Extract mask lexer for testability

3 files changed, 86 insertions, 40 deletions

diff --git a/gentoobrowse-api/service/maintenance/masksProcessor.cpp b/gentoobrowse-api/service/maintenance/masksProcessor.cpp
index 5a0bf53..c28f458 100644
--- a/gentoobrowse-api/service/maintenance/masksProcessor.cpp
+++ b/gentoobrowse-api/service/maintenance/masksProcessor.cpp
@@ -7,20 +7,11 @@
 #include "utils/dbUtils.h"
 #include <glibmm/regex.h>
 #include <fileUtils.h>
-#include <lexer-regex.h>
-#include <lexer.h>
+#include <mask.h>
 #include "sql/maintenance/masksSets.sql.h"
 #include "sql/maintenance/masksFixDates.sql.h"
 #include "sql/maintenance/masksEbuilds.sql.h"
 
-static AdHoc::Lexer::PatternPtr maskHead = AdHoc::LexerMatchers::regex(
-		"^# ([^<\n]+)? ?(<(.+?@[^\n>]+)>?)? \\((\\d{1,2} *(?:jan|feb|mar|apr|may|jun|jul|aug|sep|oct|nov|dec)\\w* \\d{4})\\)$\n",
-		(GRegexCompileFlags)(G_REGEX_OPTIMIZE | G_REGEX_CASELESS | G_REGEX_UNGREEDY | G_REGEX_MULTILINE));
-static AdHoc::Lexer::PatternPtr maskDesc = AdHoc::LexerMatchers::regex("^# *([^\n]*)$\n", G_REGEX_MULTILINE);
-static AdHoc::Lexer::PatternPtr atom = AdHoc::LexerMatchers::regex("^([^\n/]+/[^\n]*)$\n?", G_REGEX_MULTILINE);
-static AdHoc::Lexer::PatternPtr end = AdHoc::LexerMatchers::regex("^$\n", G_REGEX_MULTILINE);
-static AdHoc::Lexer::PatternPtr discard = AdHoc::LexerMatchers::regex("^([^\n]*)$\n?", G_REGEX_MULTILINE);
-
 namespace Gentoo {
 	namespace Service {
 		unsigned char MasksProcessor::phase() const { return 2; }
@@ -48,45 +39,27 @@ namespace Gentoo {
 					{ "atomSpec", "text[]" }
 				});
 
-			Glib::ustring date;
-			boost::optional<Glib::ustring> person, email;
-			std::set<std::string> atoms;
-			std::list<std::string> message;
-			int n = 0;
 			auto i = tempTable.second;
 			AdHoc::FileUtils::MemMap f(path);
-			AdHoc::Lexer l({
-				{ { AdHoc::Lexer::InitialState }, maskHead, [&](auto e)
-					{
-						person = e->pattern()->match(1);
-						email = e->pattern()->match(3);
-						date = *e->pattern()->match(4);
-						e->pushState("mask");
-					} },
-				{ { "mask" }, maskDesc, [&](auto e)
-					{
-						message.push_back(*e->pattern()->match(1));
-					} },
-				{ { "mask" }, atom, [&](auto e)
-					{
-						atoms.insert(*e->pattern()->match(1));
-					} },
-				{ { "mask" }, end, [&](auto e)
+			class MaskInserter : public Portage::Utils::Masks {
+				public:
+					MaskInserter(DB::ModifyCommand * i) : i(i) { }
+
+				protected:
+					void consume() const override
 					{
 						i->bindParamS(0, "{" + boost::algorithm::join(atoms, ",") + "}");
 						i->bindParamS(1, date);
 						if (email) i->bindParamS(2, *email); else i->bindNull(2);
 						i->bindParamS(3, boost::algorithm::join(message, " "));
-						i->bindParamI(4, ++n);
+						i->bindParamI(4, n);
 						if (person) i->bindParamS(5, *person); else i->bindNull(5);
 						i->execute();
-						atoms.clear();
-						message.clear();
-						e->popState();
-					} },
-				{ { AdHoc::Lexer::InitialState }, discard, [&](auto) { } }
-			});
-			l.extract((gchar *)f.data, f.getStat().st_size);
+					}
+					DB::ModifyCommand * i;
+			};
+			MaskInserter(tempTable.second.get())
+				.extract((gchar *)f.data, f.getStat().st_size);
 			// Dates are hand-typed and sometimes typos occur... ensure they're at
 			// least within the range of Boost ptimes.
 			while (sql::maintenance::masksFixDates.modify(dbc)->execute()) ;
diff --git a/gentoobrowse-api/service/mask.cpp b/gentoobrowse-api/service/mask.cpp
new file mode 100644
index 0000000..e3cc677
--- /dev/null
+++ b/gentoobrowse-api/service/mask.cpp
@@ -0,0 +1,46 @@
+#include <mask.h>
+#include <lexer-regex.h>
+
+namespace Portage {
+	namespace Utils {
+		static AdHoc::Lexer::PatternPtr maskHead = AdHoc::LexerMatchers::regex(
+				"^# ([^<\n]+)? ?(<(.+?@[^\n>]+)>?)? \\((\\d{1,2} *(?:jan|feb|mar|apr|may|jun|jul|aug|sep|oct|nov|dec)\\w* \\d{4})\\)$\n",
+				(GRegexCompileFlags)(G_REGEX_OPTIMIZE | G_REGEX_CASELESS | G_REGEX_UNGREEDY | G_REGEX_MULTILINE));
+		static AdHoc::Lexer::PatternPtr maskDesc = AdHoc::LexerMatchers::regex("^# *([^\n]*)$\n", G_REGEX_MULTILINE);
+		static AdHoc::Lexer::PatternPtr atom = AdHoc::LexerMatchers::regex("^([^\n/]+/[^\n]*)$\n?", G_REGEX_MULTILINE);
+		static AdHoc::Lexer::PatternPtr end = AdHoc::LexerMatchers::regex("^$\n", G_REGEX_MULTILINE);
+		static AdHoc::Lexer::PatternPtr discard = AdHoc::LexerMatchers::regex("^([^\n]*)$\n?", G_REGEX_MULTILINE);
+
+		Masks::Masks() :
+			AdHoc::Lexer({
+				{ { AdHoc::Lexer::InitialState }, maskHead, [&](auto e)
+					{
+						person = e->pattern()->match(1);
+						email = e->pattern()->match(3);
+						date = *e->pattern()->match(4);
+						e->pushState("mask");
+					} },
+					{ { "mask" }, maskDesc, [&](auto e)
+						{
+							message.push_back(*e->pattern()->match(1));
+						} },
+					{ { "mask" }, atom, [&](auto e)
+						{
+							atoms.insert(*e->pattern()->match(1));
+						} },
+					{ { "mask" }, end, [&](auto e)
+						{
+							++n;
+							consume();
+							atoms.clear();
+							message.clear();
+							e->popState();
+						} },
+					{ { AdHoc::Lexer::InitialState }, discard, [&](auto) { } }
+			}),
+			n(0)
+		{
+		}
+	}
+}
+
diff --git a/gentoobrowse-api/service/mask.h b/gentoobrowse-api/service/mask.h
new file mode 100644
index 0000000..f37b152
--- /dev/null
+++ b/gentoobrowse-api/service/mask.h
@@ -0,0 +1,27 @@
+#ifndef GENTOOBROWSE_SERVICE_MASK_H
+#define GENTOOBROWSE_SERVICE_MASK_H
+
+#include <lexer.h>
+#include <list>
+#include <glibmm/ustring.h>
+#include <boost/optional.hpp>
+
+namespace Portage {
+	namespace Utils {
+		class Masks : public AdHoc::Lexer {
+			protected:
+				Masks();
+
+				virtual void consume() const = 0;
+
+				Glib::ustring date;
+				boost::optional<Glib::ustring> person, email;
+				std::set<std::string> atoms;
+				std::list<std::string> message;
+				int n;
+		};
+	}
+}
+
+#endif
+