Merge branch 'master' of github.com:charlesreid1/centillion

* 'master' of github.com:charlesreid1/centillion:
  update config_flask.example.py to strip dc info

.gitignore

@@ -1,4 +1,4 @@
-config_*
+config_flask.py
 vp
 credentials.json
 drive*.json

.gitmodules

@@ -1,3 +1,3 @@
-[submodule "mkdocs-material"]
-	path = mkdocs-material
-	url = https://git.charlesreid1.com/charlesreid1/mkdocs-material.git
+[submodule "mkdocs-material-dib"]
+	path = mkdocs-material-dib
+	url = https://github.com/dib-lab/mkdocs-material-dib.git

Readme.md

@@ -1,18 +1,19 @@
-# The Centillion
+# Centillion
 
-**the centillion**: a pan-github-markdown-issues-google-docs search engine.
+**centillion**: a pan-github-markdown-issues-google-docs search engine.
 
 **a centillion**: a very large number consisting of a 1 with 303 zeros after it.
 
-the centillion is 3.03 log-times better than the googol.
+one centillion is 3.03 log-times better than a googol.
 
-![Screen shot of centillion](img/ss.png)
+![Screen shot of centillion](docs/images/ss.png)
 
 
 ## what is it
 
-The centillion is a search engine built using [whoosh](https://whoosh.readthedocs.io/en/latest/intro.html),
-a Python library for building search engines.
+Centillion (https://github.com/dcppc/centillion) is a search engine that can index 
+three kinds of collections: Google Documents, Github issues, and Markdown files in 
+Github repos.
 
 We define the types of documents the centillion should index,
 what info and how. The centillion then builds and
@@ -24,6 +25,30 @@ defined in `centillion.py`.
 
 The centillion keeps it simple.
 
+## authentication layer
+
+Centillion lives behind a Github authentication layer, implemented with 
+[flask-dance](https://github.com/singingwolfboy/flask-dance). When you first
+visit the site it will ask you to authenticate with Github so that it can 
+verify you have permission to access the site.
+
+## technologies
+
+Centillion is a Python program built using whoosh (search engine library). It
+indexes the full text of docx files in Google Documents, just the filenames for
+non-docx files. The full text of issues and their comments are indexed, and
+results are grouped by issue. Centillion requires Google Drive and Github OAuth
+apps. Once you provide credentials to Flask you're all set to go. 
+
+
+## control panel
+
+There's also a control panel at <https://search.nihdatacommons.us/control_panel> 
+that allows you to rebuild the search index from scratch (the Google Drive indexing 
+takes a while).
+
+![Screen shot of centillion control panel](docs/images/cp.png)
+
 
 ## quickstart (with Github auth)
 
@@ -31,6 +56,8 @@ Start by creating a Github OAuth application.
 Get the public and private application key 
 (client token and client secret token)
 from the Github application's page.
+You will also need a Github access token
+(in addition to the app tokens).
 
 When you create the application, set the callback
 URL to `/login/github/authorized`, as in:
@@ -65,11 +92,3 @@ as HTTP by Github, even though there is an HTTPS address, and
 everything else seems fine, try deleting the Github OAuth app
 and creating a new one.
 
-
-## more info
-
-For more info see the documentation: <https://charlesreid1.github.io/centillion>
-
-
-
-

centillion.py

@@ -27,10 +27,16 @@ You provide:
 
 
 class UpdateIndexTask(object):
-    def __init__(self, gh_oauth_token, diff_index=False):
+    def __init__(self, app_config, diff_index=False):
         self.diff_index = diff_index
         thread = threading.Thread(target=self.run, args=())
-        self.gh_oauth_token = gh_oauth_token
+
+        self.gh_token = app_config['GITHUB_TOKEN']
+        self.groupsio_credentials = {
+                'groupsio_token' :     app_config['GROUPSIO_TOKEN'],
+                'groupsio_username' :  app_config['GROUPSIO_USERNAME'],
+                'groupsio_password' :  app_config['GROUPSIO_PASSWORD']
+        }
         thread.daemon = True
         thread.start()
 
@@ -43,9 +49,10 @@ class UpdateIndexTask(object):
         from get_centillion_config import get_centillion_config
         config = get_centillion_config('config_centillion.json')
 
-        search.update_index_markdown(self.gh_oauth_token,config)
-        search.update_index_issues(self.gh_oauth_token,config)
-        search.update_index_gdocs(config)
+        search.update_index_groupsioemails(self.groupsio_credentials,config)
+        ###search.update_index_ghfiles(self.gh_token,config)
+        ###search.update_index_issues(self.gh_token,config)
+        ###search.update_index_gdocs(config)
 
 
 
@@ -55,11 +62,11 @@ app.wsgi_app = ProxyFix(app.wsgi_app)
 # Load default config and override config from an environment variable
 app.config.from_pyfile("config_flask.py")
 
-github_bp = make_github_blueprint()
-#github_bp = make_github_blueprint(
-#                        client_id = os.environ.get('GITHUB_OAUTH_CLIENT_ID'),
-#                        client_secret = os.environ.get('GITHUB_OAUTH_CLIENT_SECRET'),
-#                        scope='read:org')
+#github_bp = make_github_blueprint()
+github_bp = make_github_blueprint(
+                        client_id = os.environ.get('GITHUB_OAUTH_CLIENT_ID'),
+                        client_secret = os.environ.get('GITHUB_OAUTH_CLIENT_SECRET'),
+                        scope='read:org')
 
 app.register_blueprint(github_bp, url_prefix="/login")
 
@@ -172,11 +179,10 @@ def update_index():
                 mresp = github.get('/teams/%s/members/%s'%(copper_team_id,username))
                 if mresp.status_code==204:
 
-                    gh_oauth_token = github.token['access_token']
-
                     # --------------------
                     # Business as usual
-                    UpdateIndexTask(gh_oauth_token, diff_index=False)
+                    UpdateIndexTask(app.config,
+                                    diff_index=False)
                     flash("Rebuilding index, check console output")
                     return render_template("controlpanel.html", 
                                            totals={})
@@ -216,5 +222,7 @@ def oops(e):
     return contents404
 
 if __name__ == '__main__':
+    # if running local instance, set to true
+    os.environ['OAUTHLIB_INSECURE_TRANSPORT'] = 'true'
     app.run(host="0.0.0.0",port=5000)
 

centillion_search.py

@@ -1,10 +1,11 @@
 import shutil
 import html.parser
 
-from github import Github
+from github import Github, GithubException
 import base64
 
 from gdrive_util import GDrive
+from groupsio_util import GroupsIOArchivesCrawler
 from apiclient.http import MediaIoBaseDownload
 
 import mistune
@@ -128,7 +129,6 @@ class Search:
         schema = Schema(
                 id = ID(stored=True, unique=True),
                 kind = ID(stored=True),
-                #fingerprint = ID(stored=True),
 
                 created_time = ID(stored=True),
                 modified_time = ID(stored=True),
@@ -252,7 +252,6 @@ class Search:
             with open(fullpath_input, 'wb') as f:
                 f.write(r.content)
 
-
             # Try to convert docx file to plain text
             try:
                 output = pypandoc.convert_file(fullpath_input,
@@ -267,7 +266,6 @@ class Search:
 
             # If export was successful, read contents of markdown
             # into the content variable.
-            # into the content variable.
             if os.path.isfile(fullpath_output):
                 # Export was successful
                 with codecs.open(fullpath_output, encoding='utf-8') as f:
@@ -277,12 +275,14 @@ class Search:
             # No matter what happens, clean up.
             print(" > Cleaning up \"%s\""%item['name'])
 
-            subprocess.call(['rm','-fr',fullpath_output])
+            ## test
             #print(" ".join(['rm','-fr',fullpath_output]))
-
-            subprocess.call(['rm','-fr',fullpath_input])
             #print(" ".join(['rm','-fr',fullpath_input]))
 
+            # do it
+            subprocess.call(['rm','-fr',fullpath_output])
+            subprocess.call(['rm','-fr',fullpath_input])
+
             if update:
                 print(" > Removing old record")
                 writer.delete_by_term('id',item['id'])
@@ -316,7 +316,7 @@ class Search:
     # to a search index.
 
 
-    def add_issue(self, writer, issue, config, update=True):
+    def add_issue(self, writer, issue, gh_token, config, update=True):
         """
         Add a Github issue/comment to a search index.
         """
@@ -368,44 +368,58 @@ class Search:
 
 
 
-    def add_markdown(self, writer, d, config, update=True):
+
+    def add_ghfile(self, writer, d, gh_token, config, update=True):
         """
-        Use a Github markdown document API record
-        to add a markdown document's contents to
-        the search index.
+        Use a Github file API record to add a filename
+        to the search index.
         """
+        MARKDOWN_EXTS = ['.md','.markdown']
+
         repo = d['repo']
         org = d['org']
         repo_name = org + "/" + repo
         repo_url = "https://github.com/" + repo_name
 
+        try:
             fpath = d['path']
             furl = d['url']
             fsha = d['sha']
             _, fname = os.path.split(fpath)
             _, fext = os.path.splitext(fpath)
+        except:
+            print(" > XXXXXXXX Failed to find file info.")
+            return
 
-        print("Indexing markdown doc %s"%(fname))
+        indexed_time = clean_timestamp(datetime.now())
+
+        if fext in MARKDOWN_EXTS:
+            print("Indexing markdown doc %s from repo %s"%(fname,repo_name))
 
             # Unpack the requests response and decode the content
-        response = requests.get(furl)
+            # 
+            # don't forget the headers for private repos!
+            # useful: https://bit.ly/2LSAflS
+
+            headers = {'Authorization' : 'token %s'%(gh_token)}
+
+            response = requests.get(furl, headers=headers)
+            if response.status_code==200:
                 jresponse = response.json()
                 content = ""
                 try:
                     binary_content = re.sub('\n','',jresponse['content'])
                     content = base64.b64decode(binary_content).decode('utf-8')
-
                 except KeyError:
                     print(" > XXXXXXXX Failed to extract 'content' field. You probably hit the rate limit.")
-            return 
 
-        # Now create the actual search index record
-        indexed_time = clean_timestamp(datetime.now())
+            else:
+                print(" > XXXXXXXX Failed to reach file URL. There may be a problem with authentication/headers.")
+                return 
 
             usable_url = "https://github.com/%s/blob/master/%s"%(repo_name, fpath)
 
-        # Add one document per issue thread,
-        # containing entire text of thread.
+            # Now create the actual search index record
             writer.add_document(
                     id = fsha,
                     kind = 'markdown',
@@ -425,12 +439,41 @@ class Search:
                     content = content
             )
 
+        else:
+            print("Indexing github file %s from repo %s"%(fname,repo_name))
+
+            key = fname+"_"+fsha
+
+            # Now create the actual search index record
+            writer.add_document(
+                    id = key,
+                    kind = 'ghfile',
+                    created_time = '',
+                    modified_time = '',
+                    indexed_time = indexed_time,
+                    title = fname,
+                    url = repo_url,
+                    mimetype='',
+                    owner_email='',
+                    owner_name='',
+                    repo_name = repo_name,
+                    repo_url = repo_url,
+                    github_user = '',
+                    issue_title = '',
+                    issue_url = '',
+                    content = ''
+            )
+
 
 
     # ------------------------------
     # Define how to update search index
     # using different kinds of collections
 
+
+    # ------------------------------
+    # Google Drive Files/Documents
+
     def update_index_gdocs(self, 
                            config):
         """
@@ -478,7 +521,7 @@ class Search:
         remote_ids = set()
         full_items = {}
         while True:
-            ps = 12
+            ps = 100
             results = drive.list(
                     pageSize=ps,
                     pageToken=nextPageToken,
@@ -496,11 +539,11 @@ class Search:
                 # Also store the doc
                 full_items[f['id']] = f
             
-            # Shorter:
+            ## Shorter:
+            #break
+            # Longer:
+            if nextPageToken is None:
                 break
-            ## Longer:
-            #if nextPageToken is None:
-            #    break
 
 
         writer = self.ix.writer()
@@ -544,13 +587,13 @@ class Search:
         print("Done, updated %d documents in the index" % count)
 
 
+    # ------------------------------
+    # Github Issues/Comments
 
-    def update_index_issues(self, gh_oauth_token, config):
+    def update_index_issues(self, gh_token, config):
         """
         Update the search index using a collection of 
         Github repo issues and comments.
-
-        gh_oauth_token can also be an access token.
         """
         # Updated algorithm:
         # - get set of indexed ids
@@ -562,7 +605,7 @@ class Search:
         # ------
         indexed_issues = set()
         p = QueryParser("kind", schema=self.ix.schema)
-        q = p.parse("gdoc")
+        q = p.parse("issue")
         with self.ix.searcher() as s:
             results = s.search(q,limit=None)
             for result in results:
@@ -572,25 +615,29 @@ class Search:
         # Get the set of remote ids:
         # ------
         # Start with api object
-        g = Github(gh_oauth_token)
+        g = Github(gh_token)
 
         # Now index all issue threads in the user-specified repos
 
-        # Iterate over each repo 
-        list_of_repos = config['repositories']
-        for r in list_of_repos:
-
         # Start by collecting all the things
         remote_issues = set()
         full_items = {}
 
+        # Iterate over each repo 
+        list_of_repos = config['repositories']
+        for r in list_of_repos:
+
             if '/' not in r:
                 err = "Error: specify org/reponame or user/reponame in list of repos"
                 raise Exception(err)
 
             this_org, this_repo = re.split('/',r)
+            try:
                 org = g.get_organization(this_org)
                 repo = org.get_repo(this_repo)
+            except:
+                print("Error: could not gain access to repository %s"%(r))
+                continue
 
             # Iterate over each issue thread
             issues = repo.get_issues()
@@ -622,7 +669,7 @@ class Search:
             # cop out
             writer.delete_by_term('id',update_issue)
             item = full_items[update_issue]
-            self.add_issue(writer, item, config, update=True)
+            self.add_issue(writer, item, gh_token, config, update=True)
             count += 1
 
 
@@ -631,7 +678,7 @@ class Search:
         add_issues = remote_issues - indexed_issues
         for add_issue in add_issues:
             item = full_items[add_issue]
-            self.add_issue(writer, item, config, update=False)
+            self.add_issue(writer, item, gh_token, config, update=False)
             count += 1
 
 
@@ -640,16 +687,15 @@ class Search:
 
 
 
+    # ------------------------------
+    # Github Files
 
-    def update_index_markdown(self, gh_oauth_token, config): 
+    def update_index_ghfiles(self, gh_token, config): 
         """
         Update the search index using a collection of 
-        Markdown files from a Github repo.
-
-        gh_oauth_token can also be an access token.
+        files (and, separately, Markdown files) from 
+        a Github repo.
         """
-        EXT = '.md'
-
         # Updated algorithm:
         # - get set of indexed ids
         # - get set of remote ids
@@ -660,6 +706,12 @@ class Search:
         # ------
         indexed_ids = set()
         p = QueryParser("kind", schema=self.ix.schema)
+        q = p.parse("ghfiles")
+        with self.ix.searcher() as s:
+            results = s.search(q,limit=None)
+            for result in results:
+                indexed_ids.add(result['id'])
+
         q = p.parse("markdown")
         with self.ix.searcher() as s:
             results = s.search(q,limit=None)
@@ -669,62 +721,66 @@ class Search:
         # Get the set of remote ids:
         # ------
         # Start with api object
-        g = Github(gh_oauth_token)
+        g = Github(gh_token)
 
-        # Now index all markdown files
-        # in the user-specified repos
-
-        # Iterate over each repo 
-        list_of_repos = config['repositories']
-        for r in list_of_repos:
+        # Now index all the files.
 
         # Start by collecting all the things
         remote_ids = set()
         full_items = {}
 
+        # Iterate over each repo 
+        list_of_repos = config['repositories']
+        for r in list_of_repos:
+
             if '/' not in r:
                 err = "Error: specify org/reponame or user/reponame in list of repos"
                 raise Exception(err)
 
             this_org, this_repo = re.split('/',r)
+            try:
                 org = g.get_organization(this_org)
                 repo = org.get_repo(this_repo)
+            except:
+                print("Error: could not gain access to repository %s"%(r))
+                continue
 
-            # ---------
-            # begin markdown-specific code
 
             # Get head commit
             commits = repo.get_commits()
+            try:
                 last = commits[0]
                 sha = last.sha
+            except GithubException:
+                print("Error: could not get commits from repository %s"%(r))
+                continue
 
             # Get all the docs
             tree = repo.get_git_tree(sha=sha, recursive=True)
             docs = tree.raw_data['tree']
+            print("Parsing file ids from repository %s"%(r))
 
             for d in docs:
 
                 # For each doc, get the file extension
-                # If it matches EXT, download the file
+                # and decide what to do with it.
+
                 fpath = d['path']
                 _, fname = os.path.split(fpath)
                 _, fext = os.path.splitext(fpath)
 
-                if fext==EXT:
-
                 key = d['sha']
+
                 d['org'] = this_org
                 d['repo'] = this_repo
                 value = d
 
-                    # Stash the doc for later
                 remote_ids.add(key)
                 full_items[key] = value
 
         writer = self.ix.writer()
         count = 0
 
-
         # Drop any id in indexed_ids
         # not in remote_ids
         drop_ids = indexed_ids - remote_ids
@@ -736,10 +792,10 @@ class Search:
         # and in remote_ids
         update_ids = indexed_ids & remote_ids
         for update_id in update_ids:
-            # cop out
+            # cop out: just delete and re-add
             writer.delete_by_term('id',update_id)
             item = full_items[update_id]
-            self.add_markdown(writer, item, config, update=True)
+            self.add_ghfile(writer, item, gh_token, config, update=True)
             count += 1
 
 
@@ -748,15 +804,42 @@ class Search:
         add_ids = remote_ids - indexed_ids
         for add_id in add_ids:
             item = full_items[add_id]
-            self.add_markdown(writer, item, config, update=False)
+            self.add_ghfile(writer, item, gh_token, config, update=False)
             count += 1
 
 
         writer.commit()
-        print("Done, updated %d markdown documents in the index" % count)
+        print("Done, updated %d Github files in the index" % count)
 
 
 
+    # ------------------------------
+    # Groups.io Emails
+
+
+    def update_index_groupsioemails(self, groupsio_token, config):
+        """
+        Update the search index using the email archives
+        of groups.io groups.
+
+        This requires the use of a spider.
+        RELEASE THE SPIDER!!!
+        """
+        spider = GroupsIOArchivesCrawler(groupsio_token,'dcppc')
+
+        # - ask spider to crawl the archives
+        spider.crawl_group_archives()
+
+        # - ask spider for list of all email records
+        #   - 1 email = 1 dictionary
+        #   - email records compiled by the spider
+        archives = spider.get_archives()
+
+        # - email object is sent off to add email method
+
+        print("Finished indexing groups.io emails")
+
+
     # ---------------------------------
     # Search results bundler
 
@@ -864,31 +947,27 @@ class Search:
     def get_document_total_count(self):
         p = QueryParser("kind", schema=self.ix.schema)
 
-        kind_labels = {
-                "documents" : "gdoc",
-                "markdown" :  "markdown",
-                "issues" :    "issue",
-        }
         counts = {
-                "documents" : None,
+                "gdoc" : None,
+                "issue" : None,
+                "ghfile" : None,
                 "markdown" : None,
-                "issues" : None,
                 "total" : None
         }
-        for key in kind_labels:
-            kind = kind_labels[key]
-            q = p.parse(kind)
+        for key in counts.keys():
+            q = p.parse(key)
             with self.ix.searcher() as s:
                 results = s.search(q,limit=None)
                 counts[key] = len(results)
 
-        ## These two should NOT be different, but they are...
-        #counts['total'] = self.ix.searcher().doc_count_all()
-        counts['total'] = counts['documents'] + counts['markdown'] + counts['issues']
+        counts['total'] = sum(counts[k] for k in counts.keys())
 
         return counts
 
 if __name__ == "__main__":
+
+    raise Exception("Error: main method not implemented (fix groupsio credentials first)")
+
     search = Search("search_index")
 
     from get_centillion_config import get_centillion_config

config_centillion.json

@@ -1,6 +1,27 @@
 {
     "repositories" : [
+        "dcppc/project-management",
+        "dcppc/nih-demo-meetings",
+        "dcppc/internal",
+        "dcppc/organize",
+        "dcppc/dcppc-bot",
+        "dcppc/full-stacks",
+        "dcppc/design-guidelines-discuss",
+        "dcppc/dcppc-deliverables",
+        "dcppc/dcppc-milestones",
+        "dcppc/crosscut-metadata",
+        "dcppc/lucky-penny",
+        "dcppc/dcppc-workshops",
+        "dcppc/metadata-matrix",
+        "dcppc/data-stewards",
+        "dcppc/dcppc-phase1-demos",
+        "dcppc/apis",
         "dcppc/2018-june-workshop",
-        "dcppc/2018-july-workshop"
+        "dcppc/2018-july-workshop",
+        "dcppc/2018-august-workshop",
+        "dcppc/2018-september-workshop",
+        "dcppc/design-guidelines",
+        "dcppc/2018-may-workshop",
+        "dcppc/centillion"
     ]
 }

config_flask.example.py

@@ -2,8 +2,9 @@
 INDEX_DIR = "search_index"
 
 # oauth client deets
-GITHUB_OAUTH_CLIENT_ID = "63f8d49c651840cbe31e"
-GITHUB_OAUTH_CLIENT_SECRET = "36d9a4611f7427336d3c89ed041c45d086b793ee"
+GITHUB_OAUTH_CLIENT_ID = "XXX"
+GITHUB_OAUTH_CLIENT_SECRET = "YYY"
+GITHUB_TOKEN = "ZZZ"
 
 # More information footer: Repository label
 FOOTER_REPO_ORG = "charlesreid1"
@@ -12,8 +13,8 @@ FOOTER_REPO_NAME = "centillion"
 # Toggle to show Whoosh parsed query
 SHOW_PARSED_QUERY=True
 
-TAGLINE = "Search all the things"
+TAGLINE = "Search All The Things"
 
 # Flask settings
 DEBUG = True
-SECRET_KEY = '42c5a8eda356ca9d9c3ab2d149541e6b91d843fa'
+SECRET_KEY = 'WWWWW'

gdrive_util.py

@@ -29,8 +29,7 @@ class GDrive(object):
     ):
         """
         Set up the Google Drive API instance.
-        Factory method: create it and hand it over.
-        Then we're finished.
+        Factory method: create it here, hand it over in get_service().
         """
         self.credentials_file = credentials_file
         self.client_secret_file = client_secret_file
@@ -40,6 +39,9 @@ class GDrive(object):
         self.store = file.Storage(credentials_file)
 
     def get_service(self):
+        """
+        Return an instance of the Google Drive API service.
+        """
 
         creds = self.store.get()
         if not creds or creds.invalid:

groupsio_util.py

@@ -0,0 +1,382 @@
+import requests, os, re
+from bs4 import BeautifulSoup
+
+class GroupsIOArchivesCrawler(object):
+    """
+    This is a Groups.io spider
+    designed to crawl the email
+    archives of a group.
+
+    credentials (dictionary):
+        groupsio_token :     api access token
+        groupsio_username     :     username
+        groupsio_password     :     password
+    """
+    def __init__(self,
+                 credentials,
+                 group_name):
+        # template url for archives page (list of topics)
+        self.url = "https://{group}.groups.io/g/{subgroup}/topics"
+        self.login_url = "https://groups.io/login"
+
+        self.credentials = credentials
+        self.group_name = group_name
+        self.crawled_archives = False
+        self.archives = None
+
+
+    def get_archives(self):
+        """
+        Return a list of dictionaries containing 
+        information about each email topic in the 
+        groups.io email archive.
+
+        Call crawl_group_archives() first!
+        """
+        return self.archives
+
+
+    def get_subgroups_list(self):
+        """
+        Use the API to get a list of subgroups.
+        """
+        subgroups_url = 'https://api.groups.io/v1/getsubgroups'
+
+        key = self.credentials['groupsio_token']
+
+        data = [('group_name', self.group_name),
+                ('limit',100)
+        ]
+        response = requests.post(subgroups_url,
+                                 data=data,
+                                 auth=(key,''))
+        response = response.json()
+        data = response['data']
+
+        subgroups = {}
+        for group in data:
+            k = group['id']
+            v = re.sub(r'dcppc\+','',group['name'])
+            subgroups[k] = v
+
+        return subgroups
+
+
+    def crawl_group_archives(self):
+        """
+        Spider will crawl the email archives of the entire group
+        by crawling the email archives of each subgroup.
+        """
+        subgroups = self.get_subgroups_list()
+
+        # ------------------------------
+        # Start by logging in.
+
+        # Create session object to persist session data
+        session = requests.Session()
+
+        # Log in to the website
+        data = dict(email = self.credentials['groupsio_username'],
+                    password = self.credentials['groupsio_password'],
+                    timezone = 'America/Los_Angeles')
+
+        r = session.post(self.login_url,
+                         data = data)
+
+        csrf = self.get_csrf(r)
+
+        # ------------------------------
+        # For each subgroup, crawl the archives
+        # and return a list of dictionaries
+        # containing all the email threads.
+        for subgroup_id in subgroups.keys():
+            self.crawl_subgroup_archives(session,
+                                         csrf,
+                                         subgroup_id, 
+                                         subgroups[subgroup_id])
+
+        # Done. archives are now tucked away
+        # in the variable self.archives
+        # 
+        # self.archives is a list of dictionaries,
+        # with each dictionary containing info about
+        # a topic/email thread in a subgroup.
+        # ------------------------------
+
+
+
+
+    def crawl_subgroup_archives(self, session, csrf, subgroup_id, subgroup_name):
+        """
+        This kicks off the process to crawl the entire
+        archives of a given subgroup on groups.io.
+
+        For a given subgroup the url is self.url,
+        
+            https://{group}.groups.io/g/{subgroup}/topics
+
+        This is the first of a paginated list of topics.
+        Procedure is:
+        - passed a starting page (or its contents)
+        - iterate through all topics via the HTML page elements
+        - assemble a bundle of information about each topic:
+            - topic title, by, URL, date, content, permalink
+            - content filtering:
+                - ^From, Reply-To, Date, To, Subject
+                - Lines containing phone numbers
+                    - 9 digits
+                    - XXX-XXX-XXXX, (XXX) XXX-XXXX
+                    - XXXXXXXXXX, XXX XXX XXXX
+                    - ^Work: or (Work) or Work$
+                    - Home, Cell, Mobile
+                    - +1 XXX 
+                    - \w@\w
+        - while next button is not greyed out,
+        - click the next button
+
+        everything stored in self.archives:
+        list of dictionaries.
+
+        """
+        self.archives = []
+
+        prefix = "https://{group}.groups.io".format(group=self.group_name)
+
+        url = self.url.format(group=self.group_name, 
+                              subgroup=subgroup_name)
+
+        # ------------------------------
+
+        # Now get the first page
+        r = session.get(url)
+
+        # ------------------------------
+        # Fencepost algorithm:
+
+        # First page:
+
+        # Extract a list of (title, link) items
+        items = self.extract_archive_page_items_(r)
+
+        # Get the next link
+        next_url = self.get_next_url_(r)
+
+        # Now add each item to the archive of threads,
+        # then find the next button.
+        self.add_items_to_archives_(session,subgroup_name,items)
+
+        if next_url is None:
+            return
+        else:
+            full_next_url = prefix + next_url
+
+        # Now click the next button
+        next_request = requests.get(full_next_url)
+
+        while next_request.status_code==200:
+            items = self.extract_archive_page_items_(next_request)
+            next_url = self.get_next_url_(next_request)
+            self.add_items_to_archives_(session,subgroup_name,items)
+            if next_url is None:
+                return
+            else:
+                full_next_url = prefix + next_url
+            next_request = requests.get(full_next_url)
+        
+
+
+    def add_items_to_archives_(self,session,subgroup_name,items):
+        """
+        Given a set of items from a list of threads,
+        items being title and link,
+        get the page and store all info
+        in self.archives variable
+        (list of dictionaries)
+        """
+        for (title, link) in items:
+            # Get the thread page:
+            prefix = "https://{group}.groups.io".format(group=self.group_name)
+            full_link = prefix + link
+            r = session.get(full_link)
+            soup = BeautifulSoup(r.text,'html.parser')
+
+            # soup contains the entire thread
+
+            # What are we extracting:
+            # 1. thread number
+            # 2. permalink
+            # 3. content/text (filtered)
+
+            # - - - - - - - - - - - - - - 
+            # 1. topic/thread number:
+            # <a rel="nofollow" href="">
+            # where link is:
+            # https://{group}.groups.io/g/{subgroup}/topic/{topic_id}
+            # example topic id: 24209140
+            #
+            # ugly links are in the form 
+            # https://dcppc.groups.io/g/{subgroup}/topic/some_text_here/{thread_id}?p=,,,,,1,2,3,,,4,,5
+            # split at ?, 0th portion
+            # then split at /, last (-1th) portion
+            topic_id = link.split('?')[0].split('/')[-1]
+
+            # - - - - - - - - - - - - - - - 
+            # 2. permalink:
+            # - current link is ugly link
+            # - permalink is the nice one
+            # - topic id is available from the ugly link
+            # https://{group}.groups.io/g/{subgroup}/topic/{topic_id}
+
+            permalink_template = "https://{group}.groups.io/g/{subgroup}/topic/{topic_id}"
+            permalink = permalink_template.format(
+                    group = self.group_name,
+                    subgroup = subgroup_name, 
+                    topic_id = topic_id
+            )
+
+            # - - - - - - - - - - - - - - - 
+            # 3. content:
+
+            # Need to rearrange how we're assembling threads here.
+            # This is one thread, no?
+            content = []
+
+            subject = soup.find('title').text
+
+            # Extract information for the schema:
+            # - permalink for thread (done)
+            # - subject/title (done)
+            # - original sender email/name (done)
+            # - content (done)
+
+            # Groups.io pages have zero CSS classes, which makes everything
+            # a giant pain in the neck to interact with. Thanks Groups.io!
+            original_sender = ''
+            for i, tr in enumerate(soup.find_all('tr',{'class':'test'})):
+                # Every other tr row contains an email.
+                if (i+1)%2==0:
+                    # nope, no email here
+                    pass
+                else:
+                    # found an email!
+                    # this is a maze, thanks groups.io
+                    td = tr.find('td')
+                    divrow = td.find('div',{'class':'row'}).find('div',{'class':'pull-left'})
+                    if (i+1)==1:
+                        original_sender = divrow.text.strip()
+                    for div in td.find_all('div'):
+                        if div.has_attr('id'):
+
+                            # purge any signatures
+                            for x in div.find_all('div',{'id':'Signature'}):
+                                x.extract()
+
+                            # purge any headers
+                            for x in div.find_all('div'): 
+                                nonos = ['From:','Sent:','To:','Cc:','CC:','Subject:']
+                                for nono in nonos:
+                                    if nono in x.text:
+                                        x.extract()
+
+                            message_text = div.get_text()
+
+                            # More filtering:
+
+                            # phone numbers
+                            message_text = re.sub(r'[0-9]{3}-[0-9]{3}-[0-9]{4}','XXX-XXX-XXXX',message_text)
+                            message_text = re.sub(r'[0-9]\{10\}','XXXXXXXXXX',message_text)
+
+                            content.append(message_text)
+
+            full_content = "\n".join(content)
+
+            thread = {
+                    'permalink' : permalink,
+                    'subject' : subject,
+                    'original_sender' : original_sender,
+                    'content' : full_content
+            }
+            
+            print('*'*40)
+            for k in thread.keys():
+                if k=='content':
+                    pass
+                else:
+                    print("%s : %s"%(k,thread[k]))
+            print('*'*40)
+            self.archives.append(thread)
+
+
+    def extract_archive_page_items_(self, response):
+        """
+        (Private method)
+
+        Given a response from a GET request,
+        use beautifulsoup to extract all items
+        (thread titles and ugly thread links)
+        and pass them back in a list.
+        """
+        soup = BeautifulSoup(response.content,"html.parser")
+        rows = soup.find_all('tr',{'class':'test'})
+        if 'rate limited' in soup.text:
+            raise Exception("Error: rate limit in place for Groups.io")
+
+        results = []
+        for row in rows:
+            # We don't care about anything except title and ugly link
+            subject = row.find('span',{'class':'subject'})
+            title = subject.get_text()
+            link = row.find('a')['href']
+            print(title)
+            results.append((title,link))
+
+        return results
+
+
+    def get_next_url_(self, response):
+        """
+        (Private method)
+
+        Given a response (which is a list of threads),
+        find the next button and return the URL.
+
+        If no next URL, if is disabled, then return None.
+        """
+        soup = BeautifulSoup(response.text,'html.parser')
+        chevron = soup.find('i',{'class':'fa-chevron-right'})
+        try:
+            if '#' in chevron.parent['href']:
+                # empty link, abort
+                return None
+        except AttributeError:
+            # I don't even now
+            return None
+
+        if chevron.parent.parent.has_attr('class') and 'disabled' in chevron.parent.parent['class']:
+            # no next link, abort
+            return None
+
+        return chevron.parent['href']
+
+
+
+    def get_csrf(self,resp):
+        """
+        Find the CSRF token embedded in the subgroup page
+        """
+        soup = BeautifulSoup(resp.text,'html.parser')
+        csrf = ''
+        for i in soup.find_all('input'):
+            # Note that i.name is different from i['name']
+            # the first is the actual tag,
+            # the second is the attribute name="xyz"
+            if i['name']=='csrf':
+                csrf = i['value']
+        
+        if csrf=='':
+            err = "ERROR: Could not find csrf token on page."
+            raise Exception(err)
+    
+        return csrf
+
+

requirements.txt

@@ -10,3 +10,4 @@ pypandoc>=1.4
 requests>=2.19
 pandoc>=1.0
 flask-dance>=1.0.0
+beautifulsoup4>=4.6

templates/layout.html

@@ -1,5 +1,5 @@
 <!doctype html>
-<title>Markdown Search</title>
+<title>Centillion Search Engine</title>
 <link rel="stylesheet" type="text/css" href="{{ url_for('static', filename='style.css') }}">
 <link rel="stylesheet" type="text/css" href="{{ url_for('static', filename='github-markdown.css') }}">
 <link rel="stylesheet" type="text/css" href="{{ url_for('static', filename='bootstrap.min.css') }}">

templates/search.html

@@ -86,9 +86,11 @@
                     <div class="container-fluid">
                         <div class="row">
                             <div class="col-xs-12 info">
-                                <b>Indexing:</b> <span class="badge">{{totals["documents"]}}</span> Google Documents,
-                                <span class="badge">{{totals["issues"]}}</span> Github issues,
-                                <span class="badge">{{totals["markdown"]}}</span> markdown files.
+                                <b>Indexing:</b> <span
+                                    class="badge">{{totals["gdoc"]}}</span> Google Documents,
+                                <span class="badge">{{totals["issue"]}}</span> Github issues,
+                                <span class="badge">{{totals["ghfile"]}}</span> Github files,
+                                <span class="badge">{{totals["markdown"]}}</span> Github markdown files.
                             </div>
                         </div>
                 </div>
@@ -107,12 +109,19 @@
 
                     <div class="url">
                         {% if e.kind=="gdoc" %}
-                            <b>Google Drive File:</b>
+                            {% if e.mimetype=="" %}
+                                <b>Google Document:</b>
+                                <a href='{{e.url}}'>{{e.title}}</a>
+                                (Owner: {{e.owner_name}}, {{e.owner_email}})<br />
+                                <b>Document Type</b>: {{e.mimetype}}
+                            {% else %}
+                                <b>Google Drive:</b>
                                 <a href='{{e.url}}'>{{e.title}}</a>
                                 (Owner: {{e.owner_name}}, {{e.owner_email}})
+                            {% endif %}
 
                         {% elif e.kind=="issue" %}
-                            <b>Issue:</b>
+                            <b>Github Issue:</b>
                             <a href='{{e.url}}'>{{e.title}}</a>
                             {% if e.github_user %}
                             opened by <a href='https://github.com/{{e.github_user}}'>@{{e.github_user}}</a>
@@ -121,7 +130,7 @@
                             <b>Repository:</b> <a href='{{e.repo_url}}'>{{e.repo_name}}</a>
 
                         {% elif e.kind=="markdown" %}
-                            <b>Markdown:</b>
+                            <b>Github Markdown:</b>
                             <a href='{{e.url}}'>{{e.title}}</a>
                             <br/>
                             <b>Repository:</b> <a href='{{e.repo_url}}'>{{e.repo_name}}</a>
@@ -131,9 +140,15 @@
 
                         {% endif %}
                         <br />
-                        score: {{'%d'  % e.score}}
+                        Score: {{'%d'  % e.score}}
+                    </div>
+                    <div class="markdown-body">
+                        {% if e.content_highlight %}
+                            {{ e.content_highlight|safe}}
+                        {% else %}
+                        <p>(A preview of this document is not available.)</p>
+                        {% endif %}
                     </div>
-                    <div class="markdown-body">{{ e.content_highlight|safe}}</div>
 
                 </li>
             {% endfor %}