medkit.core.text.span_utils

medkit.core.text.span_utils#

Functions#

`replace`(→ tuple[str, list[medkit.core.text.span.AnySpan]])	Replace parts of a text, and update accordingly its associated spans.
`remove`(→ tuple[str, list[medkit.core.text.span.AnySpan]])	Remove parts of a text, while also removing accordingly its associated spans.
`extract`(→ tuple[str, list[medkit.core.text.span.AnySpan]])	Extract parts of a text as well as its associated spans.
`insert`(→ tuple[str, list[medkit.core.text.span.AnySpan]])	Insert strings in text, and update accordingly its associated spans.
`move`(→ tuple[str, list[medkit.core.text.span.AnySpan]])	Move part of a text to another position, also moving its associated spans.
`concatenate`(→ tuple[str, ...)	Concatenate text and span objects.
`normalize_spans`(→ list[medkit.core.text.span.Span])	Normalize spans.
`clean_up_gaps_in_normalized_spans`(spans, text[, ...])	Remove small gaps in normalized spans.

Module Contents#

medkit.core.text.span_utils.replace(text: str, spans: list[medkit.core.text.span.AnySpan], ranges: list[tuple[int, int]], replacement_texts: list[str]) → tuple[str, list[medkit.core.text.span.AnySpan]]#

Replace parts of a text, and update accordingly its associated spans.

Parameters:

textstr: The text in which some parts will be replaced
spanslist of AnySpan: The spans associated with text
rangeslist of tuple of int: The ranges of the parts that will be replaced (end excluded), sorted by ascending order
replacement_textstuple: The strings to use as replacements (must be the same length as ranges)

Returns:

textstr: The updated text
spanslist of AnySpan: The spans associated with the updated text

Examples

>>> text = "Hello, my name is John Doe."
>>> spans = [Span(0, len(text))]
>>> ranges = [(0, 5), (18, 22)]
>>> replacements = ["Hi", "Jane"]
>>> text, spans = replace(text, spans, ranges, replacements)
>>> print(text)
Hi, my name is Jane Doe.

medkit.core.text.span_utils.remove(text: str, spans: list[medkit.core.text.span.AnySpan], ranges: list[tuple[int, int]]) → tuple[str, list[medkit.core.text.span.AnySpan]]#

Remove parts of a text, while also removing accordingly its associated spans.

Parameters:

textstr: The text in which some parts will be removed
spanslist of AnySpan: The spans associated with text
rangeslist of tuple of int: The ranges of the parts that will be removed (end excluded), sorted by ascending order

Returns:

textstr: The updated text
spanslist of AnySpan: The spans associated with the updated text

medkit.core.text.span_utils.extract(text: str, spans: list[medkit.core.text.span.AnySpan], ranges: list[tuple[int, int]]) → tuple[str, list[medkit.core.text.span.AnySpan]]#

Extract parts of a text as well as its associated spans.

Parameters:

textstr: The text to extract parts from
spanslist of AnySpan: The spans associated with text
rangeslist of tuple of int: The ranges of the parts to extract (end excluded), sorted by ascending order

Returns:

textstr: The extracted text
spanslist of AnySpan: The spans associated with the extracted text

medkit.core.text.span_utils.insert(text: str, spans: list[medkit.core.text.span.AnySpan], positions: list[int], insertion_texts: list[str]) → tuple[str, list[medkit.core.text.span.AnySpan]]#

Insert strings in text, and update accordingly its associated spans.

Parameters:

textstr: The text in which some strings will be inserted
spanslist of AnySpan: The spans associated with text
positionslist of int: The positions where the strings will be inserted, sorted by ascending order
insertion_textslist of str: The strings to insert (must be the same length as positions)

Returns:

textstr: The updated text
spanslist of AnySpan: The spans associated with the updated text

Examples

>>> text = "Hello, my name is John Doe."
>>> spans = [Span(0, len(text))]
>>> positions = [5]
>>> inserts = [" everybody"]
>>> text, spans = insert(text, spans, positions, inserts)
>>> print(text)
Hello everybody, my name is John Doe."

medkit.core.text.span_utils.move(text: str, spans: list[medkit.core.text.span.AnySpan], range: tuple[int, int], destination: int) → tuple[str, list[medkit.core.text.span.AnySpan]]#

Move part of a text to another position, also moving its associated spans.

Parameters:

textstr: The text in which a part should be moved
spanslist of AnySpan: The spans associated with the input text
rangetuple of int: The range of the part to move (end excluded)
destinationint: The position where to insert the displaced range

Returns:

textstr: The updated text
spanslist of AnySpan: The spans associated with the updated text

Examples

>>> text = "Hello, my name is John Doe."
>>> spans = [Span(0, len(text))]
>>> range = (17, 22)
>>> dest = len(text) - 1
>>> text, spans = move(text, spans, range, dest)
>>> print(text)
Hi, my name is Doe John.

medkit.core.text.span_utils.concatenate(texts: list[str], all_spans: list[list[medkit.core.text.span.AnySpan]]) → tuple[str, list[medkit.core.text.span.AnySpan]]#: Concatenate text and span objects.

medkit.core.text.span_utils.normalize_spans(spans: list[medkit.core.text.span.AnySpan]) → list[medkit.core.text.span.Span]#

Normalize spans.

Return a transformed spans in which all instances of ModifiedSpan are replaced by the spans they refer to, spans are sorted and contiguous spans are merged.

Parameters:

spanslist of AnySpan: The spans associated with a text, including additional spans if insertions or replacement were performed

Returns:

normalized_spanslist of Span: Spans in spans normalized as described

Examples

>>> spans = [
...     Span(0, 10),
...     Span(20, 30),
...     ModifiedSpan(8, replaced_spans=[Span(30, 36)]),
... ]
>>> spans = normalize_spans(spans)
>>> print(spans)
>>> [Span(0, 10), Span(20, 36)]

medkit.core.text.span_utils.clean_up_gaps_in_normalized_spans(spans: list[medkit.core.text.span.Span], text: str, max_gap_length: int = 3)#

Remove small gaps in normalized spans.

This is useful for converting non-contiguous entity spans with small gaps containing only whitespace or a few meaningless characters (due to clean-up preprocessing or translation) into one unique bigger span. Gaps having less than max_gap_length will be removed by merging the spans before and after the gap.

Parameters:

spanslist of Span: The normalized spans in which to remove gaps
textstr: The text associated with spans
max_gap_lengthint, default=3: Max number of characters in gaps, after stripping leading and trailing whitespace.

Examples

>>> text = "heart failure"
>>> spans = [Span(0, 5), Span(6, 13)]
>>> spans = clean_up_gaps_in_normalized_spans(spans, text)
>>> print(spans)
>>> spans = [Span(0, 13)]