From b5ee9a89809f726b5bddb6388eafeadb341dede2 Mon Sep 17 00:00:00 2001
From: Brandon Thomas <echelon@gmail.com>
Date: Mon, 7 Nov 2016 01:18:53 -0500
Subject: [PATCH] Limit number of words or characters in the request

Config options for both.
---
 config.toml                         |  4 ++
 src/config.rs                       | 10 +++++
 src/handlers/audio_synth_handler.rs | 63 ++++++++++++++++-------------
 3 files changed, 48 insertions(+), 29 deletions(-)

diff --git a/config.toml b/config.toml
index e09dc2a900..91a4ee1ac3 100644
--- a/config.toml
+++ b/config.toml
@@ -9,6 +9,10 @@
   square_dictionary_file = "./dictionary/square-dict"
   abbreviation_file = "./dictionary/abbreviations.dict"
 
+  # Input control. Respond with 400 when over limit.
+  max_num_characters = 255
+  max_num_words = 100
+
   # Logging verbosity
   log_handler = false
   log_parsing = false
diff --git a/src/config.rs b/src/config.rs
index 9fe60921f2..92349de484 100644
--- a/src/config.rs
+++ b/src/config.rs
@@ -25,6 +25,10 @@ pub struct Config {
   /// Where the abbreviation file is found.
   pub abbreviation_file: Option<String>,
 
+  /// Input control. Respond with 400 when over limit.
+  pub max_num_characters: Option<usize>,
+  pub max_num_words: Option<usize>,
+
   /// What events to log.
   pub log_handler: Option<bool>,
   pub log_parsing: Option<bool>,
@@ -111,6 +115,12 @@ impl Config {
       abbreviation_file: self.abbreviation_file
           .clone()
           .or(other.abbreviation_file.clone()),
+      max_num_characters: self.max_num_characters
+          .clone()
+          .or(other.max_num_characters.clone()),
+      max_num_words: self.max_num_words
+          .clone()
+          .or(other.max_num_words.clone()),
       log_handler: self.log_handler
           .clone()
           .or(other.log_handler.clone()),
diff --git a/src/handlers/audio_synth_handler.rs b/src/handlers/audio_synth_handler.rs
index bd40857c3e..d6061635d5 100644
--- a/src/handlers/audio_synth_handler.rs
+++ b/src/handlers/audio_synth_handler.rs
@@ -13,6 +13,7 @@ use iron::mime::Mime;
 use iron::prelude::IronResult;
 use iron::prelude::Request;
 use iron::prelude::Response;
+use iron::status::BadRequest;
 use iron::status;
 use lang::parser::Parser;
 use speaker::Speaker;
@@ -215,7 +216,7 @@ impl SpeakRequest {
 impl Handler for AudioSynthHandler {
   /// Process request.
   fn handle(&self, req: &mut Request) -> IronResult<Response> {
-    let sentence_error = build_error(status::BadRequest,
+    let sentence_error = build_error(BadRequest,
         &format!("Missing `{}` parameter.", SENTENCE_PARAM));
 
     let request = match SpeakRequest::parse(req) {
@@ -249,8 +250,39 @@ impl Handler for AudioSynthHandler {
       }
     }
 
+    let max_characters = self.config.max_num_characters.unwrap_or(5_000);
+    if request.sentence.len() > max_characters {
+      return build_error(BadRequest, "Too many characters in request.");
+    }
+
+    // GENERATE AUDIO
+
     let start = PreciseTime::now();
-    let result = try!(self.create_audio(request));
+
+    let synth = try!(self.synthesizer.read().map_err(|_| SynthesisError::LockError));
+
+    let params = SynthesisParams {
+      use_words: request.use_words,
+      use_bare_monophones: request.use_bare_monophones,
+      use_n_phones: request.use_n_phones,
+      use_syllables: request.use_syllables,
+      use_ends: request.use_ends,
+      volume: request.volume,
+      speed: request.speed,
+      padding_between_phones: request.padding_between_phones,
+      polyphone_padding_start: request.polyphone_padding_start,
+      polyphone_padding_end: request.polyphone_padding_end,
+      word_padding_start: request.word_padding_start,
+      word_padding_end: request.word_padding_end,
+    };
+
+    let tokens = self.parser.parse(&request.sentence);
+
+    if tokens.len() > self.config.max_num_words.unwrap_or(1_000) {
+      return build_error(BadRequest, "Too many words in request.");
+    }
+
+    let result = try!(synth.generate(tokens, &request.speaker, params));
 
     info!(target: "timing",
           "Total parsing and synthesis took: {}", start.to(PreciseTime::now()));
@@ -274,33 +306,6 @@ impl AudioSynthHandler {
       config: config,
     }
   }
-
-  /// Create audio from the sentence.
-  fn create_audio(&self, request: SpeakRequest)
-                  -> Result<Vec<u8>, SynthesisError> {
-    let synth = match self.synthesizer.read() {
-      Err(_) => { return Err(SynthesisError::LockError); },
-      Ok(synth) => synth,
-    };
-
-    let params = SynthesisParams {
-      use_words: request.use_words,
-      use_bare_monophones: request.use_bare_monophones,
-      use_n_phones: request.use_n_phones,
-      use_syllables: request.use_syllables,
-      use_ends: request.use_ends,
-      volume: request.volume,
-      speed: request.speed,
-      padding_between_phones: request.padding_between_phones,
-      polyphone_padding_start: request.polyphone_padding_start,
-      polyphone_padding_end: request.polyphone_padding_end,
-      word_padding_start: request.word_padding_start,
-      word_padding_end: request.word_padding_end,
-    };
-
-    let tokens = self.parser.parse(&request.sentence);
-    synth.generate(tokens, &request.speaker, params)
-  }
 }
 
 fn get_str<'a>(params: &'a QueryParams, param_name: &str)