From 0d9ed8a2d4113c70546746f75d9ce0485dca3b1d Mon Sep 17 00:00:00 2001 From: Castro0o Date: Wed, 4 Mar 2020 11:28:20 +0100 Subject: [PATCH] WIP:multi values properties --- functions.py | 52 ++++++++++++++++++++++++------------ query2html.py | 30 ++++++++++++++++----- templates/document.html | 4 +-- templates/document_part.html | 25 +++++++++++++---- templates/index.html | 2 +- 5 files changed, 81 insertions(+), 32 deletions(-) diff --git a/functions.py b/functions.py index 5414223..29aac39 100644 --- a/functions.py +++ b/functions.py @@ -43,23 +43,41 @@ def unpack_response(response): printouts = response['printouts'] page = response['fulltext'] fullurl = response['fullurl'] - d['page'] = page - for prop in printouts: - p_item = response['printouts'][prop] - for prop_val in p_item: - if isinstance(prop_val, dict) is False: - d[prop] = prop_val - else: - # if len(prop_val) > 0: - props = list(prop_val.keys()) - if 'fulltext' in props: - val = prop_val.get('fulltext') - elif 'timestamp' in props: - val = datetime.fromtimestamp(int(prop_val.get('timestamp'))) - else: - val = list(prop_val.values())[0] - d[prop] = val - return page, d, fullurl + printouts_dumps = json.dumps(printouts) + printouts_loads = json.loads(printouts_dumps) + printouts_loads['page'] = page + # printouts_loads['Date'] = datetime.fromtimestamp( + # int(printouts_loads['Date'][0]['timestamp'])) + simplified_printouts = {} + for k, v in printouts_loads.items(): + if k == 'Date': + simplified_printouts[k] = datetime.fromtimestamp( + int(v[0]['timestamp'])) + # elif k == 'Title': + # simplified_printouts[k] = v[0]['fulltext'] + elif k in ['Part', 'Partof', 'page']: # only 1 value for each + simplified_printouts[k] = v + else: # Possibly more than 1 value for the rest of properties + simplified_printouts[k] = [] + for listitem in v: + simplified_printouts[k].append(listitem['fulltext']) + + # for prop in printouts: + # p_item = response['printouts'][prop] + # for prop_val in p_item: + # if isinstance(prop_val, dict) is False: + # d[prop] = prop_val + # else: + # # if len(prop_val) > 0: + # props = list(prop_val.keys()) + # if 'fulltext' in props: + # val = prop_val.get('fulltext') + # elif 'timestamp' in props: + # val = datetime.fromtimestamp(int(prop_val.get('timestamp'))) + # else: + # val = list(prop_val.values())[0] + # d[prop] = val + return page, simplified_printouts, fullurl def update_json(imgsjson_fn, img_dict, img_fn): diff --git a/query2html.py b/query2html.py index 018eb9a..9e234d9 100644 --- a/query2html.py +++ b/query2html.py @@ -15,7 +15,7 @@ p.add_argument("--conditions", "-c", metavar='', default='[[File:+]][[Title::+]][[Part::+]][[Date::+]]', help='The query conditions') p.add_argument("--printouts", "-p", metavar='', - default='?Title|?Date|?Part|?Partof|?Creator', + default='?Title|?Date|?Part|?Partof|?Creator|?Organization|?Format|?Event|?Topic|?Language', help='Selection of properties to printout') p.add_argument("--sort", "-s", metavar='', default='Date,Title,Part', @@ -23,6 +23,8 @@ p.add_argument("--sort", "-s", metavar='', p.add_argument("--order", "-o", metavar='', default='asc,asc,asc', help='Order of sorting conditions. Should same amount as the --sort properties') +p.add_argument('--limit', '-l', help='(optional) Limit the number of returned ' + 'items') p.add_argument('--dry', '-d', action='store_true', help='dry-run: will only show the query but not run it') @@ -33,8 +35,10 @@ if len(args.sort.split(',')) != len(args.order.split(',')): Colors.WARNING, '--sort and --order do not have the same amount of elements', Colors.ENDC) print('Script exiting now') sys.exit() - query = f'{args.conditions}|{args.printouts}|sort={args.sort}|order={args.order}' +if args.limit: + limit_str = f'|limit={args.limit}' + query += limit_str print('query:', Colors.GREEN, query, Colors.ENDC) query_unquoted = urllib.parse.quote(query) query_url = f'https://{args.host}{args.path}api.php?action=ask&query={query_unquoted}&format=json' @@ -75,7 +79,7 @@ with open(os.path.join(wd, 'templates/document_part.html')) as document_html: all_document_parts = '' # to append all content documentslist = [] for answer in site.ask(query): - publication_title = '' + # publication_title = '' # print(answer, answer.keys()) page, printout_dict, fullurl = unpack_response(answer) print(page) @@ -85,6 +89,13 @@ for answer in site.ask(query): print(Colors.WARNING, f"{printout_dict['page']} is not is missing from the local downloaded images") print(Colors.GREEN, 'run python3 download_imgs.py to fix the issue', Colors.ENDC) sys.exit() + # + # # TODO: EXTRACT PROPERTIES THROUGH THE FOLLOWING ASK QUERY + # ask_page_props = f'[[File:{printout_dict["page"]}]]|?Title|?Date|?Part|?Partof|?Creator|?Organization|?Format|?Event|?Topic|?Language' + # print(ask_page_props) + # page_props = site.ask(ask_page_props) + # print(page_props) + # import pdb; pdb.set_trace() page = site.pages[[printout_dict['page']]] # request that page from wiki pagetext = page.text() pagetext_html = pandoc(pwd=wd, content=pagetext, format_in='mediawiki', format_out='html') @@ -101,10 +112,15 @@ for answer in site.ask(query): if printout_dict['Part'] == printout_dict['Partof']: # RENDER DOCUMENT # by passing all_document_parts html to document_template content - document_html = document_template.render(title=printout_dict.get('Title'), - date=printout_dict.get('Date'), - content=all_document_parts) # render document template - htmlpage_fn = "{}.html".format(printout_dict.get('Title').replace(" ", "")) + + # TODO: EXPAND PROPERTIES IN TEMPLATE + + document_html = document_template.render( + title=printout_dict.get('Title'), + date=printout_dict.get('Date'), + content=all_document_parts) # render document template + htmlpage_fn = "{}.html".format( + printout_dict.get('Title')[0].replace(" ", "")) with open(os.path.join(static_html, htmlpage_fn), 'w') as htmlfile: htmlfile.write(document_html) all_document_parts = '' # Reset all_document_parts diff --git a/templates/document.html b/templates/document.html index 4748c77..72687fd 100644 --- a/templates/document.html +++ b/templates/document.html @@ -3,10 +3,10 @@ - {{title}} + {{ title[0] }} -

{{ title }}

+

{{ title[0] }}

{{ content }} diff --git a/templates/document_part.html b/templates/document_part.html index c435c5a..feb365a 100644 --- a/templates/document_part.html +++ b/templates/document_part.html @@ -11,17 +11,32 @@

Metadata

- {% for key, value in printout_dict.items() %} + {% for key, valuelist in printout_dict.items() %} {% if key == 'Date' %}
{{key}} - {{value.year}} {{value.month}} {{value.day}} + {{valuelist.year}} {{valuelist.month}} {{valuelist.day}} + {# Note: in Date valuelist is not a list but a dateitme object #}
- {% else %} + +
- {{key|upper}} - {{value}} + {% if key == 'Date' %} + {{key}} + {{valuelist.year}} {{valuelist.month}} {{valuelist.day}} + {% elif key == 'page' %} + {{key|upper}} + {{valuelist}} + {% else %} + {{key|upper}} + {% for val in valuelist %} + {{key|upper}} + {% endfor %} + {% endif %}
+ + {# TODO: resolve all these metadata items #} + {% endif %} {% endfor %}
diff --git a/templates/index.html b/templates/index.html index 55628bc..af2286e 100644 --- a/templates/index.html +++ b/templates/index.html @@ -8,7 +8,7 @@

Results from query:
{{query}}