extract games from playtak.com/games database

1 files changed, 91 insertions, 0 deletions

diff --git a/data/extract.sh b/data/extract.sh
new file mode 100755
index 0000000..7acb1e2
--- /dev/null
+++ b/data/extract.sh
@@ -0,0 +1,91 @@
+#!/bin/bash
+
+query() {
+    echo "SELECT $1 FROM games WHERE (size == $2) \
+and (result != '1-0') and (result != '0-1') and (result != '0-0');"
+}
+
+extract() {
+    in_file=games_anon.db
+    for size in 5 6; do
+        sqlite3 "$in_file" "$(query $1 $size)" > "playtak-$size"
+    done
+}
+
+# Correct move notation, assuming only valid stuff
+fix_move() {
+    args=($1)
+    if [ "${args[0]}" = "M" ]; then
+        result=""
+        # Annoying to compute direction
+        base_col=${args[1]:0:1}
+        base_row=${args[1]:1:1}
+        step_col=${args[2]:0:1}
+        step_row=${args[2]:1:1}
+        if [[ "$base_col" < "$step_col" ]]; then
+            result=">"
+        elif [[ "$base_col" > "$step_col" ]]; then
+            result="<"
+        elif [[ "$base_row" > "$step_row" ]]; then
+            result="-"
+        elif [[ "$base_row" < "$step_row" ]]; then
+            result="+"
+        fi
+        total=0
+        steps=${#args[@]}
+        for (( i = 3 ; i < steps ; i++ )); do
+            total=$(( total + ${args[i]} ))
+            result="$result${args[i]}"
+        done
+        echo "$total${args[1],}$result"
+    else
+        echo "$1"
+    fi;
+}
+
+# Use GNU sed to fix placement notation
+fix_place="\
+s/P \(..\) C,/C\L\1,/g;\
+s/P \(..\) W,/S\L\1,/g;\
+s/P \(..\),/\L\1,/g"
+
+# Thiss is way too slow
+extract_ptn() {
+    extract "notation, result"
+
+    size="$1"
+    dir="ptn-$size"
+    if [ ! -d "$dir" ]; then
+        mkdir "$dir";
+    fi;
+
+    # Separate each line into a file
+    count=0
+    while read game; do
+        file="$dir/$count.ptn"
+        echo "$game" | tr ',' '\n' | sed "$fix_place" > "$file"
+        # Prepare to reformat log as PTN
+        ply=1
+        prepend=""
+        new_contents=""
+        while read line; do
+            # PTN calls for numbers before every new turn in the game
+            if [ $(( $ply % 2 )) -eq 1 ]; then
+                prepend="$(( $ply / 2 + 1 )). "
+                if [ $ply -gt 1 ]; then
+                    prepend="\n$prepend"
+                fi
+            else
+                prepend=" "
+            fi
+            # We have to fix the move notation
+            line=$(fix_move "$line")
+            new_contents="$new_contents$prepend$line"
+            ply=$(( $ply + 1 ))
+        done < "$file"
+        echo -e "$new_contents" | tr '|' '\n' > "$file"
+        count=$(( $count + 1 ))
+    done < "playtak-$size"
+}
+
+extract notation