lines_style_sections {
+        for ((_, state), style_sections) in lines_and_states.iter().zip(lines_style_sections) {
+            match state {
+                State::HunkMinus(Some(raw_line)) | State::HunkPlus(Some(raw_line)) => {
+                    *style_sections = parse_style_sections(raw_line, config);
+                    continue;
+                }
+                _ => {}
+            };
             let line_has_emph_and_non_emph_sections =
                 style_sections_contain_more_than_one_style(style_sections);
             let should_update_non_emph_styles =
@@ -766,6 +751,30 @@ impl<'p> Painter<'p> {
     }
 }
 
+// Parse ANSI styles encountered in `raw_line` and apply `styles_map`.
+pub fn parse_style_sections<'a>(
+    raw_line: &'a str,
+    config: &config::Config,
+) -> LineSegments<'a, Style> {
+    let empty_map = HashMap::new();
+    let styles_map = config.styles_map.as_ref().unwrap_or(&empty_map);
+    ansi::parse_style_sections(raw_line)
+        .iter()
+        .map(|(original_style, s)| {
+            match styles_map.get(&style::ansi_term_style_equality_key(*original_style)) {
+                Some(mapped_style) => (*mapped_style, *s),
+                None => (
+                    Style {
+                        ansi_term_style: *original_style,
+                        ..Style::default()
+                    },
+                    *s,
+                ),
+            }
+        })
+        .collect()
+}
+
 #[allow(clippy::too_many_arguments)]
 pub fn paint_file_path_with_line_number(
     line_number: Option<usize>,
@@ -859,17 +868,25 @@ mod superimpose_style_sections {
     use crate::style::Style;
     use crate::utils::bat::terminal::to_ansi_color;
 
+    // We have two different annotations of the same line:
+    // `syntax_style_sections` contains foreground styles computed by syntect,
+    // and `diff_style_sections` contains styles computed by delta reflecting
+    // within-line edits. The delta styles may assign a foreground color, or
+    // they may indicate that the foreground color comes from syntax
+    // highlighting (the is_syntax_highlighting attribute on style::Style). This
+    // function takes in the two input streams and outputs one stream with a
+    // single style assigned to each character.
     pub fn superimpose_style_sections(
-        sections_1: &[(SyntectStyle, &str)],
-        sections_2: &[(Style, &str)],
+        syntax_style_sections: &[(SyntectStyle, &str)],
+        diff_style_sections: &[(Style, &str)],
         true_color: bool,
         null_syntect_style: SyntectStyle,
     ) -> Vec<(Style, String)> {
         coalesce(
             superimpose(
-                explode(sections_1)
+                explode(syntax_style_sections)
                     .iter()
-                    .zip(explode(sections_2))
+                    .zip(explode(diff_style_sections))
                     .collect::<Vec<(&(SyntectStyle, char), (Style, char))>>(),
             ),
             true_color,
diff --git a/src/style.rs b/src/style.rs
index b745cf331..047b9cc99 100644
--- a/src/style.rs
+++ b/src/style.rs
@@ -1,5 +1,6 @@
 use std::borrow::Cow;
 use std::fmt;
+use std::hash::{Hash, Hasher};
 
 use lazy_static::lazy_static;
 
@@ -210,6 +211,89 @@ pub fn ansi_term_style_equality(a: ansi_term::Style, b: ansi_term::Style) -> boo
     }
 }
 
+// TODO: The equality methods were implemented first, and the equality_key
+// methods later. The former should be re-implemented in terms of the latter.
+// But why did the former not address equality of ansi_term::Color::RGB values?
+pub struct AnsiTermStyleEqualityKey {
+    attrs_key: (bool, bool, bool, bool, bool, bool, bool, bool),
+    foreground_key: Option<(u8, u8, u8, u8)>,
+    background_key: Option<(u8, u8, u8, u8)>,
+}
+
+impl PartialEq for AnsiTermStyleEqualityKey {
+    fn eq(&self, other: &Self) -> bool {
+        let option_eq = |opt_a, opt_b| match (opt_a, opt_b) {
+            (Some(a), Some(b)) => a == b,
+            (None, None) => true,
+            _ => false,
+        };
+
+        if self.attrs_key != other.attrs_key {
+            false
+        } else {
+            option_eq(self.foreground_key, other.foreground_key)
+                && option_eq(self.background_key, other.background_key)
+        }
+    }
+}
+
+impl Eq for AnsiTermStyleEqualityKey {}
+
+impl Hash for AnsiTermStyleEqualityKey {
+    fn hash<H: Hasher>(&self, state: &mut H) {
+        self.attrs_key.hash(state);
+        self.foreground_key.hash(state);
+        self.background_key.hash(state);
+    }
+}
+
+pub fn ansi_term_style_equality_key(style: ansi_term::Style) -> AnsiTermStyleEqualityKey {
+    let attrs_key = (
+        style.is_bold,
+        style.is_dimmed,
+        style.is_italic,
+        style.is_underline,
+        style.is_blink,
+        style.is_reverse,
+        style.is_hidden,
+        style.is_strikethrough,
+    );
+    AnsiTermStyleEqualityKey {
+        attrs_key,
+        foreground_key: style.foreground.map(ansi_term_color_equality_key),
+        background_key: style.background.map(ansi_term_color_equality_key),
+    }
+}
+
+impl fmt::Debug for AnsiTermStyleEqualityKey {
+    fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
+        let is_set = |c: char, set: bool| -> String {
+            if set {
+                c.to_uppercase().to_string()
+            } else {
+                c.to_lowercase().to_string()
+            }
+        };
+
+        let (bold, dimmed, italic, underline, blink, reverse, hidden, strikethrough) =
+            self.attrs_key;
+        write!(
+            f,
+            "ansi_term::Style {{ {:?} {:?} {}{}{}{}{}{}{}{} }}",
+            self.foreground_key,
+            self.background_key,
+            is_set('b', bold),
+            is_set('d', dimmed),
+            is_set('i', italic),
+            is_set('u', underline),
+            is_set('l', blink),
+            is_set('r', reverse),
+            is_set('h', hidden),
+            is_set('s', strikethrough),
+        )
+    }
+}
+
 fn ansi_term_color_equality(a: Option<ansi_term::Color>, b: Option<ansi_term::Color>) -> bool {
     match (a, b) {
         (None, None) => true,
@@ -239,6 +323,26 @@ fn ansi_term_16_color_equality(a: ansi_term::Color, b: ansi_term::Color) -> bool
     )
 }
 
+fn ansi_term_color_equality_key(color: ansi_term::Color) -> (u8, u8, u8, u8) {
+    // Same (r, g, b, a) encoding as in utils::bat::terminal::to_ansi_color.
+    // When a = 0xFF, then a 256-color number is stored in the red channel, and
+    // the green and blue channels are meaningless. But a=0 signifies an RGB
+    // color.
+    let default = 0xFF;
+    match color {
+        ansi_term::Color::Fixed(0) | ansi_term::Color::Black => (0, default, default, default),
+        ansi_term::Color::Fixed(1) | ansi_term::Color::Red => (1, default, default, default),
+        ansi_term::Color::Fixed(2) | ansi_term::Color::Green => (2, default, default, default),
+        ansi_term::Color::Fixed(3) | ansi_term::Color::Yellow => (3, default, default, default),
+        ansi_term::Color::Fixed(4) | ansi_term::Color::Blue => (4, default, default, default),
+        ansi_term::Color::Fixed(5) | ansi_term::Color::Purple => (5, default, default, default),
+        ansi_term::Color::Fixed(6) | ansi_term::Color::Cyan => (6, default, default, default),
+        ansi_term::Color::Fixed(7) | ansi_term::Color::White => (7, default, default, default),
+        ansi_term::Color::Fixed(n) => (n, default, default, default),
+        ansi_term::Color::RGB(r, g, b) => (r, g, b, 0),
+    }
+}
+
 lazy_static! {
     pub static ref GIT_DEFAULT_MINUS_STYLE: Style = Style {
         ansi_term_style: ansi_term::Color::Red.normal(),