From 5ae2a0824e11db1ad7d296db6310fe4c9489202d Mon Sep 17 00:00:00 2001
From: Nico Wagner <n.wagner@dnb.de>
Date: Thu, 29 Aug 2024 13:40:12 +0200
Subject: [PATCH] Add group modifiers (format) (#811)

Signed-off-by: Nico Wagner <n.wagner@dnb.de>
---
 crates/pica-format/src/lib.rs           | 46 +++++++++++++++++++++++++
 crates/pica-format/src/parse.rs         | 39 ++++++++++++++++++---
 crates/pica-format/tests/integration.rs | 41 ++++++++++++++++++++++
 3 files changed, 121 insertions(+), 5 deletions(-)

diff --git a/crates/pica-format/src/lib.rs b/crates/pica-format/src/lib.rs
index 062a6197a..3c8c3c7ff 100644
--- a/crates/pica-format/src/lib.rs
+++ b/crates/pica-format/src/lib.rs
@@ -187,6 +187,36 @@ impl Formatter for Value {
 struct Group {
     fragments: Box<Fragments>,
     bounds: RangeTo<usize>,
+    modifier: Modifier,
+}
+
+#[derive(Debug, Default, Clone, PartialEq)]
+struct Modifier {
+    lowercase: bool,
+    uppercase: bool,
+    remove_ws: bool,
+    trim: bool,
+}
+
+impl Modifier {
+    pub(crate) fn lowercase(&mut self, yes: bool) -> &mut Self {
+        self.lowercase = yes;
+        self
+    }
+
+    pub(crate) fn uppercase(&mut self, yes: bool) -> &mut Self {
+        self.uppercase = yes;
+        self
+    }
+
+    pub(crate) fn remove_ws(&mut self, yes: bool) -> &mut Self {
+        self.remove_ws = yes;
+        self
+    }
+    pub(crate) fn trim(&mut self, yes: bool) -> &mut Self {
+        self.trim = yes;
+        self
+    }
 }
 
 impl Formatter for Group {
@@ -218,6 +248,22 @@ impl Formatter for Group {
             }
         }
 
+        if self.modifier.trim {
+            acc = acc.trim().to_string();
+        }
+
+        if self.modifier.remove_ws {
+            acc = acc.replace(' ', "").to_string();
+        }
+
+        if self.modifier.lowercase {
+            acc = acc.to_lowercase();
+        }
+
+        if self.modifier.uppercase {
+            acc = acc.to_uppercase();
+        }
+
         acc
     }
 }
diff --git a/crates/pica-format/src/parse.rs b/crates/pica-format/src/parse.rs
index b58eab8c4..a4749a129 100644
--- a/crates/pica-format/src/parse.rs
+++ b/crates/pica-format/src/parse.rs
@@ -14,7 +14,7 @@ use winnow::prelude::*;
 use winnow::stream::{AsChar, Compare, Stream, StreamIsPartial};
 use winnow::token::{one_of, take_till};
 
-use crate::{Format, Fragments, Group, List, Value};
+use crate::{Format, Fragments, Group, List, Modifier, Value};
 
 pub fn parse_format(i: &mut &[u8]) -> PResult<Format> {
     (
@@ -40,9 +40,9 @@ pub fn parse_format(i: &mut &[u8]) -> PResult<Format> {
 
 fn parse_fragments(i: &mut &[u8]) -> PResult<Fragments> {
     alt((
-        parse_list.map(Fragments::List),
-        parse_group.map(Fragments::Group),
-        parse_value.map(Fragments::Value),
+        ws(parse_list).map(Fragments::List),
+        ws(parse_group).map(Fragments::Group),
+        ws(parse_value).map(Fragments::Value),
     ))
     .parse_next(i)
 }
@@ -96,9 +96,37 @@ fn decrement_group_level() {
     })
 }
 
+fn parse_modifier(i: &mut &[u8]) -> PResult<Option<Modifier>> {
+    opt(preceded(
+        '?',
+        repeat(1.., alt(('L', 'U', 'T', 'W'))).map(|codes: Vec<_>| {
+            let mut modifier = Modifier::default();
+            if codes.contains(&'L') {
+                modifier.lowercase(true);
+            }
+
+            if codes.contains(&'U') {
+                modifier.uppercase(true);
+            }
+
+            if codes.contains(&'W') {
+                modifier.remove_ws(true);
+            }
+
+            if codes.contains(&'T') {
+                modifier.trim(true);
+            }
+
+            modifier
+        }),
+    ))
+    .parse_next(i)
+}
+
 fn parse_group(i: &mut &[u8]) -> PResult<Group> {
     (
         terminated(ws('('), increment_group_level),
+        parse_modifier,
         parse_fragments,
         ws(')').map(|_| decrement_group_level()),
         alt((
@@ -112,9 +140,10 @@ fn parse_group(i: &mut &[u8]) -> PResult<Group> {
             empty.value(usize::MAX),
         )),
     )
-        .map(|(_, fragments, _, end)| Group {
+        .map(|(_, modifier, fragments, _, end)| Group {
             fragments: Box::new(fragments),
             bounds: RangeTo { end },
+            modifier: modifier.unwrap_or_default(),
         })
         .parse_next(i)
 }
diff --git a/crates/pica-format/tests/integration.rs b/crates/pica-format/tests/integration.rs
index 540215cbe..31329f2fd 100644
--- a/crates/pica-format/tests/integration.rs
+++ b/crates/pica-format/tests/integration.rs
@@ -77,6 +77,47 @@ fn test_format_quantifier() -> TestResult {
     Ok(())
 }
 
+#[test]
+fn test_format_modifier_trim() -> TestResult {
+    let ada = ByteRecord::from_bytes(ada_lovelace()).expect("record");
+    let fmt = Format::from_str("042A{ (?T 'GND-SC: ' a.. ' ') }")?;
+    let result = ada.format(&fmt, &Default::default());
+    assert_eq!(result, vec!["GND-SC: 28p GND-SC: 9.5p"]);
+
+    Ok(())
+}
+
+#[test]
+fn test_format_modifier_remove_ws() -> TestResult {
+    let ada = ByteRecord::from_bytes(ada_lovelace()).expect("record");
+    let fmt = Format::from_str("028A{ (?W d) }")?;
+    let result = ada.format(&fmt, &Default::default());
+    assert_eq!(result, vec!["AdaKing"]);
+
+    Ok(())
+}
+
+#[test]
+fn test_format_modifier_uppercase() -> TestResult {
+    let ada = ByteRecord::from_bytes(ada_lovelace()).expect("record");
+    let fmt =
+        Format::from_str("028A{ (?U a) <$> (?U ', ' d <*> ' ' c) }")?;
+    let result = ada.format(&fmt, &Default::default());
+    assert_eq!(result, vec!["LOVELACE, ADA KING OF".to_string()]);
+
+    Ok(())
+}
+
+#[test]
+fn test_format_modifier_lowercase() -> TestResult {
+    let ada = ByteRecord::from_bytes(ada_lovelace()).expect("record");
+    let fmt = Format::from_str("028A{ a <$> (?L ', ' d <*> ' ' c) }")?;
+    let result = ada.format(&fmt, &Default::default());
+    assert_eq!(result, vec!["Lovelace, ada king of".to_string()]);
+
+    Ok(())
+}
+
 #[test]
 fn test_format_conference() -> TestResult {
     let fmt = Format::new(