def main(args):
"""Subcommand main.
You shouldn't need to call this yourself if you're using
`config_argparser`.
"""
output_dir = get_output_dir(args, default_overwrite=True)
# locate insertion site: target document
reader = educe.stac.Reader(args.corpus)
tgt_files = reader.filter(reader.files(), is_requested(args))
tgt_corpus = reader.slurp(tgt_files)
# TODO mark units with FIXME, optionally delete in/out relations
span = args.span
sub_text = args.sub_text
minor = args.minor
# store before/after
annos_before = []
annos_after = []
for tgt_k, tgt_doc in tgt_corpus.items():
annos_before.append(annotate_doc(tgt_doc, span=span))
# process
new_tgt_doc = replace_text_at_span(tgt_doc,
span,
sub_text,
minor=minor)
# WIP new_span, depends on the offset
offset = len(sub_text) - (span.char_end - span.char_start)
new_span = Span(span.char_start, span.char_end + offset)
# end WIP
annos_after.append(annotate_doc(new_tgt_doc, span=new_span))
# show diff and save doc
diffs = [
"======= REPLACE TEXT IN %s ========" % tgt_k,
show_diff(tgt_doc, new_tgt_doc)
]
print("\n".join(diffs).encode('utf-8'), file=sys.stderr)
save_document(output_dir, tgt_k, new_tgt_doc)
announce_output_dir(output_dir)
# commit message
tgt_k, tgt_doc = list(tgt_corpus.items())[0]
anno_str_before = annos_before[0]
anno_str_after = annos_after[0]
if tgt_k and not args.no_commit_msg:
print("-----8<------")
print(commit_msg(tgt_k, anno_str_before, anno_str_after))
def anno(doc, prefix, tspan):
"pad text segment as needed"
prefix_t = "..."\
if tspan.char_start + len(prefix) < info.span.char_start\
else ""
myspan = Span(info.span.char_start, tspan.char_end)
return "".join([prefix, prefix_t, annotate_doc(doc, span=myspan)])
def main(args):
"""Subcommand main.
You shouldn't need to call this yourself if you're using
`config_argparser`.
"""
output_dir = get_output_dir(args, default_overwrite=True)
# locate insertion site: target document
reader = educe.stac.Reader(args.corpus)
tgt_files = reader.filter(reader.files(), is_requested(args))
tgt_corpus = reader.slurp(tgt_files)
# TODO mark units with FIXME, optionally delete in/out relations
span = args.span
sub_text = args.sub_text
minor = args.minor
# store before/after
annos_before = []
annos_after = []
for tgt_k, tgt_doc in tgt_corpus.items():
annos_before.append(annotate_doc(tgt_doc, span=span))
# process
new_tgt_doc = replace_text_at_span(
tgt_doc, span, sub_text, minor=minor)
# WIP new_span, depends on the offset
offset = len(sub_text) - (span.char_end - span.char_start)
new_span = Span(span.char_start, span.char_end + offset)
# end WIP
annos_after.append(annotate_doc(new_tgt_doc, span=new_span))
# show diff and save doc
diffs = ["======= REPLACE TEXT IN %s ========" % tgt_k,
show_diff(tgt_doc, new_tgt_doc)]
print("\n".join(diffs).encode('utf-8'), file=sys.stderr)
save_document(output_dir, tgt_k, new_tgt_doc)
announce_output_dir(output_dir)
# commit message
tgt_k, tgt_doc = list(tgt_corpus.items())[0]
anno_str_before = annos_before[0]
anno_str_after = annos_after[0]
if tgt_k and not args.no_commit_msg:
print("-----8<------")
print(commit_msg(tgt_k, anno_str_before, anno_str_after))
def anno(doc, prefix, tspan):
"pad text segment as needed"
prefix_t = "..."\
if tspan.char_start + len(prefix) < info.span.char_start\
else ""
myspan = Span(info.span.char_start, tspan.char_end)
return "".join([prefix,
prefix_t,
annotate_doc(doc, span=myspan)])
def anno(doc, prefix, tspan):
"pad text segment as needed"
prefix_t = "..."\
if tspan.char_start + len(prefix) < info.span.char_start\
else ""
suffix_t = "..."\
if tspan.char_end > info.span.char_end + 1\
else ""
return "".join(
[prefix, prefix_t,
annotate_doc(doc, span=info.span), suffix_t])
def anno(doc, prefix, tspan):
"pad text segment as needed"
prefix_t = "..."\
if tspan.char_start + len(prefix) < info.span.char_start\
else ""
suffix_t = "..."\
if tspan.char_end > info.span.char_end + 1\
else ""
return "".join([prefix,
prefix_t,
annotate_doc(doc, span=info.span),
suffix_t])
def commit_msg(info):
"""
Generate a commit message describing the operation
we just did
"""
k = info.key
mini_new_doc = narrow_to_span(info.after, info.span)
lines = [
"{}_{}: split dialogue before turn {}".format(k.doc, k.subdoc,
info.tid), "",
annotate_doc(mini_new_doc), "..."
]
return "\n".join(lines)
def commit_msg(info):
"""
Generate a commit message describing the operation
we just did
"""
k = info.key
mini_new_doc = narrow_to_span(info.after, info.span)
lines = ["{}_{}: split dialogue before turn {}".format(
k.doc, k.subdoc, info.tid),
"",
annotate_doc(mini_new_doc),
"..."]
return "\n".join(lines)
def commit_msg(info):
"""
Generate a commit message describing the operation
we just did
"""
k = info.key
mini_new_doc = narrow_to_span(info.after, info.span)
lines = ["%s_%s: move turn %d %s" % (k.doc, k.subdoc,
info.tid, info.direction),
"",
annotate_doc(mini_new_doc),
"..."]
return "\n".join(lines)
def commit_msg(args, corpus, k, sought):
"""
Generate a commit message describing the dialogue merging operation
we are about to do (has to be run before merging happens)
"""
doc = corpus[k]
dstr = ", ".join(anno_id_from_tuple(x) for x in sought)
dialogues = [_get_annotation_with_id(d, doc.units) for d in sought]
if dialogues:
title_fmt = u"{doc}_{subdoc}: merge dialogues{hint}"
title_hint = " (turns %d-%d)" % tuple(args.turns) if args.turns else ""
dspan = _merge_spans(dialogues)
lines = [
title_fmt.format(doc=k.doc, subdoc=k.subdoc, hint=title_hint), "",
"Dialogues ({}), was:".format(dstr), "",
annotate_doc(doc, span=dspan)
]
return "\n".join(lines)
else:
return "(no commit message; nothing to merge)"
def commit_msg(args, corpus, k, sought):
"""
Generate a commit message describing the dialogue merging operation
we are about to do (has to be run before merging happens)
"""
doc = corpus[k]
dstr = ", ".join(anno_id_from_tuple(x) for x in sought)
dialogues = [_get_annotation_with_id(d, doc.units) for d in sought]
if dialogues:
title_fmt = u"{doc}_{subdoc}: merge dialogues{hint}"
title_hint = " (turns %d-%d)" % tuple(args.turns) if args.turns else ""
dspan = _merge_spans(dialogues)
lines = [title_fmt.format(doc=k.doc,
subdoc=k.subdoc,
hint=title_hint),
"",
"Dialogues ({}), was:".format(dstr),
"",
annotate_doc(doc, span=dspan)]
return "\n".join(lines)
else:
return "(no commit message; nothing to merge)"
