Function get_highlighted_lines

src/licensedcode/match.py:3331–3380

Yield highlighted text lines (with line returns) for the whole of the matched and unmatched text of a ``query``.

(
    match,
    query,
    stopwords=STOPWORDS,
    trace=TRACE_HIGHLIGHTED_TEXT,
)

def get_highlighted_lines(
    match,
    query,
    stopwords=STOPWORDS,
    trace=TRACE_HIGHLIGHTED_TEXT,
):
    """
    Yield highlighted text lines (with line returns) for the whole of the matched and unmatched text of a ``query``.
    """
    tokens = tokenize_matched_text(
        location=query.location,
        query_string=query.query_string,
        dictionary=query.idx.dictionary,
        start_line=match.query.start_line,
        _cache={},
    )
    tokens = tag_matched_tokens(tokens=tokens, match_qspan=match.qspan)

    if trace:
        tokens = list(tokens)
        print()
        logger_debug('get_highlighted_lines:  tokens:')
        for t in tokens:
            print('    ', t)
        print()
    header = ''&#x27;<style>
      pre.log {color: #f1f1f1; background-color: #222; font-family: monospace;}
      pre.wrap {white-space: pre-wrap;}
      span.not-matched {color:#f5bf3c;}
      span.matched {color:#000000;}
</style>

<div class="license-match"><pre>''&#x27;
    footer = '''</pre></div>'''

    yield header
    highlight_matched = '<span class="matched">{}</span>'
    highlight_not_matched = '<span class="not-matched">{}</span>'
    for token in tokens:
        val = token.value
        if token.is_text and val.lower() not in stopwords:
            if token.is_matched:
                yield highlight_matched.format(val)
            else:
                yield highlight_not_matched.format(val)
        else:
            # we do not highlight punctuation and stopwords.
            yield highlight_not_matched.format(val)

    yield footer


def tag_matched_tokens(tokens, match_qspan):