workitems: add tracking of assignee (DB schema change)

diff --git a/workitems.py b/workitems.py
index 5cad66dea718fc9f82ac776b318de6a7ed2209eb..db89f51525c203cebe69890912edf9fd4444cebd 100755 (executable)
--- a/workitems.py
+++ b/workitems.py
@@ -20,8 +20,9 @@ def get_db(dbpath):
         cur = db.cursor()
         cur.execute('''CREATE TABLE work_items (
             blueprint VARCHAR(255) NOT NULL,
-            workitem VARCHAR(255) NOT NULL,
+            workitem VARCHAR(1000) NOT NULL,
             status VARCHAR(20) NOT NULL,
+            assignee VARCHAR(200) NOT NULL,
             date TIMESTAMP NOT NULL)''')
         cur.execute('''CREATE TABLE status (
             blueprint VARCHAR(255) NOT NULL,
@@ -88,14 +89,28 @@ 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', 'status', 'assignee') tuples.
     '''
     work_items_re = re.compile('(<p>|^)work items:\s*<br />', re.I)
+    assignee_re = re.compile('<a href="https://.*launchpad.net/~([a-zA-Z0-9_-]+)" class=".*person">')
 
     found_workitems = False
+    found_assignee = False
     result = []
+    default_assignee = 'nobody'
     for l in urllib.urlopen(blueprint_url):
         end = False
+
+        if '<dt>Assignee:' in l:
+            found_assignee = True
+            continue
+
+        if found_assignee and not found_workitems:
+            m = assignee_re.search(l)
+            if m:
+                default_assignee = m.group(1)
+                found_assignee = False
+
         if not found_workitems:
             if work_items_re.search(l):
                 found_workitems = True
@@ -127,7 +142,18 @@ def get_blueprint_workitems(blueprint_url):
             print >> sys.stderr, 'ERROR: invalid state "%s" for work item "%s"' % (
                 state, desc)
             continue
-        result.append((desc, state))
+
+        if desc.startswith('['):
+            try:
+                off = desc.index(']')
+                assignee = desc[1:off]
+                desc = desc[off+1:].strip()
+            except ValueError:
+                print >> sys.stderr, 'ERROR: missing closing "]" for assignee for work item "%s"' % desc
+        else:
+            assignee = default_assignee
+
+        result.append((desc, state, assignee))
 
         if end:
             break
@@ -195,8 +221,8 @@ def dump(db):
     cur = db.cursor()
     cur.execute('SELECT * FROM work_items')
     print '== Work items: =='
-    for (blueprint, workitem, item_status, date) in cur:
-        print '%s [%s]\t%s: %s' % (date, blueprint, workitem, item_status)
+    for (blueprint, workitem, item_status, assignee, date) in cur:
+        print '%s [%s, %s]\t%s: %s' % (date, blueprint, assignee, workitem, item_status)
 
     print '\n== Status =='
     cur = db.cursor()
@@ -204,12 +230,12 @@ def dump(db):
     for (blueprint, status, date) in cur:
         print '%s: %s [%s]' % (blueprint, status, date)
 
-def add_work_item(db, blueprint, item, status):
+def add_work_item(db, blueprint, item, status, assignee):
     '''Add work item to database.'''
 
     cur = db.cursor()
-    cur.execute('INSERT INTO work_items VALUES (?, ?, ?, date(CURRENT_TIMESTAMP))',
-            (blueprint, item, status))
+    cur.execute('INSERT INTO work_items VALUES (?, ?, ?, ?, date(CURRENT_TIMESTAMP))',
+            (blueprint, item, status, assignee))
 
 def add_status(db, blueprint, status):
     '''Add blueprint status to database.'''
@@ -238,8 +264,8 @@ def import_lp(db, name_pattern, release):
         status = get_blueprint_status(bp)
         if not work_items:
             print >> sys.stderr, 'WARNING: %s has no work items' % bpname
-        for (item, state) in work_items:
-            add_work_item(db, bpname, item, state)
+        for (item, state, assignee) in work_items:
+            add_work_item(db, bpname, item, state, assignee)
         add_status(db, bpname, status)
 
 def workitems_over_time(db):
@@ -387,7 +413,7 @@ def import_moin(db, urls):
 
     for url in urls:
         for (d, s) in get_moin_workitems(url):
-            add_work_item(db, url, d, s)
+            add_work_item(db, url, d, s, 'nobody')
 
 #
 # main