Add InListAbridged to WordList.

4 files changed, 88 insertions, 0 deletions

diff --git a/doc/ScintillaHistory.html b/doc/ScintillaHistory.html
index 75f6b04d8..f97f2b8f3 100644
--- a/doc/ScintillaHistory.html
+++ b/doc/ScintillaHistory.html
@@ -499,6 +499,7 @@
       </tr><tr>
 	<td>Alexey Denisov</td>
 	<td>jedailey</td>
+	<td>oirfeodent</td>
     </tr>
     </table>
     <p>
@@ -518,6 +519,10 @@
 	Released 4 September 2016.
 	</li>
 	<li>
+	The WordList class in lexlib used by lexers adds an InListAbridged method for
+	matching keywords that have particular prefixes and/or suffixes.
+	</li>
+	<li>
 	On Cocoa, include ILexer.h in the public headers of the framework.
 	<a href="http://sourceforge.net/p/scintilla/bugs/1855/">Bug #1855</a>.
 	</li>
diff --git a/lexlib/WordList.cxx b/lexlib/WordList.cxx
index 63d22338f..b8662916c 100644
--- a/lexlib/WordList.cxx
+++ b/lexlib/WordList.cxx
@@ -236,6 +236,66 @@ bool WordList::InListAbbreviated(const char *s, const char marker) const {
 	return false;
 }
 
+/** similar to InListAbbreviated, but word s can be a abridged version of a keyword.
+* eg. the keyword is defined as "after.~:". This means the word must have a prefix (begins with) of
+* "after." and suffix (ends with) of ":" to be a keyword, Hence "after.field:" , "after.form.item:" are valid.
+* Similarly "~.is.valid" keyword is suffix only... hence "field.is.valid" , "form.is.valid" are valid.
+* The marker is ~ in this case.
+* No multiple markers check is done and wont work.
+*/
+bool WordList::InListAbridged(const char *s, const char marker) const {
+	if (0 == words)
+		return false;
+	unsigned char firstChar = s[0];
+	int j = starts[firstChar];
+	if (j >= 0) {
+		while (static_cast<unsigned char>(words[j][0]) == firstChar) {
+			const char *a = words[j];
+			const char *b = s;
+			while (*a && *a == *b) {
+				a++;
+				if (*a == marker) {
+					a++;
+					const size_t suffixLengthA = strlen(a);
+					const size_t suffixLengthB = strlen(b);
+					if (suffixLengthA >= suffixLengthB)
+						break;
+					b = b + suffixLengthB - suffixLengthA - 1;
+				}
+				b++;
+			}
+			if (!*a  && !*b)
+				return true;
+			j++;
+		}
+	}
+
+	j = starts[static_cast<unsigned int>(marker)];
+	if (j >= 0) {
+		while (words[j][0] == marker) {
+			const char *a = words[j] + 1;
+			const char *b = s;
+			const size_t suffixLengthA = strlen(a);
+			const size_t suffixLengthB = strlen(b);
+			if (suffixLengthA > suffixLengthB) {
+				j++;
+				continue;
+			}
+			b = b + suffixLengthB - suffixLengthA;
+
+			while (*a && *a == *b) {
+				a++;
+				b++;
+			}
+			if (!*a && !*b)
+				return true;
+			j++;
+		}
+	}
+
+	return false;
+}
+
 const char *WordList::WordAt(int n) const {
 	return words[n];
 }
diff --git a/lexlib/WordList.h b/lexlib/WordList.h
index 382be2812..b1f8c85b2 100644
--- a/lexlib/WordList.h
+++ b/lexlib/WordList.h
@@ -31,6 +31,7 @@ public:
 	void Set(const char *s);
 	bool InList(const char *s) const;
 	bool InListAbbreviated(const char *s, const char marker) const;
+	bool InListAbridged(const char *s, const char marker) const;
 	const char *WordAt(int n) const;
 };
 
diff --git a/test/unit/testWordList.cxx b/test/unit/testWordList.cxx
index a4ccf4d6a..e5874c01c 100644
--- a/test/unit/testWordList.cxx
+++ b/test/unit/testWordList.cxx
@@ -29,4 +29,26 @@ TEST_CASE("WordList") {
 		REQUIRE(0 == strcmp(wl.WordAt(0), "else"));
 	}
 
+	SECTION("InListAbridged") {
+		wl.Set("list w.~.active bo~k a~z ~_frozen");
+		REQUIRE(wl.InListAbridged("list", '~'));
+
+		REQUIRE(wl.InListAbridged("w.front.active", '~'));
+		REQUIRE(wl.InListAbridged("w.x.active", '~'));
+		REQUIRE(wl.InListAbridged("w..active", '~'));
+		REQUIRE(!wl.InListAbridged("w.active", '~'));
+		REQUIRE(!wl.InListAbridged("w.x.closed", '~'));
+		
+		REQUIRE(wl.InListAbridged("book", '~'));
+		REQUIRE(wl.InListAbridged("bok", '~'));
+		REQUIRE(!wl.InListAbridged("bk", '~'));
+
+		REQUIRE(wl.InListAbridged("a_frozen", '~'));
+		REQUIRE(wl.InListAbridged("_frozen", '~'));
+		REQUIRE(!wl.InListAbridged("frozen", '~'));
+
+		REQUIRE(wl.InListAbridged("abcz", '~'));
+		REQUIRE(wl.InListAbridged("abz", '~'));
+		REQUIRE(wl.InListAbridged("az", '~'));
+	}
 }