plugin_localcomments.py

# adapted from the plugin_comments provided with web2py
import re
from gluon.tools import prettydate
from gluon.contrib.markdown.markdown2 import markdown
import requests
import os.path
import urllib
from datetime import datetime
import json
from pprint import pprint


def error(): raise HTTP(404)

# builds an organized list (UL) with threaded comments
def SUL(*a,**b): return UL(*[u for u in a if u],**b)

# client-side script to manage comment display and form behavior
script=SCRIPT("""
var action = null;
var formhtml = null;
function delete_all_forms() { 
    jQuery('div.plugin_localcomments div.reply').each(function() {
        if ($(this).closest('.issue').length === 0) {
            // this is the space for new topics, with a separate link
            $(this).html('<a class="btn btn-small reply" href="#">Add a new topic</a>');
        } else {
            // this is the prompt to add comments to an existing topic
            $(this).html('<a class="btn btn-small reply" href="#">Add a comment</a>');
        }
        $(this).find('a.reply').unbind('click').click(function() {
            $formHolder = $(this).parent();
            delete_all_forms();
            $formHolder.html(formhtml); 
            capture_form(); 
            return false;
        });
    });
}
function capture_form() {
    jQuery('div.plugin_localcomments a.msg-close').unbind('click').click(function(){
        delete_all_forms();
        return false;
    });

    // adjust UI for issues versus comments
    var threadParentID = 0;
    var $parentIssueContainer = $('.plugin_localcomments form').closest('li.issue');
    if ($parentIssueContainer.length > 0) {
        threadParentID = $parentIssueContainer.find('div:eq(0)').attr('id').split('r')[1];
    }
    console.log('threadParentID:'+ threadParentID);
    var isThreadStarter = threadParentID == 0;
    if (isThreadStarter) {
        //jQuery('div.plugin_localcomments select[name=feedback_type] option:eq(1)').text( 'General comment' );
        jQuery('div.plugin_localcomments select[name=feedback_type]').show();
        jQuery('div.plugin_localcomments select[name=intended_scope]').show();
        jQuery('div.plugin_localcomments select[name=issue_title]').show();
    } else {
        //jQuery('div.plugin_localcomments select[name=feedback_type] option:eq(1)').text( 'Reply or general comment' );
        jQuery('div.plugin_localcomments select[name=feedback_type]').hide();
        jQuery('div.plugin_localcomments select[name=intended_scope]').hide();
        jQuery('div.plugin_localcomments select[name=issue_title]').hide();
    }
    // always hide expertise checkbox and surrounding label (not currently needed)
    jQuery('div.plugin_localcomments label.expertise-option').hide();

    jQuery('div.plugin_localcomments :submit').unbind('click').click(function(){
        var $form = jQuery(this).closest('form');

        // validate form fields
        var $visitorNameField = $form.find('input[name="visitor_name"]'); 
        if ($visitorNameField.is(':visible') && ($.trim($visitorNameField.val()) === '')) {
            alert("Please enter your name (and preferably an email address) so we can stay in touch.");
            return false;
        }
        var $fbTypeField = $form.find('select[name="feedback_type"]'); 
        if ($fbTypeField.is(':visible') && ($.trim($fbTypeField.val()) === '')) {
            alert("Please choose a feedback type for this topic.");
            return false;
        }
        var $titleField = $form.find('input[name="issue_title"]'); 
        if ($titleField.is(':visible') && ($.trim($titleField.val()) === '')) {
            alert("Please give this topic a title.");
            return false;
        }
        var $bodyField = $form.find('textarea[name="body"]'); 
        if ($.trim($bodyField.val()) === '') {
            alert("Please enter some text for this "+ (isThreadStarter ? 'issue' : 'comment') +".");
            return false;
        }

        jQuery.post(action,
            {
               ////$'thread_parent_id': form.find('input[name="thread_parent_id"]').val(),
               'issue_or_comment': (isThreadStarter ? 'issue' : 'comment'),
               'thread_parent_id': threadParentID,    ///$form.parent().prev().attr('id').split('r')[1], 
               'synthtree_id': $form.find('input[name="synthtree_id"]').val(),
               'synthtree_node_id': $form.find('input[name="synthtree_node_id"]').val(),
               'sourcetree_id': $form.find('input[name="sourcetree_id"]').val(),
               'ottol_id': $form.find('input[name="ottol_id"]').val(),
               'target_node_label': $form.find('input[name="target_node_label"]').val(),
               'url': $form.find('input[name="url"]').val(),
               'title': $form.find('input[name="issue_title"]').val(),
               'body': $form.find('textarea[name="body"]').val(),
               'feedback_type': $form.find('select[name="feedback_type"]').val(),
               'intended_scope': $form.find('select[name="intended_scope"]').val(),
               'claimed_expertise': $form.find(':checkbox[name="claimed_expertise"]').is(':checked'),
               'visitor_name': $form.find('input[name="visitor_name"]').val(),
               'visitor_email': $form.find('input[name="visitor_email"]').val()
            },
            function(data,r){ 
               if(data) { 
                   var $refreshArea;
                   if (isThreadStarter) {
                       $refreshArea = $form.parent().nextAll('ul');
                       // add the new comment (LI) to the top of the list
                       $refreshArea.prepend(data);
                   } else {
                       $refreshArea = $form.parent().prevAll('ul');
                       // add the new comment (LI) to the end of the list
                       $refreshArea.append(data);
                   } 
                   $form.find('textarea[name="body"]').val('');
                   //$form.find('input[name="thread_parent_id"]').val('0');
                   plugin_localcomments_init(); 
                   delete_all_forms();
               }
            },
            'html'
        );
        return false;
    });
}

function plugin_localcomments_init() {
  jQuery('div.plugin_localcomments .toggle').unbind('click').click(function(){
     var $toggle = $(this);
     var $parentIssue = $toggle.closest('li.issue');
     var $collapsibleTargets = $parentIssue.find('ul').eq(0).add($parentIssue.find('div.reply'));
     $collapsibleTargets.slideToggle(250);  // duration in ms
     if ($toggle.text().indexOf('Show') == -1) {
        $toggle.text('Show/add comments');
     } else {
        $toggle.text('Hide comments');
     }
     return false;
  });
  jQuery('div.plugin_localcomments .delete').unbind('click').click(function(){
    delete_all_forms();
    var $commentDiv = jQuery(this).closest('.msg-wrapper');
    var $msgItem = $commentDiv.closest('li');
    var issueOrComment = ($msgItem.is('.issue') ? 'issue' : 'comment');
    jQuery.post(
        action,    // WAS: action+'/delete',
        {
            'thread_parent_id': 'delete',
            'comment_id': $commentDiv.attr('id').split('r')[1],
            'issue_or_comment': issueOrComment
        },
        function(data,r){
            $msgItem.fadeOut(function() {$(this).remove();});
        }
    );
    return false;
  });
}
jQuery(document).ready(function() {
  action = jQuery('div.plugin_localcomments form').attr('action');  
  formhtml = jQuery('div.plugin_localcomments form').parent().html();
  delete_all_forms();  // important! creates .reply buttons before init() below
  plugin_localcomments_init();
});
""")

def moderation():
    comments = db().select(db.plugin_localcomments_comment.ALL, orderby=~db.plugin_localcomments_comment.created_on)  # =~ is DESCENDING ORDER
    form = SQLFORM.factory(Field('tag_name'))
    return dict(comments=comments, form=form)

def sqlform():
    # comments = db().select(db.plugin_localcomments_comment.ALL, orderby=~db.plugin_localcomments_comment.created_on)  # =~ is DESCENDING ORDER
    form = SQLFORM(db.plugin_localcomments_comment)
    return dict(form=form)
    
def show_type_icon(type):
    iconClass = "icon-comment"
    if type == 'Error in phylogeny':
        iconClass = "icon-move"
    elif type == 'Bug report':
        iconClass = "icon-warning-sign"
    elif type == 'Feature request':
        iconClass = "icon-wrench"
    elif type == 'Reply or general':
        iconClass = "icon-comment"
    return XML(I(_class=iconClass))

@auth.requires_membership(role='editor')
def grid():
    db.plugin_localcomments_comment.intended_scope.readable = True
    db.plugin_localcomments_comment.intended_scope.represent = lambda scope, row: scope and scope.capitalize() or XML(T('&mdash;'))
    db.plugin_localcomments_comment.feedback_type.represent = lambda row, value: show_type_icon(value)

    grid = SQLFORM.grid( db.plugin_localcomments_comment,

        # formstyle controls only the Add/Edit forms for individual records! not the main grid :(
        formstyle='bootstrap',
        #formstyle='table3cols',
        #formstyle='table2cols',
        #formstyle='inline',
        #formstyle='divs',
        #formstyle='ul',

        user_signature = False,  # False means *anyone* can edit, delete, etc! if they can use the method (see @auth.requires above)
        # editable=auth.has_membership(role='editor'),  # use this instead of @auth above, to allow others to search all comments

        create=False,
        deletable=False,  # we'll flip the hidden flag, but not truly delete..?
        orderby=~db.plugin_localcomments_comment.created_on,

        fields=[ 
            #db.plugin_localcomments_comment.id, 
            db.plugin_localcomments_comment.feedback_type, 
            db.plugin_localcomments_comment.body, 
            db.plugin_localcomments_comment.url, 
            db.plugin_localcomments_comment.ottol_id, 
            db.plugin_localcomments_comment.synthtree_id, 
            db.plugin_localcomments_comment.synthtree_node_id,
            db.plugin_localcomments_comment.created_on,
            db.plugin_localcomments_comment.intended_scope,
        ],
        headers = { 
            # NOTE the funky key format used here
            'plugin_localcomments_comment.feedback_type' : 'Type',
        }
        # TODO: add "virtual field" to show compound locations (treeID@nodeID), and hide underlying fields?

        # TODO: add custom rendering (via lambdas) for some fields, eg, icons for feedback_type:
        #   https://groups.google.com/forum/#!searchin/web2py/grid$20HTML/web2py/3KhSI4Ps5Tw/Ay4Nc0ti3g0J
        #   https://groups.google.com/forum/#!searchin/web2py/grid$20HTML/web2py/4-rgcM9FNcA/NFpIyZdj4OkJ
        #   http://web2py.com/books/default/chapter/29/06?search=represent#Record-representation

    )
    return locals()
    
def smartgrid():
    # comments = db().select(db.plugin_localcomments_comment.ALL, orderby=~db.plugin_localcomments_comment.created_on)  # =~ is DESCENDING ORDER
    grid = SQLFORM.smartgrid(db.plugin_localcomments_comment)
    return locals()
    

def index():
    # this is a tricky function that does simple display, handles POSTed comments, moderation, etc.

    # TODO: break this up into more sensible functions, and refactor
    # display/markup generation to shared code?

    synthtree_id = request.vars['synthtree_id']
    synthtree_node_id = request.vars['synthtree_node_id']
    sourcetree_id = request.vars['sourcetree_id']
    ottol_id = request.vars['ottol_id']
    target_node_label = request.vars['target_node_label']
    url = request.vars['url'] or request.get('env').get('http_referer')

    filter = request.vars['filter']

    # if anonymous user submitted identifying information, remember it
    visitor_name = request.vars['visitor_name'] 
    if visitor_name:
        session['visitor_name'] = visitor_name
    visitor_email = request.vars['visitor_email']
    if visitor_email:
        session['visitor_email'] = visitor_email

    issue_or_comment = request.vars['issue_or_comment']
    thread_parent_id = request.vars['thread_parent_id'] # can be None
    comment_id = request.vars['comment_id'] # used for some operations (eg, delete)
    feedback_type = request.vars['feedback_type'] # used for new comments
    intended_scope = request.vars['intended_scope'] # used for new comments
    issue_title = request.vars['title'] # used for new issues (threads)
    claims_expertise = request.vars['claimed_expertise'] # used for new comments
    threads = [ ]
    def node(comment):
        ##print("building node for comment id={0}...".format(comment.get('number', comment['id'])))
        # preload its comments (a separate API call)
        child_comments = [ ]
        if comment.get('comments') and comment.get('comments') > 0:
            get_children_url = comment['comments_url']
            resp = requests.get( get_children_url, headers=GH_GET_HEADERS)
            resp.raise_for_status()
            try:
                child_comments = resp.json()
            except:
                child_comments = resp.json
            print("found {0} child comments".format(len(child_comments)))

        metadata = parse_comment_metadata(comment['body'])
        ##print(metadata)

        # Examine the comment metadata (if any) to get the best display name
        # and URL for its author. Guests should appear here as the name and
        # email address they entered when creating a comment, rather than the
        # 'opentreeapi' bot user.
        #
        # Default values are what we can fetch from the issues API
        author_display_name = comment['user']['login']
        author_link = comment['user']['html_url']
        # Now let's try for something more friendly...
        if metadata:
            meta_author_info = metadata.get('Author', None)
            if meta_author_info:
                # Try to parse this fron a Markdown hyperlink. Typical values include:
                #   u'opentreeapi'
                #   u'11'
                #   u'[Jim Allman](https://github.com/jimallman)'
                #   u'[John Smith](mailto:example.guest@gmail.com)'
                regex = re.compile(r'\[(.*)\]\((.*)\)')
                markdown_fields = regex.findall(meta_author_info)
                if len(markdown_fields) > 0:
                    # look for parts of a markdown link
                    author_display_name, author_link = markdown_fields[0]
                else:
                    # it's not a markdown link, just a bare name or numeric userid
                    if meta_author_info.isdigit():
                        # ignore ugly userid (login is better)
                        pass
                    else:
                        author_display_name = meta_author_info

        # Is this node for an issue (thread starter) or a comment (reply)?
        issue_node = 'number' in comment

        # Is the current user logged in? If so, what is their GitHub ID (login)?
        current_user_id = auth.user and auth.user.github_login or None

        # Cook up some reasonably strong regular expressions to detect bare
        # URLs and wrap them in hyperlinks. Adapted from 
        # http://stackoverflow.com/questions/1071191/detect-urls-in-a-string-and-wrap-with-a-href-tag
        link_regex = re.compile(  r'''
                             (?x)( # verbose identify URLs within text
                      (http|https) # make sure we find a resource type
                               :// # ...needs to be followed by colon-slash-slash
                    (\w+[:.]?){2,} # at least two domain groups, e.g. (gnosis.)(cx)
                              (/?| # could be just the domain name (maybe w/ slash)
                        [^ \n\r"]+ # or stuff then space, newline, tab, quote
                            [\w/]) # resource name ends in alphanumeric or slash
             (?=([\s\.,>)'"\]]|$)) # assert: followed by white or clause ending OR end of line
                                 ) # end of match group
                                   ''')
        # link_replace = r'<a href="\1" />\1</a>'
        # let's try this do-nothing version
        link_replace = r'\1'
        # NOTE the funky constructor required to use this below

        try:   # TODO: if not comment.deleted:
            markup = LI(
                    DIV(##T('posted by %(first_name)s %(last_name)s',comment.created_by),
                    # not sure why this doesn't work... db.auth record is not a mapping!?
                    ('title' in comment) and DIV( comment['title'], A(T('on GitHub'), _href=comment['html_url'], _target='_blank'), _class='topic-title') or '',
                    DIV( XML(markdown(get_visible_comment_body(comment['body'] or ''), extras={'link-patterns':None}, link_patterns=[(link_regex, link_replace)]).encode('utf-8'), sanitize=False),_class=(issue_node and 'body issue-body' or 'body comment-body')),
                    DIV(
                        A(T(author_display_name), _href=author_link, _target='_blank'),
                        # SPAN(' [local expertise]',_class='badge') if comment.claimed_expertise else '',
                        SPAN(' ',metadata.get('Feedback type'),' ',_class='badge') if metadata.get('Feedback type') else '',
                        SPAN(' ',metadata.get('Intended scope'),' ',_class='badge') if metadata.get('Intended scope') else '',
                        T(' - %s',prettydate(utc_to_local(datetime.strptime(comment['created_at'], GH_DATETIME_FORMAT)),T)),
                        SPAN(
                            issue_node and A(T(child_comments and 'Hide comments' or 'Show/add comments'),_class='toggle',_href='#') or '',
                            issue_node and comment['user']['login'] == current_user_id and SPAN(' | ') or '',
                            A(T('Delete'),_class='delete',_href='#') if comment['user']['login'] == current_user_id else '',
                        _class='controls'),
                    _class='byline'),
                    _id='r%s' % comment.get('number', comment['id']),
                    _class='msg-wrapper'),
                # child messages (toggle hides/shows these)
                issue_node and SUL(*[node(comment) for comment in child_comments], _style=("" if child_comments else "display: none;")) or '',
                issue_node and DIV(_class='reply', _style=("" if child_comments else "display: none;")) or '',
                _class=(issue_node and 'issue' or 'comment'))
            return markup
        except:
            import sys
            print "Unexpected error:", sys.exc_info()[0]
            raise

    if thread_parent_id == 'delete':
        # delete the specified comment or close an issue...
        try:
            if issue_or_comment == 'issue':
                print("CLOSING ISSUE: ")
                print(comment_id)
                close_issue(comment_id)
                return 'closed'
            else:
                print("DELETING COMMENT: ")
                print(comment_id)
                delete_comment(comment_id)
                return 'deleted'
        except:
            return error()
    elif thread_parent_id:
        # add a new comment using the submitted vars
        if not request.vars.body:
            print('MISSING BODY:')
            print(request.vars.body)
            return error()

        if not (visitor_name or auth.user):
            print('MISSING USER-ID:')
            print('  visitor_name:')
            print(visitor_name)
            print('  auth.user:')
            print(auth.user)
            return error()

        # build useful links for some footer fields
        if auth.user:
            author_link = '[{0}]({1})'.format(auth.user.name, auth.user.github_url)
        elif visitor_name and visitor_email: 
            author_link = '[{0}](mailto:{1})'.format(visitor_name, visitor_email)
        elif visitor_name: 
            # no email provided
            author_link = visitor_name
        elif visitor_email:
            # no name provided
            author_link = '[{0}](mailto:{1})'.format(visitor_email, visitor_email)
        else:
            # no identifying information provided
            author_link = 'Anonymous'

        if (thread_parent_id == '0'):
            # create a new issue (thread starter)
            msg_body = request.vars.body
            if len(re.compile('\s+').sub('',msg_body))<1:
                return ''

            # more useful links for some footer fields
            if url.startswith('http'):
                # truncate visible link
                url_link = '[{0}]({1})'.format(url.split('//')[1], url)
            else:
                # expand hidden link
                url_link = '[{0}]({1}{2})'.format(url, request.get('env').get('http_origin'), url)

            # add full metadata for an issue 
            footer = build_comment_metadata_footer(metadata={
                "Author": author_link,
                "Upvotes": 0,
                "URL": url_link,
                "Target node label": target_node_label,
                "Synthetic tree id": synthtree_id,
                "Synthetic tree node id": synthtree_node_id,
                "Source tree id": sourcetree_id,
                "Open Tree Taxonomy id": ottol_id,
                "Intended scope": intended_scope
            })
            msg_data = {
                "title": issue_title,
                "body": "{0}\n{1}".format(msg_body, footer),
                "labels": [ ]
            }
            if feedback_type:
                # omit an empty value here!
                msg_data['labels'].append(feedback_type)
            new_msg = add_or_update_issue(msg_data)
        else:
            # attach this comment to an existing issue
            ##print("ADD A COMMENT")
            msg_body = request.vars.body
            if len(re.compile('\s+').sub('',msg_body))<1:
                return ''
            # add abbreviated metadata for a comment
            footer = build_comment_metadata_footer(metadata={
                "Author" : author_link,
                "Upvotes" : 0,
            })
            print(footer)
            msg_data = {
                "body": "{0}\n{1}".format(msg_body, footer)
            }
            new_msg = add_or_update_comment(msg_data, parent_issue_id=thread_parent_id)
        return node(new_msg)                

    # retrieve related comments, based on the chosen filter
    print("retrieving local comments using this filter:")
    print(filter)
    if filter == 'synthtree_id,synthtree_node_id':
        comments = get_local_comments({
            "Synthetic tree id": synthtree_id, 
            "Synthetic tree node id": synthtree_node_id})
    elif filter == 'sourcetree_id':
        comments = get_local_comments({"Source tree id(s)": sourcetree_id})
    elif filter == 'ottol_id':
        comments = get_local_comments({"Open Tree Taxonomy id": ottol_id})
    else:   # fall back to url
        comments = get_local_comments({"URL": url})

    #pprint(comments) 

    for comment in comments:
        #thread[comment.thread_parent_id] = thread.get(comment.thread_parent_id,[])+[comment]
        threads.append(comment)
    pprint('{0} threads loaded'.format(len(threads)))
    return DIV(script,
               DIV(FORM(# anonymous users should see be encouraged to login or add a name-or-email to their comments
                        '' if auth.user_id else A(T('Login'),_href=URL(r=request,c='default',f='user',args=['login']),_class='login-logout reply'),
                        '' if auth.user_id else T(' or '),
                        '' if auth.user_id else INPUT(_type='text',_id='visitor_name',_name='visitor_name',_value=session.get('visitor_name',''),_placeholder="Enter your name"),
                        '' if auth.user_id else T(' '),
                        '' if auth.user_id else INPUT(_type='text',_id='visitor_email',_name='visitor_email',_value=session.get('visitor_email',''),_placeholder="Your email (visible on GitHub)"),
                        '' if auth.user_id else BR(),
                        SELECT(
                            OPTION('What kind of feedback is this?', _value=''),
                            OPTION('General comment', _value='General comment'),
                            OPTION('Reporting an error in phylogeny', _value='Error in phylogeny'),
                            OPTION('Bug report (website behavior)', _value='Bug report'),
                            OPTION('New feature request', _value='Feature request'),
                        _name='feedback_type',value='',_style='width: auto;'),
                        T(' '),
                        SELECT(
                            OPTION('about node placement in the synthetic tree', _value='Re: synthetic tree'),
                            OPTION('about node placement in the source tree', _value='Re: source tree'),
                            OPTION('about taxon data in OTT', _value='Re: OTT taxon'),
                            OPTION('general feedback (none of the above)', _value=''),
                        _name='intended_scope',value='',_style='width: auto;'),
                        LABEL(INPUT(_type='checkbox',_name=T('claimed_expertise')), T(' I claim expertise in this area'),_style='float: right;',_class='expertise-option'),
                        INPUT(_type='text',_id='issue_title',_name='issue_title',_value='',_placeholder="Give this topic a title"),   # should appear for proper issues only
                        TEXTAREA(_name='body',_placeholder="Add more to this topic, using Markdown (click 'Markdown help' below to learn more)."),
                        INPUT(_type='hidden',_name='synthtree_id',_value=synthtree_id),
                        INPUT(_type='hidden',_name='synthtree_node_id',_value=synthtree_node_id),
                        INPUT(_type='hidden',_name='sourcetree_id',_value=sourcetree_id),
                        INPUT(_type='hidden',_name='ottol_id',_value=ottol_id),
                        INPUT(_type='hidden',_name='target_node_label',_value=target_node_label),
                        INPUT(_type='hidden',_name='url',_value=url),
                        # INPUT(_type='text',_name='thread_parent_id',_value=0),   # we'll get this from a nearby id, eg 'r8'
                        DIV(A(T('Close'),_class='msg-close',_href='#',_style='margin-right: 6px'),
                            SPAN(' | ',_style='margin-right: 6px'),
                            A(T('Markdown help'),_href='https://help.github.com/articles/markdown-basics',
                              _target='_blank',_style='margin-right: 10px'),
                            INPUT(_type='submit',_value=T('Post'),_class='btn btn-small',_style=''), 
                            _class='msg-footer'),
                        _method='post',_action=URL(r=request,args=[])),_class='reply'),
               SUL(*[node(comment) for comment in threads]),_class='plugin_localcomments')

#
# Perform basic CRUD for local comments, using GitHub Issues API
#
GH_BASE_URL = 'https://api.github.com'
oauth_token_path = os.path.expanduser('~/.ssh/OPENTREEAPI_OAUTH_TOKEN')
try:
    OPENTREEAPI_AUTH_TOKEN = open(oauth_token_path).read().strip()
except:
    OPENTREEAPI_AUTH_TOKEN = ''
    print("OAuth token (%s) not found!" % oauth_token_path)

# if the current user is logged in, use their auth token instead
USER_AUTH_TOKEN = auth.user and auth.user.github_auth_token or None

# Specify the media-type from GitHub, to freeze v3 API responses and get
# the comment body as markdown (vs. plaintext or HTML)
PREFERRED_MEDIA_TYPE = 'application/vnd.github.v3.raw+json'
# to get markdown AND html body, use 'application/vnd.github.v3.full+json'

GH_DATETIME_FORMAT = '%Y-%m-%dT%H:%M:%SZ'
GH_GET_HEADERS = {'Authorization': ('token %s' % (USER_AUTH_TOKEN or OPENTREEAPI_AUTH_TOKEN)),
                  'Accept': PREFERRED_MEDIA_TYPE}
GH_POST_HEADERS = {'Authorization': ('token %s' % (USER_AUTH_TOKEN or OPENTREEAPI_AUTH_TOKEN)),
                   'Content-Type': 'application/json',
                   'Accept': PREFERRED_MEDIA_TYPE}

def add_or_update_issue(msg_data, issue_id=None):
    # WATCH for accidental creation of bogus labels!
    if issue_id:
        # edit an existing issue via the GitHub API
        url = '{0}/repos/OpenTreeOfLife/feedback/issues/{1}'.format(GH_BASE_URL)
        resp = requests.patch( url, 
            headers=GH_POST_HEADERS,
            data=json.dumps(msg_data)
        )
    else:
        # create a new issue
        url = '{0}/repos/OpenTreeOfLife/feedback/issues'.format(GH_BASE_URL)
        resp = requests.post( url, 
            headers=GH_POST_HEADERS,
            data=json.dumps(msg_data)
        )
    try:
        new_msg = resp.json()
    except:
        new_msg = resp.json
    resp.raise_for_status()
    return new_msg


def add_or_update_comment(msg_data, comment_id=None, parent_issue_id=None ):
    # comment on an existing issue via the GitHub API
    if comment_id:
        # edit an existing comment
        url = '{0}/repos/OpenTreeOfLife/feedback/issues/comments/{1}'.format(GH_BASE_URL, comment_id)
        print('URL for editing an existing comment:')
        print(url)
        resp = requests.patch( url, 
            headers=GH_POST_HEADERS,
            data=json.dumps(msg_data)
        )
    else:
        # create a new comment
        url = '{0}/repos/OpenTreeOfLife/feedback/issues/{1}/comments'.format(GH_BASE_URL, parent_issue_id)
        print('URL for adding a new comment:')
        print(url)
        resp = requests.post( url, 
            headers=GH_POST_HEADERS,
            data=json.dumps(msg_data)
        )
    ##pprint(resp)
    resp.raise_for_status()
    try:
        new_msg = resp.json()
    except:
        new_msg = resp.json
    return new_msg

def close_issue(issue_id):
    # close a thread (issue) on GitHub
    url = '{0}/repos/OpenTreeOfLife/feedback/issues/{1}'.format(GH_BASE_URL, issue_id)
    print('URL for closing an existing issue:')
    print(url)
    resp = requests.patch( url, 
        headers=GH_POST_HEADERS,
        data=json.dumps({"state":"closed"})
    )
    ##pprint(resp)
    resp.raise_for_status()
    try:
        resp_json = resp.json()
    except:
        resp_json = resp.json
    return resp_json

def delete_comment(comment_id):
    # delete a comment on GitHub
    url = '{0}/repos/OpenTreeOfLife/feedback/issues/comments/{1}'.format(GH_BASE_URL, comment_id)
    print('URL for deleting an existing comment:')
    print(url)
    resp = requests.delete( url, 
        headers=GH_GET_HEADERS
    )
    ##pprint(resp)
    resp.raise_for_status()
    try:
        resp_json = resp.json()
    except:
        resp_json = resp.json
    # clobber all cached comments (since we have no metadata)
    clear_matching_cache_keys("^localcomments:")
    return resp_json

def build_localcomments_key(request):
    return 'localcomments:'+ request.url +'?'+ repr(request.vars)

def clear_matching_cache_keys(key_pattern):
    # ASSUMES we're working with RAM cache
    # NOTE that we apparently need to "clear" (using a bogus regex) to get a fresh view of the cache
    cache.ram.clear(regex='^_BOGUS_CACHE_KEY_$')
    item_count_before = len(cache.ram.storage.keys())
    pprint("=== %d RAM cache keys BEFORE clearing: ===" % item_count_before)
    for k in cache.ram.storage.keys():
        pprint(k)
    pprint("===")
    pprint("> clearing cached items matching [%s]" % key_pattern)
    cache.ram.clear(regex=key_pattern)
    item_count_after = len(cache.ram.storage.keys())
    pprint("=== %d RAM cache keys AFTER clearing: ===" % item_count_after)
    for k in cache.ram.storage.keys():
        pprint(k)
    pprint("===")
    pprint("  %d items removed" % (item_count_before - item_count_after,))

@cache(key=build_localcomments_key(request), 
       time_expire=60*5, 
       cache_model=cache.ram)
def get_local_comments(location={}):
    # Use the Search API to get all comments for this location. 
    # See https://developer.github.com/v3/search/#search-issues
    # build and encode search text (location "filter")
    print('>> its cache key would be:')
    print(build_localcomments_key(request))
    search_text = '' 
    for k,v in location.items():
        search_text = '{0}"{1} | {2} " '.format( search_text, k, v )
    search_text = urllib.quote_plus(search_text.encode('utf-8'), safe='~')
    #print search_text
    print('>> calling GitHub API for local issues...')
    url = '{0}/search/issues?q={1}repo:OpenTreeOfLife%2Ffeedback+state:open&sort=created&order=desc'
    ##TODO: search only within body?
    ## url = '{0}/search/issues?q={1}repo:OpenTreeOfLife%2Ffeedback+in:body+state:open&sort=created&order=asc'
    url = url.format(GH_BASE_URL, search_text)
    ##print(url)
    resp = requests.get( url, headers=GH_GET_HEADERS)
    ##print(resp)
    resp.raise_for_status()
    try:
        results = resp.json()
    except:
        results = resp.json
    ##pprint(results)
    ##print("Returned {0} issues ({1})".format(
    ##  results["total_count"],
    ##  results["incomplete_results"] and 'INCOMPLETE' or 'COMPLETE'
    ##  ))
    return results['items']

def clear_local_comments():
    # Examine the JSON payload (now in request.vars) to see if we can clear
    # only the affected localcomments. If not, play it safe and clear all 
    # comments in the cache.
    trigger_action = request.vars.get('action', None)
    print(">>> clear_local_comments(): trigger_action is [%s]" % trigger_action)

    if trigger_action in ['created', 'TODO']:
        issue_with_metadata = request.vars.issue
    elif trigger_action in ['FOO', 'BAR']:
        # different payload structure for some events?
        issue_with_metadata = request.vars.issue
    else:
        # fall back to most likely payload structure
        print(">>> Unexpected trigger_action [%s]!" % trigger_action)
        issue_with_metadata = request.vars.issue

    metadata = parse_comment_metadata(issue_with_metadata.get('body', ''))
    if metadata:
        # Clobber any cached comment keyed to its metadata. N.B. that we err on
        # the side of clobbering, since reloading is no big deal, but we definitely
        # don't want to show stale comments from the cache.
        if metadata['URL']:
            # Extract a root-relative URL from markdown strings like 
            # "[devtree.opentreeoflife.org/opentree/argus/otol.draft.22@132](http://devtree.opentreeoflife.org/opentree/argus/otol.draft.22@132)"
            markdown_url = metadata['URL']
            print('markdown_url:')
            print(markdown_url)
            parts = markdown_url.split('[')
            if len(parts) > 1:
                markdown_url = parts[1]
                parts = markdown_url.split(']')
                markdown_url = parts[0]
                parts = markdown_url.split('/')[1:]
                root_relative_url = '/' + '/'.join(parts)
            else:
                # assume we have an absolute URL, and remove three slashes
                parts = markdown_url.split('/')[3:]
                root_relative_url = '/' + '/'.join(parts)
            print('root_relative_url:')
            print(root_relative_url)
            clear_matching_cache_keys("^localcomments:.*'url': '%s'.*" % root_relative_url)
        if metadata['Open Tree Taxonomy id']:
            clear_matching_cache_keys("^localcomments:.*'ottol_id': '%s'.*" % metadata['Open Tree Taxonomy id'])
        if metadata['Synthetic tree node id']:
            clear_matching_cache_keys("^localcomments:.*'synthtree_node_id': '%s'.*" % metadata['Synthetic tree node id'])
    else:
        # Play it safe and clobber *all* local comments in cache.
        print(">>> No metadata found. CLEARING ALL cached localcomments!")
        clear_matching_cache_keys("^localcomments:")


# Build and parse metadata for comments (stored as markdown in GitHub). 
# The full footer is used for a thread starter (GitHub issue), while replies
# (appended GitHub comments) use an abbreviated version.

full_footer = """
================================================
Metadata   |   Do not edit below this line
:------------|:----------
Author   |   %(Author)s 
Upvotes   |   %(Upvotes)d 
URL   |   %(URL)s 
Target node label   |   %(Target node label)s 
Synthetic tree id   |   %(Synthetic tree id)s 
Synthetic tree node id   |   %(Synthetic tree node id)s 
Source tree id(s)   |   %(Source tree id)s 
Open Tree Taxonomy id   |   %(Open Tree Taxonomy id)s 
Intended scope   |   %(Intended scope)s 
"""
reply_footer = """
================================================
Metadata   |   Do not edit below this line
:------------|:----------
Author   |   %(Author)s 
Upvotes   |   %(Upvotes)s 
"""

# TODO: Restore the expertise flag to both footers?
#   Claimed Expertise   |   %(Claimed Expertise)s 
# TODO: Move 'Feedback type' from labels to footer?
#   Feedback type   |   %(Feedback type)s 

def build_comment_metadata_footer(metadata={}):
    # build full footer (for starter) or abbreviated (for replies), 
    # and return the string
    if 'Intended scope' in metadata:
        # it's a thread starter (a proper GitHub issue)
        footer_template = full_footer
    else:
        # it's a reply (GitHub comment)
        footer_template = reply_footer
    return footer_template % metadata

def parse_comment_metadata(comment_body):
    # extract metadata from comment body, return as dict
    metadata = { }
    looking_for_footer = True
    for line in comment_body.split('\n'):
        if looking_for_footer:
            if line.startswith('Metadata   |   '):
                looking_for_footer = False
        else:
            try:
                key, value = line.split('|')
            except ValueError:
                # we're past the footer?
                break
            key = key.strip()
            value = value.strip()
            if key.startswith(':---'):
                # skip this divider row
                continue
            metadata[key] = value
    return metadata

def get_visible_comment_body(comment_body):
    # discard the footer (starting at line '=========...')
    visible_lines = [ ]
    for line in comment_body.split('\n'):
        if line.startswith('======='):
            break
        visible_lines.append(line)
    return '\n'.join(visible_lines)
    
# Time-zone converstion from UTC to local time (needed for GitHub date-strings),
# adapted from code found here: http://stackoverflow.com/a/13287083
import calendar
from datetime import datetime, timedelta
def utc_to_local(utc_dt):
    # get integer timestamp to avoid precision lost
    timestamp = calendar.timegm(utc_dt.timetuple())
    local_dt = datetime.fromtimestamp(timestamp)
    assert utc_dt.resolution >= timedelta(microseconds=1)
    return local_dt.replace(microsecond=utc_dt.microsecond)