From 4e04bb47d53970dc0f7c8763d5f5c6ec45306d49 Mon Sep 17 00:00:00 2001
From: "Elf M. Sternberg" <elf.sternberg@gmail.com>
Date: Tue, 13 Oct 2020 17:23:24 -0700
Subject: [PATCH] FEAT: Reference parser is now working.

It's probably not the fastest thing in the world, but
it's going to be enough for now.
---
 server/nm-store/Cargo.toml              |   1 +
 server/nm-store/src/lib.rs              |   1 +
 server/nm-store/src/reference_parser.rs | 119 ++++++++++++++++++++++++
 3 files changed, 121 insertions(+)
 create mode 100644 server/nm-store/src/reference_parser.rs

diff --git a/server/nm-store/Cargo.toml b/server/nm-store/Cargo.toml
index d0d9960..795fa77 100644
--- a/server/nm-store/Cargo.toml
+++ b/server/nm-store/Cargo.toml
@@ -15,6 +15,7 @@ friendly_id = "0.3.0"
 thiserror = "1.0.20"
 derive_builder = "0.9.0"
 lazy_static = "1.4.0"
+comrak = "0.8.2"
 regex = "1.3.9"
 slug = "0.1.4"
 tokio = { version = "0.2.22", features = ["rt-threaded", "blocking"] }
diff --git a/server/nm-store/src/lib.rs b/server/nm-store/src/lib.rs
index 0fa19c4..0ce5da3 100644
--- a/server/nm-store/src/lib.rs
+++ b/server/nm-store/src/lib.rs
@@ -2,6 +2,7 @@ mod errors;
 mod row_structs;
 mod store;
 mod structs;
+mod reference_parser;
 
 pub use crate::errors::NoteStoreError;
 pub use crate::store::NoteStore;
diff --git a/server/nm-store/src/reference_parser.rs b/server/nm-store/src/reference_parser.rs
new file mode 100644
index 0000000..0dbd19b
--- /dev/null
+++ b/server/nm-store/src/reference_parser.rs
@@ -0,0 +1,119 @@
+use comrak::nodes::{AstNode, NodeValue};
+use comrak::{parse_document, Arena, ComrakOptions};
+use lazy_static::lazy_static;
+use regex::bytes::Regex as BytesRegex;
+use regex::Regex;
+
+pub struct Finder(pub Vec<String>);
+
+impl Finder {
+    pub fn new() -> Self {
+        Finder(Vec::new())
+    }
+
+    fn iter_nodes<'a, F>(&mut self, node: &'a AstNode<'a>, f: &F)
+    where
+        F: Fn(&'a AstNode<'a>) -> Option<Vec<String>>,
+    {
+        if let Some(mut v) = f(node) {
+            self.0.append(&mut v);
+        }
+        for c in node.children() {
+            self.iter_nodes(c, f);
+        }
+    }
+}
+
+/// Given a content block, return a list of all the page references found
+/// within the block.  The references may need further massaging.
+pub(crate) fn find_links(document: &str) -> Vec<String> {
+    let arena = Arena::new();
+    let mut finder = Finder::new();
+    let root = parse_document(&arena, document, &ComrakOptions::default());
+
+    finder.iter_nodes(root, &|node| {
+        lazy_static! {
+            static ref RE_REFERENCES: BytesRegex = BytesRegex::new(r"(\[\[([^\]]+)\]\]|(\#[\w\-]+))").unwrap();
+        }
+
+        match &node.data.borrow().value {
+            &NodeValue::Text(ref text) => Some(
+                RE_REFERENCES
+                    .captures_iter(text)
+                    .map(|t| String::from_utf8_lossy(&t.get(1).unwrap().as_bytes()).to_string())
+                    .collect(),
+            ),
+            _ => None,
+        }
+    });
+
+    finder.0
+}
+
+fn recase(title: &str) -> String {
+    lazy_static! {
+        static ref RE_PASS1: Regex = Regex::new(r"(?P<s>.)(?P<n>[A-Z][a-z]+)").unwrap();
+        static ref RE_PASS2: Regex = Regex::new(r"(?P<s>[[:lower:]]|\d)(?P<n>[[:upper:]])").unwrap();
+        static ref RE_PASS4: Regex = Regex::new(r"(?P<s>[a-z])(?P<n>\d)").unwrap();
+        static ref RE_PASS3: Regex = Regex::new(r"(_|-| )+").unwrap();
+    }
+
+    // This should panic if misused, so... :-)
+    let pass = title.to_string();
+    let pass = pass.strip_prefix("#").unwrap();
+
+    let pass = RE_PASS1.replace_all(&pass, "$s $n");
+    let pass = RE_PASS4.replace_all(&pass, "$s $n");
+    let pass = RE_PASS2.replace_all(&pass, "$s $n");
+    RE_PASS3.replace_all(&pass, " ").to_string()
+}
+
+fn build_page_titles(references: &Vec<String>) -> Vec<String> {
+    references
+        .iter()
+        .map(|s| {
+            let c = s.chars().nth(0);
+            match c {
+                Some('#') => recase(s),
+                Some('[') => s.strip_prefix("[[").unwrap().strip_suffix("]]").unwrap().to_string(),
+                Some(_) => s.clone(),
+                _ => "".to_string(),
+            }
+        })
+        .collect()
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    #[test]
+    fn finds_expected() {
+        let sample = r###"
+# Header
+- NotATest 1
+- [[Test 2]]
+- #Test3
+- #TestFourAndSo
+- #Test-six-is-six
+- #
+- [[]]
+
+But *[[Test Seven]]* isn't.  And *#Test_Eight____is_Messed-up*
+And [[Test Bite Me]] is the worst.
+Right? [[
+]]
+"###;
+        let res = build_page_titles(&find_links(sample));
+        let expected = vec![
+            "Test 2",
+            "Test 3",
+            "Test Four And So",
+            "Test six is six",
+            "Test Seven",
+            "Test Eight is Messed up",
+            "Test Bite Me",
+        ];
+        assert!(res.iter().eq(expected.iter()));
+    }
+}