Don't fail if no initial epsilons! But do fail on initial whitespace

apertium · Aug 28, 2020 · 340887b · 340887b
1 parent c016057
commit 340887b
Show file tree

Hide file tree

Showing 7 changed files with 60 additions and 12 deletions.
diff --git a/lttoolbox/alphabet.cc b/lttoolbox/alphabet.cc
@@ -261,10 +261,10 @@ Alphabet::decode(int const code) const
 }
 
 set<int>
-Alphabet::getLeftEpsilons() const {
+Alphabet::symbolsWhereLeftIs(wchar_t l) const {
   set<int> eps;
   for(const auto& sp: spair) {  // [(l, r) : tag]
-    if(sp.first.first == 0) {
+    if(sp.first.first == l) {
       eps.insert(sp.second);
     }
   }

diff --git a/lttoolbox/alphabet.h b/lttoolbox/alphabet.h
@@ -175,7 +175,10 @@ class Alphabet
    */
   pair<int, int> const & decode(int const code) const;
 
-  set<int> getLeftEpsilons() const;
+  /**
+   * Get all symbols where the left-hand side of the symbol-pair is l.
+   */
+  set<int> symbolsWhereLeftIs(wchar_t l) const;
 
   enum Side
   {

diff --git a/lttoolbox/compiler.cc b/lttoolbox/compiler.cc
@@ -138,16 +138,22 @@ Compiler::parse(string const &file, wstring const &dir)
 bool
 Compiler::valid(wstring const& dir) const
 {
+  const wstring side = dir == COMPILER_RESTRICTION_RL_VAL ? L"right" : L"left";
+  const set<int> epsilonSymbols = alphabet.symbolsWhereLeftIs(0);
+  const set<int> spaceSymbols = alphabet.symbolsWhereLeftIs(L' ');
   for (auto &section : sections) {
     auto &fst = section.second;
-    set<int> initialClosure = fst.closure(fst.getInitial(), alphabet.getLeftEpsilons());
     auto finals = fst.getFinals();
-    for(const auto i : initialClosure) {
+    auto initial = fst.getInitial();
+    for(const auto i : fst.closure(initial, epsilonSymbols)) {
       if (finals.count(i)) {
-        const wstring side = dir == COMPILER_RESTRICTION_RL_VAL ? L"right" : L"left";
         wcerr << L"Error: Invalid dictionary (hint: the " << side << " side of an entry is empty)" << endl;
         return false;
       }
+      if(fst.closure(i, spaceSymbols).size() > 1) { // >1 since closure always includes self
+        wcerr << L"Error: Invalid dictionary (hint: entry on the " << side << " beginning with whitespace)" << endl;
+        return false;
+      }
     }
   }
   return true;

diff --git a/lttoolbox/transducer.cc b/lttoolbox/transducer.cc
@@ -220,13 +220,17 @@ Transducer::closure(int const state, set<int> const &epsilon_tags) const
   while (nonvisited.size() > 0) {
     int auxest = *nonvisited.begin();
     for (const int epsilon_tag : epsilon_tags) {
-      auto range = transitions.at(auxest).equal_range(epsilon_tag);
-      while (range.first != range.second) {
-        if (result.find(range.first->second.first) == result.end()) {
-          result.insert(range.first->second.first);
-          nonvisited.insert(range.first->second.first);
+      try {
+        auto range = transitions.at(auxest).equal_range(epsilon_tag);
+        while (range.first != range.second) {
+          if (result.find(range.first->second.first) == result.end()) {
+            result.insert(range.first->second.first);
+            nonvisited.insert(range.first->second.first);
+          }
+          range.first++;
         }
-        range.first++;
+      } catch (out_of_range const &e) {
+        // No transition from any of the epsilon_tags – this is fine
       }
     }
     nonvisited.erase(auxest);

diff --git a/tests/data/lhs-ws-mono.dix b/tests/data/lhs-ws-mono.dix
@@ -0,0 +1,12 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<dictionary>
+  <alphabet>ABCDEFGHIJKLMNOPQRSTUVWXYZÀÁÂÄÅÆÇÈÉÊËÍÑÒÓÔÕÖØÙÚÜČĐŊŠŦŽabcdefghijklmnopqrstuvwxyzàáâäåæçèéêëíñòóôõöøùúüčđŋšŧž-</alphabet>
+  <sdefs>
+    <sdef n="n" 	c="Noun"/>
+  </sdefs>
+  <pardefs>
+  </pardefs>
+  <section id="main" type="standard">
+    <e><p><l><b/>a</l><r>a<s n="n"/></r></p></e>
+  </section>
+</dictionary>
diff --git a/tests/data/rhs-ws-mono.dix b/tests/data/rhs-ws-mono.dix
@@ -0,0 +1,12 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<dictionary>
+  <alphabet>ABCDEFGHIJKLMNOPQRSTUVWXYZÀÁÂÄÅÆÇÈÉÊËÍÑÒÓÔÕÖØÙÚÜČĐŊŠŦŽabcdefghijklmnopqrstuvwxyzàáâäåæçèéêëíñòóôõöøùúüčđŋšŧž-</alphabet>
+  <sdefs>
+    <sdef n="n" 	c="Noun"/>
+  </sdefs>
+  <pardefs>
+  </pardefs>
+  <section id="main" type="standard">
+    <e><p><l>a<s n="n"/></l><r><b/>a</r></p></e>
+  </section>
+</dictionary>
diff --git a/tests/lt_comp/__init__.py b/tests/lt_comp/__init__.py
@@ -17,3 +17,14 @@ class CompEmptyRhsShouldError(unittest.TestCase, ProcTest):
     procdir = "rl"
     procdix = "data/rhs-empty-mono.dix"
     expectedCompRetCodeFail = True
+
+
+class CompLhsInitialSpaceShouldError(unittest.TestCase, ProcTest):
+    procdix = "data/lhs-ws-mono.dix"
+    expectedCompRetCodeFail = True
+
+
+class CompRhsInitialSpaceShouldError(unittest.TestCase, ProcTest):
+    procdix = "data/rhs-ws-mono.dix"
+    procdir = "rl"
+    expectedCompRetCodeFail = True