piecrust2: piecrust/importing/wordpress.py annotate

annotate piecrust/importing/wordpress.py @ 1188:a7c43131d871

bake: Fix file write flushing problem with Python 3.8+ Writing the cache files fails in Python 3.8 because it looks like flushing behaviour has changed. We need to explicitly flush. And even then, in very rare occurrences, it looks like it can still run into racing conditions, so we do a very hacky and ugly "retry" loop when fetching cached data :(

author	Ludovic Chabant <ludovic@chabant.com>
date	Tue, 15 Jun 2021 22:36:23 -0700
parents	4850f8c21b6e
children

rev	line source
300 2daa05a21026 import: Add an XML-based Wordpress importer. Ludovic Chabant <ludovic@chabant.com> parents: diff changeset	1 import os.path
2daa05a21026 import: Add an XML-based Wordpress importer. Ludovic Chabant <ludovic@chabant.com> parents: diff changeset	2 import logging
2daa05a21026 import: Add an XML-based Wordpress importer. Ludovic Chabant <ludovic@chabant.com> parents: diff changeset	3 import datetime
2daa05a21026 import: Add an XML-based Wordpress importer. Ludovic Chabant <ludovic@chabant.com> parents: diff changeset	4 import yaml
302 103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	5 from collections import OrderedDict
300 2daa05a21026 import: Add an XML-based Wordpress importer. Ludovic Chabant <ludovic@chabant.com> parents: diff changeset	6 from piecrust import CONFIG_PATH
302 103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	7 from piecrust.configuration import (
852 4850f8c21b6e core: Start of the big refactor for PieCrust 3.0. Ludovic Chabant <ludovic@chabant.com> parents: 363 diff changeset	8 ConfigurationLoader, ConfigurationDumper, merge_dicts)
300 2daa05a21026 import: Add an XML-based Wordpress importer. Ludovic Chabant <ludovic@chabant.com> parents: diff changeset	9 from piecrust.importing.base import Importer, create_page, download_asset
2daa05a21026 import: Add an XML-based Wordpress importer. Ludovic Chabant <ludovic@chabant.com> parents: diff changeset	10
2daa05a21026 import: Add an XML-based Wordpress importer. Ludovic Chabant <ludovic@chabant.com> parents: diff changeset	11
2daa05a21026 import: Add an XML-based Wordpress importer. Ludovic Chabant <ludovic@chabant.com> parents: diff changeset	12 logger = logging.getLogger(__name__)
2daa05a21026 import: Add an XML-based Wordpress importer. Ludovic Chabant <ludovic@chabant.com> parents: diff changeset	13
2daa05a21026 import: Add an XML-based Wordpress importer. Ludovic Chabant <ludovic@chabant.com> parents: diff changeset	14
302 103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	15 class WordpressImporterBase(Importer):
300 2daa05a21026 import: Add an XML-based Wordpress importer. Ludovic Chabant <ludovic@chabant.com> parents: diff changeset	16 def setupParser(self, parser, app):
2daa05a21026 import: Add an XML-based Wordpress importer. Ludovic Chabant <ludovic@chabant.com> parents: diff changeset	17 parser.add_argument(
852 4850f8c21b6e core: Start of the big refactor for PieCrust 3.0. Ludovic Chabant <ludovic@chabant.com> parents: 363 diff changeset	18 '--pages-source',
4850f8c21b6e core: Start of the big refactor for PieCrust 3.0. Ludovic Chabant <ludovic@chabant.com> parents: 363 diff changeset	19 default="pages",
4850f8c21b6e core: Start of the big refactor for PieCrust 3.0. Ludovic Chabant <ludovic@chabant.com> parents: 363 diff changeset	20 help="The source to store pages in.")
300 2daa05a21026 import: Add an XML-based Wordpress importer. Ludovic Chabant <ludovic@chabant.com> parents: diff changeset	21 parser.add_argument(
852 4850f8c21b6e core: Start of the big refactor for PieCrust 3.0. Ludovic Chabant <ludovic@chabant.com> parents: 363 diff changeset	22 '--posts-source',
4850f8c21b6e core: Start of the big refactor for PieCrust 3.0. Ludovic Chabant <ludovic@chabant.com> parents: 363 diff changeset	23 default="posts",
4850f8c21b6e core: Start of the big refactor for PieCrust 3.0. Ludovic Chabant <ludovic@chabant.com> parents: 363 diff changeset	24 help="The source to store posts in.")
300 2daa05a21026 import: Add an XML-based Wordpress importer. Ludovic Chabant <ludovic@chabant.com> parents: diff changeset	25 parser.add_argument(
852 4850f8c21b6e core: Start of the big refactor for PieCrust 3.0. Ludovic Chabant <ludovic@chabant.com> parents: 363 diff changeset	26 '--default-post-layout',
4850f8c21b6e core: Start of the big refactor for PieCrust 3.0. Ludovic Chabant <ludovic@chabant.com> parents: 363 diff changeset	27 help="The default layout to use for posts.")
300 2daa05a21026 import: Add an XML-based Wordpress importer. Ludovic Chabant <ludovic@chabant.com> parents: diff changeset	28 parser.add_argument(
852 4850f8c21b6e core: Start of the big refactor for PieCrust 3.0. Ludovic Chabant <ludovic@chabant.com> parents: 363 diff changeset	29 '--default-post-category',
4850f8c21b6e core: Start of the big refactor for PieCrust 3.0. Ludovic Chabant <ludovic@chabant.com> parents: 363 diff changeset	30 help="The default category to use for posts.")
300 2daa05a21026 import: Add an XML-based Wordpress importer. Ludovic Chabant <ludovic@chabant.com> parents: diff changeset	31 parser.add_argument(
852 4850f8c21b6e core: Start of the big refactor for PieCrust 3.0. Ludovic Chabant <ludovic@chabant.com> parents: 363 diff changeset	32 '--default-page-layout',
4850f8c21b6e core: Start of the big refactor for PieCrust 3.0. Ludovic Chabant <ludovic@chabant.com> parents: 363 diff changeset	33 help="The default layout to use for pages.")
300 2daa05a21026 import: Add an XML-based Wordpress importer. Ludovic Chabant <ludovic@chabant.com> parents: diff changeset	34 parser.add_argument(
852 4850f8c21b6e core: Start of the big refactor for PieCrust 3.0. Ludovic Chabant <ludovic@chabant.com> parents: 363 diff changeset	35 '--default-page-category',
4850f8c21b6e core: Start of the big refactor for PieCrust 3.0. Ludovic Chabant <ludovic@chabant.com> parents: 363 diff changeset	36 help="The default category to use for pages.")
300 2daa05a21026 import: Add an XML-based Wordpress importer. Ludovic Chabant <ludovic@chabant.com> parents: diff changeset	37
2daa05a21026 import: Add an XML-based Wordpress importer. Ludovic Chabant <ludovic@chabant.com> parents: diff changeset	38 def importWebsite(self, app, args):
302 103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	39 impl = self._getImplementation(app, args)
300 2daa05a21026 import: Add an XML-based Wordpress importer. Ludovic Chabant <ludovic@chabant.com> parents: diff changeset	40 return impl.importWebsite()
2daa05a21026 import: Add an XML-based Wordpress importer. Ludovic Chabant <ludovic@chabant.com> parents: diff changeset	41
302 103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	42 def _getImplementation(self, app, args):
103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	43 raise NotImplementedError()
300 2daa05a21026 import: Add an XML-based Wordpress importer. Ludovic Chabant <ludovic@chabant.com> parents: diff changeset	44
302 103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	45
103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	46 class _ImporterBase(object):
103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	47 def __init__(self, app, args):
103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	48 self.app = app
103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	49 self._cat_map = {}
103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	50 self._author_map = {}
103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	51 self._pages_source = app.getSource(args.pages_source)
103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	52 self._posts_source = app.getSource(args.posts_source)
103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	53
103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	54 def importWebsite(self):
103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	55 ctx = self._open()
103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	56
103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	57 # Site configuration.
103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	58 logger.info("Generating site configuration...")
103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	59 site_config = self._getSiteConfig(ctx)
103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	60 site_config.setdefault('site', {})
103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	61 site_config['site'].update({
852 4850f8c21b6e core: Start of the big refactor for PieCrust 3.0. Ludovic Chabant <ludovic@chabant.com> parents: 363 diff changeset	62 'post_url': '%year%/%month%/%slug%',
4850f8c21b6e core: Start of the big refactor for PieCrust 3.0. Ludovic Chabant <ludovic@chabant.com> parents: 363 diff changeset	63 'category_url': 'category/%category%'})
302 103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	64
103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	65 site_config_path = os.path.join(self.app.root_dir, CONFIG_PATH)
103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	66 with open(site_config_path, 'r') as fp:
103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	67 cfg_data = yaml.load(fp, Loader=ConfigurationLoader)
103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	68
103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	69 cfg_data = cfg_data or {}
103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	70 merge_dicts(cfg_data, site_config)
103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	71
103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	72 with open(site_config_path, 'w') as fp:
103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	73 yaml.dump(cfg_data, fp, default_flow_style=False,
103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	74 allow_unicode=True,
103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	75 Dumper=ConfigurationDumper)
103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	76
103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	77 # Content
103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	78 for p in self._getPosts(ctx):
103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	79 if p['type'] == 'attachment':
103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	80 self._createAsset(p)
103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	81 else:
103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	82 self._createPost(p)
103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	83
103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	84 self._close(ctx)
103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	85
103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	86 def _open(self):
103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	87 raise NotImplementedError()
103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	88
103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	89 def _close(self, ctx):
103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	90 pass
103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	91
103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	92 def _getSiteConfig(self, ctx):
103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	93 raise NotImplementedError()
103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	94
103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	95 def _getPosts(self, ctx):
103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	96 raise NotImplementedError()
103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	97
103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	98 def _createAsset(self, asset_info):
103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	99 download_asset(self.app, asset_info['url'])
103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	100
103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	101 def _createPost(self, post_info):
103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	102 post_dt = post_info['datetime']
103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	103 finder = {
852 4850f8c21b6e core: Start of the big refactor for PieCrust 3.0. Ludovic Chabant <ludovic@chabant.com> parents: 363 diff changeset	104 'year': post_dt.year,
4850f8c21b6e core: Start of the big refactor for PieCrust 3.0. Ludovic Chabant <ludovic@chabant.com> parents: 363 diff changeset	105 'month': post_dt.month,
4850f8c21b6e core: Start of the big refactor for PieCrust 3.0. Ludovic Chabant <ludovic@chabant.com> parents: 363 diff changeset	106 'day': post_dt.day,
4850f8c21b6e core: Start of the big refactor for PieCrust 3.0. Ludovic Chabant <ludovic@chabant.com> parents: 363 diff changeset	107 'slug': post_info['slug']}
302 103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	108 if post_info['type'] == 'post':
103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	109 source = self._posts_source
103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	110 elif post_info['type'] == 'page':
103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	111 source = self._pages_source
103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	112 else:
103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	113 raise Exception("Unknown post type: %s" % post_info['type'])
363 dd25bd3ce1f9 serve: Refactoring and fixes to be able to serve taxonomy pages. Ludovic Chabant <ludovic@chabant.com> parents: 314 diff changeset	114 factory = source.findPageFactory(finder, MODE_CREATING)
302 103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	115
103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	116 metadata = post_info['metadata'].copy()
103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	117 for name in ['title', 'author', 'status', 'post_id', 'post_guid',
103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	118 'description', 'categories']:
103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	119 val = post_info.get(name)
103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	120 if val is not None:
103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	121 metadata[name] = val
103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	122
103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	123 content = post_info['content']
103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	124 excerpt = post_info['excerpt']
103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	125 text = content
103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	126 if excerpt is not None and excerpt.strip() != '':
103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	127 text = "%s\n\n---excerpt---\n\n%s" % (content, excerpt)
103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	128
314 1ddd18ad5e76 import: Wordpress importer puts drafts in a `draft` folder. Ignore other statuses. Ludovic Chabant <ludovic@chabant.com> parents: 302 diff changeset	129 status = metadata.get('status')
1ddd18ad5e76 import: Wordpress importer puts drafts in a `draft` folder. Ignore other statuses. Ludovic Chabant <ludovic@chabant.com> parents: 302 diff changeset	130 if status == 'publish':
363 dd25bd3ce1f9 serve: Refactoring and fixes to be able to serve taxonomy pages. Ludovic Chabant <ludovic@chabant.com> parents: 314 diff changeset	131 path = factory.path
314 1ddd18ad5e76 import: Wordpress importer puts drafts in a `draft` folder. Ignore other statuses. Ludovic Chabant <ludovic@chabant.com> parents: 302 diff changeset	132 create_page(self.app, path, metadata, text)
1ddd18ad5e76 import: Wordpress importer puts drafts in a `draft` folder. Ignore other statuses. Ludovic Chabant <ludovic@chabant.com> parents: 302 diff changeset	133 elif status == 'draft':
1ddd18ad5e76 import: Wordpress importer puts drafts in a `draft` folder. Ignore other statuses. Ludovic Chabant <ludovic@chabant.com> parents: 302 diff changeset	134 filename = '-'.join(metadata['title'].split(' ')) + '.html'
1ddd18ad5e76 import: Wordpress importer puts drafts in a `draft` folder. Ignore other statuses. Ludovic Chabant <ludovic@chabant.com> parents: 302 diff changeset	135 path = os.path.join(self.app.root_dir, 'drafts', filename)
1ddd18ad5e76 import: Wordpress importer puts drafts in a `draft` folder. Ignore other statuses. Ludovic Chabant <ludovic@chabant.com> parents: 302 diff changeset	136 create_page(self.app, path, metadata, text)
1ddd18ad5e76 import: Wordpress importer puts drafts in a `draft` folder. Ignore other statuses. Ludovic Chabant <ludovic@chabant.com> parents: 302 diff changeset	137 else:
1ddd18ad5e76 import: Wordpress importer puts drafts in a `draft` folder. Ignore other statuses. Ludovic Chabant <ludovic@chabant.com> parents: 302 diff changeset	138 logger.warning("Ignoring post with status: %s" % status)
302 103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	139
103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	140
103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	141 class _XmlImporter(_ImporterBase):
300 2daa05a21026 import: Add an XML-based Wordpress importer. Ludovic Chabant <ludovic@chabant.com> parents: diff changeset	142 ns_wp = {'wp': 'http://wordpress.org/export/1.2/'}
2daa05a21026 import: Add an XML-based Wordpress importer. Ludovic Chabant <ludovic@chabant.com> parents: diff changeset	143 ns_dc = {'dc': "http://purl.org/dc/elements/1.1/"}
2daa05a21026 import: Add an XML-based Wordpress importer. Ludovic Chabant <ludovic@chabant.com> parents: diff changeset	144 ns_excerpt = {'excerpt': "http://wordpress.org/export/1.2/excerpt/"}
2daa05a21026 import: Add an XML-based Wordpress importer. Ludovic Chabant <ludovic@chabant.com> parents: diff changeset	145 ns_content = {'content': "http://purl.org/rss/1.0/modules/content/"}
2daa05a21026 import: Add an XML-based Wordpress importer. Ludovic Chabant <ludovic@chabant.com> parents: diff changeset	146
2daa05a21026 import: Add an XML-based Wordpress importer. Ludovic Chabant <ludovic@chabant.com> parents: diff changeset	147 def __init__(self, app, args):
302 103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	148 super(_XmlImporter, self).__init__(app, args)
103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	149 self.path = args.xml_path
300 2daa05a21026 import: Add an XML-based Wordpress importer. Ludovic Chabant <ludovic@chabant.com> parents: diff changeset	150
302 103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	151 def _open(self):
300 2daa05a21026 import: Add an XML-based Wordpress importer. Ludovic Chabant <ludovic@chabant.com> parents: diff changeset	152 if not os.path.exists(self.path):
2daa05a21026 import: Add an XML-based Wordpress importer. Ludovic Chabant <ludovic@chabant.com> parents: diff changeset	153 raise Exception("No such file: %s" % self.path)
2daa05a21026 import: Add an XML-based Wordpress importer. Ludovic Chabant <ludovic@chabant.com> parents: diff changeset	154
2daa05a21026 import: Add an XML-based Wordpress importer. Ludovic Chabant <ludovic@chabant.com> parents: diff changeset	155 try:
2daa05a21026 import: Add an XML-based Wordpress importer. Ludovic Chabant <ludovic@chabant.com> parents: diff changeset	156 import xml.etree.ElementTree as ET
2daa05a21026 import: Add an XML-based Wordpress importer. Ludovic Chabant <ludovic@chabant.com> parents: diff changeset	157 except ImportError:
2daa05a21026 import: Add an XML-based Wordpress importer. Ludovic Chabant <ludovic@chabant.com> parents: diff changeset	158 logger.error("You don't seem to have any support for ElementTree "
2daa05a21026 import: Add an XML-based Wordpress importer. Ludovic Chabant <ludovic@chabant.com> parents: diff changeset	159 "XML parsing.")
2daa05a21026 import: Add an XML-based Wordpress importer. Ludovic Chabant <ludovic@chabant.com> parents: diff changeset	160 return 1
2daa05a21026 import: Add an XML-based Wordpress importer. Ludovic Chabant <ludovic@chabant.com> parents: diff changeset	161
2daa05a21026 import: Add an XML-based Wordpress importer. Ludovic Chabant <ludovic@chabant.com> parents: diff changeset	162 with open(self.path, 'r', encoding='utf8') as fp:
2daa05a21026 import: Add an XML-based Wordpress importer. Ludovic Chabant <ludovic@chabant.com> parents: diff changeset	163 xml = fp.read()
2daa05a21026 import: Add an XML-based Wordpress importer. Ludovic Chabant <ludovic@chabant.com> parents: diff changeset	164 xml = xml.replace(chr(0x1e), '')
2daa05a21026 import: Add an XML-based Wordpress importer. Ludovic Chabant <ludovic@chabant.com> parents: diff changeset	165 xml = xml.replace(chr(0x10), '')
2daa05a21026 import: Add an XML-based Wordpress importer. Ludovic Chabant <ludovic@chabant.com> parents: diff changeset	166 tree = ET.fromstring(xml)
2daa05a21026 import: Add an XML-based Wordpress importer. Ludovic Chabant <ludovic@chabant.com> parents: diff changeset	167 channel = tree.find('channel')
2daa05a21026 import: Add an XML-based Wordpress importer. Ludovic Chabant <ludovic@chabant.com> parents: diff changeset	168
302 103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	169 return channel
103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	170
103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	171 def _getSiteConfig(self, channel):
300 2daa05a21026 import: Add an XML-based Wordpress importer. Ludovic Chabant <ludovic@chabant.com> parents: diff changeset	172 # Get basic site information
2daa05a21026 import: Add an XML-based Wordpress importer. Ludovic Chabant <ludovic@chabant.com> parents: diff changeset	173 title = find_text(channel, 'title')
2daa05a21026 import: Add an XML-based Wordpress importer. Ludovic Chabant <ludovic@chabant.com> parents: diff changeset	174 description = find_text(channel, 'description')
302 103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	175 site_config = OrderedDict({
852 4850f8c21b6e core: Start of the big refactor for PieCrust 3.0. Ludovic Chabant <ludovic@chabant.com> parents: 363 diff changeset	176 'site': {
4850f8c21b6e core: Start of the big refactor for PieCrust 3.0. Ludovic Chabant <ludovic@chabant.com> parents: 363 diff changeset	177 'title': title,
4850f8c21b6e core: Start of the big refactor for PieCrust 3.0. Ludovic Chabant <ludovic@chabant.com> parents: 363 diff changeset	178 'description': description}
4850f8c21b6e core: Start of the big refactor for PieCrust 3.0. Ludovic Chabant <ludovic@chabant.com> parents: 363 diff changeset	179 })
300 2daa05a21026 import: Add an XML-based Wordpress importer. Ludovic Chabant <ludovic@chabant.com> parents: diff changeset	180
2daa05a21026 import: Add an XML-based Wordpress importer. Ludovic Chabant <ludovic@chabant.com> parents: diff changeset	181 # Get authors' names.
2daa05a21026 import: Add an XML-based Wordpress importer. Ludovic Chabant <ludovic@chabant.com> parents: diff changeset	182 authors = {}
2daa05a21026 import: Add an XML-based Wordpress importer. Ludovic Chabant <ludovic@chabant.com> parents: diff changeset	183 for a in channel.findall('wp:author', self.ns_wp):
2daa05a21026 import: Add an XML-based Wordpress importer. Ludovic Chabant <ludovic@chabant.com> parents: diff changeset	184 login = find_text(a, 'wp:author_login', self.ns_wp)
2daa05a21026 import: Add an XML-based Wordpress importer. Ludovic Chabant <ludovic@chabant.com> parents: diff changeset	185 authors[login] = {
852 4850f8c21b6e core: Start of the big refactor for PieCrust 3.0. Ludovic Chabant <ludovic@chabant.com> parents: 363 diff changeset	186 'email': find_text(a, 'wp:author_email', self.ns_wp),
4850f8c21b6e core: Start of the big refactor for PieCrust 3.0. Ludovic Chabant <ludovic@chabant.com> parents: 363 diff changeset	187 'display_name': find_text(a, 'wp:author_display_name',
4850f8c21b6e core: Start of the big refactor for PieCrust 3.0. Ludovic Chabant <ludovic@chabant.com> parents: 363 diff changeset	188 self.ns_wp),
4850f8c21b6e core: Start of the big refactor for PieCrust 3.0. Ludovic Chabant <ludovic@chabant.com> parents: 363 diff changeset	189 'first_name': find_text(a, 'wp:author_first_name',
4850f8c21b6e core: Start of the big refactor for PieCrust 3.0. Ludovic Chabant <ludovic@chabant.com> parents: 363 diff changeset	190 self.ns_wp),
4850f8c21b6e core: Start of the big refactor for PieCrust 3.0. Ludovic Chabant <ludovic@chabant.com> parents: 363 diff changeset	191 'last_name': find_text(a, 'wp:author_last_name',
4850f8c21b6e core: Start of the big refactor for PieCrust 3.0. Ludovic Chabant <ludovic@chabant.com> parents: 363 diff changeset	192 self.ns_wp),
4850f8c21b6e core: Start of the big refactor for PieCrust 3.0. Ludovic Chabant <ludovic@chabant.com> parents: 363 diff changeset	193 'author_id': find_text(a, 'wp:author_id',
4850f8c21b6e core: Start of the big refactor for PieCrust 3.0. Ludovic Chabant <ludovic@chabant.com> parents: 363 diff changeset	194 self.ns_wp)}
300 2daa05a21026 import: Add an XML-based Wordpress importer. Ludovic Chabant <ludovic@chabant.com> parents: diff changeset	195 site_config['site']['authors'] = authors
2daa05a21026 import: Add an XML-based Wordpress importer. Ludovic Chabant <ludovic@chabant.com> parents: diff changeset	196
302 103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	197 return site_config
300 2daa05a21026 import: Add an XML-based Wordpress importer. Ludovic Chabant <ludovic@chabant.com> parents: diff changeset	198
302 103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	199 def _getPosts(self, channel):
300 2daa05a21026 import: Add an XML-based Wordpress importer. Ludovic Chabant <ludovic@chabant.com> parents: diff changeset	200 for i in channel.findall('item'):
2daa05a21026 import: Add an XML-based Wordpress importer. Ludovic Chabant <ludovic@chabant.com> parents: diff changeset	201 post_type = find_text(i, 'wp:post_type', self.ns_wp)
2daa05a21026 import: Add an XML-based Wordpress importer. Ludovic Chabant <ludovic@chabant.com> parents: diff changeset	202 if post_type == 'attachment':
302 103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	203 yield self._getAssetInfo(i)
103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	204 else:
103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	205 yield self._getPostInfo(i)
103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	206
103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	207 def _getAssetInfo(self, node):
103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	208 url = find_text(node, 'wp:attachment_url', self.ns_wp)
103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	209 return {'type': 'attachment', 'url': url}
300 2daa05a21026 import: Add an XML-based Wordpress importer. Ludovic Chabant <ludovic@chabant.com> parents: diff changeset	210
302 103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	211 def _getPostInfo(self, node):
103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	212 post_date_str = find_text(node, 'wp:post_date', self.ns_wp)
103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	213 post_date = datetime.datetime.strptime(post_date_str,
103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	214 '%Y-%m-%d %H:%M:%S')
103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	215 post_name = find_text(node, 'wp:post_name', self.ns_wp)
103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	216 post_type = find_text(node, 'wp:post_type', self.ns_wp)
103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	217 post_info = {
852 4850f8c21b6e core: Start of the big refactor for PieCrust 3.0. Ludovic Chabant <ludovic@chabant.com> parents: 363 diff changeset	218 'type': post_type,
4850f8c21b6e core: Start of the big refactor for PieCrust 3.0. Ludovic Chabant <ludovic@chabant.com> parents: 363 diff changeset	219 'slug': post_name,
4850f8c21b6e core: Start of the big refactor for PieCrust 3.0. Ludovic Chabant <ludovic@chabant.com> parents: 363 diff changeset	220 'datetime': post_date}
300 2daa05a21026 import: Add an XML-based Wordpress importer. Ludovic Chabant <ludovic@chabant.com> parents: diff changeset	221
2daa05a21026 import: Add an XML-based Wordpress importer. Ludovic Chabant <ludovic@chabant.com> parents: diff changeset	222 title = find_text(node, 'title')
2daa05a21026 import: Add an XML-based Wordpress importer. Ludovic Chabant <ludovic@chabant.com> parents: diff changeset	223 creator = find_text(node, 'dc:creator', self.ns_dc)
2daa05a21026 import: Add an XML-based Wordpress importer. Ludovic Chabant <ludovic@chabant.com> parents: diff changeset	224 status = find_text(node, 'wp:status', self.ns_wp)
2daa05a21026 import: Add an XML-based Wordpress importer. Ludovic Chabant <ludovic@chabant.com> parents: diff changeset	225 post_id = find_text(node, 'wp:post_id', self.ns_wp)
2daa05a21026 import: Add an XML-based Wordpress importer. Ludovic Chabant <ludovic@chabant.com> parents: diff changeset	226 guid = find_text(node, 'guid')
2daa05a21026 import: Add an XML-based Wordpress importer. Ludovic Chabant <ludovic@chabant.com> parents: diff changeset	227 description = find_text(node, 'description')
2daa05a21026 import: Add an XML-based Wordpress importer. Ludovic Chabant <ludovic@chabant.com> parents: diff changeset	228 # TODO: menu order, parent, password, sticky
302 103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	229 post_info.update({
852 4850f8c21b6e core: Start of the big refactor for PieCrust 3.0. Ludovic Chabant <ludovic@chabant.com> parents: 363 diff changeset	230 'title': title,
4850f8c21b6e core: Start of the big refactor for PieCrust 3.0. Ludovic Chabant <ludovic@chabant.com> parents: 363 diff changeset	231 'author': creator,
4850f8c21b6e core: Start of the big refactor for PieCrust 3.0. Ludovic Chabant <ludovic@chabant.com> parents: 363 diff changeset	232 'status': status,
4850f8c21b6e core: Start of the big refactor for PieCrust 3.0. Ludovic Chabant <ludovic@chabant.com> parents: 363 diff changeset	233 'post_id': post_id,
4850f8c21b6e core: Start of the big refactor for PieCrust 3.0. Ludovic Chabant <ludovic@chabant.com> parents: 363 diff changeset	234 'post_guid': guid,
4850f8c21b6e core: Start of the big refactor for PieCrust 3.0. Ludovic Chabant <ludovic@chabant.com> parents: 363 diff changeset	235 'description': description})
300 2daa05a21026 import: Add an XML-based Wordpress importer. Ludovic Chabant <ludovic@chabant.com> parents: diff changeset	236
302 103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	237 categories = []
103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	238 for c in node.findall('category'):
103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	239 nicename = str(c.attrib.get('nicename'))
103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	240 categories.append(nicename)
103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	241 post_info['categories'] = categories
103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	242
103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	243 metadata = {}
300 2daa05a21026 import: Add an XML-based Wordpress importer. Ludovic Chabant <ludovic@chabant.com> parents: diff changeset	244 for m in node.findall('wp:postmeta', self.ns_wp):
2daa05a21026 import: Add an XML-based Wordpress importer. Ludovic Chabant <ludovic@chabant.com> parents: diff changeset	245 key = find_text(m, 'wp:meta_key', self.ns_wp)
2daa05a21026 import: Add an XML-based Wordpress importer. Ludovic Chabant <ludovic@chabant.com> parents: diff changeset	246 metadata[key] = find_text(m, 'wp:meta_value', self.ns_wp)
302 103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	247 post_info['metadata'] = metadata
300 2daa05a21026 import: Add an XML-based Wordpress importer. Ludovic Chabant <ludovic@chabant.com> parents: diff changeset	248
2daa05a21026 import: Add an XML-based Wordpress importer. Ludovic Chabant <ludovic@chabant.com> parents: diff changeset	249 content = find_text(node, 'content:encoded', self.ns_content)
2daa05a21026 import: Add an XML-based Wordpress importer. Ludovic Chabant <ludovic@chabant.com> parents: diff changeset	250 excerpt = find_text(node, 'excerpt:encoded', self.ns_excerpt)
302 103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	251 post_info.update({
852 4850f8c21b6e core: Start of the big refactor for PieCrust 3.0. Ludovic Chabant <ludovic@chabant.com> parents: 363 diff changeset	252 'content': content,
4850f8c21b6e core: Start of the big refactor for PieCrust 3.0. Ludovic Chabant <ludovic@chabant.com> parents: 363 diff changeset	253 'excerpt': excerpt})
300 2daa05a21026 import: Add an XML-based Wordpress importer. Ludovic Chabant <ludovic@chabant.com> parents: diff changeset	254
302 103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	255 return post_info
300 2daa05a21026 import: Add an XML-based Wordpress importer. Ludovic Chabant <ludovic@chabant.com> parents: diff changeset	256
2daa05a21026 import: Add an XML-based Wordpress importer. Ludovic Chabant <ludovic@chabant.com> parents: diff changeset	257
302 103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	258 class WordpressXmlImporter(WordpressImporterBase):
103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	259 name = 'wordpress-xml'
103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	260 description = "Imports a Wordpress blog from an exported XML archive."
300 2daa05a21026 import: Add an XML-based Wordpress importer. Ludovic Chabant <ludovic@chabant.com> parents: diff changeset	261
302 103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	262 def setupParser(self, parser, app):
103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	263 super(WordpressXmlImporter, self).setupParser(parser, app)
103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	264 parser.add_argument(
103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	265 'xml_path',
103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	266 help="The path to the exported XML archive file.")
103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	267
103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	268 def _getImplementation(self, app, args):
103abb08755e import: Make the Wordpress importer extendable, rename it to `wordpressxml`. Ludovic Chabant <ludovic@chabant.com> parents: 300 diff changeset	269 return _XmlImporter(app, args)
300 2daa05a21026 import: Add an XML-based Wordpress importer. Ludovic Chabant <ludovic@chabant.com> parents: diff changeset	270
2daa05a21026 import: Add an XML-based Wordpress importer. Ludovic Chabant <ludovic@chabant.com> parents: diff changeset	271
2daa05a21026 import: Add an XML-based Wordpress importer. Ludovic Chabant <ludovic@chabant.com> parents: diff changeset	272 def find_text(parent, child_name, namespaces=None):
2daa05a21026 import: Add an XML-based Wordpress importer. Ludovic Chabant <ludovic@chabant.com> parents: diff changeset	273 return str(parent.find(child_name, namespaces).text)
2daa05a21026 import: Add an XML-based Wordpress importer. Ludovic Chabant <ludovic@chabant.com> parents: diff changeset	274

Mercurial > piecrust2

annotate piecrust/importing/wordpress.py @ 1188:a7c43131d871