Moved part of keyword matching into a separate function; it was way too long

src/data/keyword.cpp

@@ -393,6 +393,7 @@ void dump(int i, KeywordTrie* t) {
 String KeywordDatabase::expand(const String& text,
                                const ScriptValueP& expand_default,
                                const ScriptValueP& combine_script,
+                               bool case_sensitive,
                                Context& ctx) const {
 	assert(combine_script);
 	
@@ -408,19 +409,19 @@ String KeywordDatabase::expand(const String& text,
 	}
 	
 	// Remove all old reminder texts
-	String s = remove_tag_contents(text, _("<atom-reminder"));
-	s = remove_tag_contents(s, _("<atom-keyword")); // OLD, TODO: REMOVEME
-	s = remove_tag_contents(s, _("<atom-kwpph>"));
-	s = remove_tag(s, _("<keyword-param"));
-	s = remove_tag(s, _("<param-"));
-	String untagged = untag_no_escape(s);
+	String tagged = remove_tag_contents(text, _("<atom-reminder"));
+	tagged = remove_tag_contents(tagged, _("<atom-keyword")); // OLD, TODO: REMOVEME
+	tagged = remove_tag_contents(tagged, _("<atom-kwpph>"));
+	tagged = remove_tag(tagged, _("<keyword-param"));
+	tagged = remove_tag(tagged, _("<param-"));
+	String untagged = untag_no_escape(tagged);
 	
-	if (!root) return s;
+	if (!root) return tagged;
 	
 	String result;
 	
 	// Find keywords
-	while (!s.empty()) {
+	while (!tagged.empty()) {
 		vector<KeywordTrie*> current; // current location(s) in the trie
 		vector<KeywordTrie*> next;    // location(s) after this step
 		set<const Keyword*>  used;    // keywords already investigated
@@ -433,20 +434,20 @@ String KeywordDatabase::expand(const String& text,
 		                        //  - 'a' = reminder text in default state, hidden
 		                        //  - 'A' = reminder text in default state, shown
 		
-		for (size_t i = 0 ; i < s.size() ;) {
-			Char c = s.GetChar(i);
+		for (size_t i = 0 ; i < tagged.size() ;) {
+			Char c = tagged.GetChar(i);
 			// tag?
 			if (c == _('<')) {
-				if (is_substr(s, i, _("<kw-")) && i + 4 < s.size()) {
-					expand_type = s.GetChar(i + 4); // <kw-?>
-					s = s.erase(i, skip_tag(s,i)-i); // remove the tag from the string
-				} else if (is_substr(s, i, _("</kw-"))) {
+				if (is_substr(tagged, i, _("<kw-")) && i + 4 < tagged.size()) {
+					expand_type = tagged.GetChar(i + 4); // <kw-?>
+					tagged = tagged.erase(i, skip_tag(tagged,i)-i); // remove the tag from the string
+				} else if (is_substr(tagged, i, _("</kw-"))) {
 					expand_type = 'a';
-					s = s.erase(i, skip_tag(s,i)-i); // remove the tag from the string
-				} else if (is_substr(s, i, _("<atom"))) {
-					i = match_close_tag_end(s, i); // skip <atom>s
+					tagged = tagged.erase(i, skip_tag(tagged,i)-i); // remove the tag from the string
+				} else if (is_substr(tagged, i, _("<atom"))) {
+					i = match_close_tag_end(tagged, i); // skip <atom>s
 				} else {
-					i = skip_tag(s, i);
+					i = skip_tag(tagged, i);
 				}
 				continue;
 			} else {
@@ -473,32 +474,69 @@ String KeywordDatabase::expand(const String& text,
 			FOR_EACH(n, current) {
 				FOR_EACH(f, n->finished) {
 					const Keyword* kw = f;
-					if (used.insert(kw).second) {
-						// we have found a possible match, which we have not seen before
 					assert(kw->match_re.IsValid());
+					if (used.insert(kw).second) {
+						// we have found a possible match, for a keyword which we have not seen before
+						if (tryExpand(*kw, i, tagged, untagged, result, expand_type,
+						              expand_default, combine_script, case_sensitive, ctx,
+						              stat, stat_key))
+						{
+							// it matches
+							used.clear();
+							expand_type = _('a');
+							goto matched_keyword;
+						}
+					}
+				}
+			}
+		}
+		// Remainder of the string
+		result += tagged;
+		tagged.clear();
+		
+		matched_keyword:;
+	}
+	
+	return result;
+}
 
-						// try to match it against the *untagged* string
-						if (kw->match_re.Matches(untagged)) {
-							// Everything before the keyword
+bool KeywordDatabase::tryExpand(const Keyword& kw,
+                                size_t expand_type_known_upto,
+                                String& tagged,
+                                String& untagged,
+                                String& result,
+                                char expand_type,
+                                const ScriptValueP& expand_default,
+                                const ScriptValueP& combine_script,
+                                bool case_sensitive,
+                                Context& ctx,
+                                KeywordUsageStatistics* stat,
+                                Value* stat_key) const
+{
+	// try to match regex against the *untagged* string
+	if (!kw.match_re.Matches(untagged)) return false;
+	
+	// Find match position
 	size_t start_u, len_u;
-							kw->match_re.GetMatch(&start_u, &len_u, 0);
-							size_t start = untagged_to_index(s, start_u, true),
-							       end   = untagged_to_index(s, start_u + len_u, true);
-							if (start == end) continue; // don't match empty keywords
+	kw.match_re.GetMatch(&start_u, &len_u, 0);
+	size_t start = untagged_to_index(tagged, start_u, true),
+	       end   = untagged_to_index(tagged, start_u + len_u, true);
+	if (start == end) return false; // don't match empty keywords
+	
 	// copy text before keyword
-							result += remove_tag(s.substr(0, start), _("<kw-"));
+	result += remove_tag(tagged.substr(0, start), _("<kw-"));
 	
-							// a part of s has not been searched for <kw- tags
+	// a part of tagged has not been searched for <kw- tags
 	// this can happen when the trie incorrectly matches too early
-							for (size_t j = i+1 ; j < start ;) {
-								Char c = s.GetChar(j);
+	for (size_t j = expand_type_known_upto+1 ; j < start ;) {
+		Char c = tagged.GetChar(j);
 		if (c == _('<')) {
-									if (is_substr(s, j, _("<kw-")) && j + 4 < s.size()) {
-										expand_type = s.GetChar(j + 4); // <kw-?>
-									} else if (is_substr(s, j, _("</kw-"))) {
+			if (is_substr(tagged, j, _("<kw-")) && j + 4 < tagged.size()) {
+				expand_type = tagged.GetChar(j + 4); // <kw-?>
+			} else if (is_substr(tagged, j, _("</kw-"))) {
 				expand_type = 'a';
 			}
-									j = skip_tag(s, j);
+			j = skip_tag(tagged, j);
 		} else {
 			++j;
 		}
@@ -506,22 +544,22 @@ String KeywordDatabase::expand(const String& text,
 	
 	// Split the keyword, set parameters in context
 	String total; // the total keyword
-							size_t match_count = kw->match_re.GetMatchCount();
-							assert(match_count - 1 == 1 + 2 * kw->parameters.size());
+	size_t match_count = kw.match_re.GetMatchCount();
+	assert(match_count - 1 == 1 + 2 * kw.parameters.size());
 	for (size_t j = 1 ; j < match_count ; ++j) {
 		// we start counting at 1, so
 		// j = 1 mod 2 -> text
 		// j = 0 mod 2 -> parameter  #((j-1)/2) == (j/2-1)
 		size_t start_u, len_u;
-								kw->match_re.GetMatch(&start_u, &len_u, j);
+		kw.match_re.GetMatch(&start_u, &len_u, j);
 		// note: start_u can be (uint)-1 when len_u == 0
-								size_t part_end = len_u > 0 ? untagged_to_index(s, start_u + len_u, true) : start;
-								String part = s.substr(start, part_end - start);
+		size_t part_end = len_u > 0 ? untagged_to_index(tagged, start_u + len_u, true) : start;
+		String part = tagged.substr(start, part_end - start);
 		// strip left over </kw tags
 		part = remove_tag(part,_("</kw-"));
 		if ((j % 2) == 0) {
 			// parameter
-									KeywordParam& kwp = *kw->parameters[j/2-1];
+			KeywordParam& kwp = *kw.parameters[j/2-1];
 			String param = untagged.substr(start_u, len_u); // untagged version
 			// strip separator_before
 			String separator_before, separator_after;
@@ -579,7 +617,7 @@ String KeywordDatabase::expand(const String& text,
 		total += part;
 		start = part_end;
 	}
-							ctx.setVariable(_("mode"), to_script(kw->mode));
+	ctx.setVariable(_("mode"), to_script(kw.mode));
 	
 	// Show reminder text?
 	bool expand = expand_type == _('1');
@@ -593,9 +631,9 @@ String KeywordDatabase::expand(const String& text,
 	if (expand) {
 		String reminder;
 		try {
-									reminder = kw->reminder.invoke(ctx)->toString();
+			reminder = kw.reminder.invoke(ctx)->toString();
 		} catch (const Error& e) {
-									handle_error(_ERROR_2_("in keyword reminder", e.what(), kw->keyword), true, false);
+			handle_error(_ERROR_2_("in keyword reminder", e.what(), kw.keyword), true, false);
 		}
 		ctx.setVariable(_("keyword"),  to_script(total));
 		ctx.setVariable(_("reminder"), to_script(reminder));
@@ -610,28 +648,14 @@ String KeywordDatabase::expand(const String& text,
 	
 	// Add to usage statistics
 	if (stat && stat_key) {
-								stat->push_back(make_pair(stat_key, kw));
+		stat->push_back(make_pair(stat_key, &kw));
 	}
 	
 	// After keyword
-							s        = s.substr(end);
+	tagged   = tagged.substr(end);
 	untagged = untagged.substr(start_u + len_u);
-							used.clear();
-							expand_type = _('a');
-							goto matched_keyword;
-						}
 	
-					}
-				}
-			}
-		}
-		// Remainder of the string
-		result += s; s.clear();
-		
-		matched_keyword:;
-	}
-	
-	return result;
+	return true;
 }
 
 // ----------------------------------------------------------------------------- : KeywordParamValue

src/data/keyword.hpp

@@ -152,12 +152,23 @@ class KeywordDatabase {
 	/// Expand/update all keywords in the given string.
 	/** @param expand_default script function indicating whether reminder text should be shown by default
 	 *  @param combine_script script function to combine keyword and reminder text in some way
+	 *  @param case_sensitive case sensitive matching of keywords?
 	 *  @param ctx            context for evaluation of scripts
 	 */
-	String expand(const String& text, const ScriptValueP& expand_default, const ScriptValueP& combine_script, Context& ctx) const;
+	String expand(const String& text, const ScriptValueP& expand_default, const ScriptValueP& combine_script, bool case_sensitive, Context& ctx) const;
 	
   private:
 	KeywordTrie* root;	///< Data structure for finding keywords
+	
+	/// (try to) expand a single keyword
+	/** If the keyword matches:
+	 *    - add the result to out
+	 *    - advance the tagged and untagged string by dropping a part from the front
+	 *    - return true
+	 */
+	bool tryExpand(const Keyword& kw, size_t pos, String& tagged, String& untagged, String& out, char expand_type,
+	               const ScriptValueP& expand_default, const ScriptValueP& combine_script, bool case_sensitive, Context& ctx,
+	               KeywordUsageStatistics* stat, Value* stat_key) const;
 };
 
 // ----------------------------------------------------------------------------- : Processing parameters

src/script/functions/basic.cpp

@@ -553,7 +553,7 @@ SCRIPT_RULE_2_N_DEP(expand_keywords, ScriptValueP, _("default expand"), default_
 	SCRIPT_OPTIONAL_PARAM_C_(CardP, card);
 	WITH_DYNAMIC_ARG(keyword_usage_statistics, card ? &card->keyword_usage : nullptr);
 	try {
-		SCRIPT_RETURN(db.expand(input, default_expand, combine, ctx));
+		SCRIPT_RETURN(db.expand(input, default_expand, combine, true, ctx));
 	} catch (const Error& e) {
 		throw ScriptError(_ERROR_2_("in function", e.what(), _("expand_keywords")));
 	}