MoinMoin: attachment:1.patch of MoinMoinPatch/DocbookParserCaching

   1 changeset:   4206:0274b9f1343b
   2 branch:      docbook
   3 parent:      4199:f414aece63e0
   4 user:        matthijs@stdin.nl
   5 date:        Mon Nov 24 22:40:14 2008 +0100
   6 summary:     docbook parser: Use the caching framework for the compiled XSL file.
   7 
   8 diff -r f414aece63e0 -r 0274b9f1343b MoinMoin/config/multiconfig.py
   9 --- a/MoinMoin/config/multiconfig.py	Tue Nov 11 22:00:33 2008 +0100
  10 +++ b/MoinMoin/config/multiconfig.py	Mon Nov 24 22:40:14 2008 +0100
  11 @@ -900,8 +900,10 @@
  12      ('plugin_dir', None, "Plugin directory, by default computed to be `data_dir`/plugin."),
  13      ('plugin_dirs', [], "Additional plugin directories."),
  14  
  15 -    ('docbook_html_dir', r"/usr/share/xml/docbook/stylesheet/nwalsh/html/",
  16 -     'Path to the directory with the Docbook to HTML XSLT files (optional, used by the docbook parser). The default value is correct for Debian Etch.'),
  17 +    ('docbook_html_dir', None,
  18 +     'Path to the docbook.xsl file. Deprecated, use docbook_xsl instead.'),
  19 +    ('docbook_xsl', r"/usr/share/xml/docbook/stylesheet/nwalsh/html/docbook.xsl",
  20 +     'Path to the HTML XSLT file (optional, used by the docbook parser). The default value is correct for Debian Etch.'),
  21      ('shared_intermap', None,
  22       "Path to a file containing global InterWiki definitions (or a list of such filenames)"),
  23    )),
  24 diff -r f414aece63e0 -r 0274b9f1343b MoinMoin/parser/text_docbook.py
  25 --- a/MoinMoin/parser/text_docbook.py	Tue Nov 11 22:00:33 2008 +0100
  26 +++ b/MoinMoin/parser/text_docbook.py	Mon Nov 24 22:40:14 2008 +0100
  27 @@ -35,6 +35,9 @@
  28  import re
  29  
  30  from MoinMoin import  Page
  31 +from MoinMoin import caching
  32 +from MoinMoin import log
  33 +logging = log.getLogger(__name__)
  34  from MoinMoin.parser.text_xslt import Parser as XsltParser
  35  from MoinMoin.parser.text_moin_wiki import Parser as WikiParser
  36  
  37 @@ -52,31 +55,97 @@
  38          XsltParser.__init__(self, raw, request)
  39  
  40          # relative path to docbook.xsl and compiled_xsl
  41 -        docbook_html_directory = request.cfg.docbook_html_dir
  42 -        self.db_xsl = os.path.join(docbook_html_directory, 'docbook.xsl')
  43 -        self.db_compiled_xsl = os.path.join(docbook_html_directory, 'db_compiled.dat')
  44 +        if not request.cfg.docbook_html_dir is None:
  45 +            self.db_xsl = os.path.join(request.cfg.docbook_html, 'docbook.xsl')
  46 +        else:
  47 +            self.db_xsl = request.cfg.docbook_xsl
  48  
  49          self.wikiParser = WikiParser(raw=self.raw, request=self.request, pretty_url=1)
  50          self.key = 'docbook'
  51  
  52 +        # We save the compiled version of the XSL file in the cache, to
  53 +        # speed up rendering. We have a separate entry for the filename,
  54 +        # so we can properly refresh the cache when the docbook_xsl
  55 +        # configuration value changes.
  56 +        self.cached = caching.CacheEntry(self.request, 
  57 +                                    arena='docbook',
  58 +                                    key='compiled_xsl',
  59 +                                    scope='farm',
  60 +                                    use_pickle=True)
  61 +
  62 +        self.cached_name = caching.CacheEntry(self.request, 
  63 +                                    arena='docbook',
  64 +                                    key='docbook_xsl_filename',
  65 +                                    scope='farm',
  66 +                                    use_pickle=True)
  67 +        
  68 +
  69      def format(self, formatter):
  70          self.wikiParser.formatter = formatter
  71          XsltParser.format(self, formatter)
  72  
  73 +    def get_cached_stylesheet(self, abs_db_xsl, log=True):
  74 +        """ Try to get the compiled xsl file from the cache.
  75 +            
  76 +        @param abs_db_xsl: The input xsl file of which we should find
  77 +        the compiled version.
  78 +        @param errors: Should we do logging?
  79 +        """
  80 +        try:
  81 +            if self.cached_name.content() != abs_db_xsl:
  82 +                if log:
  83 +                    logging.debug(
  84 +                        "Docbook XSL file configuration changed from %s to %s" % 
  85 +                        (cached_name.content(), abs_db_xsl)
  86 +                    )
  87 +            elif self.cached.needsUpdate(abs_db_xsl):
  88 +                if log:
  89 +                    logging.debug("Docbook XSL file changed")
  90 +            else:
  91 +                # Good, we got a cache hit!
  92 +                compiled = self.cached.content()
  93 +                if log:
  94 +                    logging.debug("Got compiled Docbook XSL file from cache")
  95 +                return compiled
  96 +        except caching.CacheError:
  97 +            if log:
  98 +                logging.debug("Got cache error for compiled XSL file")
  99 +        return None
 100 +
 101      def append_stylesheet(self):
 102          """"
 103              virtual function, for docbook parser
 104          """
 105          abs_db_xsl = os.path.abspath(self.db_xsl)
 106 -        abs_db_compiled_xsl = os.path.abspath(self.db_compiled_xsl)
 107  
 108 -        # same as path.exists, but also test if it is a file
 109 -        if not os.path.isfile(abs_db_compiled_xsl):
 110 -            _compile_xsl(abs_db_xsl, abs_db_compiled_xsl)
 111 +        compiled = self.get_cached_stylesheet(abs_db_xsl)
 112  
 113 -        assert os.path.isfile(abs_db_compiled_xsl)
 114 +        if compiled is None:
 115 +            # Recompile the XSL file
 116 +            logging.debug("(Re)compiling Docbook XSL file: '%s'" % (abs_db_xsl))
 117 +            compiled = _compile_xsl(abs_db_xsl)
 118 +            # Check the cache again (another thread might have filled
 119 +            # the cache by now) and update the cache if that's not the
 120 +            # case. Don't do any logging of the result, since that might
 121 +            # be confusing.
 122 +            if (self.get_cached_stylesheet(abs_db_xsl, log=False) is None):
 123 +                # We first remove the old cached_name, to prevent an
 124 +                # inconsistent situation and invalid cache hits after
 125 +                # changing the xsl filename configuration. This does
 126 +                # allow for possibly indefinite recompiling, if
 127 +                # everytime a new thread triggers recompilation just
 128 +                # between the remove and update of cached_name. However,
 129 +                # the recheck of the cache combined with the long time
 130 +                # needed to compile the xsl should make this case
 131 +                # extremely unlikely.
 132 +                self.cached_name.remove()
 133 +                self.cached.update(compiled)
 134 +                self.cached_name.update(abs_db_xsl)
 135 +                logging.debug("Saved compiled Docbook XSL file to cache")
 136 +            else:
 137 +                logging.debug("Skipping saving of compiled Docbook XSL file to cache")
 138  
 139 -        self.processor.appendStylesheetInstance(cPickle.load(file(abs_db_compiled_xsl, 'rb')))
 140 +        self.processor.appendStylesheetInstance(compiled)
 141  
 142      def parse_result(self, result):
 143          """
 144 @@ -153,7 +222,7 @@
 145  
 146  
 147  
 148 -def _compile_xsl(XSLT_FILE, XSLT_COMPILED_FILE):
 149 +def _compile_xsl(XSLT_FILE):
 150      """
 151          compiling docbook stylesheet
 152  
 153 @@ -176,9 +245,8 @@
 154  
 155      # Pickled stylesheet will be self.abs_db_compiled_xsl file
 156      db_root = db_processor.stylesheet.root
 157 -    fw = file(XSLT_COMPILED_FILE, 'wb')
 158 -    cPickle.dump(db_root, fw) # , protocol=2)
 159 -    fw.close()
 160 +
 161 +    return db_root
 162  
 163  
 164  def _splitResult(iterator, result):