From 126c719d2366008bf857b6a40dabd63c8e48b627 Mon Sep 17 00:00:00 2001
From: Wilson Lin <code@wilsonl.in>
Date: Sun, 29 Dec 2019 21:51:25 +1100
Subject: [PATCH] Merge entity code; simplify build IO

---
 build.rs           | 43 ++++++++++++++++++++++++-------------------
 src/spec/entity.rs |  8 --------
 src/spec/mod.rs    |  1 -
 src/unit/entity.rs | 15 +++++++++++----
 4 files changed, 35 insertions(+), 32 deletions(-)
 delete mode 100644 src/spec/entity.rs

diff --git a/build.rs b/build.rs
index b5fdc9a..b734ea1 100644
--- a/build.rs
+++ b/build.rs
@@ -17,6 +17,20 @@ fn create_byte_string_literal(bytes: &[u8]) -> String {
         .collect::<String>())
 }
 
+fn read_json<T>(name: &str) -> T
+    where for<'de> T: Deserialize<'de> {
+    let patterns_path = Path::new("gen").join(format!("{}.json", name));
+    let patterns_file = File::open(patterns_path).unwrap();
+    serde_json::from_reader(patterns_file).unwrap()
+}
+
+fn write_rs(name: &str, code: String) -> () {
+    let out_dir = env::var("OUT_DIR").unwrap();
+    let dest_path = Path::new(&out_dir).join(format!("gen_{}.rs", name));
+    let mut dest_file = File::create(&dest_path).unwrap();
+    dest_file.write_all(code.as_bytes()).unwrap();
+}
+
 struct AutoIncrement {
     next_val: usize,
 }
@@ -107,8 +121,8 @@ fn build_pattern(pattern: String) -> String {
     };
 
     format!("SinglePattern {{ seq: {}, table: &[{}] }}",
-            create_byte_string_literal(pattern.as_bytes()),
-            table.iter().map(|v| v.to_string()).collect::<Vec<String>>().join(", "))
+        create_byte_string_literal(pattern.as_bytes()),
+        table.iter().map(|v| v.to_string()).collect::<Vec<String>>().join(", "))
 }
 
 #[derive(Serialize, Deserialize, Debug)]
@@ -119,9 +133,7 @@ struct Entity {
 
 fn generate_entities() {
     // Read named entities map from JSON file.
-    let entities_path = Path::new("gen").join("entities.json");
-    let entities_file = File::open(entities_path).unwrap();
-    let entities: HashMap<String, Entity> = serde_json::from_reader(entities_file).unwrap();
+    let entities: HashMap<String, Entity> = read_json("entities");
 
     // Add entities to trie builder.
     let mut trie_builder = TrieBuilderNode::new();
@@ -133,20 +145,16 @@ fn generate_entities() {
     let trie_root_id = trie_builder.build(&mut AutoIncrement::new(), "&'static [u8]", &mut trie_code);
 
     // Write trie code to output Rust file.
-    let out_dir = env::var("OUT_DIR").unwrap();
-    let dest_path = Path::new(&out_dir).join("gen_entities.rs");
-    let mut dest_file = File::create(&dest_path).unwrap();
-    dest_file.write_all(trie_code
-        // Make trie root public and use proper variable name.
-        .replace(format!("static N{}:", trie_root_id).as_str(), "pub static ENTITY_REFERENCES:")
-        .as_bytes()).unwrap();
+    // Make trie root public and use proper variable name.
+    write_rs("entities", trie_code.replace(
+        format!("static N{}:", trie_root_id).as_str(),
+        "pub static ENTITY_REFERENCES:",
+    ));
 }
 
 fn generate_patterns() {
     // Read named entities map from JSON file.
-    let patterns_path = Path::new("gen").join("patterns.json");
-    let patterns_file = File::open(patterns_path).unwrap();
-    let patterns: HashMap<String, String> = serde_json::from_reader(patterns_file).unwrap();
+    let patterns: HashMap<String, String> = read_json("patterns");
 
     // Add entities to trie builder.
     let mut code = String::new();
@@ -155,10 +163,7 @@ fn generate_patterns() {
     };
 
     // Write trie code to output Rust file.
-    let out_dir = env::var("OUT_DIR").unwrap();
-    let dest_path = Path::new(&out_dir).join("gen_patterns.rs");
-    let mut dest_file = File::create(&dest_path).unwrap();
-    dest_file.write_all(code.as_bytes()).unwrap();
+    write_rs("patterns", code);
 }
 
 fn main() {
diff --git a/src/spec/entity.rs b/src/spec/entity.rs
deleted file mode 100644
index 2347667..0000000
--- a/src/spec/entity.rs
+++ /dev/null
@@ -1,8 +0,0 @@
-use phf::phf_map;
-use crate::pattern::TrieNode;
-
-include!(concat!(env!("OUT_DIR"), "/gen_entities.rs"));
-
-pub fn is_valid_entity_reference_name_char(c: u8) -> bool {
-    c >= b'0' && c <= b'9' || c >= b'A' && c <= b'Z' || c >= b'a' && c <= b'z'
-}
diff --git a/src/spec/mod.rs b/src/spec/mod.rs
index 013caa6..55b1257 100644
--- a/src/spec/mod.rs
+++ b/src/spec/mod.rs
@@ -1,3 +1,2 @@
 pub mod codepoint;
-pub mod entity;
 pub mod tag;
diff --git a/src/unit/entity.rs b/src/unit/entity.rs
index c3c57be..902961f 100644
--- a/src/unit/entity.rs
+++ b/src/unit/entity.rs
@@ -1,3 +1,9 @@
+use crate::err::ProcessingResult;
+use crate::proc::{Processor, ProcessorRange};
+use crate::spec::codepoint::{is_digit, is_hex_digit, is_lower_hex_digit, is_upper_hex_digit};
+use phf::phf_map;
+use crate::pattern::TrieNode;
+
 // The minimum length of any entity is 3, which is a character entity reference
 // with a single character name. The longest UTF-8 representation of a Unicode
 // code point is 4 bytes. Because there are no character entity references with
@@ -17,10 +23,11 @@
 // - Names must match case sensitively.
 // - Entities that don't have a semicolon do work e.g. `&amp2` => `&2`.
 
-use crate::err::ProcessingResult;
-use crate::proc::{Processor, ProcessorRange};
-use crate::spec::codepoint::{is_digit, is_hex_digit, is_lower_hex_digit, is_upper_hex_digit};
-use crate::spec::entity::{ENTITY_REFERENCES, is_valid_entity_reference_name_char};
+include!(concat!(env!("OUT_DIR"), "/gen_entities.rs"));
+
+fn is_valid_entity_reference_name_char(c: u8) -> bool {
+    c >= b'0' && c <= b'9' || c >= b'A' && c <= b'Z' || c >= b'a' && c <= b'z'
+}
 
 #[derive(Clone, Copy)]
 pub enum EntityType {