Scan for XML files when generating the docset index

generate_index.py scanned a fixed range of xeps.  The code has been
reworked slightly to scan the xeps dir for the right XML files, and
generate records for every one.

Test-Information:
Checked that it generated SQL for every XEP in xeps

generate_index.py

@@ -1,3 +1,6 @@
+import os
+import os.path
+import re
 import xml.dom.minidom
 
 print 'CREATE TABLE IF NOT EXISTS searchIndex(id INTEGER PRIMARY KEY);'
@@ -6,27 +9,33 @@
 print 'CREATE UNIQUE INDEX anchor ON searchIndex (name, type, path);'
 
 def putSQL(name, type, path):
-  print "INSERT OR IGNORE INTO searchIndex('name', 'type', 'path') VALUES ('%s', '%s', '%s');" % (name, type, path)
+    print "INSERT OR IGNORE INTO searchIndex('name', 'type', 'path') VALUES ('%s', '%s', '%s');" % (name, type, path)
 
-for i in range(1,380):
-  path = "xep-%04d.html" % i
-  type = 'Extension'
-  name = "xep%d" % i
-  putSQL(name, type, path)
-  name = "xep%04d" % i
-  putSQL(name, type, path)
-  dom = xml.dom.minidom.parse("xeps/xep-%04d.xml" % i)
-  header = dom.getElementsByTagName('header')
-  title = header[0].getElementsByTagName('title')[0]
-  for child in title.childNodes:
-    if child.nodeType == child.TEXT_NODE:
-      title = child.data
-  if title != "N/A":
-    putSQL(title, type, path)
-  shortname = header[0].getElementsByTagName('shortname')
-  if len(shortname) > 0:
-    for child in shortname[0].childNodes:
-      if child.nodeType == child.TEXT_NODE:
-        shortname = child.data
-    if shortname != 'NOT_YET_ASSIGNED':
-      putSQL(shortname, type, path)
+xepre = re.compile('xep-0*([1-9][0-9]*).xml')
+for root, dirs, files in os.walk('xeps'):
+    for path in files:
+        match = xepre.match(path)
+        if not match:
+            continue
+
+        i = int(match.group(1))
+        type = 'Extension'
+        name = "xep%d" % i
+        putSQL(name, type, path)
+        name = "xep%04d" % i
+        putSQL(name, type, path)
+        dom = xml.dom.minidom.parse(os.path.join(root, path))
+        header = dom.getElementsByTagName('header')
+        title = header[0].getElementsByTagName('title')[0]
+        for child in title.childNodes:
+            if child.nodeType == child.TEXT_NODE:
+                title = child.data
+        if title != "N/A":
+            putSQL(title, type, path)
+        shortname = header[0].getElementsByTagName('shortname')
+        if len(shortname) > 0:
+            for child in shortname[0].childNodes:
+                if child.nodeType == child.TEXT_NODE:
+                    shortname = child.data
+            if shortname != 'NOT_YET_ASSIGNED':
+                putSQL(shortname, type, path)