reapack

commit 6b37b62dabd37a0604cbb1a1fa8ec8ff131895d5
parent b71e5f0c8123cc3d8c6a19e1761ef0e9c4a644cd
Author: cfillion <cfillion@users.noreply.github.com>
Date:   Thu,  5 Jan 2017 03:03:33 -0500

filter: implement exact word matching [t=185894]

Diffstat:
M
src/filter.cpp
 | 
34
++++++++++++++++++++++++++--------
M
src/filter.hpp
 | 
1
+
M
test/filter.cpp
 | 
23
++++++++++++++++++++++-

3 files changed, 49 insertions(+), 9 deletions(-)
diff --git a/src/filter.cpp b/src/filter.cpp
@@ -47,6 +47,7 @@ void Filter::set(const string &input)
         state = DoubleQuote;
 
       flags |= Node::QuotedFlag;
+      continue;
     }
     else if(c == '\'' && state != DoubleQuote) {
       if(state == SingleQuote)
@@ -55,13 +56,19 @@ void Filter::set(const string &input)
         state = SingleQuote;
 
       flags |= Node::QuotedFlag;
+      continue;
     }
-    else if(c == '\x20' && state == Default) {
-      group = group->push(buf, &flags);
-      buf.clear();
+    else if(c == '\x20') {
+      if(state == Default) {
+        group = group->push(buf, &flags);
+        buf.clear();
+        continue;
+      }
+      else
+        flags |= Node::PhraseFlag;
     }
-    else
-      buf += c;
+
+    buf += c;
   }
 
   group->push(buf, &flags);
@@ -197,12 +204,23 @@ bool Filter::Token::match(const vector<string> &rows) const
 bool Filter::Token::matchRow(const string &str) const
 {
   const size_t pos = str.find(m_buf);
+  const bool isStart = pos == 0, isEnd = pos + m_buf.size() == str.size();
   const bool fail = test(NotFlag);
 
-  if(test(StartAnchorFlag) && pos != 0)
+  if(test(StartAnchorFlag) && !isStart)
     return fail;
-  if(test(EndAnchorFlag) && pos + m_buf.size() != str.size())
+  if(test(EndAnchorFlag) && !isEnd)
     return fail;
 
-  return (pos != string::npos) ^ fail;
+  if(pos == string::npos)
+    return fail;
+
+  if(test(QuotedFlag) && !test(PhraseFlag)) {
+    return fail ^ (
+      (isStart || !isalnum(str[pos - 1])) &&
+      (isEnd || !isalnum(str[pos + m_buf.size()]))
+    );
+  }
+
+  return !fail;
 }
diff --git a/src/filter.hpp b/src/filter.hpp
@@ -43,6 +43,7 @@ private:
       EndAnchorFlag   = 1<<1,
       QuotedFlag      = 1<<2,
       NotFlag         = 1<<3,
+      PhraseFlag      = 1<<4,
     };
 
     Node(int flags) : m_flags(flags) {}
diff --git a/test/filter.cpp b/test/filter.cpp
@@ -59,7 +59,7 @@ TEST_CASE("word matching", M) {
   REQUIRE(f.match({"hello test world"}));
 }
 
-TEST_CASE("quote matching", M) {
+TEST_CASE("quote phrase matching", M) {
   Filter f;
 
   SECTION("double quotes")
@@ -68,10 +68,26 @@ TEST_CASE("quote matching", M) {
     f.set("'hello world'");
 
   REQUIRE(f.match({"hello world"}));
+  REQUIRE(f.match({"BEFOREhello worldAFTER"}));
   REQUIRE_FALSE(f.match({"helloworld"}));
   REQUIRE_FALSE(f.match({"hello test world"}));
 }
 
+TEST_CASE("quote word matching", M) {
+  Filter f;
+
+  SECTION("double quotes")
+    f.set("\"word\"");
+  SECTION("single quotes")
+    f.set("'word'");
+
+  REQUIRE(f.match({"BEFORE word AFTER"}));
+  REQUIRE(f.match({"_word_"}));
+  REQUIRE_FALSE(f.match({"BEFOREword"}));
+  REQUIRE_FALSE(f.match({"wordAFTER"}));
+  REQUIRE_FALSE(f.match({"BEFOREwordAFTER"}));
+}
+
 TEST_CASE("mixing quotes", M) {
   Filter f;
 
@@ -251,6 +267,11 @@ TEST_CASE("NOT operator", M) {
     REQUIRE(f.match({"hello", "bacon"}));
   }
 
+  SECTION("quote word matching") {
+    f.set("NOT 'hello'");
+    REQUIRE(f.match({"hellobacon"}));
+  }
+
   SECTION("NOT NOT") {
     f.set("NOT NOT hello");
     REQUIRE(f.match({"hello"}));