From 9ea3577a3b12312da4bd759b9f7c367ffe0654b1 Mon Sep 17 00:00:00 2001
From: "Christoph J. Scherr" <Christoph.Scherr@newtec.de>
Date: Thu, 25 Jul 2024 17:02:09 +0200
Subject: [PATCH] feat(german): allow loading any wordlist file, bundle english
 and german

---
 src/error.rs         | 19 +++++++++++++++
 src/wlist/builtin.rs | 56 ++++++++++++++++++++++++++++++++++++++++----
 2 files changed, 71 insertions(+), 4 deletions(-)

diff --git a/src/error.rs b/src/error.rs
index 0b10b8c..e2fac23 100644
--- a/src/error.rs
+++ b/src/error.rs
@@ -13,6 +13,11 @@ pub enum Error {
         #[from]
         source: GameError,
     },
+    #[error("Wordlist Error")]
+    WordlistError {
+        #[from]
+        source: WordlistError,
+    },
     #[error("Solver Error")]
     SolverError {
         #[from]
@@ -68,3 +73,17 @@ pub enum SolverError {
     #[error("Wordlist has no matches for the gamestate")]
     NoMatches,
 }
+
+#[derive(Debug, Error)]
+pub enum WordlistError {
+    #[error("Wordlist has no matches for the gamestate")]
+    BadFormat {
+        #[from]
+        source: serde_json::Error,
+    },
+    #[error("I/O error (probably filesystem related)")]
+    IO {
+        #[from]
+        source: std::io::Error,
+    },
+}
diff --git a/src/wlist/builtin.rs b/src/wlist/builtin.rs
index 116e66f..0b9eaba 100644
--- a/src/wlist/builtin.rs
+++ b/src/wlist/builtin.rs
@@ -1,16 +1,66 @@
 use std::fmt::{write, Debug, Display};
+use std::path::{Path, PathBuf};
 
 use serde_json;
 
+use crate::error::{WResult, WordlistError};
+
 use super::{Word, WordList};
 
-const RAW_WORDLIST_FILE: &str = include_str!("../../data/wordlists/en_US_3b1b_freq_map.json");
+pub const RAW_WORDLIST_BUNDLED_ENGLISH: &str =
+    include_str!("../../data/wordlists/en_US_3b1b_freq_map.json");
+pub const RAW_WORDLIST_BUNDLED_GERMAN_SMALL: &str =
+    include_str!("../../data/wordlists/german_SUBTLEX-DE_small.json");
+pub const RAW_WORDLIST_PATH_ENGLISH: &str = "../../data/wordlists/en_US_3b1b_freq_map.json";
+pub const RAW_WORDLIST_PATH_GERMAN_FULL: &str = "../../data/wordlists/german_SUBTLEX-DE_full.json";
+pub const RAW_WORDLIST_PATH_GERMAN_SMALL: &str = "../../data/wordlists/german_SUBTLEX-DE_full.json";
 
 #[derive(Clone)]
 pub struct BuiltinWList {
     words: super::WordMap,
 }
 
+impl BuiltinWList {
+    /// load a wordlist from file
+    ///
+    /// Wordlist files are expected to have the following format:
+    ///
+    /// ```json
+    /// {
+    ///     "word": 0.001
+    /// }
+    /// ```
+    ///
+    /// Where the number is the frequency. Higher/Lower case is ignored.
+    ///
+    /// ## Errors
+    ///
+    /// Will fail if the file path cannot be read or the format is wrong.
+    pub fn load<P: AsRef<std::path::Path>>(wl_path: P) -> Result<Self, WordlistError> {
+        let path: &Path = wl_path.as_ref();
+        let file = std::fs::File::open(path)?;
+
+        // don't load the whole string into memory
+        let reader = std::io::BufReader::new(file);
+        let words: super::WordMap = serde_json::from_reader(reader)?;
+
+        Ok(Self { words })
+    }
+
+    pub fn english() -> Self {
+        let words: super::WordMap = serde_json::from_str(RAW_WORDLIST_BUNDLED_ENGLISH).unwrap();
+
+        Self { words }
+    }
+
+    pub fn german() -> Self {
+        let words: super::WordMap =
+            serde_json::from_str(RAW_WORDLIST_BUNDLED_GERMAN_SMALL).unwrap();
+
+        Self { words }
+    }
+}
+
 impl super::WordList for BuiltinWList {
     fn length_range(&self) -> impl std::ops::RangeBounds<usize> {
         5..5
@@ -25,9 +75,7 @@ impl super::WordList for BuiltinWList {
 
 impl Default for BuiltinWList {
     fn default() -> Self {
-        let words: super::WordMap = serde_json::from_str(RAW_WORDLIST_FILE).unwrap();
-
-        Self { words }
+        Self::english()
     }
 }