chore: add configurable indentation handling to parser context

Author: gvozdvmozgu

crates/lib/src/core/config.rs

@@ -457,6 +457,17 @@ impl Index<&str> for Value {
 }
 
 impl Value {
+    pub fn to_bool(&self) -> bool {
+        match *self {
+            Value::Int(v) => v != 0,
+            Value::Bool(v) => v,
+            Value::Float(v) => v != 0.0,
+            Value::String(ref v) => !v.is_empty(),
+            Value::Map(ref v) => !v.is_empty(),
+            Value::None => false,
+        }
+    }
+
     pub fn as_map(&self) -> Option<&AHashMap<String, Value>> {
         if let Self::Map(map) = self { Some(map) } else { None }
     }

crates/lib/src/core/parser/context.rs

@@ -1,4 +1,5 @@
 use ahash::AHashMap;
+use itertools::Itertools;
 
 use super::match_result::MatchResult;
 use super::matchable::Matchable;
@@ -14,12 +15,13 @@ pub struct ParseContext {
     match_stack: Vec<String>,
     match_depth: usize,
     track_progress: bool,
-    pub terminators: Vec<Box<dyn Matchable>>,
+    pub(crate) terminators: Vec<Box<dyn Matchable>>,
     parse_cache: AHashMap<((String, (usize, usize), &'static str, usize), String), MatchResult>,
+    pub(crate) indentation_config: AHashMap<String, bool>,
 }
 
 impl ParseContext {
-    pub fn new(dialect: Dialect) -> Self {
+    pub fn new(dialect: Dialect, indentation_config: AHashMap<String, bool>) -> Self {
         Self {
             dialect,
             tqdm: None,
@@ -29,16 +31,21 @@ impl ParseContext {
             track_progress: true,
             terminators: Vec::new(),
             parse_cache: AHashMap::new(),
+            indentation_config,
         }
     }
 
     pub fn dialect(&self) -> &Dialect {
         &self.dialect
     }
 
-    pub fn from_config(_config: FluffConfig) -> Self {
+    pub fn from_config(config: FluffConfig) -> Self {
         let dialect = dialect_selector("ansi").unwrap();
-        Self::new(dialect)
+        let indentation_config = config.raw["indentation"].as_map().unwrap().clone();
+        let indentation_config: AHashMap<_, _> =
+            indentation_config.into_iter().map(|(key, value)| (key, value.to_bool())).collect();
+
+        Self::new(dialect, indentation_config)
     }
 
     pub fn progress_bar<T>(&mut self, mut f: impl FnMut(&mut Self) -> T) -> T {

crates/lib/src/core/parser/grammar/anyof.rs

@@ -332,7 +332,7 @@ mod tests {
                 g.disallow_gaps();
             }
 
-            let mut ctx = ParseContext::new(fresh_ansi_dialect());
+            let mut ctx = ParseContext::new(fresh_ansi_dialect(), <_>::default());
 
             // Check directly
             let mut segments = g.match_segments(&test_segments(), &mut ctx).unwrap();
@@ -347,7 +347,7 @@ mod tests {
 
     #[test]
     fn test__parser__grammar_oneof_templated() {
-        let mut ctx = ParseContext::new(fresh_ansi_dialect());
+        let mut ctx = ParseContext::new(fresh_ansi_dialect(), <_>::default());
 
         let bs = StringParser::new(
             "bar",
@@ -420,7 +420,7 @@ mod tests {
         ];
 
         let segments = generate_test_segments_func(vec!["a", " ", "b", " ", "c", "d", " ", "d"]);
-        let mut parse_cx = ParseContext::new(fresh_ansi_dialect());
+        let mut parse_cx = ParseContext::new(fresh_ansi_dialect(), <_>::default());
 
         for (mode, sequence, terminators, output, max_times) in cases {
             let elements = sequence
@@ -514,7 +514,7 @@ mod tests {
             None,
         );
 
-        let mut ctx = ParseContext::new(fresh_ansi_dialect());
+        let mut ctx = ParseContext::new(fresh_ansi_dialect(), <_>::default());
         let g = AnyNumberOf::new(vec![Box::new(bar), Box::new(foo)]);
         let result = g.match_segments(&segments, &mut ctx).unwrap().matched_segments;
 
@@ -558,7 +558,7 @@ mod tests {
         let g1 = one_of(vec![Box::new(foo_regex.clone()), Box::new(foo.clone())]);
         let g2 = one_of(vec![Box::new(foo), Box::new(foo_regex)]);
 
-        let mut ctx = ParseContext::new(fresh_ansi_dialect());
+        let mut ctx = ParseContext::new(fresh_ansi_dialect(), <_>::default());
 
         for segment in g1.match_segments(&segments, &mut ctx).unwrap().matched_segments.iter() {
             assert_eq!(segment.get_raw().unwrap(), "foo");

crates/lib/src/core/parser/grammar/base.rs

@@ -1,3 +1,6 @@
+use std::borrow::Cow;
+use std::ops::Deref;
+
 use ahash::AHashSet;
 use itertools::enumerate;
 use uuid::Uuid;
@@ -99,7 +102,7 @@ impl Matchable for BaseGrammar {
 
 #[derive(Clone, Hash)]
 pub struct Ref {
-    reference: String,
+    reference: Cow<'static, str>,
     exclude: Option<Box<dyn Matchable>>,
     terminators: Vec<Box<dyn Matchable>>,
     reset_terminators: bool,
@@ -116,9 +119,9 @@ impl std::fmt::Debug for Ref {
 
 impl Ref {
     // Constructor function
-    pub fn new(reference: impl ToString) -> Self {
+    pub fn new(reference: impl Into<Cow<'static, str>>) -> Self {
         Ref {
-            reference: reference.to_string(),
+            reference: reference.into(),
             exclude: None,
             terminators: Vec::new(),
             reset_terminators: false,
@@ -145,7 +148,7 @@ impl Ref {
 
     // Static method to create a Ref instance for a keyword
     pub fn keyword(keyword: &str) -> Self {
-        let name = format!("{}KeywordSegment", capitalize(keyword));
+        let name = capitalize(keyword) + "KeywordSegment";
         Ref::new(name)
     }
 }
@@ -178,7 +181,7 @@ impl Matchable for Ref {
         crumbs: Option<Vec<&str>>,
     ) -> Option<(AHashSet<String>, AHashSet<String>)> {
         if let Some(ref c) = crumbs {
-            if c.contains(&self.reference.as_str()) {
+            if c.contains(&self.reference.deref()) {
                 let loop_string = c.join(" -> ");
                 panic!("Self referential grammar detected: {}", loop_string);
             }
@@ -450,7 +453,7 @@ mod tests {
         // Assuming 'generate_test_segments' and 'fresh_ansi_dialect' are implemented
         // elsewhere
         let ts = generate_test_segments_func(vec!["ABS", "ABSOLUTE"]);
-        let mut ctx = ParseContext::new(fresh_ansi_dialect());
+        let mut ctx = ParseContext::new(fresh_ansi_dialect(), <_>::default());
 
         // Assert ABS does not match, due to the exclude
         assert!(ni.match_segments(&[ts[0].clone()], &mut ctx).unwrap().matched_segments.is_empty());
@@ -463,7 +466,7 @@ mod tests {
 
     #[test]
     fn test_parser_grammar_nothing() {
-        let mut ctx = ParseContext::new(fresh_ansi_dialect());
+        let mut ctx = ParseContext::new(fresh_ansi_dialect(), <_>::default());
 
         assert!(
             Nothing::new()
@@ -488,7 +491,7 @@ mod tests {
             (0..2, "bar", true, (0..2).into()),
         ];
 
-        let mut ctx = ParseContext::new(fresh_ansi_dialect());
+        let mut ctx = ParseContext::new(fresh_ansi_dialect(), <_>::default());
         for (segments_slice, matcher_keyword, trim_noncode, result_slice) in cases {
             let matchers = vec![
                 StringParser::new(
@@ -560,7 +563,7 @@ mod tests {
             Box::new(Sequence::new(vec![bs, fs])),
         ];
 
-        let mut ctx = ParseContext::new(fresh_ansi_dialect());
+        let mut ctx = ParseContext::new(fresh_ansi_dialect(), <_>::default());
         // Matching the first element of the list
         let (match_result, matcher) =
             longest_trimmed_match(&test_segments(), matchers.clone(), &mut ctx, true).unwrap();

crates/lib/src/core/parser/grammar/conditional.rs

@@ -1 +1,96 @@
+use crate::core::errors::SQLParseError;
+use crate::core::parser::context::ParseContext;
+use crate::core::parser::match_result::MatchResult;
+use crate::core::parser::matchable::Matchable;
+use crate::core::parser::segments::base::{ErasedSegment, Segment};
+use crate::core::parser::segments::meta::Indent;
+use crate::helpers::ToErasedSegment;
 
+#[derive(Clone, Debug, Hash, PartialEq)]
+pub struct Conditional {
+    meta: Indent,
+    indented_joins: bool,
+    indented_using_on: bool,
+    indented_on_contents: bool,
+    indented_then: bool,
+}
+
+impl Conditional {
+    pub fn new(meta: Indent) -> Self {
+        Self {
+            meta,
+            indented_joins: false,
+            indented_using_on: false,
+            indented_on_contents: false,
+            indented_then: false,
+        }
+    }
+
+    pub fn indented_joins(mut self) -> Self {
+        self.indented_joins = true;
+        self
+    }
+
+    pub fn indented_using_on(mut self) -> Self {
+        self.indented_using_on = true;
+        self
+    }
+
+    pub fn indented_on_contents(mut self) -> Self {
+        self.indented_on_contents = true;
+        self
+    }
+
+    pub fn indented_then(mut self) -> Self {
+        self.indented_then = true;
+        self
+    }
+
+    pub fn indented_then_contents(mut self) -> Self {
+        self
+    }
+
+    fn is_enabled(&self, parse_context: &mut ParseContext) -> bool {
+        macro_rules! check_config_match {
+            ($self:expr, $parse_context:expr, $field:ident) => {{
+                let config_value = $parse_context
+                    .indentation_config
+                    .get(stringify!($field))
+                    .copied()
+                    .unwrap_or_default();
+
+                if $self.$field && $self.$field != config_value {
+                    return false;
+                }
+            }};
+        }
+
+        check_config_match!(self, parse_context, indented_joins);
+        check_config_match!(self, parse_context, indented_using_on);
+        check_config_match!(self, parse_context, indented_on_contents);
+        check_config_match!(self, parse_context, indented_then);
+
+        true
+    }
+}
+
+impl Segment for Conditional {}
+
+impl Matchable for Conditional {
+    fn match_segments(
+        &self,
+        segments: &[ErasedSegment],
+        parse_context: &mut ParseContext,
+    ) -> Result<MatchResult, SQLParseError> {
+        if !self.is_enabled(parse_context) {
+            return Ok(MatchResult::from_unmatched(segments.to_vec()));
+        }
+
+        dbg!(self);
+
+        Ok(MatchResult {
+            matched_segments: vec![self.meta.clone().to_erased_segment()],
+            unmatched_segments: segments.to_vec(),
+        })
+    }
+}

crates/lib/src/core/parser/grammar/delimited.rs

@@ -306,7 +306,7 @@ mod tests {
             (2.into(), true, false, vec!["bar", ".", "bar", "foo"], 0),
         ];
 
-        let mut ctx = ParseContext::new(fresh_ansi_dialect());
+        let mut ctx = ParseContext::new(fresh_ansi_dialect(), <_>::default());
 
         for (min_delimiters, allow_gaps, allow_trailing, token_list, match_len) in cases {
             let test_segments = generate_test_segments_func(token_list);
@@ -349,7 +349,7 @@ mod tests {
 
     #[test]
     fn test__parser__grammar_anything_bracketed() {
-        let mut ctx = ParseContext::new(fresh_ansi_dialect());
+        let mut ctx = ParseContext::new(fresh_ansi_dialect(), <_>::default());
         let foo = StringParser::new(
             "foo",
             |segment| {
@@ -393,7 +393,7 @@ mod tests {
         for (terminators, match_length) in cases {
             let _panic = enter_panic(terminators.join(" "));
 
-            let mut cx = ParseContext::new(fresh_ansi_dialect());
+            let mut cx = ParseContext::new(fresh_ansi_dialect(), <_>::default());
             let terms = terminators
                 .iter()
                 .map(|it| {

crates/lib/src/core/parser/grammar/noncode.rs

@@ -55,7 +55,7 @@ mod tests {
     #[test]
     fn test__parser__grammar_noncode() {
         let dialect = fresh_ansi_dialect(); // Assuming this function exists and returns a Dialect
-        let mut ctx = ParseContext::new(dialect);
+        let mut ctx = ParseContext::new(dialect, <_>::default());
 
         let matcher = NonCodeMatcher;
         let test_segments = test_segments(); // Assuming this function exists and generates test segments