Merge pull request #29 from OpenBookPublishers/feature/write_urls

Write Landing Page and Full Text URLs to Thoth automatically on chapter creation

README.md

@@ -20,8 +20,9 @@ The entry point is `.src/main.py`, which can be called with the following argume
 | `-u`      | URL path of this book                                                     |
 | `-k`      | Copyright file                                                            |
 | `-a`      | Donation link                                                             |
-| `-m`      | MathJax support                                                             |
-| `-p`      | Privacy policy URL                                                             |
+| `-m`      | MathJax support                                                           |
+| `-p`      | Privacy policy URL                                                        |
+| `-w`      | Write chapter URLs to Thoth (True/False)                                  |
 
 
 ## Thoth Wrapper (Optional)
@@ -38,12 +39,15 @@ docker run --rm \
            -v /path/to/output:/ebook_automation/output \
            -e MATHJAX=False \
            -e PRIVACYPOLICY_URL=https://example.org \
+           -e [email protected] \
+           -e THOTH_PWD=password \
            openbookpublishers/epublius \
 	   thoth_wrapper.py /ebook_automation/epub_file.epub \
                             --doi 10.11647/obp.0275
 ```
 
 The environment variable MATHJAX enables or disable MathJax support
+The environment variables THOTH_EMAIL and THOTH_PWD allow use of the `--write-urls` option by providing Thoth credentials
 
 ## Contributing
 

src/epublius/epublius.py

@@ -73,6 +73,10 @@ def parse_args(self, argv=None):
         parser.add_argument('-p', '--privacy-policy',
                             help = 'Privacy policy URL')
 
+        parser.add_argument('-w', '--write-urls',
+                            help = 'Write HTML chapter URLs to Thoth',
+                            default = False)
+
         return parser.parse_args(argv)
 
     def unzip_epub(self, prefix):

src/epublius/metadata.py

@@ -126,6 +126,23 @@ def get_chapter_title(self):
 
         return html.escape(ch_title)
 
+    def get_chapter_doi(self):
+        '''
+        Retrieve chapter DOI based on the text of <p class=doi>
+        (this contains both copyright statement and DOI link)
+        '''
+        # Not all chapters will have DOIs
+        doi = None
+
+        doi_node = self.soup.find('p', class_='doi')
+
+        if (doi_node is not None):
+            doi_link = doi_node.a
+            if (doi_link is not None) and (doi_link.string is not None):
+                doi = doi_link.string
+
+        return doi
+
     def get_css(self):
         '''
         Return a str with the CSS information of a file

src/epublius/thoth.py

@@ -0,0 +1,72 @@
+#!/usr/bin/env python3
+
+from os import getenv
+from thothlibrary import ThothClient
+import urllib.parse
+
+
+class Thoth:
+    '''
+    Module to handle Thoth interactions
+    '''
+
+    def __init__(self):
+        self.client = ThothClient()
+        self.logged_in = self.login()
+
+    def login(self):
+        username = getenv('THOTH_EMAIL')
+        password = getenv('THOTH_PWD')
+        if username is None:
+            print('[WARNING] No Thoth username provided '
+                '(THOTH_EMAIL environment variable not set)')
+            return False
+        if password is None:
+            print('[WARNING] No Thoth password provided '
+                '(THOTH_PWD environment variable not set)')
+            return False
+        try:
+            self.client.login(username, password)
+            return True
+        except:
+            return False
+
+    def write_urls(self, metadata, book_doi):
+        '''
+        Write chapter Landing Page and Full Text URLs
+        to Thoth in standard OBP format
+        '''
+        chapter_doi_full = metadata.get_chapter_doi()
+
+        # Skip chapters that don't have a DOI
+        if chapter_doi_full is not None:
+            work_id = self.client.work_by_doi(chapter_doi_full).workId
+            chapter_doi = chapter_doi_full.split('doi.org/')[-1].lower()
+            landing_page_root = (
+                'https://www.openbookpublishers.com/books/'
+                '{book_doi}/chapters/{chapter_doi}')
+
+            publication = {"workId":          work_id,
+                           "publicationType": "HTML",
+                           "isbn":            None,
+                           "widthMm":         None,
+                           "widthIn":         None,
+                           "heightMm":        None,
+                           "heightIn":        None,
+                           "depthMm":         None,
+                           "depthIn":         None,
+                           "weightG":         None,
+                           "weightOz":        None}
+            publication_id = self.client.create_publication(publication)
+
+            location = {"publicationId":    publication_id,
+                        "landingPage":      landing_page_root.format(
+                            book_doi=book_doi, chapter_doi=chapter_doi),
+                        "fullTextUrl":      urllib.parse.unquote_plus(
+                            metadata.get_page_url()),
+                        "locationPlatform": "OTHER",
+                        "canonical":        "true"}
+            self.client.create_location(location)
+
+            print('{}: URLs written to Thoth'.format(
+                metadata.contents[metadata.index]))

src/main.py

@@ -5,6 +5,8 @@
 from epublius.epublius import Epublius
 from epublius.metadata import Metadata
 from epublius.output import Output
+from epublius.thoth import Thoth
+from thothlibrary import ThothError
 
 
 def main():
@@ -15,6 +17,11 @@ def main():
         # Create instances
         epublius = Epublius(work_dir)
         output = Output(os.path.abspath('assets/template.xhtml'))
+        thoth = Thoth()
+
+        # Warn if user requested to write URLs to Thoth but Thoth login failed
+        if epublius.argv.write_urls and not thoth.logged_in:
+            print('[WARNING] Thoth login failed; URLs will not be written')
 
         # Get book contents
         contents = epublius.get_contents()
@@ -57,6 +64,13 @@ def main():
             file_path = os.path.join(output_directory, section)
             output.write_file(processed_content, file_path)
 
+            if epublius.argv.write_urls and thoth.logged_in:
+                # Write chapter URL metadata to Thoth
+                try:
+                    thoth.write_urls(metadata, epublius.argv.doi)
+                except (KeyError, ThothError) as e:
+                    # Continue on error, but display warning
+                    print('[WARNING] Error writing URLs to Thoth for {}: {}'.format(section, e))
 
         # Duplicate TOC file to output_directory/main.html
         epublius.duplicate_contents(TOC_filepath.get('TOC_filepath'))

src/thoth_wrapper.py

@@ -12,7 +12,7 @@
 def query_thoth(doi_url):
     thoth = ThothClient()
     return thoth.query('workByDoi', {'doi': f'"{doi_url}"'})
-    
+
 def get_title(thoth_data):
     return thoth_data["fullTitle"]
 
@@ -37,9 +37,9 @@ def get_html_pub_url(thoth_data):
 def run():
     parser = argparse.ArgumentParser(description='Thoth wrapper')
     parser.add_argument('epub_path', help='Path to epub file')
-    parser.add_argument('-d', '--doi', help='Work DOI (registered in Thoth)')
+    parser.add_argument('-d', '--doi', help='Work DOI (registered in Thoth)', required=True)
     args = parser.parse_args()
-    
+
     doi_url = urllib.parse.urljoin('https://doi.org/', args.doi)
 
     thoth_data = query_thoth(doi_url)
@@ -57,7 +57,8 @@ def run():
             "-t", os.path.join(epublius_dir, ""),
             "-d", args.doi,
             "-m", MATHJAX,
-            "-p", os.getenv('PRIVACYPOLICY_URL', '#')]
+            "-p", os.getenv('PRIVACYPOLICY_URL', '#'),
+            "-w", 'True']
 
     os.execvp(sys.executable, [exe] + args)