Merge branch 'develop' into 'master'

[feature][WDT-116] Data provider for sphinx html



See merge request !1

conf_publisher/data_providers/sphinx_base_data_provider.py

@@ -0,0 +1,42 @@
+import os
+
+from . import DataProvider
+
+
+class SphinxBaseDataProvider(DataProvider):
+
+    DEFAULT_SOURCE_EXT = None
+    DEFAULT_ROOT_DIR = './'
+    DEFAULT_SOURCE_DIR = 'docs/build/result'
+    DEFAULT_IMAGES_DIR = '_images'
+    DEFAULT_DOWNLOADS_DIR = '_downloads'
+
+    def __init__(self, root_dir=None, base_dir=None, downloads_dir=None, images_dir=None, source_ext=None):
+        self._root_dir = os.path.abspath(root_dir or './')
+
+        self._source_dir = base_dir or self.DEFAULT_SOURCE_DIR
+        if not os.path.isabs(self._source_dir):
+            self._source_dir = os.path.join(self._root_dir, self._source_dir)
+
+        self._downloads_dir = downloads_dir or self.DEFAULT_DOWNLOADS_DIR
+        if not os.path.isabs(self._downloads_dir):
+            self._downloads_dir = os.path.join(self._source_dir, self._downloads_dir)
+
+        self._images_dir = images_dir or self.DEFAULT_IMAGES_DIR
+        if not os.path.isabs(self._images_dir):
+            self._images_dir = os.path.join(self._source_dir, self._images_dir)
+
+        self._source_ext = source_ext or self.DEFAULT_SOURCE_EXT
+
+    def get_source(self, filename):
+        print self._source_dir, self._source_ext
+        return os.path.join(self._source_dir, filename + self._source_ext)
+
+    def get_source_data(self, filename):
+        raise NotImplementedError()
+
+    def get_image(self, filename):
+        return os.path.join(self._images_dir, filename)
+
+    def get_attachment(self, filename):
+        return os.path.join(self._downloads_dir, filename)

conf_publisher/data_providers/sphinx_fjson_data_provider.py

@@ -1,38 +1,18 @@
 import os
-
-from . import DataProvider
+from .sphinx_base_data_provider import SphinxBaseDataProvider
 from ..serializers import json_serializer
 
 
-class SphinxFJsonDataProvider(DataProvider):
-    def __init__(self, root_dir=None, base_dir=None, downloads_dir=None, images_dir=None, source_ext=None):
-        self._root_dir = os.path.abspath(root_dir or './')
-
-        self._source_dir = base_dir or 'docs/build/json'
-        if not os.path.isabs(self._source_dir):
-            self._source_dir = os.path.join(self._root_dir, self._source_dir)
-
-        self._downloads_dir = downloads_dir or '_downloads'
-        if not os.path.isabs(self._downloads_dir):
-            self._downloads_dir = os.path.join(self._source_dir, self._downloads_dir)
-
-        self._images_dir = images_dir or '_images'
-        if not os.path.isabs(self._images_dir):
-            self._images_dir = os.path.join(self._source_dir, self._images_dir)
-
-        self._source_ext = source_ext or '.fjson'
-
-    def get_source(self, filename):
-        return os.path.join(self._source_dir, filename + self._source_ext)
+class SphinxFJsonDataProvider(SphinxBaseDataProvider):
+    DEFAULT_SOURCE_EXT = '.fjson'
+    DEFAULT_SOURCE_DIR = 'docs/build/json'
 
     def get_source_data(self, filename):
-        with open(filename, 'r') as f:
+        if os.path.isabs(filename):
+            filepath = filename
+        else:
+            filepath = self.get_source(filename)
+        with open(filepath, 'r') as f:
             content = json_serializer.load(f)
 
         return content['title'], content['body']
-
-    def get_image(self, filename):
-        return os.path.join(self._images_dir, filename)
-
-    def get_attachment(self, filename):
-        return os.path.join(self._downloads_dir, filename)

conf_publisher/data_providers/sphinx_html_data_provider.py

@@ -0,0 +1,16 @@
+from .sphinx_base_data_provider import SphinxBaseDataProvider
+import re
+import codecs
+
+
+class SphinxHTMLDataProvider(SphinxBaseDataProvider):
+    DEFAULT_SOURCE_EXT = '.html'
+    DEFAULT_SOURCE_DIR = 'docs/build/html'
+
+    def get_source_data(self, filename):
+        with codecs.open(filename, 'r', encoding='utf-8') as f:
+            content = f.read()
+            title = re.findall(r'<title.*?>(.+?)</title>', content)[0]
+            body = re.findall(r'<body.*?>(.+?)</body>', content, re.DOTALL)[0].strip()
+
+        return title, body

conf_publisher/publish.py

@@ -7,15 +7,30 @@
 from .config import ConfigLoader, flatten_page_config_list, PageImageAattachmentConfig
 from .constants import DEFAULT_CONFLUENCE_API_VERSION, DEFAULT_WATERMARK_CONTENT
 from .data_providers.sphinx_fjson_data_provider import SphinxFJsonDataProvider
+from .data_providers.sphinx_html_data_provider import SphinxHTMLDataProvider
 from .mutators.page_mutator import WatermarkPageMutator, LinkPageMutator
 
 
+def get_data_provider_class(config):
+    if config.source_ext == '.html':
+        data_provider_class = SphinxHTMLDataProvider
+    else:
+        data_provider_class = SphinxFJsonDataProvider
+    return data_provider_class
+
+
 def create_publisher(config, confluence_api):
     page_manager = ConfluencePageManager(confluence_api)
     attachment_publisher = AttachmentPublisher(confluence_api)
-    data_provider = SphinxFJsonDataProvider(base_dir=config.base_dir, downloads_dir=config.downloads_dir,
-                                            images_dir=config.images_dir, source_ext=config.source_ext)
 
+    data_provider_class = get_data_provider_class(config)
+
+    data_provider = data_provider_class(
+        base_dir=config.base_dir,
+        downloads_dir=config.downloads_dir,
+        images_dir=config.images_dir,
+        source_ext=config.source_ext
+    )
     return Publisher(config, data_provider, page_manager, attachment_publisher)
 
 
@@ -32,7 +47,7 @@ def _page_title(current_title, new_title, config_title=None, hold_current=False)
             return current_title
         if config_title:
             return config_title
-        return new_title
+        return new_title if new_title else current_title
 
     def _page(self, current_page, source):
         page = copy.copy(current_page)