Implement blueprint status tracking by parsing Status: in the whiteboard

diff --git a/workitems.py b/workitems.py
index d5621062e2a58e07da80ac6575bf17b9cfcaeeaf..928c941b1e0f969f5761527f2657690fef720b5b 100755 (executable)
--- a/workitems.py
+++ b/workitems.py
@@ -23,6 +23,10 @@ def get_db(dbpath):
             workitem VARCHAR(255) NOT NULL,
             status VARCHAR(20) NOT NULL,
             date TIMESTAMP NOT NULL)''')
+        cur.execute('''CREATE TABLE statuses (
+            blueprint VARCHAR(255) NOT NULL,
+            status VARCHAR(20) NOT NULL,
+            date TIMESTAMP NOT NULL)''')
         db.commit()
 
     return db
@@ -84,7 +88,7 @@ def get_blueprints(url, name_pattern):
 def get_blueprint_workitems(blueprint_url):
     '''Collect work items from a particular blueprint URL.
 
-    This will return a list of ('item', 'status') pairs.
+    This will return a list of ('item', 'state') pairs.
     '''
     work_items_re = re.compile('(<p>|^)work items:\s*<br />', re.I)
 
@@ -130,12 +134,42 @@ def get_blueprint_workitems(blueprint_url):
 
     return result
 
+def get_blueprint_status(blueprint_url):
+    '''Collect status from a particular blueprint URL.
+
+    This will return a list of lines.
+    '''
+    status_re = re.compile('(<p>|^)status:\s*<br />', re.I)
+
+    found_status = False
+    result = []
+    for l in urllib.urlopen(blueprint_url):
+        end = False
+        if not found_status:
+            if status_re.search(l):
+                found_status = True
+            continue
+
+        if '</p>' in l:
+            end = True
+        l = l.replace('<br />', '').replace('</div>', '').replace('</p>', '').strip()
+
+        if not l:
+            break
+
+        result.append(l.strip())
+
+        if end:
+            break
+
+    return "\n".join(result)
+
 def get_moin_workitems(url):
     '''Collect work items from a moin wiki URL.
 
     Every line starting with "|| " is treated as a work item.
 
-    Return a list of ('item', 'status') pairs.
+    Return a list of ('item', 'state') pairs.
     '''
     result = []
     for line in urllib.urlopen(url):
@@ -159,34 +193,46 @@ def dump(db):
     '''Dump database contents.'''
 
     cur = db.cursor()
-    cur.execute('SELECT * FROM work_items')
+    cur.execute('SELECT w.*, s.status FROM work_items w LEFT JOIN statuses s on w.blueprint = s.blueprint')
     for (blueprint, workitem, status, date) in cur:
         print '%s [%s]\t%s: %s' % (date, blueprint, workitem, status)
 
-def add_work_item(db, blueprint, item, status):
+def add_work_item(db, blueprint, item, state):
     '''Add work item to database.'''
 
     cur = db.cursor()
     cur.execute('INSERT INTO work_items VALUES (?, ?, ?, date(CURRENT_TIMESTAMP))',
-            (blueprint, item, status))
+            (blueprint, item, state))
+
+def add_status(db, blueprint, status):
+    '''Add blueprint status to database.'''
+
+    cur = db.cursor()
+    cur.execute('INSERT INTO statuses VALUES (?, ?, date(CURRENT_TIMESTAMP))',
+            (blueprint, status))
 
 def import_lp(db, name_pattern, release):
-    '''Collect blueprint work items from Launchpad into DB.'''
+    '''Collect blueprint work items and status from Launchpad into DB.'''
 
     blueprints = get_blueprints('%s//ubuntu/%s/+specs?batch=300' % (blueprints_base_url,
         opts.release), name_pattern)
 
     cur = db.cursor()
     cur.execute('DELETE FROM work_items WHERE date = date(CURRENT_TIMESTAMP)')
+    cur.execute('DELETE FROM statuses WHERE date = date(CURRENT_TIMESTAMP)')
 
     for bp in blueprints:
         #print 'Checking', bp
         bpname = bp.split('/')[-1]
         work_items = get_blueprint_workitems(bp)
+        status = get_blueprint_status(bp)
         if not work_items:
             print >> sys.stderr, 'WARNING: %s has no work items' % bpname
-        for (item, status) in work_items:
-            add_work_item(db, bpname, item, status)
+        if not status:
+            print >> sys.stderr, 'WARNING: %s has no status' % bpname
+        for (item, state) in work_items:
+            add_work_item(db, bpname, item, state)
+        add_status(db, bpname, status)
 
 def workitems_over_time(db):
     '''Calculate work item development over time.
@@ -202,8 +248,8 @@ def workitems_over_time(db):
             data.setdefault(date, {})[s] = num
     return data
 
-def blueprint_status(db):
-    '''Determine current blueprint status.
+def blueprint_stats(db):
+    '''Determine current blueprint stats.
 
     Return blueprint -> [todo, done, postponed] mapping.
     '''
@@ -217,17 +263,19 @@ def blueprint_status(db):
     index = 0
     for s in valid_states:
         cur = db.cursor()
-        cur.execute('SELECT blueprint, count(workitem) FROM work_items '
-                'WHERE status = ? AND date = ? GROUP BY blueprint', 
+        cur.execute('SELECT w.blueprint, count(w.workitem), s.status FROM work_items w '
+                'LEFT JOIN statuses s ON w.blueprint = s.blueprint '
+                'WHERE w.status = ? AND w.date = ? GROUP BY w.blueprint',
                 (s, last_date))
-        for (bp, num) in cur:
-            data.setdefault(bp, [0, 0, 0])[index] = num
+        for (bp, num, status) in cur:
+            data.setdefault(bp, [0, 0, 0, ""])[index] = num
+            data[bp][-1] = status
         index += 1
 
     return data
 
 def text(db):
-    '''Print work item status as text.'''
+    '''Print work item stats as text.'''
 
     data = workitems_over_time(db)
 
@@ -235,9 +283,10 @@ def text(db):
     for d in sorted(data.keys()):
         print d, data[d]
 
-    print '\nBlueprint status:'
-    data = blueprint_status(db)
-    for (bp, (todo, done, postponed)) in data.iteritems():
+    print '\nBlueprint stats:'
+    data = blueprint_stats(db)
+    for (bp, (todo, done, postponed, status)) in data.iteritems():
+        # TODO print status
         print '%s: %i/%i (%i%%)' % (bp, postponed+done, todo+done+postponed, 
                 int(float(postponed+done)/(todo+done+postponed)*100 + 0.5))
 
@@ -297,22 +346,23 @@ def html(db):
 
 <h1>Status by blueprint</h1>
 <table>
-  <tr><th>Blueprint</th> <th>todo/postponed/done</th> <th>Completion</th></tr>
+  <tr><th>Blueprint</th> <th>todo/postponed/done</th> <th>Completion</th> <th>Status</th></tr>
 '''
 
-    data = blueprint_status(db)
+    data = blueprint_stats(db)
 
     completion = []
-    for (bp, (todo, done, postponed)) in data.iteritems():
+    for (bp, (todo, done, postponed, status)) in data.iteritems():
         completion.append((bp,
             int(float(postponed+done)/(todo+done+postponed)*100 + 0.5)))
 
     completion.sort(key=lambda k: k[1], reverse=True)
 
     for (bp, percent) in completion:
-        print '  <tr><td><a href="%s/ubuntu/+spec/%s">%s</a></td> <td>%i/%i/%i</td> <td>%i%%</td></tr>' % (
+        print '  <tr><td><a href="%s/ubuntu/+spec/%s">%s</a></td> <td>%i/%i/%i</td> <td>%i%%</td> <td>%s</td></tr>' % (
                 blueprints_base_url, bp, bp, data[bp][0], data[bp][2],
-                data[bp][1], percent)
+                data[bp][1], percent,
+                data[bp][-1])
 
     print '</table>'