def explainAnchorTextZoning(guid, max_inlinks=10, use_full_text=False):
"""
This generates a clipping collection file, including all the citation
contexts of other files to this file
"""
meta=cp.Corpus.getMetadataByGUID(guid)
all_html=["""<h1 class="title">%s</h1><span>Inlink context summary for %s</span>""" % (meta["title"],formatCitation(meta))]
global CURRENT_CITATION
CURRENT_CITATION=re.escape(formatCitation(meta))
for index, link in enumerate(meta["inlinks"]):
if index == max_inlinks:
break
print("Processing anchor text from %s" % link)
doc=cp.Corpus.loadSciDoc(link)
if not use_full_text:
trimDocToRelevantBits(doc, guid)
renderer=SciDocRenderer(doc)
html=renderer.prettyPrintDocumentHTML(
formatspans=True,
include_bibliography=use_full_text,
wrap_with_HTML_tags=False,
extra_attribute_function=extraAttributes,
citation_formatting_function=citationFormatting,
reference_formatting_function=referenceFormatting)
all_html.append(html)
html=padWithHTML(" ".join(all_html))
writeFileText(html,os.path.join(cp.Corpus.paths.output,guid+"_ilc_zoning.html"))
def explainInFileZoning(guid):
"""
Given a guid, generates a HTML visualization where sentences are tagged
with their class for the contents of that file
"""
doc=cp.Corpus.loadSciDoc(guid)
renderer=SciDocRenderer(doc)
html=renderer.prettyPrintDocumentHTML(True,True, False, extra_attribute_function=extraAttributes, wrap_with_HTML_tags=False,
reference_formatting_function=referenceFormatting)
html=padWithHTML(html)
writeFileText(html,os.path.join(cp.Corpus.paths.output,guid+"_text_zoning.html"))
