Mercurial > piecrust2
annotate piecrust/processing/sitemap.py @ 182:a54d3c0b5f4a
tests: Patch `os.path.exists` and improve patching for `open`.
You can specify additional modules for which to patch `open`.
Also, it was incorrectly updating the opened file, even when it was opened
for read only. Now it only updates the contents if the file was opened for
write, and supports appending to the end.
Last, it supports opening text files in binary mode.
| author | Ludovic Chabant <ludovic@chabant.com> |
|---|---|
| date | Sun, 04 Jan 2015 14:55:41 -0800 |
| parents | ab6e7e0e9d44 |
| children | 1c4078ec3011 |
| rev | line source |
|---|---|
| 34 | 1 import time |
| 2 import logging | |
| 3 import yaml | |
| 4 from piecrust.processing.base import SimpleFileProcessor | |
| 5 | |
| 6 | |
| 7 logger = logging.getLogger(__name__) | |
| 8 | |
| 9 | |
| 10 SITEMAP_HEADER = \ | |
| 11 """<?xml version="1.0" encoding="utf-8"?> | |
| 12 <urlset | |
| 13 xmlns="http://www.sitemaps.org/schemas/sitemap/0.9"> | |
| 14 """ | |
| 15 SITEMAP_FOOTER = "</urlset>\n" | |
| 16 | |
| 17 SITEURL_HEADER = " <url>\n" | |
| 18 SITEURL_LOC = " <loc>%s</loc>\n" | |
| 19 SITEURL_LASTMOD = " <lastmod>%s</lastmod>\n" | |
| 20 SITEURL_CHANGEFREQ = " <changefreq>%s</changefreq>\n" | |
| 21 SITEURL_PRIORITY = " <priority>%f</priority>\n" | |
| 22 SITEURL_FOOTER = " </url>\n" | |
| 23 | |
| 24 | |
| 25 class SitemapProcessor(SimpleFileProcessor): | |
| 26 PROCESSOR_NAME = 'sitemap' | |
| 27 | |
| 28 def __init__(self): | |
| 29 super(SitemapProcessor, self).__init__({'sitemap': 'xml'}) | |
| 30 self._start_time = None | |
| 31 | |
| 32 def onPipelineStart(self, pipeline): | |
| 33 self._start_time = time.time() | |
| 34 | |
| 35 def _doProcess(self, in_path, out_path): | |
| 36 with open(in_path, 'r') as fp: | |
| 37 sitemap = yaml.load(fp) | |
| 38 | |
| 39 with open(out_path, 'w') as fp: | |
| 40 fp.write(SITEMAP_HEADER) | |
| 41 self._writeManualLocs(sitemap, fp) | |
| 42 self._writeAutoLocs(sitemap, fp) | |
| 43 fp.write(SITEMAP_FOOTER) | |
| 44 | |
| 45 return True | |
| 46 | |
| 47 def _writeManualLocs(self, sitemap, fp): | |
| 48 locs = sitemap.setdefault('locations', None) | |
| 49 if not locs: | |
| 50 return | |
| 51 | |
| 52 logger.debug("Generating manual sitemap entries.") | |
| 53 for loc in locs: | |
| 54 self._writeEntry(loc, fp) | |
| 55 | |
| 56 def _writeAutoLocs(self, sitemap, fp): | |
| 57 source_names = sitemap.setdefault('autogen', None) | |
| 58 if not source_names: | |
| 59 return | |
| 60 | |
| 61 for name in source_names: | |
| 62 logger.debug("Generating automatic sitemap entries for '%s'." % | |
| 63 name) | |
| 64 source = self.app.getSource(name) | |
| 65 if source is None: | |
| 66 raise Exception("No such source: %s" % name) | |
| 67 | |
|
114
371a6c879ab9
When possible, try and batch-load pages so we only lock once.
Ludovic Chabant <ludovic@chabant.com>
parents:
34
diff
changeset
|
68 for page in source.getPages(): |
|
371a6c879ab9
When possible, try and batch-load pages so we only lock once.
Ludovic Chabant <ludovic@chabant.com>
parents:
34
diff
changeset
|
69 route = self.app.getRoute(source.name, page.source_metadata) |
|
147
ab6e7e0e9d44
Pass date information to routing when building URLs.
Ludovic Chabant <ludovic@chabant.com>
parents:
114
diff
changeset
|
70 uri = route.getUri(page.source_metadata, page) |
| 34 | 71 |
| 72 t = page.datetime.timestamp() | |
| 73 sm_cfg = page.config.get('sitemap') | |
| 74 | |
| 75 args = {'url': uri, 'lastmod': strftime_iso8601(t)} | |
| 76 if sm_cfg: | |
| 77 args.update(cm_cfg) | |
| 78 | |
| 79 self._writeEntry(args, fp) | |
| 80 | |
| 81 def _writeEntry(self, args, fp): | |
| 82 fp.write(SITEURL_HEADER) | |
| 83 fp.write(SITEURL_LOC % args['url']) | |
| 84 if 'lastmod' in args: | |
| 85 fp.write(SITEURL_LASTMOD % args['lastmod']) | |
| 86 if 'changefreq' in args: | |
| 87 fp.write(SITEURL_CHANGEFREQ % args['changefreq']) | |
| 88 if 'priority' in args: | |
| 89 fp.write(SITEURL_PRIORITY % args['priority']) | |
| 90 fp.write(SITEURL_FOOTER) | |
| 91 | |
| 92 | |
| 93 def strftime_iso8601(t): | |
| 94 return time.strftime('%Y-%m-%dT%H:%M:%SZ', time.gmtime(t)) | |
| 95 |