From d69f7f31e4dd19ea69407ee44910bf6db81a03ef Mon Sep 17 00:00:00 2001
From: Aster <aster@vers.site>
Date: Fri, 21 Jun 2024 18:30:52 +0800
Subject: [PATCH] Check the paragraph break

---
 projects/pex-core/src/helpers/comment/mod.rs | 11 +++++++
 projects/pex-core/src/helpers/mod.rs         |  3 +-
 projects/pex-core/src/helpers/string/mod.rs  | 32 ++++++++++++++++++++
 3 files changed, 45 insertions(+), 1 deletion(-)

diff --git a/projects/pex-core/src/helpers/comment/mod.rs b/projects/pex-core/src/helpers/comment/mod.rs
index 376b613..d65ad70 100644
--- a/projects/pex-core/src/helpers/comment/mod.rs
+++ b/projects/pex-core/src/helpers/comment/mod.rs
@@ -51,6 +51,17 @@ impl<'i> FnOnce<(ParseState<'i>,)> for CommentLine {
     }
 }
 
+/// Parse the rest of the line, note this does not catch the newline,
+pub fn rest_of_line(input: ParseState) -> ParseResult<StringView> {
+    let offset = match input.residual.find(&['\r', '\n']) {
+        Some(s) => s,
+        None => input.residual.len(),
+    };
+    // SAFETY: find offset always valid
+    let body = unsafe { input.residual.get_unchecked(0..offset) };
+    input.advance(offset).finish(StringView::new(body, input.start_offset))
+}
+
 /// Parse the comment block
 ///
 /// # Patterns
diff --git a/projects/pex-core/src/helpers/mod.rs b/projects/pex-core/src/helpers/mod.rs
index 2d881cf..65887fe 100644
--- a/projects/pex-core/src/helpers/mod.rs
+++ b/projects/pex-core/src/helpers/mod.rs
@@ -14,7 +14,8 @@ pub use self::{
     comment::{CommentBlock, CommentLine},
     number::*,
     string::{
-        quotation_pair, quotation_pair_escaped, quotation_pair_nested, surround_pair_with_escaper, unescape_us, UnicodeUnescape,
+        paragraph_break, quotation_pair, quotation_pair_escaped, quotation_pair_nested, surround_pair_with_escaper,
+        unescape_us, UnicodeUnescape,
     },
     surround_pair::{SurroundPair, SurroundPattern},
     trie_set::CharactersTrie,
diff --git a/projects/pex-core/src/helpers/string/mod.rs b/projects/pex-core/src/helpers/string/mod.rs
index 64a515a..4fa945f 100644
--- a/projects/pex-core/src/helpers/string/mod.rs
+++ b/projects/pex-core/src/helpers/string/mod.rs
@@ -219,3 +219,35 @@ pub fn unescape_us(input: ParseState) -> ParseResult<char> {
         None => StopBecause::custom_error("Characters must not beyond U+10FFFF", start.start_offset, state.start_offset)?,
     }
 }
+
+/// A period of whitespace with more than two newlines, and terminated by a newline
+pub fn paragraph_break<'i>(input: ParseState<'i>) -> ParseResult<&'i str> {
+    let mut offset = 0;
+    // Capture all newlines and spaces
+    for c in input.residual.chars() {
+        if c.is_whitespace() {
+            offset += c.len_utf8();
+        }
+        else {
+            break;
+        }
+    }
+    let text = unsafe { input.residual.get_unchecked(..offset) };
+    // Fallback for spaces that don't have to be captured
+    for c in text.chars().rev() {
+        if c == ' ' {
+            offset -= c.len_utf8();
+        }
+        else {
+            break;
+        }
+    }
+    if offset == 0 {
+        StopBecause::missing_string("PARAGRAPH_LINE", input.start_offset)?;
+    }
+    let newlines = text.chars().filter(|c| *c == '\n').count();
+    if newlines <= 1 {
+        StopBecause::missing_string("PARAGRAPH_BREAK", input.start_offset)?;
+    }
+    input.advance_view(offset)
+}