From 55481cd69d21b31bfdcc6e33853ea475ed16f622 Mon Sep 17 00:00:00 2001
From: "Steinar H. Gunderson" <sgunderson@bigfoot.com>
Date: Fri, 12 Dec 2014 22:56:30 +0100
Subject: [PATCH] Store and handle the PGN start position.

---
 binloader.cpp                               | 11 ++++++++++-
 binlookup.cpp                               |  5 +++--
 binmerger.cpp                               |  2 ++
 count.proto                                 |  4 +++-
 partition-pgn.sh => find-pgn-split-point.sh |  0
 opening-stats.pl                            |  2 +-
 www/opening-stats.pl                        |  2 +-
 7 files changed, 20 insertions(+), 6 deletions(-)
 rename partition-pgn.sh => find-pgn-split-point.sh (100%)

diff --git a/binloader.cpp b/binloader.cpp
index 6472f10..5bf2502 100644
--- a/binloader.cpp
+++ b/binloader.cpp
@@ -25,6 +25,7 @@ struct Element {
 	Result result;
 	int opening_num, white_elo, black_elo;
 	time_t timestamp;
+	long start_position;
 
 	bool operator< (const Element& other) const {
 		return bpfen < other.bpfen;
@@ -68,6 +69,7 @@ int main(int argc, char **argv)
 
 			int opening_num, white_elo, black_elo;
 			time_t timestamp;
+			long start_position;
 			if (fread(&white_elo, sizeof(white_elo), 1, fp) != 1) {
 				perror("fread()");
 				//exit(1);
@@ -88,6 +90,12 @@ int main(int argc, char **argv)
 				//exit(1);
 				break;
 			}
+			if (fread(&start_position, sizeof(start_position), 1, fp) != 1) {
+				perror("fread()");
+				//exit(1);
+				break;
+			}
+
 
 			l = getc(fp);
 			if (l == -1) {
@@ -102,7 +110,7 @@ int main(int argc, char **argv)
 			}
 
 			int bucket = hash_key_to_bucket(bpfen.data(), bpfen.size(), num_buckets);
-			elems[bucket].emplace_back(Element {std::move(bpfen), std::move(move), Result(r), opening_num, white_elo, black_elo, timestamp});
+			elems[bucket].emplace_back(Element {std::move(bpfen), std::move(move), Result(r), opening_num, white_elo, black_elo, timestamp, start_position});
 			++num_elems;
 		}
 		fclose(fp);
@@ -145,6 +153,7 @@ int main(int argc, char **argv)
 					c.set_first_timestamp(e.timestamp);
 				}
 				c.set_opening_num(e.opening_num);
+				c.set_pgn_start_position(e.start_position);
 			}
 			if (!moves.count(e.move)) {
 				moves.insert(e.move);
diff --git a/binlookup.cpp b/binlookup.cpp
index 958a7bf..179515b 100644
--- a/binlookup.cpp
+++ b/binlookup.cpp
@@ -61,11 +61,12 @@ int main(int argc, char **argv)
 		while (mtbl_iter_next(it, &key, &len_key, &val, &len_val)) {
 			Count c;
 			c.ParseFromArray(val, len_val);
-			printf("%d %d %d %u %f %f %d %ld",
+			printf("%d %d %d %u %f %f %d %ld %ld",
 				c.white(), c.draw(), c.black(), c.opening_num(),
 				double(c.sum_white_elo()) / c.num_elo(),
 				double(c.sum_black_elo()) / c.num_elo(),
-				c.num_elo(), c.first_timestamp());
+				c.num_elo(), c.first_timestamp(),
+				c.pgn_start_position());
 			for (int j = 0; j < c.move_size(); ++j) {
 				printf(" %s", c.move(j).c_str());
 			}
diff --git a/binmerger.cpp b/binmerger.cpp
index 9bd969e..7e34159 100644
--- a/binmerger.cpp
+++ b/binmerger.cpp
@@ -32,11 +32,13 @@ void merge_count(void* userdata,
 		if (c0.has_first_timestamp()) {
 			c.set_first_timestamp(c0.first_timestamp());
 		}
+		c.set_pgn_start_position(c0.pgn_start_position());
 	} else {
 		c.set_opening_num(c1.opening_num());
 		if (c1.has_first_timestamp()) {
 			c.set_first_timestamp(c1.first_timestamp());
 		}
+		c.set_pgn_start_position(c1.pgn_start_position());
 	}
 
 	// Merge the moves, with deduplication.
diff --git a/count.proto b/count.proto
index 7712f97..9ee2859 100644
--- a/count.proto
+++ b/count.proto
@@ -12,8 +12,10 @@ message Count {
 	optional int64 sum_black_elo = 6;
 	optional int32 num_elo = 7;
 
-	// First timestamp this position/move was seen.	
+	// First timestamp this position/move was seen, and the byte offset
+	// into the PGN	file for that game.
 	optional int64 first_timestamp = 8 [default=32503680000];
+	optional int64 pgn_start_position = 10;
 
 	// Moves seen from this position.
 	repeated string move = 9;
diff --git a/partition-pgn.sh b/find-pgn-split-point.sh
similarity index 100%
rename from partition-pgn.sh
rename to find-pgn-split-point.sh
diff --git a/opening-stats.pl b/opening-stats.pl
index 324b158..a90ca0b 100755
--- a/opening-stats.pl
+++ b/opening-stats.pl
@@ -21,7 +21,7 @@ my $line = <$chld_out>;
 print $line;
 chomp $line;
 
-my ($white, $draw, $black, $opening_num, $white_avg_elo, $black_avg_elo, $num_elo, $timestamp, @moves) = split / /, $line;
+my ($white, $draw, $black, $opening_num, $white_avg_elo, $black_avg_elo, $num_elo, $timestamp, $pgn_start_position, @moves) = split / /, $line;
 
 # Explore one move out.
 for my $move (@moves) {
diff --git a/www/opening-stats.pl b/www/opening-stats.pl
index f9e8548..3a02cf9 100755
--- a/www/opening-stats.pl
+++ b/www/opening-stats.pl
@@ -21,7 +21,7 @@ my $hex = unpack('H*', $pos->bitpacked_fen);
 print $chld_in $hex, "\n";
 chomp (my $line = <$chld_out>);
 
-my ($white, $draw, $black, $opening_num, $white_avg_elo, $black_avg_elo, $num_elo, $timestamp, @moves) = split / /, $line;
+my ($white, $draw, $black, $opening_num, $white_avg_elo, $black_avg_elo, $num_elo, $timestamp, $pgn_start_position, @moves) = split / /, $line;
 my $opening = $openings{$opening_num} // 'A00: Start position';
 
 # Explore one move out.
-- 
2.39.2