Validator: Fix compiler warnings (signed/unsigned)

This also fixes a regression in validate_grapheme_test introduced
by commit 32e9d7c.

Signed-off-by: Stefan Weil <[email protected]>

Author: stweil

src/training/validate_indic.cpp

@@ -120,7 +120,7 @@ bool ValidateIndic::ConsumeViramaIfValid(IndicPair joiner, bool post_matra) {
         ASSERT_HOST(!CodeOnlyToOutput());
       } else {
         // Half-form with optional Nukta.
-        int len = output_.size() + 1 - output_used_;
+        unsigned len = output_.size() + 1 - output_used_;
         if (UseMultiCode(len)) return true;
       }
       if (codes_used_ < num_codes &&
@@ -179,7 +179,7 @@ bool ValidateIndic::ConsumeConsonantHeadIfValid() {
     CodeOnlyToOutput();
     // Special Sinhala case of [H Z Yayana/Rayana].
     int index = output_.size() - 3;
-    if (output_used_ <= index &&
+    if (output_used_ + 3 <= output_.size() &&
         (output_.back() == kYayana || output_.back() == kRayana) &&
         IsVirama(output_[index]) && output_[index + 1] == kZeroWidthJoiner) {
       MultiCodePart(3);
@@ -192,7 +192,7 @@ bool ValidateIndic::ConsumeConsonantHeadIfValid() {
     }
     // Test for subscript conjunct.
     index = output_.size() - 2 - have_nukta;
-    if (output_used_ <= index && IsSubscriptScript() &&
+    if (output_used_ + 2 + have_nukta <= output_.size() && IsSubscriptScript() &&
         IsVirama(output_[index])) {
       // Output previous virama, consonant + optional nukta.
       MultiCodePart(2 + have_nukta);

src/training/validate_javanese.cpp

@@ -2,7 +2,6 @@
  * File:        validate_javanese.cpp
  * Description: Text validator for Javanese Script - aksara jawa.
  * Author:      Shree Devi Kumar
- * Created:     August 03, 2018
  *
  * Licensed under the Apache License, Version 2.0 (the "License");
  * you may not use this file except in compliance with the License.
@@ -90,7 +89,7 @@ bool ValidateJavanese::ConsumeViramaIfValid(IndicPair joiner, bool post_matra) {
         ASSERT_HOST(!CodeOnlyToOutput());
       } else {
         // Half-form with optional Nukta.
-        int len = output_.size() + 1 - output_used_;
+        unsigned len = output_.size() + 1 - output_used_;
         if (UseMultiCode(len)) return true;
       }
       if (codes_used_ < num_codes &&
@@ -149,7 +148,7 @@ bool ValidateJavanese::ConsumeConsonantHeadIfValid() {
     CodeOnlyToOutput();
     // Special Sinhala case of [H Z Yayana/Rayana].
     int index = output_.size() - 3;
-    if (output_used_ <= index &&
+    if (output_used_ + 3 <= output_.size() &&
         (output_.back() == kPengkal || output_.back() == kCakra) &&
         IsVirama(output_[index]) && output_[index + 1] == kZeroWidthJoiner) {
       MultiCodePart(3);
@@ -162,7 +161,7 @@ bool ValidateJavanese::ConsumeConsonantHeadIfValid() {
     }
     // Test for subscript conjunct.
     index = output_.size() - 2 - have_nukta;
-    if (output_used_ <= index && IsSubscriptScript() &&
+    if (output_used_ + 2 + have_nukta <= output_.size() && IsSubscriptScript() &&
         IsVirama(output_[index])) {
       // Output previous virama, consonant + optional nukta.
       MultiCodePart(2 + have_nukta);

src/training/validate_khmer.cpp

@@ -45,7 +45,7 @@ bool ValidateKhmer::ConsumeGraphemeIfValid() {
       if (UseMultiCode(1)) return true;
     }
   }
-  int num_matra_parts = 0;
+  unsigned num_matra_parts = 0;
   if (codes_[codes_used_].second == kZeroWidthJoiner ||
       codes_[codes_used_].second == kZeroWidthNonJoiner) {
     if (CodeOnlyToOutput()) {

src/training/validator.h

@@ -3,7 +3,6 @@
  * Description: Base class for various text validators. Intended mainly for
  *              scripts that use a virama character.
  * Author:      Ray Smith
- * Created:     Tue May 23 2017
  *
  * (C) Copyright 2017, Google Inc.
  * Licensed under the Apache License, Version 2.0 (the "License");
@@ -179,7 +178,7 @@ class Validator {
   // output_, adds unicodes as single-element vectors to parts_ to catch
   // output_used_ up to output->size() - length before adding the length-element
   // vector.
-  void MultiCodePart(int length) {
+  void MultiCodePart(unsigned length) {
     while (output_used_ + length < output_.size()) {
       parts_.emplace_back(
           std::initializer_list<char32>{output_[output_used_++]});
@@ -193,7 +192,7 @@ class Validator {
   // Helper function appends the next element of codes_ to output_, and then
   // calls MultiCodePart to add the appropriate components to parts_.
   // Returns true at the end of codes_.
-  bool UseMultiCode(int length) {
+  bool UseMultiCode(unsigned length) {
     output_.push_back(codes_[codes_used_].second);
     MultiCodePart(length);
     return ++codes_used_ == codes_.size();