From 09216eca5de3a8d499cf465e727ceb7b58896565 Mon Sep 17 00:00:00 2001
From: twanvl <twanvl@0fc631ac-6414-0410-93d0-97cfa31319b6>
Date: Mon, 29 Dec 2008 22:24:16 +0000
Subject: [PATCH] punctuation breaking is now handled by check_spelling
 function. Symbols are now handled correctly (i.e. they are considered to be
 always correct)

git-svn-id: svn://svn.code.sf.net/p/magicseteditor/code/trunk@1268 0fc631ac-6414-0410-93d0-97cfa31319b6
---
 src/script/functions/spelling.cpp | 65 +++++++++++++++++++------------
 src/util/spell_checker.cpp        | 39 ++++++++++++++-----
 src/util/string.cpp               |  9 +++++
 src/util/string.hpp               |  3 ++
 src/util/tagged_string.cpp        |  4 +-
 src/util/tagged_string.hpp        |  4 +-
 6 files changed, 87 insertions(+), 37 deletions(-)

diff --git a/src/script/functions/spelling.cpp b/src/script/functions/spelling.cpp
index 0d295670..31c840f9 100644
--- a/src/script/functions/spelling.cpp
+++ b/src/script/functions/spelling.cpp
@@ -15,14 +15,34 @@
 
 // ----------------------------------------------------------------------------- : Functions
 
-void check_word(const String& input, String& out, size_t start, size_t end, SpellChecker& checker, bool must_be_empty) {
-	if (start >= end) return;
+bool spelled_correctly(const String& input, size_t start, size_t end, SpellChecker& checker) {
+	// untag
 	String word = untag(input.substr(start,end-start));
-	// TODO: handle keywords and cardname references
-	bool error = !word.empty() && (must_be_empty || !checker.spell_with_punctuation(word));
-	if (error) out += _("<error-spelling>");
+	if (word.empty()) return true;
+	// remove punctuation
+	size_t start_u = 0, end_u = String::npos;
+	trim_punctuation(word, start_u, end_u);
+	if (start_u >= end_u) {
+		// punctuation only, but not empty => error
+		return false;
+	}
+	// find the tagged text without punctuation
+	size_t start2 = untagged_to_index(input, start_u, true, start);
+	size_t end2   = untagged_to_index(input, end_u,   true, start);
+	if (in_tag(input,_("<sym"),start2,end2) != String::npos) {
+		// symbols are always spelled correctly
+		return true;
+	}
+	// run through spellchecker
+	return checker.spell(word.substr(start_u,end_u));
+}
+
+void check_word(const String& input, String& out, size_t start, size_t end, SpellChecker& checker) {
+	if (start >= end) return;
+	bool good = spelled_correctly(input, start, end, checker);
+	if (!good) out += _("<error-spelling>");
 	out.append(input, start, end-start);
-	if (error) out += _("</error-spelling>");
+	if (!good) out += _("</error-spelling>");
 }
 
 SCRIPT_FUNCTION(check_spelling) {
@@ -37,37 +57,32 @@ SCRIPT_FUNCTION(check_spelling) {
 	input = remove_tag(input, _("<error-spelling"));
 	// now walk over the words in the input, and mark misspellings
 	String result;
-	size_t word_start = 0, pos = 0;
-	bool must_be_empty = false; // must this word be empty?
+	size_t word_start = 0, word_end = 0, pos = 0;
 	while (pos < input.size()) {
 		Char c = input.GetChar(pos);
 		if (c == _('<')) {
-			if (is_substr(input,pos,_("<sym"))) {
-				// before symbols should be empty
-				check_word(input,result, word_start,pos, checker, true);
-				// don't spellcheck symbols
-				word_start = pos;
-				pos = min(input.size(), match_close_tag_end(input,pos));
-				result.append(input, word_start, pos-word_start);
-				word_start = pos;
-				must_be_empty = true; // need a space after symbols
+			if (word_start == pos) {
+				// prefer to place word start inside tags
+				pos = skip_tag(input,pos);
+				result.append(input, word_start, pos - word_start);
+				word_end = word_start = pos;
 			} else {
 				pos = skip_tag(input,pos);
 			}
-		} else if (isSpace(c)) {
+		} else if (isSpace(c) || c == EM_DASH || c == EN_DASH) {
 			// word boundary -> check word
-			check_word(input,result, word_start,pos, checker, must_be_empty);
+			check_word(input, result, word_start, word_end, checker);
+			// non-word characters
+			result.append(input, word_end, pos - word_end + 1);
 			// next
-			result += c;
-			pos++;
-			word_start = pos;
-			must_be_empty = false;
+			word_start = word_end = pos = pos + 1;
 		} else {
-			pos++;
+			word_end = pos = pos + 1;
 		}
 	}
 	// last word
-	check_word(input,result, word_start,input.size(), checker, must_be_empty);
+	check_word(input, result, word_start, word_end, checker);
+	result.append(input, word_end, String::npos);
 	// done
 	SCRIPT_RETURN(result);
 }
diff --git a/src/util/spell_checker.cpp b/src/util/spell_checker.cpp
index 38bac732..125d0dae 100644
--- a/src/util/spell_checker.cpp
+++ b/src/util/spell_checker.cpp
@@ -8,6 +8,7 @@
 
 #include <util/prec.hpp>
 #include <util/spell_checker.hpp>
+#include <util/string.hpp>
 #include <util/io/package_manager.hpp>
 
 // ----------------------------------------------------------------------------- : Spell checker : construction
@@ -47,10 +48,33 @@ void SpellChecker::destroyAll() {
 
 bool SpellChecker::spell(const String& word) {
 	if (word.empty()) return true; // empty word is okay
+	// fix curly quotes, especially apstrophes
+	String fixed;
+	FOR_EACH_CONST(c,word) {
+		if (c == LEFT_SINGLE_QUOTE || c == RIGHT_SINGLE_QUOTE) {
+			fixed += _('\'');
+		} else if (c == LEFT_DOUBLE_QUOTE || c == RIGHT_DOUBLE_QUOTE) {
+			fixed += _('\"');
+		} else if (c == 0x00C6) {
+			// expand ligatures, TODO: put this in a better place
+			fixed += _("Ae");
+		} else if (c == 0x0132) {
+			fixed += _("IJ");
+		} else if (c == 0x0152) {
+			fixed += _("Oe");
+		} else if (c == 0xFB01) {
+			fixed += _("fi");
+		} else if (c == 0xFB02) {
+			fixed += _("fl");
+		} else {
+			fixed += c;
+		}
+	}
+	// convert encoding
 	#ifdef UNICODE
-		wxCharBuffer str = word.mb_str(encoding);
+		wxCharBuffer str = fixed.mb_str(encoding);
 	#else
-		wxCharBuffer str = word.mb_str(encoding);
+		wxCharBuffer str = fixed.mb_str(encoding);
 	#endif
 	if (*str == '\0') {
 		// If encoding fails we get an empty string, since the word was not empty this can never happen
@@ -60,12 +84,9 @@ bool SpellChecker::spell(const String& word) {
 	return Hunspell::spell(str);
 }
 
-const String word_start = String(_("[({\"\'")) + LEFT_SINGLE_QUOTE + LEFT_DOUBLE_QUOTE;
-const String word_end   = String(_("])}.,;:?!\"\'")) + RIGHT_SINGLE_QUOTE + RIGHT_DOUBLE_QUOTE;
-
 bool SpellChecker::spell_with_punctuation(const String& word) {
-	size_t first = word.find_first_not_of(word_start);
-	size_t last  = word.find_last_not_of(word_end);
-	if (first > last) return false; // just punctuation is incorrect
-	return spell(word.substr(first, last-first+1));
+	size_t start = 0, end = String::npos;
+	trim_punctuation(word, start, end);
+	if (start >= end) return true; // just punctuation is wrong
+	return spell(word.substr(start,end-start));
 }
diff --git a/src/util/string.cpp b/src/util/string.cpp
index 3e172b06..8944f50a 100644
--- a/src/util/string.cpp
+++ b/src/util/string.cpp
@@ -122,6 +122,15 @@ String strip_last_word(const String& s) {
 	}
 }
 
+const String word_start = String(_("[({\"\'")) + LEFT_SINGLE_QUOTE + LEFT_DOUBLE_QUOTE;
+const String word_end   = String(_("])}.,;:?!\"\'")) + RIGHT_SINGLE_QUOTE + RIGHT_DOUBLE_QUOTE;
+
+void trim_punctuation(const String& str, size_t& start, size_t& end) {
+	start = str.find_first_not_of(word_start, start);
+	end   = str.find_last_not_of(word_end,    min(end,str.size()-1)) + 1;
+	if (start >= end) start = end;
+}
+
 // ----------------------------------------------------------------------------- : Caseing
 
 /// Quick check to see if the substring starting at the given iterator is equal to some given string
diff --git a/src/util/string.hpp b/src/util/string.hpp
index b022840a..5fdaadd1 100644
--- a/src/util/string.hpp
+++ b/src/util/string.hpp
@@ -136,6 +136,9 @@ String last_word(const String&);
 /// Remove the last word from a string, leaves whitespace before that word
 String strip_last_word(const String&);
 
+/// Trim punctuation at the start/end of a word in the range [start..end)
+void trim_punctuation(const String&, size_t& start, size_t& end);
+
 // ----------------------------------------------------------------------------- : Caseing
 
 /// Make each word in a string start with an upper case character.
diff --git a/src/util/tagged_string.cpp b/src/util/tagged_string.cpp
index d6f0373f..4932f810 100644
--- a/src/util/tagged_string.cpp
+++ b/src/util/tagged_string.cpp
@@ -378,8 +378,8 @@ String untag_for_cursor(const String& str) {
 
 // ----------------------------------------------------------------------------- : Untagged position
 
-size_t untagged_to_index(const String& str, size_t pos, bool inside) {
-	size_t i = 0, p = 0;
+size_t untagged_to_index(const String& str, size_t pos, bool inside, size_t start_index) {
+	size_t i = start_index, p = 0;
 	while (i < str.size()) {
 		Char c = str.GetChar(i);
 		if (c == _('<')) {
diff --git a/src/util/tagged_string.hpp b/src/util/tagged_string.hpp
index eaf8d6c6..e2513fb5 100644
--- a/src/util/tagged_string.hpp
+++ b/src/util/tagged_string.hpp
@@ -128,8 +128,10 @@ String untag_for_cursor(const String& str);
 /// Find the tagged position corresponding to the given untagged position.
 /** An untagged position in str is a position in untag(str).
  *  @param inside if inside then it prefers to find positions inside tags (after open tags, before close tags)
+ *  @param start_index is the position to start at in the string,
+ *                     untagged_to_index(s,_,_,i) == untagged_to_index(s.substr(i),_,_)+i
  */
-size_t untagged_to_index(const String& str, size_t pos, bool inside);
+size_t untagged_to_index(const String& str, size_t pos, bool inside, size_t start_index = 0);
 
 /// Find the untagged position corresponding to the given tagged position.
 /** An untagged position in str is a position in untag(str).