X-Git-Url: https://git.mdrn.pl/wolnelektury.git/blobdiff_plain/87996744612ff81cf1d041f745a0b6a6b42c5f03..9b31d2bc42b0e1239609cd92c532dfd345e48d0f:/src/catalogue/api/tojson.py diff --git a/src/catalogue/api/tojson.py b/src/catalogue/api/tojson.py index 3d47cd995..d3b7ab539 100644 --- a/src/catalogue/api/tojson.py +++ b/src/catalogue/api/tojson.py @@ -3,6 +3,7 @@ import json import re from sys import argv from lxml import etree +from catalogue.utils import absolute_url, gallery_url tags = { 'utwor': ('_pass', False, None, None, None), @@ -54,7 +55,7 @@ tags = { 'osoba': ('em', True, {'class': 'osoba'}, None, None), 'didaskalia': ('div', True, {'class': 'didaskalia'}, None, None), 'kwestia': ('div', False, {'class': 'kwestia'}, None, None), - 'didask_tekst': ('em', False, {'class': 'didask_tekst'}, None, None), + 'didask_tekst': ('em', True, {'class': 'didask_tekst'}, None, None), 'naglowek_czesc': ('h2', True, None, None, None), 'naglowek_akt': ('h2', True, None, None, None), @@ -89,6 +90,45 @@ tags = { 'pr': ('a', True, {'class': 'footnote footnote-pr'}, None, False), 'pt': ('a', True, {'class': 'footnote footnote-pt'}, None, False), 'ptrad': ('a', True, {'class': 'footnote footnote-ptrad'}, None, False), + + 'werset': ('p', True, {'class': 'werset'}, None, True), + 'br': ('br', False, None, None, None), + 'indeks_dolny': ('em', True, {'class': 'indeks_dolny'}, None, False), + 'mat': ('span', True, {'class': 'mat'}, None, False), + + 'mfenced': ('math_mfenced', True, None, None, False), + 'mfrac': ('math_mfrac', True, None, None, False), + 'mrow': ('math_mrow', True, None, None, False), + 'mi': ('math_mi', True, None, None, False), + 'mn': ('math_mn', True, None, None, False), + 'mo': ('math_mo', True, None, None, False), + 'msup': ('math_msup', True, None, None, False), + + 'list': ('blockquote', False, {'class': 'list'}, None, None), + 'wywiad_pyt': ('blockquote', False, {'class': 'wywiad_pyt'}, None, None), + 'wywiad_odp': ('blockquote', False, {'class': 'wywiad_odp'}, None, None), + 'rownolegle': ('blockquote', False, {'class': 'rownolegle'}, None, None), + 'animacja': ('div', False, {'class': 'animacja'}, None, None), + 'data': ('div', True, {'class': 'data'}, None, None), + 'podpis': ('div', True, {'class': 'podpis'}, None, None), + 'naglowek_listu': ('div', True, {'class': 'naglowek_listu'}, None, None), + 'pozdrowienie': ('div', True, {'class': 'pozdrowienie'}, None, None), + 'adresat': ('div', True, {'class': 'adresat'}, None, None), + 'tytul_oryg': ('div', True, {'class': 'tytul_oryg'}, None, None), + 'miejsce_data': ('div', True, {'class': 'miejsce_data'}, None, None), + 'audio': ('_ignore', False, None, None, None), + 'www': ('a', True, {'class': 'www'}, {'href': '.text'}, False), + + 'tabela': ('table', False, None, None, None), + 'tabelka': ('table', False, None, None, None), + 'wiersz': ('tr', False, None, None, None), + 'kol': ('td', True, None, None, None), + + 'ilustr': ('img', False, None, { + 'src': lambda e, S: absolute_url(gallery_url(S['slug'])) + e.attrib['src'].rsplit('/', 1)[-1], + }, False), + 'tab': ('span', False, {'class': 'tab'}, {'szer': 'szer'}, False), + } id_prefixes = { @@ -153,7 +193,13 @@ def toj(elem, S): if attr_map: output.setdefault('attr', {}) for k, v in attr_map.items(): - output['attr'][k] = elem.attrib[v] + if v == '.text': + val = elem.text + elif callable(v): + val = v(elem, S) + else: + val = elem.attrib[v] + output['attr'][k] = val output['contents'] = contents output = [output] if elem.tag == 'strofa': @@ -204,7 +250,7 @@ def toj(elem, S): return [] return output -def conv(tree): +def conv(tree, slug): S = { 'index': 0, 'vindex': 0, @@ -212,6 +258,7 @@ def conv(tree): 'stack': [], 'front1': [], 'front2': [], + 'slug': slug, } output = toj(tree.getroot(), S) if not len(output): return {}