Initial commit.

Board scraper:
 * takes traveller file
 * complete produces PBN for all boards in segment

Scores scraper:
 * takes board file, round, segment and board number (1..n, not physical board number)
 * produces SQL that UPDATES scores table (so scores needs to have rows)
 * does not support ARB/Axx scores
 * probably won't support wrong lines

1 files changed, 51 insertions, 0 deletions

diff --git a/scrape-boards.py b/scrape-boards.py
new file mode 100644
index 0000000..f9e02cd
--- /dev/null
+++ b/scrape-boards.py
@@ -0,0 +1,51 @@
+from bs4 import BeautifulSoup as bs
+import bs4
+import os
+import sys
+
+traveller_file = file(sys.argv[1])
+traveller = bs(traveller_file, 'lxml')
+
+print '% PBN 1.0'
+print '[Generator "JFRTeamy-restorerer"]'
+print '[Event "%s"]' % (traveller_file.name)
+
+board_links = traveller.select('td.bdcc a.zb')
+for board_link in board_links:
+    if board_link.has_attr('href'):
+        board_number = board_link.text.strip()
+        board_file = open(
+            os.path.join(
+                os.path.dirname(traveller_file.name),
+                board_link['href']
+            )
+        )
+        board = bs(board_file, 'lxml')
+        conditions = [
+            c for c in
+            board.select('td[valign="top"] h4')[0].contents
+            if type(c) == bs4.element.NavigableString
+        ]
+        dealer = conditions[0]
+        vulnerability = conditions[1].title()
+        if len(vulnerability) < 3:
+            vulnerability = vulnerability.upper()
+        card_cells = board.select('td.w')
+        if len(card_cells) == 4:
+            cards = [
+                [
+                    line.replace('10', 'T').replace(' ', '').strip()
+                    for line in c
+                    if type(line) == bs4.element.NavigableString
+                ][1::2] for c in card_cells
+            ]
+            print '[Board "%s"]' % board_number
+            print '[Dealer "%s"]' % dealer
+            print '[Vulnerable "%s"]' % vulnerability
+            print '[Deal "N:%s %s %s %s"]' % (
+                '.'.join(cards[0]),
+                '.'.join(cards[2]),
+                '.'.join(cards[3]),
+                '.'.join(cards[1])
+            )
+            print